Synced with 3 months of Google-internal development.

Major changes:
- Got rid of all bytestream interfaces in favor of
  using regular handlers.
- new Pipeline object represents a upb pipeline, does
  bump allocation internally to manage memory.
- proto2 support now can handle extensions.

bindings/cpp/upb/pb/decoder.hpp

@@ -1,81 +0,0 @@
-//
-// upb - a minimalist implementation of protocol buffers.
-//
-// Copyright (c) 2011 Google Inc.  See LICENSE for details.
-// Author: Josh Haberman <jhaberman@gmail.com>
-//
-// upb::Decoder is a high performance, streaming decoder for protobuf
-// data that works by getting its input data from a ubp::ByteRegion and calling
-// into a upb::Handlers.
-//
-// A DecoderPlan contains whatever data structures and generated (JIT-ted) code
-// are necessary to decode protobuf data of a specific type to a specific set
-// of handlers.  By generating the plan ahead of time, we avoid having to
-// redo this work every time we decode.
-//
-// A DecoderPlan is threadsafe, meaning that it can be used concurrently by
-// different upb::Decoders in different threads.  However, the upb::Decoders are
-// *not* thread-safe.
-
-#ifndef UPB_PB_DECODER_HPP
-#define UPB_PB_DECODER_HPP
-
-#include "upb/pb/decoder.h"
-
-#include "upb/bytestream.h"
-#include "upb/upb.h"
-
-namespace upb {
-
-class DecoderPlan : public upb_decoderplan {
- public:
-  static DecoderPlan* New(const Handlers* h, bool allow_jit) {
-    return static_cast<DecoderPlan*>(upb_decoderplan_new(h, allow_jit));
-  }
-  void Unref() { upb_decoderplan_unref(this); }
-
-  // Returns true if the plan contains JIT-ted code.  This may not be the same
-  // as the "allowjit" parameter to the constructor if support for JIT-ting was
-  // not compiled in.
-  bool HasJitCode() { return upb_decoderplan_hasjitcode(this); }
-
- private:
-  DecoderPlan() {}  // Only constructed by New
-};
-
-class Decoder : public upb_decoder {
- public:
-  Decoder() { upb_decoder_init(this); }
-  ~Decoder() { upb_decoder_uninit(this); }
-
-  // Resets the plan that the decoder will parse from.  This will also reset the
-  // decoder's input to be uninitialized -- ResetInput() must be called before
-  // parsing can occur.  The plan must live until the decoder is destroyed or
-  // reset to a different plan.
-  //
-  // Must be called before ResetInput() or Decode().
-  void ResetPlan(DecoderPlan* plan) { upb_decoder_resetplan(this, plan); }
-
-  // Resets the input of the decoder.  This puts it in a state where it has not
-  // seen any data, and expects the next data to be from the beginning of a new
-  // protobuf.
-  //
-  // ResetInput() must be called before Decode() but may be called more than
-  // once.  "input" must live until the decoder destroyed or ResetInput is
-  // called again.  "c" is the closure that will be passed to the handlers.
-  void ResetInput(ByteRegion* byte_region, void* c) {
-    upb_decoder_resetinput(this, byte_region, c);
-  }
-
-  // Decodes serialized data (calling Handlers as the data is parsed) until
-  // error or EOF (see status() for details).
-  Status::Success Decode() { return upb_decoder_decode(this); }
-
-  const upb::Status& status() {
-    return static_cast<const upb::Status&>(*upb_decoder_status(this));
-  }
-};
-
-}  // namespace upb
-
-#endif

bindings/lua/table.c

@@ -31,16 +31,16 @@ static void lupbtable_setnum(lua_State *L, int tab, const char *key,
   lua_setfield(L, tab - 1, key);
 }
 
-static void lupbtable_pushval(lua_State *L, upb_value val, upb_ctype_t type) {
+static void lupbtable_pushval(lua_State *L, _upb_value val, upb_ctype_t type) {
   switch (type) {
     case UPB_CTYPE_INT32:
-      lua_pushnumber(L, upb_value_getint32(val));
+      lua_pushnumber(L, val.int32);
       break;
     case UPB_CTYPE_PTR:
-      lupb_def_pushwrapper(L, upb_value_getptr(val), NULL);
+      lupb_def_pushwrapper(L, val.ptr, NULL);
       break;
     case UPB_CTYPE_CSTR:
-      lua_pushstring(L, upb_value_getcstr(val));
+      lua_pushstring(L, val.cstr);
       break;
     default:
       luaL_error(L, "Unexpected type: %d", type);

bindings/lua/test.lua

@@ -13,6 +13,7 @@ function test_fielddef()
   assert_false(f:is_frozen())
   assert_nil(f:number())
   assert_nil(f:name())
+  assert_nil(f:type())
   assert_equal(upb.LABEL_OPTIONAL, f:label())
 
   f:set_name("foo_field")

bindings/lua/upb.c

@@ -14,7 +14,7 @@
 #include <string.h>
 #include "lauxlib.h"
 #include "bindings/lua/upb.h"
-#include "upb/bytestream.h"
+#include "upb/handlers.h"
 #include "upb/pb/glue.h"
 
 // Lua metatable types.
@@ -75,7 +75,7 @@ static uint32_t lupb_checkint32(lua_State *L, int narg, const char *name) {
 // Converts a number or bool from Lua -> upb_value.
 static upb_value lupb_getvalue(lua_State *L, int narg, upb_fieldtype_t type) {
   upb_value val;
-  if (type == UPB_TYPE(BOOL)) {
+  if (type == UPB_TYPE_BOOL) {
     if (!lua_isboolean(L, narg))
       luaL_error(L, "Must explicitly pass true or false for boolean fields");
     upb_value_setbool(&val, lua_toboolean(L, narg));
@@ -83,41 +83,35 @@ static upb_value lupb_getvalue(lua_State *L, int narg, upb_fieldtype_t type) {
     // Numeric type.
     lua_Number num = luaL_checknumber(L, narg);
     switch (type) {
-      case UPB_TYPE(INT32):
-      case UPB_TYPE(SINT32):
-      case UPB_TYPE(SFIXED32):
-      case UPB_TYPE(ENUM):
+      case UPB_TYPE_INT32:
+      case UPB_TYPE_ENUM:
         if (num > INT32_MAX || num < INT32_MIN || num != rint(num))
           luaL_error(L, "Cannot convert %f to 32-bit integer", num);
         upb_value_setint32(&val, num);
         break;
-      case UPB_TYPE(INT64):
-      case UPB_TYPE(SINT64):
-      case UPB_TYPE(SFIXED64):
+      case UPB_TYPE_INT64:
         if (num > INT64_MAX || num < INT64_MIN || num != rint(num))
           luaL_error(L, "Cannot convert %f to 64-bit integer", num);
         upb_value_setint64(&val, num);
         break;
-      case UPB_TYPE(UINT32):
-      case UPB_TYPE(FIXED32):
+      case UPB_TYPE_UINT32:
         if (num > UINT32_MAX || num < 0 || num != rint(num))
           luaL_error(L, "Cannot convert %f to unsigned 32-bit integer", num);
         upb_value_setuint32(&val, num);
         break;
-      case UPB_TYPE(UINT64):
-      case UPB_TYPE(FIXED64):
+      case UPB_TYPE_UINT64:
         if (num > UINT64_MAX || num < 0 || num != rint(num))
           luaL_error(L, "Cannot convert %f to unsigned 64-bit integer", num);
         upb_value_setuint64(&val, num);
         break;
-      case UPB_TYPE(DOUBLE):
+      case UPB_TYPE_DOUBLE:
         if (num > DBL_MAX || num < -DBL_MAX) {
           // This could happen if lua_Number was long double.
           luaL_error(L, "Cannot convert %f to double", num);
         }
         upb_value_setdouble(&val, num);
         break;
-      case UPB_TYPE(FLOAT):
+      case UPB_TYPE_FLOAT:
         if (num > FLT_MAX || num < -FLT_MAX)
           luaL_error(L, "Cannot convert %f to float", num);
         upb_value_setfloat(&val, num);
@@ -131,34 +125,21 @@ static upb_value lupb_getvalue(lua_State *L, int narg, upb_fieldtype_t type) {
 // Converts a upb_value -> Lua value.
 static void lupb_pushvalue(lua_State *L, upb_value val, upb_fieldtype_t type) {
   switch (type) {
-    case UPB_TYPE(INT32):
-    case UPB_TYPE(SINT32):
-    case UPB_TYPE(SFIXED32):
-    case UPB_TYPE(ENUM):
+    case UPB_TYPE_INT32:
+    case UPB_TYPE_ENUM:
       lua_pushnumber(L, upb_value_getint32(val)); break;
-    case UPB_TYPE(INT64):
-    case UPB_TYPE(SINT64):
-    case UPB_TYPE(SFIXED64):
+    case UPB_TYPE_INT64:
       lua_pushnumber(L, upb_value_getint64(val)); break;
-    case UPB_TYPE(UINT32):
-    case UPB_TYPE(FIXED32):
+    case UPB_TYPE_UINT32:
       lua_pushnumber(L, upb_value_getuint32(val)); break;
-    case UPB_TYPE(UINT64):
-    case UPB_TYPE(FIXED64):
+    case UPB_TYPE_UINT64:
       lua_pushnumber(L, upb_value_getuint64(val)); break;
-    case UPB_TYPE(DOUBLE):
+    case UPB_TYPE_DOUBLE:
       lua_pushnumber(L, upb_value_getdouble(val)); break;
-    case UPB_TYPE(FLOAT):
+    case UPB_TYPE_FLOAT:
       lua_pushnumber(L, upb_value_getfloat(val)); break;
-    case UPB_TYPE(BOOL):
+    case UPB_TYPE_BOOL:
       lua_pushboolean(L, upb_value_getbool(val)); break;
-    case UPB_TYPE(STRING):
-    case UPB_TYPE(BYTES): {
-      const upb_byteregion *r = upb_value_getbyteregion(val);
-      size_t len;
-      const char *str = upb_byteregion_getptr(r, 0, &len);
-      lua_pushlstring(L, str, len);
-    }
     default: luaL_error(L, "internal error");
   }
 }
@@ -341,7 +322,7 @@ static void lupb_fielddef_dosetdefault(lua_State *L, upb_fielddef *f,
   int type = lua_type(L, narg);
   upb_fieldtype_t upbtype = upb_fielddef_type(f);
   if (type == LUA_TSTRING) {
-    if (!upb_fielddef_isstring(f) && upbtype != UPB_TYPE(ENUM))
+    if (!upb_fielddef_isstring(f) && upbtype != UPB_TYPE_ENUM)
       luaL_argerror(L, narg, "field does not expect a string default");
     size_t len;
     const char *str = lua_tolstring(L, narg, &len);
@@ -387,6 +368,21 @@ static void lupb_fielddef_dosettype(lua_State *L, upb_fielddef *f, int narg) {
     luaL_argerror(L, narg, "invalid field type");
 }
 
+static void lupb_fielddef_dosetintfmt(lua_State *L, upb_fielddef *f, int narg) {
+  int32_t intfmt = luaL_checknumber(L, narg);
+  if (!upb_fielddef_settype(f, intfmt))
+    luaL_argerror(L, narg, "invalid field intfmt");
+}
+
+static void lupb_fielddef_dosettagdelim(lua_State *L, upb_fielddef *f,
+                                        int narg) {
+  if (!lua_isboolean(L, narg))
+    luaL_argerror(L, narg, "tagdelim value must be boolean");
+  int32_t tagdelim = luaL_checknumber(L, narg);
+  if (!upb_fielddef_settagdelim(f, tagdelim))
+    luaL_argerror(L, narg, "invalid field tagdelim");
+}
+
 // Setter API calls.  These use the setter functions above.
 
 static int lupb_fielddef_setdefault(lua_State *L) {
@@ -425,6 +421,18 @@ static int lupb_fielddef_settype(lua_State *L) {
   return 0;
 }
 
+static int lupb_fielddef_setintfmt(lua_State *L) {
+  upb_fielddef *f = lupb_fielddef_checkmutable(L, 1);
+  lupb_fielddef_dosetintfmt(L, f, 2);
+  return 0;
+}
+
+static int lupb_fielddef_settagdelim(lua_State *L) {
+  upb_fielddef *f = lupb_fielddef_checkmutable(L, 1);
+  lupb_fielddef_dosettagdelim(L, f, 2);
+  return 0;
+}
+
 // Constructor and other methods.
 
 static int lupb_fielddef_new(lua_State *L) {
@@ -470,11 +478,22 @@ static int lupb_fielddef_default(lua_State *L) {
   const upb_fielddef *f = lupb_fielddef_check(L, 1);
   upb_fieldtype_t type = upb_fielddef_type(f);
   if (upb_fielddef_default_is_symbolic(f))
-    type = UPB_TYPE(STRING);
+    type = UPB_TYPE_STRING;
   lupb_pushvalue(L, upb_fielddef_default(f), type);
   return 1;
 }
 
+static int lupb_fielddef_getsel(lua_State *L) {
+  const upb_fielddef *f = lupb_fielddef_check(L, 1);
+  upb_selector_t sel;
+  if (upb_getselector(f, luaL_checknumber(L, 2), &sel)) {
+    lua_pushnumber(L, sel);
+    return 1;
+  } else {
+    return 0;
+  }
+}
+
 static int lupb_fielddef_label(lua_State *L) {
   const upb_fielddef *f = lupb_fielddef_check(L, 1);
   lua_pushnumber(L, upb_fielddef_label(f));
@@ -530,7 +549,22 @@ static int lupb_fielddef_subdefname(lua_State *L) {
 
 static int lupb_fielddef_type(lua_State *L) {
   const upb_fielddef *f = lupb_fielddef_check(L, 1);
-  lua_pushnumber(L, upb_fielddef_type(f));
+  if (upb_fielddef_typeisset(f))
+    lua_pushnumber(L, upb_fielddef_type(f));
+  else
+    lua_pushnil(L);
+  return 1;
+}
+
+static int lupb_fielddef_intfmt(lua_State *L) {
+  const upb_fielddef *f = lupb_fielddef_check(L, 1);
+  lua_pushnumber(L, upb_fielddef_intfmt(f));
+  return 1;
+}
+
+static int lupb_fielddef_istagdelim(lua_State *L) {
+  const upb_fielddef *f = lupb_fielddef_check(L, 1);
+  lua_pushboolean(L, upb_fielddef_istagdelim(f));
   return 1;
 }
 
@@ -545,7 +579,10 @@ static const struct luaL_Reg lupb_fielddef_m[] = {
   LUPB_COMMON_DEF_METHODS
 
   {"default", lupb_fielddef_default},
+  {"getsel", lupb_fielddef_getsel},
   {"has_subdef", lupb_fielddef_hassubdef},
+  {"intfmt", lupb_fielddef_intfmt},
+  {"istagdelim", lupb_fielddef_istagdelim},
   {"label", lupb_fielddef_label},
   {"msgdef", lupb_fielddef_msgdef},
   {"name", lupb_def_fullname},  // name() is just an alias for fullname()
@@ -561,6 +598,8 @@ static const struct luaL_Reg lupb_fielddef_m[] = {
   {"set_subdef", lupb_fielddef_setsubdef},
   {"set_subdef_name", lupb_fielddef_setsubdefname},
   {"set_type", lupb_fielddef_settype},
+  {"set_intfmt", lupb_fielddef_setintfmt},
+  {"set_tagdelim", lupb_fielddef_settagdelim},
 
   // Internal-only.
   {"_selector_base", lupb_fielddef_selectorbase},
@@ -1040,28 +1079,44 @@ int luaopen_upb(lua_State *L) {
   lua_call(L, 1, 0);
 
   // Register constants.
-  lupb_setfieldi(L, "LABEL_OPTIONAL", UPB_LABEL(OPTIONAL));
-  lupb_setfieldi(L, "LABEL_REQUIRED", UPB_LABEL(REQUIRED));
-  lupb_setfieldi(L, "LABEL_REPEATED", UPB_LABEL(REPEATED));
-
-  lupb_setfieldi(L, "TYPE_DOUBLE",    UPB_TYPE(DOUBLE));
-  lupb_setfieldi(L, "TYPE_FLOAT",     UPB_TYPE(FLOAT));
-  lupb_setfieldi(L, "TYPE_INT64",     UPB_TYPE(INT64));
-  lupb_setfieldi(L, "TYPE_UINT64",    UPB_TYPE(UINT64));
-  lupb_setfieldi(L, "TYPE_INT32",     UPB_TYPE(INT32));
-  lupb_setfieldi(L, "TYPE_FIXED64",   UPB_TYPE(FIXED64));
-  lupb_setfieldi(L, "TYPE_FIXED32",   UPB_TYPE(FIXED32));
-  lupb_setfieldi(L, "TYPE_BOOL",      UPB_TYPE(BOOL));
-  lupb_setfieldi(L, "TYPE_STRING",    UPB_TYPE(STRING));
-  lupb_setfieldi(L, "TYPE_GROUP",     UPB_TYPE(GROUP));
-  lupb_setfieldi(L, "TYPE_MESSAGE",   UPB_TYPE(MESSAGE));
-  lupb_setfieldi(L, "TYPE_BYTES",     UPB_TYPE(BYTES));
-  lupb_setfieldi(L, "TYPE_UINT32",    UPB_TYPE(UINT32));
-  lupb_setfieldi(L, "TYPE_ENUM",      UPB_TYPE(ENUM));
-  lupb_setfieldi(L, "TYPE_SFIXED32",  UPB_TYPE(SFIXED32));
-  lupb_setfieldi(L, "TYPE_SFIXED64",  UPB_TYPE(SFIXED64));
-  lupb_setfieldi(L, "TYPE_SINT32",    UPB_TYPE(SINT32));
-  lupb_setfieldi(L, "TYPE_SINT64",    UPB_TYPE(SINT64));
+  lupb_setfieldi(L, "LABEL_OPTIONAL", UPB_LABEL_OPTIONAL);
+  lupb_setfieldi(L, "LABEL_REQUIRED", UPB_LABEL_REQUIRED);
+  lupb_setfieldi(L, "LABEL_REPEATED", UPB_LABEL_REPEATED);
+
+  lupb_setfieldi(L, "TYPE_DOUBLE",    UPB_TYPE_DOUBLE);
+  lupb_setfieldi(L, "TYPE_FLOAT",     UPB_TYPE_FLOAT);
+  lupb_setfieldi(L, "TYPE_INT64",     UPB_TYPE_INT64);
+  lupb_setfieldi(L, "TYPE_UINT64",    UPB_TYPE_UINT64);
+  lupb_setfieldi(L, "TYPE_INT32",     UPB_TYPE_INT32);
+  lupb_setfieldi(L, "TYPE_BOOL",      UPB_TYPE_BOOL);
+  lupb_setfieldi(L, "TYPE_STRING",    UPB_TYPE_STRING);
+  lupb_setfieldi(L, "TYPE_MESSAGE",   UPB_TYPE_MESSAGE);
+  lupb_setfieldi(L, "TYPE_BYTES",     UPB_TYPE_BYTES);
+  lupb_setfieldi(L, "TYPE_UINT32",    UPB_TYPE_UINT32);
+  lupb_setfieldi(L, "TYPE_ENUM",      UPB_TYPE_ENUM);
+
+  lupb_setfieldi(L, "INTFMT_VARIABLE",  UPB_INTFMT_VARIABLE);
+  lupb_setfieldi(L, "INTFMT_FIXED",     UPB_INTFMT_FIXED);
+  lupb_setfieldi(L, "INTFMT_ZIGZAG",    UPB_INTFMT_ZIGZAG);
+
+  lupb_setfieldi(L, "DESCRIPTOR_TYPE_DOUBLE",    UPB_DESCRIPTOR_TYPE_DOUBLE);
+  lupb_setfieldi(L, "DESCRIPTOR_TYPE_FLOAT",     UPB_DESCRIPTOR_TYPE_FLOAT);
+  lupb_setfieldi(L, "DESCRIPTOR_TYPE_INT64",     UPB_DESCRIPTOR_TYPE_INT64);
+  lupb_setfieldi(L, "DESCRIPTOR_TYPE_UINT64",    UPB_DESCRIPTOR_TYPE_UINT64);
+  lupb_setfieldi(L, "DESCRIPTOR_TYPE_INT32",     UPB_DESCRIPTOR_TYPE_INT32);
+  lupb_setfieldi(L, "DESCRIPTOR_TYPE_FIXED64",   UPB_DESCRIPTOR_TYPE_FIXED64);
+  lupb_setfieldi(L, "DESCRIPTOR_TYPE_FIXED32",   UPB_DESCRIPTOR_TYPE_FIXED32);
+  lupb_setfieldi(L, "DESCRIPTOR_TYPE_BOOL",      UPB_DESCRIPTOR_TYPE_BOOL);
+  lupb_setfieldi(L, "DESCRIPTOR_TYPE_STRING",    UPB_DESCRIPTOR_TYPE_STRING);
+  lupb_setfieldi(L, "DESCRIPTOR_TYPE_GROUP",     UPB_DESCRIPTOR_TYPE_GROUP);
+  lupb_setfieldi(L, "DESCRIPTOR_TYPE_MESSAGE",   UPB_DESCRIPTOR_TYPE_MESSAGE);
+  lupb_setfieldi(L, "DESCRIPTOR_TYPE_BYTES",     UPB_DESCRIPTOR_TYPE_BYTES);
+  lupb_setfieldi(L, "DESCRIPTOR_TYPE_UINT32",    UPB_DESCRIPTOR_TYPE_UINT32);
+  lupb_setfieldi(L, "DESCRIPTOR_TYPE_ENUM",      UPB_DESCRIPTOR_TYPE_ENUM);
+  lupb_setfieldi(L, "DESCRIPTOR_TYPE_SFIXED32",  UPB_DESCRIPTOR_TYPE_SFIXED32);
+  lupb_setfieldi(L, "DESCRIPTOR_TYPE_SFIXED64",  UPB_DESCRIPTOR_TYPE_SFIXED64);
+  lupb_setfieldi(L, "DESCRIPTOR_TYPE_SINT32",    UPB_DESCRIPTOR_TYPE_SINT32);
+  lupb_setfieldi(L, "DESCRIPTOR_TYPE_SINT64",    UPB_DESCRIPTOR_TYPE_SINT64);
 
   lupb_setfieldi(L, "DEF_MSG",      UPB_DEF_MSG);
   lupb_setfieldi(L, "DEF_FIELD",    UPB_DEF_FIELD);
@@ -1069,6 +1124,21 @@ int luaopen_upb(lua_State *L) {
   lupb_setfieldi(L, "DEF_SERVICE",  UPB_DEF_SERVICE);
   lupb_setfieldi(L, "DEF_ANY",      UPB_DEF_ANY);
 
+  lupb_setfieldi(L, "UPB_HANDLER_INT32",       UPB_HANDLER_INT32);
+  lupb_setfieldi(L, "UPB_HANDLER_INT64",       UPB_HANDLER_INT64);
+  lupb_setfieldi(L, "UPB_HANDLER_UINT32",      UPB_HANDLER_UINT32);
+  lupb_setfieldi(L, "UPB_HANDLER_UINT64",      UPB_HANDLER_UINT64);
+  lupb_setfieldi(L, "UPB_HANDLER_FLOAT",       UPB_HANDLER_FLOAT);
+  lupb_setfieldi(L, "UPB_HANDLER_DOUBLE",      UPB_HANDLER_DOUBLE);
+  lupb_setfieldi(L, "UPB_HANDLER_BOOL",        UPB_HANDLER_BOOL);
+  lupb_setfieldi(L, "UPB_HANDLER_STARTSTR",    UPB_HANDLER_STARTSTR);
+  lupb_setfieldi(L, "UPB_HANDLER_STRING",      UPB_HANDLER_STRING);
+  lupb_setfieldi(L, "UPB_HANDLER_ENDSTR",      UPB_HANDLER_ENDSTR);
+  lupb_setfieldi(L, "UPB_HANDLER_STARTSUBMSG", UPB_HANDLER_STARTSUBMSG);
+  lupb_setfieldi(L, "UPB_HANDLER_ENDSUBMSG",   UPB_HANDLER_ENDSUBMSG);
+  lupb_setfieldi(L, "UPB_HANDLER_STARTSEQ",    UPB_HANDLER_STARTSEQ);
+  lupb_setfieldi(L, "UPB_HANDLER_ENDSEQ",      UPB_HANDLER_ENDSEQ);
+
   return 1;  // Return package table.
 }
 

tests/test_cpp.cc

@@ -10,9 +10,10 @@
 #include <stdio.h>
 #include <string.h>
 #include <iostream>
-#include "upb/bytestream.h"
 #include "upb/def.h"
+#include "upb/descriptor/reader.h"
 #include "upb/handlers.h"
+#include "upb/pb/decoder.h"
 #include "upb/pb/glue.h"
 #include "upb_test.h"
 #include "upb/upb.h"
@@ -31,16 +32,6 @@ static void TestSymbolTable(const char *descriptor_file) {
   md->Unref(&md);
 }
 
-static void TestByteStream() {
-  upb::StringSource stringsrc;
-  stringsrc.Reset("testing", 7);
-  upb::ByteRegion* byteregion = stringsrc.AllBytes();
-  ASSERT(byteregion->FetchAll() == UPB_BYTE_OK);
-  char* str = byteregion->StrDup();
-  ASSERT(strcmp(str, "testing") == 0);
-  free(str);
-}
-
 extern "C" {
 
 int run_tests(int argc, char *argv[]) {
@@ -49,7 +40,6 @@ int run_tests(int argc, char *argv[]) {
     return 1;
   }
   TestSymbolTable(argv[1]);
-  TestByteStream();
   return 0;
 }
 

tests/test_decoder.cc

@@ -31,11 +31,12 @@
 #include <stdint.h>
 #include <stdlib.h>
 #include <string.h>
+#include "upb/bytestream.h"
 #include "upb/handlers.h"
 #include "upb/pb/decoder.h"
 #include "upb/pb/varint.h"
-#include "upb/upb.h"
 #include "upb_test.h"
+#include "upb/upb.h"
 #include "third_party/upb/tests/test_decoder_schema.upb.h"
 
 uint32_t filter_hash = 0;
@@ -186,16 +187,13 @@ void indentbuf(buffer *buf, int depth) {
     buf->append("  ", 2);
 }
 
-void indent(void *depth) {
-  indentbuf(&output, *(int*)depth);
-}
-
 #define NUMERIC_VALUE_HANDLER(member, ctype, fmt) \
-  bool value_ ## member(void *closure, void *fval, ctype val) {       \
-    indent(closure);                                                  \
-    uint32_t *num = static_cast<uint32_t*>(fval);                     \
-    output.appendf("%" PRIu32 ":%" fmt "\n", *num, val);              \
-    return true;                                                      \
+  bool value_ ## member(const upb::SinkFrame *frame, ctype val) {  \
+    int *depth = (int*)frame->userdata();                                      \
+    indentbuf(&output, *depth);                                                \
+    uint32_t *num = static_cast<uint32_t*>(frame->handler_data());             \
+    output.appendf("%" PRIu32 ":%" fmt "\n", *num, val);                       \
+    return true;                                                               \
   }
 
 NUMERIC_VALUE_HANDLER(uint32, uint32_t, PRIu32)
@@ -205,68 +203,73 @@ NUMERIC_VALUE_HANDLER(int64,  int64_t,  PRId64)
 NUMERIC_VALUE_HANDLER(float,  float,    "g")
 NUMERIC_VALUE_HANDLER(double, double,   "g")
 
-bool value_bool(void *closure, void *fval, bool val) {
-  indent(closure);
-  uint32_t *num = static_cast<uint32_t*>(fval);
+bool value_bool(const upb::SinkFrame *frame, bool val) {
+  int *depth = (int*)frame->userdata();
+  indentbuf(&output, *depth);
+  uint32_t *num = static_cast<uint32_t*>(frame->handler_data());
   output.appendf("%" PRIu32 ":%s\n", *num, val ? "true" : "false");
   return true;
 }
 
-void* startstr(void *closure, void *fval, size_t size_hint) {
-  indent(closure);
-  uint32_t *num = static_cast<uint32_t*>(fval);
+void* startstr(const upb::SinkFrame* frame, size_t size_hint) {
+  int *depth = (int*)frame->userdata();
+  indentbuf(&output, *depth);
+  uint32_t *num = static_cast<uint32_t*>(frame->handler_data());
   output.appendf("%" PRIu32 ":(%zu)\"", *num, size_hint);
-  return ((int*)closure) + 1;
+  return depth + 1;
 }
 
-size_t value_string(void *closure, void *fval, const char *buf, size_t n) {
+size_t value_string(const upb::SinkFrame* frame, const char* buf, size_t n) {
   output.append(buf, n);
   return n;
 }
 
-bool endstr(void *closure, void *fval) {
-  UPB_UNUSED(fval);
+bool endstr(const upb::SinkFrame* frame) {
   output.append("\"\n");
   return true;
 }
 
-void* startsubmsg(void *closure, void *fval) {
-  indent(closure);
-  uint32_t *num = static_cast<uint32_t*>(fval);
+void* startsubmsg(const upb::SinkFrame* frame) {
+  int *depth = (int*)frame->userdata();
+  indentbuf(&output, *depth);
+  uint32_t *num = static_cast<uint32_t*>(frame->handler_data());
   output.appendf("%" PRIu32 ":{\n", *num);
-  return ((int*)closure) + 1;
+  return depth + 1;
 }
 
-bool endsubmsg(void *closure, void *fval) {
-  UPB_UNUSED(fval);
-  indent(closure);
+bool endsubmsg(const upb::SinkFrame* frame) {
+  int *depth = (int*)frame->userdata();
+  indentbuf(&output, *depth);
   output.append("}\n");
   return true;
 }
 
-void* startseq(void *closure, void *fval) {
-  indent(closure);
-  uint32_t *num = static_cast<uint32_t*>(fval);
+void* startseq(const upb::SinkFrame* frame) {
+  int *depth = (int*)frame->userdata();
+  indentbuf(&output, *depth);
+  uint32_t *num = static_cast<uint32_t*>(frame->handler_data());
   output.appendf("%" PRIu32 ":[\n", *num);
-  return ((int*)closure) + 1;
+  return depth + 1;
 }
 
-bool endseq(void *closure, void *fval) {
-  UPB_UNUSED(fval);
-  indent(closure);
+bool endseq(const upb::SinkFrame* frame) {
+  int *depth = (int*)frame->userdata();
+  indentbuf(&output, *depth);
   output.append("]\n");
   return true;
 }
 
-bool startmsg(void *closure) {
-  indent(closure);
+bool startmsg(const upb::SinkFrame* frame) {
+  int *depth = (int*)frame->userdata();
+  indentbuf(&output, *depth);
   output.append("<\n");
   return true;
 }
 
-void endmsg(void *closure, upb_status *status) {
+void endmsg(const upb::SinkFrame* frame, upb_status* status) {
   (void)status;
-  indent(closure);
+  int *depth = (int*)frame->userdata();
+  indentbuf(&output, *depth);
   output.append(">\n");
 }
 
@@ -299,7 +302,7 @@ uint32_t rep_fn(uint32_t fn) {
 #define UNKNOWN_FIELD 666
 
 template <class T>
-void reg(upb_handlers *h, upb_fieldtype_t type,
+void reg(upb_handlers *h, upb_descriptortype_t type,
          typename upb::Handlers::Value<T>::Handler *handler) {
   // We register both a repeated and a non-repeated field for every type.
   // For the non-repeated field we make the field number the same as the
@@ -346,167 +349,111 @@ void reghandlers(upb_handlers *h) {
   upb_handlers_setendmsg(h, &endmsg);
 
   // Register handlers for each type.
-  reg<double>  (h, UPB_TYPE(DOUBLE),   &value_double);
-  reg<float>   (h, UPB_TYPE(FLOAT),    &value_float);
-  reg<int64_t> (h, UPB_TYPE(INT64),    &value_int64);
-  reg<uint64_t>(h, UPB_TYPE(UINT64),   &value_uint64);
-  reg<int32_t> (h, UPB_TYPE(INT32) ,   &value_int32);
-  reg<uint64_t>(h, UPB_TYPE(FIXED64),  &value_uint64);
-  reg<uint32_t>(h, UPB_TYPE(FIXED32),  &value_uint32);
-  reg<bool>    (h, UPB_TYPE(BOOL),     &value_bool);
-  reg<uint32_t>(h, UPB_TYPE(UINT32),   &value_uint32);
-  reg<int32_t> (h, UPB_TYPE(ENUM),     &value_int32);
-  reg<int32_t> (h, UPB_TYPE(SFIXED32), &value_int32);
-  reg<int64_t> (h, UPB_TYPE(SFIXED64), &value_int64);
-  reg<int32_t> (h, UPB_TYPE(SINT32),   &value_int32);
-  reg<int64_t> (h, UPB_TYPE(SINT64),   &value_int64);
-
-  reg_str(h, UPB_TYPE(STRING));
-  reg_str(h, UPB_TYPE(BYTES));
-  reg_str(h, rep_fn(UPB_TYPE(STRING)));
-  reg_str(h, rep_fn(UPB_TYPE(BYTES)));
+  reg<double>  (h, UPB_DESCRIPTOR_TYPE_DOUBLE,   &value_double);
+  reg<float>   (h, UPB_DESCRIPTOR_TYPE_FLOAT,    &value_float);
+  reg<int64_t> (h, UPB_DESCRIPTOR_TYPE_INT64,    &value_int64);
+  reg<uint64_t>(h, UPB_DESCRIPTOR_TYPE_UINT64,   &value_uint64);
+  reg<int32_t> (h, UPB_DESCRIPTOR_TYPE_INT32 ,   &value_int32);
+  reg<uint64_t>(h, UPB_DESCRIPTOR_TYPE_FIXED64,  &value_uint64);
+  reg<uint32_t>(h, UPB_DESCRIPTOR_TYPE_FIXED32,  &value_uint32);
+  reg<bool>    (h, UPB_DESCRIPTOR_TYPE_BOOL,     &value_bool);
+  reg<uint32_t>(h, UPB_DESCRIPTOR_TYPE_UINT32,   &value_uint32);
+  reg<int32_t> (h, UPB_DESCRIPTOR_TYPE_ENUM,     &value_int32);
+  reg<int32_t> (h, UPB_DESCRIPTOR_TYPE_SFIXED32, &value_int32);
+  reg<int64_t> (h, UPB_DESCRIPTOR_TYPE_SFIXED64, &value_int64);
+  reg<int32_t> (h, UPB_DESCRIPTOR_TYPE_SINT32,   &value_int32);
+  reg<int64_t> (h, UPB_DESCRIPTOR_TYPE_SINT64,   &value_int64);
+
+  reg_str(h, UPB_DESCRIPTOR_TYPE_STRING);
+  reg_str(h, UPB_DESCRIPTOR_TYPE_BYTES);
+  reg_str(h, rep_fn(UPB_DESCRIPTOR_TYPE_STRING));
+  reg_str(h, rep_fn(UPB_DESCRIPTOR_TYPE_BYTES));
 
   // Register submessage/group handlers that are self-recursive
   // to this type, eg: message M { optional M m = 1; }
-  reg_subm(h, UPB_TYPE(MESSAGE));
-  reg_subm(h, rep_fn(UPB_TYPE(MESSAGE)));
+  reg_subm(h, UPB_DESCRIPTOR_TYPE_MESSAGE);
+  reg_subm(h, rep_fn(UPB_DESCRIPTOR_TYPE_MESSAGE));
 
   // For NOP_FIELD we register no handlers, so we can pad a proto freely without
   // changing the output.
 }
 
 
-/* Custom bytesrc that can insert buffer seams in arbitrary places ************/
-
-typedef struct {
-  upb_bytesrc bytesrc;
-  const char *str;
-  size_t len, seam1, seam2;
-  upb_byteregion byteregion;
-} upb_seamsrc;
-
-size_t upb_seamsrc_avail(const upb_seamsrc *src, size_t ofs) {
-  if (ofs < src->seam1) return src->seam1 - ofs;
-  if (ofs < src->seam2) return src->seam2 - ofs;
-  return src->len - ofs;
-}
-
-upb_bytesuccess_t upb_seamsrc_fetch(void *_src, uint64_t ofs, size_t *read) {
-  upb_seamsrc *src = (upb_seamsrc*)_src;
-  assert(ofs < src->len);
-  if (ofs == src->len) {
-    upb_status_seteof(&src->bytesrc.status);
-    return UPB_BYTE_EOF;
-  }
-  *read = upb_seamsrc_avail(src, ofs);
-  return UPB_BYTE_OK;
-}
-
-void upb_seamsrc_copy(const void *_src, uint64_t ofs,
-                      size_t len, char *dst) {
-  const upb_seamsrc *src = (const upb_seamsrc*)_src;
-  assert(ofs + len <= src->len);
-  memcpy(dst, src->str + ofs, len);
-}
-
-void upb_seamsrc_discard(void *src, uint64_t ofs) {
-  (void)src;
-  (void)ofs;
-}
-
-const char *upb_seamsrc_getptr(const void *_s, uint64_t ofs, size_t *len) {
-  const upb_seamsrc *src = (const upb_seamsrc*)_s;
-  *len = upb_seamsrc_avail(src, ofs);
-  return src->str + ofs;
-}
-
-void upb_seamsrc_init(upb_seamsrc *s, const char *str, size_t len) {
-  static upb_bytesrc_vtbl vtbl = {
-    &upb_seamsrc_fetch,
-    &upb_seamsrc_discard,
-    &upb_seamsrc_copy,
-    &upb_seamsrc_getptr,
-  };
-  upb_bytesrc_init(&s->bytesrc, &vtbl);
-  s->seam1 = 0;
-  s->seam2 = 0;
-  s->str = str;
-  s->len = len;
-  s->byteregion.bytesrc = &s->bytesrc;
-  s->byteregion.toplevel = true;
-  s->byteregion.start = 0;
-  s->byteregion.end = len;
-}
-
-void upb_seamsrc_resetseams(upb_seamsrc *s, size_t seam1, size_t seam2) {
-  assert(seam1 <= seam2);
-  s->seam1 = seam1;
-  s->seam2 = seam2;
-  s->byteregion.discard = 0;
-  s->byteregion.fetch = 0;
-}
-
-void upb_seamsrc_uninit(upb_seamsrc *s) { (void)s; }
-
-upb_bytesrc *upb_seamsrc_bytesrc(upb_seamsrc *s) {
-  return &s->bytesrc;
-}
-
-// Returns the top-level upb_byteregion* for this seamsrc.  Invalidated when
-// the seamsrc is reset.
-upb_byteregion *upb_seamsrc_allbytes(upb_seamsrc *s) {
-  return &s->byteregion;
-}
-
-
 /* Running of test cases ******************************************************/
 
-upb_decoderplan *plan;
+const upb::Handlers *handlers;
+const upb::Handlers *plan;
 
 uint32_t Hash(const buffer& proto, const buffer* expected_output) {
   uint32_t hash = MurmurHash2(proto.buf(), proto.len(), 0);
   if (expected_output)
     hash = MurmurHash2(expected_output->buf(), expected_output->len(), hash);
-  bool hasjit = upb_decoderplan_hasjitcode(plan);
+  bool hasjit = upb::pb::HasJitCode(plan);
   hash = MurmurHash2(&hasjit, 1, hash);
   return hash;
 }
 
+bool parse(
+    upb_sink *s, const char *buf, size_t start, size_t end, size_t *ofs) {
+  start = UPB_MAX(start, *ofs);
+  if (start <= end) {
+    size_t len = end - start;
+    size_t parsed =
+        s->PutStringBuffer(UPB_BYTESTREAM_BYTES_STRING, buf + start, len);
+    if (s->pipeline()->status().ok() != (parsed >= len)) {
+      ASSERT(false);
+    }
+    if (!s->pipeline()->status().ok())
+      return false;
+    *ofs += parsed;
+  }
+  return true;
+}
+
 #define LINE(x) x "\n"
 void run_decoder(const buffer& proto, const buffer* expected_output) {
   testhash = Hash(proto, expected_output);
   if (filter_hash && testhash != filter_hash) return;
-  upb_seamsrc src;
-  upb_seamsrc_init(&src, proto.buf(), proto.len());
-  upb_decoder d;
-  upb_decoder_init(&d);
-  upb_decoder_resetplan(&d, plan);
+  upb::Pipeline pipeline(NULL, 0, upb_realloc, NULL);
+  upb::Sink* sink = pipeline.NewSink(handlers);
+  upb::Sink* decoder_sink = pipeline.NewSink(plan);
+  upb::pb::Decoder* d = decoder_sink->base()->GetUserdata<upb::pb::Decoder>();
+  upb::pb::ResetDecoderSink(d, sink);
   for (size_t i = 0; i < proto.len(); i++) {
     for (size_t j = i; j < UPB_MIN(proto.len(), i + 5); j++) {
-      upb_seamsrc_resetseams(&src, i, j);
-      upb_byteregion *input = upb_seamsrc_allbytes(&src);
+      pipeline.Reset();
       output.clear();
-      upb_decoder_resetinput(&d, input, &closures[0]);
-      upb_success_t success = upb_decoder_decode(&d);
-      ASSERT(upb_ok(upb_decoder_status(&d)) == (success == UPB_OK));
+      sink->Reset(&closures[0]);
+      size_t ofs = 0;
+      bool ok =
+          decoder_sink->StartMessage() &&
+          decoder_sink->StartString(
+              UPB_BYTESTREAM_BYTES_STARTSTR, proto.len()) &&
+          parse(decoder_sink, proto.buf(), 0, i, &ofs) &&
+          parse(decoder_sink, proto.buf(), i, j, &ofs) &&
+          parse(decoder_sink, proto.buf(), j, proto.len(), &ofs) &&
+          ofs == proto.len() &&
+          decoder_sink->EndString(UPB_BYTESTREAM_BYTES_ENDSTR);
+      if (ok) decoder_sink->EndMessage();
       if (expected_output) {
-        ASSERT_STATUS(success == UPB_OK, upb_decoder_status(&d));
-        // The input should be fully consumed.
-        ASSERT(upb_byteregion_fetchofs(input) == upb_byteregion_endofs(input));
-        ASSERT(upb_byteregion_discardofs(input) ==
-               upb_byteregion_endofs(input));
         if (!output.eql(*expected_output)) {
           fprintf(stderr, "Text mismatch: '%s' vs '%s'\n",
                   output.buf(), expected_output->buf());
         }
+        if (!ok) {
+          fprintf(stderr, "Failed: %s\n", pipeline.status().GetString());
+        }
+        ASSERT(ok);
         ASSERT(output.eql(*expected_output));
       } else {
-        ASSERT(success == UPB_ERROR);
+        if (ok) {
+          fprintf(stderr, "Didn't expect ok result, but got output: '%s'\n",
+                  output.buf());
+        }
+        ASSERT(!ok);
       }
     }
   }
-  upb_decoder_uninit(&d);
-  upb_seamsrc_uninit(&src);
   testhash = 0;
 }
 
@@ -540,7 +487,7 @@ void assert_does_not_parse(const buffer& proto) {
 
 /* The actual tests ***********************************************************/
 
-void test_premature_eof_for_type(upb_fieldtype_t type) {
+void test_premature_eof_for_type(upb_descriptortype_t type) {
   // Incomplete values for each wire type.
   static const buffer incompletes[6] = {
     buffer("\x80"),     // UPB_WIRE_TYPE_VARINT
@@ -590,10 +537,10 @@ void test_premature_eof_for_type(upb_fieldtype_t type) {
     assert_does_not_parse_at_eof(
         cat( tag(UNKNOWN_FIELD, wire_type), varint(1) ));
 
-    if (type == UPB_TYPE(MESSAGE)) {
+    if (type == UPB_DESCRIPTOR_TYPE_MESSAGE) {
       // Submessage ends in the middle of a value.
       buffer incomplete_submsg =
-          cat ( tag(UPB_TYPE(INT32), UPB_WIRE_TYPE_VARINT),
+          cat ( tag(UPB_DESCRIPTOR_TYPE_INT32, UPB_WIRE_TYPE_VARINT),
                 incompletes[UPB_WIRE_TYPE_VARINT] );
       assert_does_not_parse(
           cat( tag(fieldnum, UPB_WIRE_TYPE_DELIMITED),
@@ -615,7 +562,7 @@ void test_premature_eof_for_type(upb_fieldtype_t type) {
 
 // "33" and "66" are just two random values that all numeric types can
 // represent.
-void test_valid_data_for_type(upb_fieldtype_t type,
+void test_valid_data_for_type(upb_descriptortype_t type,
                               const buffer& enc33, const buffer& enc66) {
   uint32_t fieldnum = type;
   uint32_t rep_fieldnum = rep_fn(type);
@@ -653,7 +600,7 @@ void test_valid_data_for_type(upb_fieldtype_t type,
       LINE(">"), rep_fieldnum, rep_fieldnum, rep_fieldnum);
 }
 
-void test_valid_data_for_signed_type(upb_fieldtype_t type,
+void test_valid_data_for_signed_type(upb_descriptortype_t type,
                                      const buffer& enc33, const buffer& enc66) {
   uint32_t fieldnum = type;
   uint32_t rep_fieldnum = rep_fn(type);
@@ -694,22 +641,22 @@ void test_valid_data_for_signed_type(upb_fieldtype_t type,
 // Test that invalid protobufs are properly detected (without crashing) and
 // have an error reported.  Field numbers match registered handlers above.
 void test_invalid() {
-  test_premature_eof_for_type(UPB_TYPE(DOUBLE));
-  test_premature_eof_for_type(UPB_TYPE(FLOAT));
-  test_premature_eof_for_type(UPB_TYPE(INT64));
-  test_premature_eof_for_type(UPB_TYPE(UINT64));
-  test_premature_eof_for_type(UPB_TYPE(INT32));
-  test_premature_eof_for_type(UPB_TYPE(FIXED64));
-  test_premature_eof_for_type(UPB_TYPE(FIXED32));
-  test_premature_eof_for_type(UPB_TYPE(BOOL));
-  test_premature_eof_for_type(UPB_TYPE(STRING));
-  test_premature_eof_for_type(UPB_TYPE(BYTES));
-  test_premature_eof_for_type(UPB_TYPE(UINT32));
-  test_premature_eof_for_type(UPB_TYPE(ENUM));
-  test_premature_eof_for_type(UPB_TYPE(SFIXED32));
-  test_premature_eof_for_type(UPB_TYPE(SFIXED64));
-  test_premature_eof_for_type(UPB_TYPE(SINT32));
-  test_premature_eof_for_type(UPB_TYPE(SINT64));
+  test_premature_eof_for_type(UPB_DESCRIPTOR_TYPE_DOUBLE);
+  test_premature_eof_for_type(UPB_DESCRIPTOR_TYPE_FLOAT);
+  test_premature_eof_for_type(UPB_DESCRIPTOR_TYPE_INT64);
+  test_premature_eof_for_type(UPB_DESCRIPTOR_TYPE_UINT64);
+  test_premature_eof_for_type(UPB_DESCRIPTOR_TYPE_INT32);
+  test_premature_eof_for_type(UPB_DESCRIPTOR_TYPE_FIXED64);
+  test_premature_eof_for_type(UPB_DESCRIPTOR_TYPE_FIXED32);
+  test_premature_eof_for_type(UPB_DESCRIPTOR_TYPE_BOOL);
+  test_premature_eof_for_type(UPB_DESCRIPTOR_TYPE_STRING);
+  test_premature_eof_for_type(UPB_DESCRIPTOR_TYPE_BYTES);
+  test_premature_eof_for_type(UPB_DESCRIPTOR_TYPE_UINT32);
+  test_premature_eof_for_type(UPB_DESCRIPTOR_TYPE_ENUM);
+  test_premature_eof_for_type(UPB_DESCRIPTOR_TYPE_SFIXED32);
+  test_premature_eof_for_type(UPB_DESCRIPTOR_TYPE_SFIXED64);
+  test_premature_eof_for_type(UPB_DESCRIPTOR_TYPE_SINT32);
+  test_premature_eof_for_type(UPB_DESCRIPTOR_TYPE_SINT64);
 
   // EOF inside a tag's varint.
   assert_does_not_parse_at_eof( buffer("\x80") );
@@ -734,31 +681,47 @@ void test_invalid() {
 
   // Test exceeding the resource limit of stack depth.
   buffer buf;
-  for (int i = 0; i < UPB_MAX_NESTING; i++) {
-    buf.assign(submsg(UPB_TYPE(MESSAGE), buf));
+  for (int i = 0; i <= UPB_MAX_NESTING; i++) {
+    buf.assign(submsg(UPB_DESCRIPTOR_TYPE_MESSAGE, buf));
   }
   assert_does_not_parse(buf);
 }
 
 void test_valid() {
-  test_valid_data_for_signed_type(UPB_TYPE(DOUBLE), dbl(33), dbl(-66));
-  test_valid_data_for_signed_type(UPB_TYPE(FLOAT), flt(33), flt(-66));
-  test_valid_data_for_signed_type(UPB_TYPE(INT64), varint(33), varint(-66));
-  test_valid_data_for_signed_type(UPB_TYPE(INT32), varint(33), varint(-66));
-  test_valid_data_for_signed_type(UPB_TYPE(ENUM), varint(33), varint(-66));
-  test_valid_data_for_signed_type(UPB_TYPE(SFIXED32), uint32(33), uint32(-66));
-  test_valid_data_for_signed_type(UPB_TYPE(SFIXED64), uint64(33), uint64(-66));
-  test_valid_data_for_signed_type(UPB_TYPE(SINT32), zz32(33), zz32(-66));
-  test_valid_data_for_signed_type(UPB_TYPE(SINT64), zz64(33), zz64(-66));
-
-  test_valid_data_for_type(UPB_TYPE(UINT64), varint(33), varint(66));
-  test_valid_data_for_type(UPB_TYPE(UINT32), varint(33), varint(66));
-  test_valid_data_for_type(UPB_TYPE(FIXED64), uint64(33), uint64(66));
-  test_valid_data_for_type(UPB_TYPE(FIXED32), uint32(33), uint32(66));
+  test_valid_data_for_signed_type(UPB_DESCRIPTOR_TYPE_DOUBLE,
+                                  dbl(33),
+                                  dbl(-66));
+  test_valid_data_for_signed_type(UPB_DESCRIPTOR_TYPE_FLOAT, flt(33), flt(-66));
+  test_valid_data_for_signed_type(UPB_DESCRIPTOR_TYPE_INT64,
+                                  varint(33),
+                                  varint(-66));
+  test_valid_data_for_signed_type(UPB_DESCRIPTOR_TYPE_INT32,
+                                  varint(33),
+                                  varint(-66));
+  test_valid_data_for_signed_type(UPB_DESCRIPTOR_TYPE_ENUM,
+                                  varint(33),
+                                  varint(-66));
+  test_valid_data_for_signed_type(UPB_DESCRIPTOR_TYPE_SFIXED32,
+                                  uint32(33),
+                                  uint32(-66));
+  test_valid_data_for_signed_type(UPB_DESCRIPTOR_TYPE_SFIXED64,
+                                  uint64(33),
+                                  uint64(-66));
+  test_valid_data_for_signed_type(UPB_DESCRIPTOR_TYPE_SINT32,
+                                  zz32(33),
+                                  zz32(-66));
+  test_valid_data_for_signed_type(UPB_DESCRIPTOR_TYPE_SINT64,
+                                  zz64(33),
+                                  zz64(-66));
+
+  test_valid_data_for_type(UPB_DESCRIPTOR_TYPE_UINT64, varint(33), varint(66));
+  test_valid_data_for_type(UPB_DESCRIPTOR_TYPE_UINT32, varint(33), varint(66));
+  test_valid_data_for_type(UPB_DESCRIPTOR_TYPE_FIXED64, uint64(33), uint64(66));
+  test_valid_data_for_type(UPB_DESCRIPTOR_TYPE_FIXED32, uint32(33), uint32(66));
 
   // Test implicit startseq/endseq.
-  uint32_t repfl_fn = rep_fn(UPB_TYPE(FLOAT));
-  uint32_t repdb_fn = rep_fn(UPB_TYPE(DOUBLE));
+  uint32_t repfl_fn = rep_fn(UPB_DESCRIPTOR_TYPE_FLOAT);
+  uint32_t repdb_fn = rep_fn(UPB_DESCRIPTOR_TYPE_DOUBLE);
   assert_successful_parse(
       cat( tag(repfl_fn, UPB_WIRE_TYPE_32BIT), flt(33),
            tag(repdb_fn, UPB_WIRE_TYPE_64BIT), dbl(66) ),
@@ -772,7 +735,7 @@ void test_valid() {
       LINE(">"), repfl_fn, repfl_fn, repdb_fn, repdb_fn);
 
   // Submessage tests.
-  uint32_t msg_fn = UPB_TYPE(MESSAGE);
+  uint32_t msg_fn = UPB_DESCRIPTOR_TYPE_MESSAGE;
   assert_successful_parse(
       submsg(msg_fn, submsg(msg_fn, submsg(msg_fn, buffer()))),
       LINE("<")
@@ -790,7 +753,7 @@ void test_valid() {
       LINE("}")
       LINE(">"), msg_fn, msg_fn, msg_fn);
 
-  uint32_t repm_fn = rep_fn(UPB_TYPE(MESSAGE));
+  uint32_t repm_fn = rep_fn(UPB_DESCRIPTOR_TYPE_MESSAGE);
   assert_successful_parse(
       submsg(repm_fn, submsg(repm_fn, buffer())),
       LINE("<")
@@ -813,11 +776,11 @@ void test_valid() {
   buffer textbuf;
   int total = UPB_MAX_NESTING - 1;
   for (int i = 0; i < total; i++) {
-    buf.assign(submsg(UPB_TYPE(MESSAGE), buf));
+    buf.assign(submsg(UPB_DESCRIPTOR_TYPE_MESSAGE, buf));
     indentbuf(&textbuf, i);
     textbuf.append("<\n");
     indentbuf(&textbuf, i);
-    textbuf.appendf("%u:{\n", UPB_TYPE(MESSAGE));
+    textbuf.appendf("%u:{\n", UPB_DESCRIPTOR_TYPE_MESSAGE);
   }
   indentbuf(&textbuf, total);
   textbuf.append("<\n");
@@ -848,35 +811,36 @@ int run_tests(int argc, char *argv[]) {
 
   // Create an empty handlers to make sure that the decoder can handle empty
   // messages.
-  upb_handlers *h = upb_handlers_new(UPB_TEST_DECODER_EMPTYMESSAGE, &h);
-  bool ok = upb_handlers_freeze(&h, 1, NULL);
+  upb::Handlers *h = upb_handlers_new(UPB_TEST_DECODER_EMPTYMESSAGE, NULL, &h);
+  bool ok = upb::Handlers::Freeze(&h, 1, NULL);
   ASSERT(ok);
-  plan = upb_decoderplan_new(h, true);
-  upb_handlers_unref(h, &h);
-  upb_decoderplan_unref(plan);
+  plan = upb::pb::GetDecoderHandlers(h, true, &plan);
+  h->Unref(&h);
+  plan->Unref(&plan);
 
   // Construct decoder plan.
-  h = upb_handlers_new(UPB_TEST_DECODER_DECODERTEST, &h);
+  h = upb::Handlers::New(UPB_TEST_DECODER_DECODERTEST, NULL, &handlers);
   reghandlers(h);
-  ok = upb_handlers_freeze(&h, 1, NULL);
+  ok = upb::Handlers::Freeze(&h, 1, NULL);
+  handlers = h;
 
   // Test without JIT.
-  plan = upb_decoderplan_new(h, false);
-  ASSERT(!upb_decoderplan_hasjitcode(plan));
+  plan = upb::pb::GetDecoderHandlers(handlers, false, &plan);
+  ASSERT(!upb::pb::HasJitCode(plan));
   run_tests();
-  upb_decoderplan_unref(plan);
+  plan->Unref(&plan);
 
 #ifdef UPB_USE_JIT_X64
   // Test JIT.
-  plan = upb_decoderplan_new(h, true);
-  ASSERT(upb_decoderplan_hasjitcode(plan));
+  plan = upb::pb::GetDecoderHandlers(handlers, true, &plan);
+  ASSERT(upb::pb::HasJitCode(plan));
   run_tests();
-  upb_decoderplan_unref(plan);
+  plan->Unref(&plan);
 #endif
 
   plan = NULL;
   printf("All tests passed, %d assertions.\n", num_assertions);
-  upb_handlers_unref(h, &h);
+  handlers->Unref(&handlers);
   return 0;
 }
 

tests/test_def.c

@@ -88,15 +88,15 @@ static void test_fielddef_accessors() {
   ASSERT(!upb_fielddef_isfrozen(f1));
   upb_fielddef_setname(f1, "f1");
   upb_fielddef_setnumber(f1, 1937);
-  upb_fielddef_settype(f1, UPB_TYPE(FIXED64));
-  upb_fielddef_setlabel(f1, UPB_LABEL(REPEATED));
+  upb_fielddef_settype(f1, UPB_TYPE_INT64);
+  upb_fielddef_setlabel(f1, UPB_LABEL_REPEATED);
   ASSERT(upb_fielddef_number(f1) == 1937);
 
   ASSERT(!upb_fielddef_isfrozen(f2));
   upb_fielddef_setname(f2, "f2");
   upb_fielddef_setnumber(f2, 1572);
-  upb_fielddef_settype(f2, UPB_TYPE(BYTES));
-  upb_fielddef_setlabel(f2, UPB_LABEL(REPEATED));
+  upb_fielddef_settype(f2, UPB_TYPE_BYTES);
+  upb_fielddef_setlabel(f2, UPB_LABEL_REPEATED);
   ASSERT(upb_fielddef_number(f2) == 1572);
 
   upb_fielddef_unref(f1, &f1);
@@ -104,7 +104,7 @@ static void test_fielddef_accessors() {
 
   // Test that we don't leak an unresolved subdef name.
   f1 = upb_fielddef_new(&f1);
-  upb_fielddef_settype(f1, UPB_TYPE(MESSAGE));
+  upb_fielddef_settype(f1, UPB_TYPE_MESSAGE);
   upb_fielddef_setsubdefname(f1, "YO");
   upb_fielddef_unref(f1, &f1);
 }
@@ -127,7 +127,7 @@ static upb_msgdef *upb_msgdef_newnamed(const char *name, void *owner) {
   return m;
 }
 
-INLINE upb_enumdef *upb_enumdef_newnamed(const char *name, void *owner) {
+static upb_enumdef *upb_enumdef_newnamed(const char *name, void *owner) {
   upb_enumdef *e = upb_enumdef_new(owner);
   upb_def_setfullname(upb_upcast(e), name);
   return e;
@@ -138,7 +138,7 @@ static void test_replacement() {
 
   upb_msgdef *m = upb_msgdef_newnamed("MyMessage", &s);
   upb_msgdef_addfield(m, newfield(
-      "field1", 1, UPB_TYPE(ENUM), UPB_LABEL(OPTIONAL), ".MyEnum", &s), &s);
+      "field1", 1, UPB_TYPE_ENUM, UPB_LABEL_OPTIONAL, ".MyEnum", &s), &s);
   upb_msgdef *m2 = upb_msgdef_newnamed("MyMessage2", &s);
   upb_enumdef *e = upb_enumdef_newnamed("MyEnum", &s);
 

tests/test_pipeline.c

@@ -0,0 +1,116 @@
+/*
+ * upb - a minimalist implementation of protocol buffers.
+ *
+ * Copyright (c) 2013 Google Inc.  See LICENSE for details.
+ *
+ * Test of upb_pipeline.
+ */
+
+#include "upb/sink.h"
+#include "tests/upb_test.h"
+
+static void *count_realloc(void *ud, void *ptr, size_t size) {
+  int *count = ud;
+  *count += 1;
+  return upb_realloc(ud, ptr, size);
+}
+
+static void test_empty() {
+  // A pipeline with no initial memory or allocation function should return
+  // NULL from attempts to allocate.
+  upb_pipeline pipeline;
+  upb_pipeline_init(&pipeline, NULL, 0, NULL, NULL);
+  ASSERT(upb_pipeline_alloc(&pipeline, 1) == NULL);
+  ASSERT(upb_pipeline_alloc(&pipeline, 1) == NULL);
+  ASSERT(upb_pipeline_realloc(&pipeline, NULL, 0, 1) == NULL);
+  upb_pipeline_uninit(&pipeline);
+}
+
+static void test_only_initial() {
+  upb_pipeline pipeline;
+  char initial[152];  // 128 + a conservative 24 bytes overhead.
+  upb_pipeline_init(&pipeline, initial, sizeof(initial), NULL, NULL);
+  void *p1 = upb_pipeline_alloc(&pipeline, 64);
+  void *p2 = upb_pipeline_alloc(&pipeline, 64);
+  void *p3 = upb_pipeline_alloc(&pipeline, 64);
+  ASSERT(p1);
+  ASSERT(p2);
+  ASSERT(!p3);
+  ASSERT(p1 != p2);
+  ASSERT((void*)initial <= p1);
+  ASSERT(p1 < p2);
+  ASSERT(p2 < (void*)(initial + sizeof(initial)));
+  upb_pipeline_uninit(&pipeline);
+}
+
+static void test_with_alloc_func() {
+  upb_pipeline pipeline;
+  char initial[152];  // 128 + a conservative 24 bytes overhead.
+  int count = 0;
+  upb_pipeline_init(&pipeline, initial, sizeof(initial), count_realloc, &count);
+  void *p1 = upb_pipeline_alloc(&pipeline, 64);
+  void *p2 = upb_pipeline_alloc(&pipeline, 64);
+  ASSERT(p1);
+  ASSERT(p2);
+  ASSERT(p1 != p2);
+  ASSERT(count == 0);
+
+  void *p3 = upb_pipeline_alloc(&pipeline, 64);
+  ASSERT(p3);
+  ASSERT(p3 != p2);
+  ASSERT(count == 1);
+
+  // Allocation larger than internal block size should force another alloc.
+  char *p4 = upb_pipeline_alloc(&pipeline, 16384);
+  ASSERT(p4);
+  p4[16383] = 1;  // Verify memory is writable without crashing.
+  ASSERT(p4[16383] == 1);
+  ASSERT(count == 2);
+
+  upb_pipeline_uninit(&pipeline);
+  ASSERT(count == 4);  // From two calls to free the memory.
+}
+
+static void test_realloc() {
+  upb_pipeline pipeline;
+  char initial[152];  // 128 + a conservative 24 bytes overhead.
+  int count = 0;
+  upb_pipeline_init(&pipeline, initial, sizeof(initial), count_realloc, &count);
+  void *p1 = upb_pipeline_alloc(&pipeline, 64);
+  // This realloc should work in-place.
+  void *p2 = upb_pipeline_realloc(&pipeline, p1, 64, 128);
+  ASSERT(p1);
+  ASSERT(p2);
+  ASSERT(p1 == p2);
+  ASSERT(count == 0);
+
+  // This realloc will *not* work in place, due to size.
+  void *p3 = upb_pipeline_realloc(&pipeline, p2, 128, 256);
+  ASSERT(p3);
+  ASSERT(p3 != p2);
+  ASSERT(count == 1);
+
+  void *p4 = upb_pipeline_alloc(&pipeline, 64);
+  void *p5 = upb_pipeline_alloc(&pipeline, 64);
+  // This realloc will *not* work in place because it was not the last
+  // allocation.
+  void *p6 = upb_pipeline_realloc(&pipeline, p4, 64, 128);
+  ASSERT(p4);
+  ASSERT(p5);
+  ASSERT(p6);
+  ASSERT(p4 != p6);
+  ASSERT(p4 < p5);
+  ASSERT(p5 < p6);
+  ASSERT(count == 1);  // These should all fit in the first dynamic block.
+
+  upb_pipeline_uninit(&pipeline);
+  ASSERT(count == 2);
+}
+
+int run_tests(int argc, char *argv[]) {
+  test_empty();
+  test_only_initial();
+  test_with_alloc_func();
+  test_realloc();
+  return 0;
+}

tests/test_table.cc

@@ -46,13 +46,14 @@ void test_strtable(const vector<std::string>& keys, uint32_t num_to_insert) {
   /* Test correctness. */
   for(uint32_t i = 0; i < keys.size(); i++) {
     const std::string& key = keys[i];
-    const upb_value *v = upb_strtable_lookup(&table, key.c_str());
+    upb_value v;
+    bool found = upb_strtable_lookup(&table, key.c_str(), &v);
     if(m.find(key) != m.end()) { /* Assume map implementation is correct. */
-      ASSERT(v);
-      ASSERT(upb_value_getint32(*v) == key[0]);
+      ASSERT(found);
+      ASSERT(upb_value_getint32(v) == key[0]);
       ASSERT(m[key] == key[0]);
     } else {
-      ASSERT(v == NULL);
+      ASSERT(!found);
     }
   }
 
@@ -88,14 +89,15 @@ void test_inttable(int32_t *keys, uint16_t num_entries, const char *desc) {
 
   /* Test correctness. */
   for(uint32_t i = 0; i <= largest_key; i++) {
-    const upb_value *v = upb_inttable_lookup(&table, i);
+    upb_value v;
+    bool found = upb_inttable_lookup(&table, i, &v);
     if(m.find(i) != m.end()) { /* Assume map implementation is correct. */
-      ASSERT(v);
-      ASSERT(upb_value_getuint32(*v) == i*2);
+      ASSERT(found);
+      ASSERT(upb_value_getuint32(v) == i*2);
       ASSERT(m[i] == i*2);
       ASSERT(hm[i] == i*2);
     } else {
-      ASSERT(v == NULL);
+      ASSERT(!found);
     }
   }
 
@@ -112,28 +114,30 @@ void test_inttable(int32_t *keys, uint16_t num_entries, const char *desc) {
 
   /* Test correctness. */
   for(uint32_t i = 0; i <= largest_key; i++) {
-    const upb_value *v = upb_inttable_lookup(&table, i);
+    upb_value v;
+    bool found = upb_inttable_lookup(&table, i, &v);
     if(m.find(i) != m.end()) { /* Assume map implementation is correct. */
-      ASSERT(v);
-      ASSERT(upb_value_getuint32(*v) == i*2);
+      ASSERT(found);
+      ASSERT(upb_value_getuint32(v) == i*2);
       ASSERT(m[i] == i*2);
       ASSERT(hm[i] == i*2);
     } else {
-      ASSERT(v == NULL);
+      ASSERT(!found);
     }
   }
 
   // Compact and test correctness again.
   upb_inttable_compact(&table);
   for(uint32_t i = 0; i <= largest_key; i++) {
-    const upb_value *v = upb_inttable_lookup(&table, i);
+    upb_value v;
+    bool found = upb_inttable_lookup(&table, i, &v);
     if(m.find(i) != m.end()) { /* Assume map implementation is correct. */
-      ASSERT(v);
-      ASSERT(upb_value_getuint32(*v) == i*2);
+      ASSERT(found);
+      ASSERT(upb_value_getuint32(v) == i*2);
       ASSERT(m[i] == i*2);
       ASSERT(hm[i] == i*2);
     } else {
-      ASSERT(v == NULL);
+      ASSERT(!found);
     }
   }
 
@@ -172,8 +176,9 @@ void test_inttable(int32_t *keys, uint16_t num_entries, const char *desc) {
   for(i = 0; true; i++) {
     MAYBE_BREAK;
     int32_t key = keys[i & mask];
-    const upb_value *v = upb_inttable_lookup32(&table, key);
-    x += (uintptr_t)v;
+    upb_value v;
+    bool ok = upb_inttable_lookup32(&table, key, &v);
+    x += (uintptr_t)ok;
   }
   double total = get_usertime() - before;
   printf("%s/s\n", eng(i/total, 3, false));
@@ -184,8 +189,9 @@ void test_inttable(int32_t *keys, uint16_t num_entries, const char *desc) {
   for(i = 0; true; i++) {
     MAYBE_BREAK;
     int32_t key = keys[rand_order[i & mask]];
-    const upb_value *v = upb_inttable_lookup32(&table, key);
-    x += (uintptr_t)v;
+    upb_value v;
+    bool ok = upb_inttable_lookup32(&table, key, &v);
+    x += (uintptr_t)ok;
   }
   total = get_usertime() - before;
   printf("%s/s\n", eng(i/total, 3, false));
@@ -232,6 +238,7 @@ void test_inttable(int32_t *keys, uint16_t num_entries, const char *desc) {
     x += hm[key];
   }
   total = get_usertime() - before;
+  if (x == INT_MAX) abort();
   printf("%s/s\n\n", eng(i/total, 3, false));
   upb_inttable_uninit(&table);
   delete rand_order;

tests/test_vs_proto2.cc

@@ -10,6 +10,7 @@
 #define __STDC_LIMIT_MACROS  // So we get UINT32_MAX
 #include <assert.h>
 #include <google/protobuf/descriptor.h>
+#include <google/protobuf/dynamic_message.h>
 #include <google/protobuf/message.h>
 #include <google/protobuf/wire_format_lite.h>
 #include <inttypes.h>
@@ -17,10 +18,11 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include "benchmarks/google_messages.pb.h"
-#include "bindings/cpp/upb/pb/decoder.hpp"
+#include "upb/bytestream.h"
 #include "upb/def.h"
 #include "upb/google/bridge.h"
 #include "upb/handlers.h"
+#include "upb/pb/decoder.h"
 #include "upb/pb/glue.h"
 #include "upb/pb/varint.h"
 #include "upb_test.h"
@@ -36,25 +38,36 @@ void compare_metadata(const google::protobuf::Descriptor* d,
     ASSERT(proto2_f);
     ASSERT(upb_f->number() == proto2_f->number());
     ASSERT(std::string(upb_f->name()) == proto2_f->name());
-    ASSERT(upb_f->type() == static_cast<upb::FieldDef::Type>(proto2_f->type()));
+    ASSERT(upb_f->descriptor_type() ==
+           static_cast<upb::FieldDef::DescriptorType>(proto2_f->type()));
     ASSERT(upb_f->IsSequence() == proto2_f->is_repeated());
   }
 }
 
-void parse_and_compare(MESSAGE_CIDENT *msg1, MESSAGE_CIDENT *msg2,
-                       const upb::Handlers *handlers,
+void parse_and_compare(google::protobuf::Message *msg1,
+                       google::protobuf::Message *msg2,
+                       const upb::Handlers *protomsg_handlers,
                        const char *str, size_t len, bool allow_jit) {
   // Parse to both proto2 and upb.
   ASSERT(msg1->ParseFromArray(str, len));
 
-  upb::DecoderPlan* plan = upb::DecoderPlan::New(handlers, allow_jit);
-  upb::StringSource src(str, len);
-  upb::Decoder decoder;
-  decoder.ResetPlan(plan);
-  decoder.ResetInput(src.AllBytes(), msg2);
+  const upb::Handlers* decoder_handlers = upb::pb::GetDecoderHandlers(
+      protomsg_handlers, allow_jit, &decoder_handlers);
+
+  upb::Pipeline pipeline(NULL, 0, upb_realloc, NULL);
+  pipeline.DonateRef(decoder_handlers, &decoder_handlers);
+  upb::Sink* protomsg_sink = pipeline.NewSink(protomsg_handlers);
+  upb::Sink* decoder_sink = pipeline.NewSink(decoder_handlers);
+
+  protomsg_sink->Reset(msg2);
+  upb::pb::Decoder* decoder =
+      decoder_sink->base()->GetUserdata<upb::pb::Decoder>();
+  upb::pb::ResetDecoderSink(decoder, protomsg_sink);
+
   msg2->Clear();
-  ASSERT(decoder.Decode() == UPB_OK);
-  plan->Unref();
+  bool ok = upb::PutStringToBytestream(decoder_sink, str, len);
+  ASSERT(ok);
+  ASSERT(pipeline.status().ok());
 
   // Would like to just compare the message objects themselves,  but
   // unfortunately MessageDifferencer is not part of the open-source release of
@@ -110,13 +123,29 @@ int run_tests(int argc, char *argv[])
   parse_and_compare(&msg1, &msg2, h, str, len, true);
   parse_and_compare(&msg1, &msg2, h, str, len, false);
   parse_and_compare(&msg1, &msg2, h, str, len, true);
-  printf("All tests passed, %d assertions.\n", num_assertions);
+  h->Unref(&h);
 
+  // Test with DynamicMessage.
+  google::protobuf::DynamicMessageFactory* factory =
+      new google::protobuf::DynamicMessageFactory;
+  const google::protobuf::Message* prototype =
+      factory->GetPrototype(msg1.descriptor());
+  google::protobuf::Message* dyn_msg1 = prototype->New();
+  google::protobuf::Message* dyn_msg2 = prototype->New();
+  h = upb::google::NewWriteHandlers(*dyn_msg1, &h);
+  parse_and_compare(dyn_msg1, dyn_msg2, h, str, len, false);
+  parse_and_compare(dyn_msg1, dyn_msg2, h, str, len, true);
+  delete dyn_msg1;
+  delete dyn_msg2;
+  delete factory;
   h->Unref(&h);
+
   free((void*)str);
 
   test_zig_zag();
 
+  printf("All tests passed, %d assertions.\n", num_assertions);
+
   google::protobuf::ShutdownProtobufLibrary();
   return 0;
 }

tools/dump_cinit.lua

@@ -45,7 +45,28 @@ function const(obj, name)
       return "UPB_" .. k
     end
   end
-  assert(false, "Couldn't find constant")
+  assert(false, "Couldn't find UPB_" .. string.upper(name) ..
+                " constant for value: " .. val)
+end
+
+function constlist(pattern)
+  local ret = {}
+  for k, v in pairs(upb) do
+    if string.find(k, "^UPB_" .. pattern) then
+      ret[k] = v
+    end
+  end
+  return ret
+end
+
+function boolstr(val)
+  if val == true then
+    return "true"
+  elseif val == false then
+    return "false"
+  else
+    assert(false, "Bad bool value: " .. tostring(val))
+  end
 end
 
 --[[
@@ -128,11 +149,11 @@ function Dumper:new(linktab)
   return obj
 end
 
--- Dumps a upb_value, eg:
+-- Dumps a _upb_value, eg:
 --   UPB_VALUE_INIT_INT32(5)
-function Dumper:value(val, upbtype)
+function Dumper:_value(val, upbtype)
   if type(val) == "nil" then
-    return "UPB_VALUE_INIT_NONE"
+    return "UPB__VALUE_INIT_NONE"
   elseif type(val) == "number" then
     -- Use upbtype to disambiguate what kind of number it is.
     if upbtype == upbtable.CTYPE_INT32 then
@@ -164,7 +185,7 @@ end
 -- Dumps a table entry.
 function Dumper:tabent(ent)
   local key = self:tabkey(ent.key)
-  local val = self:value(ent.value, ent.valtype)
+  local val = self:_value(ent.value, ent.valtype)
   local next = self.linktab:addr(ent.next)
   return string.format('  {%s, %s, %s},\n', key, val, next)
 end
@@ -173,7 +194,7 @@ end
 -- except that nil values have a special value to indicate "empty".
 function Dumper:arrayval(val)
   if val.val then
-    return string.format("  %s,\n", self:value(val.val, val.valtype))
+    return string.format("  %s,\n", self:_value(val.val, val.valtype))
   else
     return "  UPB_ARRAY_EMPTYENT,\n"
   end
@@ -283,7 +304,7 @@ local function dump_defs_c(symtab, basename, append)
   append("const upb_enumdef %s;\n", linktab:cdecl(upb.DEF_ENUM))
   append("const upb_tabent %s;\n", linktab:cdecl("strentries"))
   append("const upb_tabent %s;\n", linktab:cdecl("intentries"))
-  append("const upb_value %s;\n", linktab:cdecl("arrays"))
+  append("const _upb_value %s;\n", linktab:cdecl("arrays"))
   append("\n")
 
   -- Emit defs.
@@ -307,13 +328,23 @@ local function dump_defs_c(symtab, basename, append)
     if f:has_subdef() then
       subdef = string.format("upb_upcast(%s)", linktab:addr(f:subdef()))
     end
-    -- UPB_FIELDDEF_INIT(label, type, name, num, msgdef, subdef,
-    --                   selector_base, default_value)
-    append('  UPB_FIELDDEF_INIT(%s, %s, "%s", %d, %s, %s, %d, %s),\n',
-           const(f, "label"), const(f, "type"), f:name(),
+    local intfmt
+    if f:type() == upb.TYPE_UINT32 or
+       f:type() == upb.TYPE_INT32 or
+       f:type() == upb.TYPE_UINT64 or
+       f:type() == upb.TYPE_INT64 then
+      intfmt = const(f, "intfmt")
+    else
+      intfmt = "0"
+    end
+    -- UPB_FIELDDEF_INIT(label, type, intfmt, tagdelim, name, num, msgdef,
+    --                   subdef, selector_base, default_value)
+    append('  UPB_FIELDDEF_INIT(%s, %s, %s, %s, "%s", %d, %s, %s, %d, ' ..
+           'UPB_VALUE_INIT_NONE),\n',  -- TODO: support default value
+           const(f, "label"), const(f, "type"), intfmt,
+           boolstr(f:istagdelim()), f:name(),
            f:number(), linktab:addr(f:msgdef()), subdef,
-           f:_selector_base(),
-           dumper:value(nil) -- TODO
+           f:_selector_base()
            )
   end
   append("};\n\n")
@@ -343,7 +374,7 @@ local function dump_defs_c(symtab, basename, append)
   end
   append("};\n\n");
 
-  append("const upb_value %s = {\n", linktab:cdecl("arrays"))
+  append("const _upb_value %s = {\n", linktab:cdecl("arrays"))
   for ent in linktab:objs("arrays") do
     append(dumper:arrayval(ent))
   end
@@ -400,6 +431,21 @@ local function dump_defs_h(symtab, basename, append, linktab)
   end
   append("\n")
 
+  append("// Selector definitions.\n")
+  local selector_types = constlist("HANDLER_")
+  for f in linktab:objs(upb.DEF_FIELD) do
+    for sel_type_name, sel_type_value in pairs(selector_types) do
+      sel_type_name = sel_type_name:gsub("UPB_HANDLER_", "")
+      local sel = f:getsel(sel_type_value)
+      if sel then
+        local symname = f:msgdef():full_name() .. "." .. f:name() ..
+                        "." .. sel_type_name
+        append("#define %s %d\n", to_preproc(symname), sel)
+      end
+    end
+  end
+  append("\n")
+
   append('#ifdef __cplusplus\n')
   append('};  // extern "C"\n')
   append('#endif\n\n')

upb/bytestream.c

@@ -1,176 +0,0 @@
-/*
- * upb - a minimalist implementation of protocol buffers.
- *
- * Copyright (c) 2010 Google Inc.  See LICENSE for details.
- * Author: Josh Haberman <jhaberman@gmail.com>
- */
-
-#include "upb/bytestream.h"
-
-#include <stdlib.h>
-#include <string.h>
-
-
-/* upb_byteregion *************************************************************/
-
-char *upb_byteregion_strdup(const upb_byteregion *r) {
-  char *ret = malloc(upb_byteregion_len(r) + 1);
-  upb_byteregion_copyall(r, ret);
-  ret[upb_byteregion_len(r)] = '\0';
-  return ret;
-}
-
-upb_byteregion *upb_byteregion_new(const void *str) {
-  return upb_byteregion_newl(str, strlen(str));
-}
-
-upb_byteregion *upb_byteregion_newl(const void *str, size_t len) {
-  upb_stringsrc *src = malloc(sizeof(*src));
-  upb_stringsrc_init(src);
-  char *ptr = malloc(len + 1);
-  memcpy(ptr, str, len);
-  ptr[len] = '\0';
-  upb_stringsrc_reset(src, ptr, len);
-  return upb_stringsrc_allbytes(src);
-}
-
-void upb_byteregion_free(upb_byteregion *r) {
-  if (!r) return;
-  size_t len;
-  free((char*)upb_byteregion_getptr(r, 0, &len));
-  upb_stringsrc_uninit((upb_stringsrc*)r->bytesrc);
-  free(r->bytesrc);
-}
-
-void upb_bytesink_init(upb_bytesink *sink, upb_bytesink_vtbl *vtbl) {
-  sink->vtbl = vtbl;
-  upb_status_init(&sink->status);
-}
-
-void upb_bytesink_uninit(upb_bytesink *sink) {
-  upb_status_uninit(&sink->status);
-}
-
-void upb_byteregion_reset(upb_byteregion *r, const upb_byteregion *src,
-                          uint64_t ofs, uint64_t len) {
-  assert(ofs >= upb_byteregion_startofs(src));
-  assert(len <= upb_byteregion_remaining(src, ofs));
-  r->bytesrc = src->bytesrc;
-  r->toplevel = false;
-  r->start = ofs;
-  r->discard = ofs;
-  r->end = ofs + len;
-  r->fetch = UPB_MIN(src->fetch, r->end);
-}
-
-upb_bytesuccess_t upb_byteregion_fetch(upb_byteregion *r) {
-  uint64_t fetchable = upb_byteregion_remaining(r, r->fetch);
-  if (fetchable == 0) return UPB_BYTE_EOF;
-  size_t fetched;
-  upb_bytesuccess_t ret = upb_bytesrc_fetch(r->bytesrc, r->fetch, &fetched);
-  if (ret != UPB_BYTE_OK) return false;
-  r->fetch += UPB_MIN(fetched, fetchable);
-  return UPB_BYTE_OK;
-}
-
-
-/* upb_stringsrc **************************************************************/
-
-upb_bytesuccess_t upb_stringsrc_fetch(void *_src, uint64_t ofs, size_t *read) {
-  upb_stringsrc *src = _src;
-  assert(ofs < src->len);
-  if (ofs == src->len) {
-    upb_status_seteof(&src->bytesrc.status);
-    return UPB_BYTE_EOF;
-  }
-  *read = src->len - ofs;
-  return UPB_BYTE_OK;
-}
-
-void upb_stringsrc_copy(const void *_src, uint64_t ofs,
-                        size_t len, char *dst) {
-  const upb_stringsrc *src = _src;
-  assert(ofs + len <= src->len);
-  memcpy(dst, src->str + ofs, len);
-}
-
-void upb_stringsrc_discard(void *src, uint64_t ofs) {
-  (void)src;
-  (void)ofs;
-}
-
-const char *upb_stringsrc_getptr(const void *_s, uint64_t ofs, size_t *len) {
-  const upb_stringsrc *src = _s;
-  *len = src->len - ofs;
-  return src->str + ofs;
-}
-
-void upb_stringsrc_init(upb_stringsrc *s) {
-  static upb_bytesrc_vtbl vtbl = {
-    &upb_stringsrc_fetch,
-    &upb_stringsrc_discard,
-    &upb_stringsrc_copy,
-    &upb_stringsrc_getptr,
-  };
-  upb_bytesrc_init(&s->bytesrc, &vtbl);
-  s->str = NULL;
-  s->byteregion.bytesrc = &s->bytesrc;
-  s->byteregion.toplevel = true;
-}
-
-void upb_stringsrc_reset(upb_stringsrc *s, const char *str, size_t len) {
-  s->str = str;
-  s->len = len;
-  s->byteregion.start = 0;
-  s->byteregion.discard = 0;
-  s->byteregion.fetch = 0;
-  s->byteregion.end = len;
-}
-
-void upb_stringsrc_uninit(upb_stringsrc *s) { (void)s; }
-
-/* upb_stringsink *************************************************************/
-
-void upb_stringsink_uninit(upb_stringsink *s) {
-  free(s->str);
-}
-
-void upb_stringsink_reset(upb_stringsink *s, char *str, size_t size) {
-  free(s->str);
-  s->str = str;
-  s->len = 0;
-  s->size = size;
-}
-
-upb_bytesink *upb_stringsink_bytesink(upb_stringsink *s) {
-  return &s->bytesink;
-}
-
-static int32_t upb_stringsink_vprintf(void *_s, const char *fmt, va_list args) {
-  // TODO: detect realloc() errors.
-  upb_stringsink *s = _s;
-  int ret = upb_vrprintf(&s->str, &s->size, s->len, fmt, args);
-  if (ret >= 0) s->len += ret;
-  return ret;
-}
-
-int upb_stringsink_write(void *_s, const void *buf, int len) {
-  // TODO: detect realloc() errors.
-  upb_stringsink *s = _s;
-  if (s->len + len > s->size) {
-    while(s->len + len > s->size) s->size *= 2;
-    s->str = realloc(s->str, s->size);
-  }
-  memcpy(s->str + s->len, buf, len);
-  s->len += len;
-  return len;
-}
-
-void upb_stringsink_init(upb_stringsink *s) {
-  static upb_bytesink_vtbl vtbl = {
-    upb_stringsink_write,
-    upb_stringsink_vprintf
-  };
-  upb_bytesink_init(&s->bytesink, &vtbl);
-  s->str = NULL;
-}

upb/bytestream.h

@@ -1,629 +1,50 @@
 /*
  * upb - a minimalist implementation of protocol buffers.
  *
- * Copyright (c) 2011 Google Inc.  See LICENSE for details.
+ * Copyright (c) 2013 Google Inc.  See LICENSE for details.
  * Author: Josh Haberman <jhaberman@gmail.com>
  *
- * This file defines three core interfaces:
- * - upb_bytesink: for writing streams of data.
- * - upb_bytesrc: for reading streams of data.
- * - upb_byteregion: for reading from a specific region of a upb_bytesrc;
- *   should be used by decoders instead of using upb_bytesrc directly.
+ * This file contains the standard ByteStream msgdef and some useful routines
+ * surrounding it.
  *
- * These interfaces are used by streaming encoders and decoders: for example, a
- * protobuf parser gets its input from a upb_byteregion.  They are virtual base
- * classes so concrete implementations can get the data from a fd, a FILE*, a
- * string, etc.
+ * This is a mixed C/C++ interface that offers a full API to both languages.
+ * See the top-level README for more information.
  */
 
-// A upb_byteregion represents a region of data from a bytesrc.
-//
-// Parsers get data from this interface instead of a bytesrc because we often
-// want to parse only a specific region of the input.  For example, if we parse
-// a string from our input but know that the string represents a protobuf, we
-// can pass its upb_byteregion to an appropriate protobuf parser.
-//
-// Since the bytes may be coming from a file or network socket, bytes must be
-// fetched before they can be read (though in some cases this fetch may be a
-// no-op).  "fetch" is the only operation on a byteregion that could fail or
-// block, because it is the only operation that actually performs I/O.
-//
-// Bytes can be discarded when they are no longer needed.  Parsers should
-// always discard bytes they no longer need, both so the buffers can be freed
-// when possible and to give better visibility into what bytes the parser is
-// still using.
-//
-// start      discard                     read             fetch             end
-// ofs          ofs                       ofs               ofs              ofs
-// |             |--->discard()            |                 |--->fetch()      |
-// V             V                         V                 V                 V
-// +-------------+-------------------------+-----------------+-----------------+
-// |  discarded  |                         |                 |    fetchable    |
-// +-------------+-------------------------+-----------------+-----------------+
-//               | <------------- loaded ------------------> |
-//                                         | <- available -> |
-//                                         | <---------- remaining ----------> |
-//
-// Note that the start offset may be something other than zero!  A byteregion
-// is a view into an underlying bytesrc stream, and the region may start
-// somewhere other than the beginning of that stream.
-//
-// The region can be either delimited or nondelimited.  A non-delimited region
-// will keep returning data until the underlying data source returns EOF.  A
-// delimited region will return EOF at a predetermined offset.
-//
-//                       end
-//                       ofs
-//                         |
-//                         V
-// +-----------------------+
-// |  delimited region     |   <-- hard EOF, even if data source has more data.
-// +-----------------------+
-//
-// +------------------------
-// | nondelimited region   Z   <-- won't return EOF until data source hits EOF.
-// +------------------------
-//
-// TODO: if 64-bit math for stream offsets is a performance issue on
-// non-64-bit machines, we could introduce a upb_off_t typedef that can be
-// defined as a 32-bit type for applications that don't need to handle
-// streams longer than 4GB.
+#ifndef UPB_BYTESTREAM_H_
+#define UPB_BYTESTREAM_H_
 
+#include "upb/sink.h"
+#include "upb/bytestream.upb.h"
 
-#ifndef UPB_BYTESTREAM_H
-#define UPB_BYTESTREAM_H
+#define UPB_BYTESTREAM_BYTES &upb_bytestream_fields[0]
 
-#include "upb.h"
-
-#ifdef __cplusplus
-namespace upb {
-class ByteRegion;
-class StringSource;
-}  // namespace upb
-typedef upb::StringSource upb_stringsrc;
-extern "C" {
-#else
-struct upb_stringsrc;
-typedef struct upb_stringsrc upb_stringsrc;
-#endif
-
-typedef enum {
-  UPB_BYTE_OK = UPB_OK,
-  UPB_BYTE_WOULDBLOCK = UPB_SUSPENDED,
-  UPB_BYTE_ERROR = UPB_ERROR,
-  UPB_BYTE_EOF
-} upb_bytesuccess_t;
-
-/* upb_bytesrc ****************************************************************/
-
-// A upb_bytesrc allows the consumer of a stream of bytes to obtain buffers as
-// they become available, and to preserve some trailing amount of data before
-// it is discarded.  Consumers should not use upb_bytesrc directly, but rather
-// should use a upb_byteregion (which allows access to a region of a bytesrc).
-//
-// upb_bytesrc is a virtual base class with implementations that get data from
-// eg. a string, a cord, a file descriptor, a FILE*, etc.
-
-typedef upb_bytesuccess_t upb_bytesrc_fetch_func(void*, uint64_t, size_t*);
-typedef void upb_bytesrc_discard_func(void*, uint64_t);
-typedef void upb_bytesrc_copy_func(const void*, uint64_t, size_t, char*);
-typedef const char *upb_bytesrc_getptr_func(const void*, uint64_t, size_t*);
-typedef struct _upb_bytesrc_vtbl {
-  upb_bytesrc_fetch_func     *fetch;
-  upb_bytesrc_discard_func   *discard;
-  upb_bytesrc_copy_func      *copy;
-  upb_bytesrc_getptr_func    *getptr;
-} upb_bytesrc_vtbl;
-
-typedef struct {
-  const upb_bytesrc_vtbl  *vtbl;
-  upb_status status;
-} upb_bytesrc;
-
-INLINE void upb_bytesrc_init(upb_bytesrc *src, const upb_bytesrc_vtbl *vtbl) {
-  src->vtbl = vtbl;
-  upb_status_init(&src->status);
-}
-
-INLINE void upb_bytesrc_uninit(upb_bytesrc *src) {
-  upb_status_uninit(&src->status);
-}
-
-// Fetches at least one byte starting at ofs, returning the success or failure
-// of the operation.  If UPB_BYTE_OK is returned, *read indicates the number of
-// of bytes successfully fetched; any error or EOF status will be reflected in
-// upb_bytesrc_status().  It is valid for bytes to be fetched multiple times,
-// as long as the bytes have not been previously discarded.
-INLINE upb_bytesuccess_t upb_bytesrc_fetch(upb_bytesrc *src, uint64_t ofs,
-                                           size_t *read) {
-  return src->vtbl->fetch(src, ofs, read);
-}
-
-// Discards all data prior to ofs (except data that is pinned, if pinning
-// support is added -- see TODO below).
-INLINE void upb_bytesrc_discard(upb_bytesrc *src, uint64_t ofs) {
-  src->vtbl->discard(src, ofs);
-}
-
-// Copies "len" bytes of data from ofs to "dst", which must be at least "len"
-// bytes long.  The given region must not be discarded.
-INLINE void upb_bytesrc_copy(const upb_bytesrc *src, uint64_t ofs, size_t len,
-                             char *dst) {
-  src->vtbl->copy(src, ofs, len, dst);
-}
-
-// Returns a pointer to the bytesrc's internal buffer, storing in *len how much
-// data is available.  The given offset must not be discarded.  The returned
-// buffer is valid for as long as its bytes are not discarded (in the case that
-// part of the returned buffer is discarded, only the non-discarded bytes
-// remain valid).
-INLINE const char *upb_bytesrc_getptr(const upb_bytesrc *src, uint64_t ofs,
-                                      size_t *len) {
-  return src->vtbl->getptr(src, ofs, len);
-}
-
-// TODO: Add if/when there is a demonstrated need:
-//
-// // When the caller pins a region (which must not be already discarded), it
-// // is guaranteed that the region will not be discarded (nor will the bytesrc
-// // be destroyed) until the region is unpinned.  However, not all bytesrc's
-// // support pinning; a false return indicates that a pin was not possible.
-// INLINE bool upb_bytesrc_pin(upb_bytesrc *src, uint64_t ofs, size_t len) {
-//   return src->vtbl->refregion(src, ofs, len);
-// }
-//
-// // Releases some number of pinned bytes from the beginning of a pinned
-// // region (which may be fewer than the total number of bytes pinned).
-// INLINE void upb_bytesrc_unpin(upb_bytesrc *src, uint64_t ofs, size_t len,
-//                               size_t bytes_to_release) {
-//   src->vtbl->unpin(src, ofs, len);
-// }
-//
-// Adding pinning support would also involve adding a "pin_ofs" parameter to
-// upb_bytesrc_fetch, so that the fetch can extend an already-pinned region.
-
-
-/* upb_byteregion *************************************************************/
-
-#define UPB_NONDELIMITED (0xffffffffffffffffULL)
-
-#ifdef __cplusplus
-}  // extern "C"
-
-class upb::ByteRegion {
- public:
-  static const uint64_t kNondelimited = UPB_NONDELIMITED;
-  typedef upb_bytesuccess_t ByteSuccess;
-
-  // Accessors for the regions bounds -- the meaning of these is described in
-  // the diagram above.
-  uint64_t start_ofs() const;
-  uint64_t discard_ofs() const;
-  uint64_t fetch_ofs() const;
-  uint64_t end_ofs() const;
-
-  // Returns how many bytes are fetched and available for reading starting from
-  // offset "offset".
-  uint64_t BytesAvailable(uint64_t offset) const;
-
-  // Returns the total number of bytes remaining after offset "offset", or
-  // kNondelimited if the byteregion is non-delimited.
-  uint64_t BytesRemaining(uint64_t offset) const;
-
-  uint64_t Length() const;
-
-  // Sets the value of this byteregion to be a subset of the given byteregion's
-  // data.  The caller is responsible for releasing this region before the src
-  // region is released (unless the region is first pinned, if pinning support
-  // is added.  see below).
-  void Reset(const upb_byteregion *src, uint64_t ofs, uint64_t len);
-  void Release();
-
-  // Attempts to fetch more data, extending the fetched range of this
-  // byteregion.  Returns true if the fetched region was extended by at least
-  // one byte, false on EOF or error (see *s for details).
-  ByteSuccess Fetch();
-
-  // Fetches all remaining data, returning false if the operation failed (see
-  // *s for details).  May only be used on delimited byteregions.
-  ByteSuccess FetchAll();
-
-  // Discards bytes from the byteregion up until ofs (which must be greater or
-  // equal to discard_ofs()).  It is valid to discard bytes that have not been
-  // fetched (such bytes will never be fetched) but it is an error to discard
-  // past the end of a delimited byteregion.
-  void Discard(uint64_t ofs);
-
-  // Copies "len" bytes of data into "dst", starting at ofs.  The specified
-  // region must be available.
-  void Copy(uint64_t ofs, size_t len, char *dst) const;
-
-  // Copies all bytes from the byteregion into dst.  Requires that the entire
-  // byteregion is fetched and that none has been discarded.
-  void CopyAll(char *dst) const;
-
-  // Returns a pointer to the internal buffer for the byteregion starting at
-  // offset "ofs." Stores the number of bytes available in this buffer in *len.
-  // The returned buffer is invalidated when the byteregion is reset or
-  // released, or when the bytes are discarded.  If the byteregion is not
-  // currently pinned, the pointer is only valid for the lifetime of the parent
-  // byteregion.
-  const char *GetPtr(uint64_t ofs, size_t *len) const;
-
-  // Copies the contents of the byteregion into a newly-allocated,
-  // NULL-terminated string.  Requires that the byteregion is fully fetched.
-  char *StrDup() const;
-
-  template <typename T> void AssignToString(T* str);
-
-#else
-struct upb_byteregion {
-#endif
-  uint64_t start;
-  uint64_t discard;
-  uint64_t fetch;
-  uint64_t end;         // UPB_NONDELIMITED if nondelimited.
-  upb_bytesrc *bytesrc;
-  bool toplevel;        // If true, discards hit the underlying bytesrc.
-};
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-// Native C API.
-INLINE uint64_t upb_byteregion_startofs(const upb_byteregion *r) {
-  return r->start;
-}
-INLINE uint64_t upb_byteregion_discardofs(const upb_byteregion *r) {
-  return r->discard;
-}
-INLINE uint64_t upb_byteregion_fetchofs(const upb_byteregion *r) {
-  return r->fetch;
-}
-INLINE uint64_t upb_byteregion_endofs(const upb_byteregion *r) {
-  return r->end;
-}
-INLINE uint64_t upb_byteregion_available(const upb_byteregion *r, uint64_t o) {
-  assert(o >= upb_byteregion_discardofs(r));
-  assert(o <= r->fetch);  // Could relax this.
-  return r->fetch - o;
-}
-INLINE uint64_t upb_byteregion_remaining(const upb_byteregion *r, uint64_t o) {
-  return r->end == UPB_NONDELIMITED ? UPB_NONDELIMITED : r->end - o;
-}
-
-INLINE uint64_t upb_byteregion_len(const upb_byteregion *r) {
-  return upb_byteregion_remaining(r, r->start);
-}
-void upb_byteregion_reset(upb_byteregion *r, const upb_byteregion *src,
-                          uint64_t ofs, uint64_t len);
-void upb_byteregion_release(upb_byteregion *r);
-upb_bytesuccess_t upb_byteregion_fetch(upb_byteregion *r);
-INLINE upb_bytesuccess_t upb_byteregion_fetchall(upb_byteregion *r) {
-  assert(upb_byteregion_len(r) != UPB_NONDELIMITED);
-  upb_bytesuccess_t ret;
-  do {
-    ret = upb_byteregion_fetch(r);
-  } while (ret == UPB_BYTE_OK);
-  return ret == UPB_BYTE_EOF ? UPB_BYTE_OK : ret;
-}
-INLINE void upb_byteregion_discard(upb_byteregion *r, uint64_t ofs) {
-  assert(ofs >= upb_byteregion_discardofs(r));
-  assert(ofs <= upb_byteregion_endofs(r));
-  r->discard = ofs;
-  if (ofs > r->fetch) r->fetch = ofs;
-  if (r->toplevel) upb_bytesrc_discard(r->bytesrc, ofs);
-}
-INLINE void upb_byteregion_copy(const upb_byteregion *r, uint64_t ofs,
-                                size_t len, char *dst) {
-  assert(ofs >= upb_byteregion_discardofs(r));
-  assert(len <= upb_byteregion_available(r, ofs));
-  upb_bytesrc_copy(r->bytesrc, ofs, len, dst);
-}
-INLINE void upb_byteregion_copyall(const upb_byteregion *r, char *dst) {
-  assert(r->start == r->discard && r->end == r->fetch);
-  upb_byteregion_copy(r, r->start, upb_byteregion_len(r), dst);
-}
-INLINE const char *upb_byteregion_getptr(const upb_byteregion *r,
-                                         uint64_t ofs, size_t *len) {
-  assert(ofs >= upb_byteregion_discardofs(r));
-  const char *ret = upb_bytesrc_getptr(r->bytesrc, ofs, len);
-  *len = UPB_MIN(*len, upb_byteregion_available(r, ofs));
+// A convenience method that handles the start/end calls and tracks overall
+// success.
+UPB_INLINE bool upb_bytestream_putstr(upb_sink *s, const char *buf, size_t n) {
+  bool ret =
+      upb_sink_startmsg(s) &&
+      upb_sink_startstr(s, UPB_BYTESTREAM_BYTES_STARTSTR, n) &&
+      upb_sink_putstring(s, UPB_BYTESTREAM_BYTES_STRING, buf, n) == n &&
+      upb_sink_endstr(s, UPB_BYTESTREAM_BYTES_ENDSTR);
+  if (ret) upb_sink_endmsg(s);
   return ret;
 }
 
-// TODO: add if/when there is a demonstrated need.
-//
-// // Pins this byteregion's bytes in memory, allowing it to outlive its parent
-// // byteregion.  Normally a byteregion may only be used while its parent is
-// // still valid, but a pinned byteregion may continue to be used until it is
-// // reset or released.  A byteregion must be fully fetched to be pinned
-// // (this implies that the byteregion must be delimited).
-// //
-// // In some cases this operation may cause the input data to be copied.
-// //
-// // void upb_byteregion_pin(upb_byteregion *r);
-
-// Convenience functions for creating and destroying a byteregion with a simple
-// string as its data.  These are relatively inefficient compared with creating
-// your own bytesrc (they call malloc() and copy the string data) so should not
-// be used on any critical path.
-//
-// The string data in the returned region is guaranteed to be contiguous and
-// NULL-terminated.
-upb_byteregion *upb_byteregion_new(const void *str);
-upb_byteregion *upb_byteregion_newl(const void *str, size_t len);
-// May *only* be called on a byteregion created with upb_byteregion_new[l]()!
-void upb_byteregion_free(upb_byteregion *r);
-
-// Copies the contents of the byteregion into a newly-allocated, NULL-terminated
-// string.  Requires that the byteregion is fully fetched.
-char *upb_byteregion_strdup(const upb_byteregion *r);
-
-
-/* upb_bytesink ***************************************************************/
-
-// A bytesink is an interface that allows the caller to push byte-wise data.
-// It is very simple -- the only special capability is the ability to "rewind"
-// the stream, which is really only a mechanism of having the bytesink ignore
-// some subsequent calls.
-typedef int upb_bytesink_write_func(void*, const void*, int);
-typedef int upb_bytesink_vprintf_func(void*, const char *fmt, va_list args);
-
-typedef struct {
-  upb_bytesink_write_func   *write;
-  upb_bytesink_vprintf_func *vprintf;
-} upb_bytesink_vtbl;
-
-typedef struct {
-  upb_bytesink_vtbl *vtbl;
-  upb_status status;
-  uint64_t offset;
-} upb_bytesink;
-
-// Should be called by derived classes.
-void upb_bytesink_init(upb_bytesink *sink, upb_bytesink_vtbl *vtbl);
-void upb_bytesink_uninit(upb_bytesink *sink);
-
-INLINE int upb_bytesink_write(upb_bytesink *s, const void *buf, int len) {
-  return s->vtbl->write(s, buf, len);
-}
-
-#define upb_bytesink_writestr(s, buf) upb_bytesink_write(s, buf, strlen(buf))
-
-// Returns the number of bytes written or -1 on error.
-INLINE int upb_bytesink_printf(upb_bytesink *sink, const char *fmt, ...) {
-  va_list args;
-  va_start(args, fmt);
-  uint32_t ret = sink->vtbl->vprintf(sink, fmt, args);
-  va_end(args);
-  return ret;
-}
-
-INLINE int upb_bytesink_putc(upb_bytesink *sink, char ch) {
-  return upb_bytesink_write(sink, &ch, 1);
-}
-
-INLINE int upb_bytesink_putrepeated(upb_bytesink *sink, char ch, int len) {
-  for (int i = 0; i < len; i++)
-    if (upb_bytesink_write(sink, &ch, 1) < 0)
-      return -1;
-  return len;
-}
-
-INLINE uint64_t upb_bytesink_getoffset(upb_bytesink *sink) {
-  return sink->offset;
-}
-
-// Rewinds the stream to the given offset.  This cannot actually "unput" any
-// data, it is for situations like:
-//
-// // If false is returned (because of error), call again later to resume.
-// bool write_some_data(upb_bytesink *sink, int indent) {
-//   uint64_t start_offset = upb_bytesink_getoffset(sink);
-//   if (upb_bytesink_writestr(sink, "Some data") < 0) goto err;
-//   if (upb_bytesink_putrepeated(sink, ' ', indent) < 0) goto err;
-//   return true;
-//  err:
-//   upb_bytesink_rewind(sink, start_offset);
-//   return false;
-// }
-//
-// The subsequent bytesink writes *must* be identical to the writes that were
-// rewinded past.
-INLINE void upb_bytesink_rewind(upb_bytesink *sink, uint64_t offset) {
-  // TODO
-  (void)sink;
-  (void)offset;
-}
-
-// OPT: add getappendbuf()
-// OPT: add writefrombytesrc()
-// TODO: add flush()
-
-
-/* upb_stringsrc **************************************************************/
-
-// bytesrc/bytesink for a simple contiguous string.
-
 #ifdef __cplusplus
-}  // extern "C"
-
-class upb::StringSource {
- public:
-  StringSource();
-  template <typename T> explicit StringSource(const T& str);
-  StringSource(const char *data, size_t len);
-  ~StringSource();
-
-  // Resets the stringsrc to a state where it will vend the given string.  The
-  // string data must be valid until the stringsrc is reset again or destroyed.
-  void Reset(const char* data, size_t len);
-  template <typename T> void Reset(const T& str);
-
-  // Returns the top-level upb_byteregion* for this stringsrc.  Invalidated
-  // when the stringsrc is reset.
-  ByteRegion* AllBytes();
-
-  upb_bytesrc* ByteSource();
-
-#else
-struct upb_stringsrc {
-#endif
-  upb_bytesrc bytesrc;
-  const char *str;
-  size_t len;
-  upb_byteregion byteregion;
-};
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-// Native C API.
-void upb_stringsrc_init(upb_stringsrc *s);
-void upb_stringsrc_uninit(upb_stringsrc *s);
-void upb_stringsrc_reset(upb_stringsrc *s, const char *str, size_t len);
-INLINE upb_bytesrc *upb_stringsrc_bytesrc(upb_stringsrc *s) {
-  return &s->bytesrc;
-}
-INLINE upb_byteregion *upb_stringsrc_allbytes(upb_stringsrc *s) {
-  return &s->byteregion;
-}
-
-
-/* upb_stringsink *************************************************************/
-
-struct _upb_stringsink {
-  upb_bytesink bytesink;
-  char *str;
-  size_t len, size;
-};
-typedef struct _upb_stringsink upb_stringsink;
-
-// Create/free a stringsrc.
-void upb_stringsink_init(upb_stringsink *s);
-void upb_stringsink_uninit(upb_stringsink *s);
-
-// Resets the sink's string to "str", which the sink takes ownership of.
-// "str" may be NULL, which will make the sink allocate a new string.
-void upb_stringsink_reset(upb_stringsink *s, char *str, size_t len);
-
-// Releases ownership of the returned string (which is "len" bytes long) and
-// resets the internal string to be empty again (as if reset were called with
-// NULL).
-const char *upb_stringsink_release(upb_stringsink *s, size_t *len);
-
-// Returns the upb_bytesink* for this stringsrc.  Invalidated by reset above.
-upb_bytesink *upb_stringsink_bytesink(upb_stringsink *s);
-
-#ifdef __cplusplus
-}  // extern "C"
 
 namespace upb {
 
-inline uint64_t ByteRegion::start_ofs() const {
-  return upb_byteregion_startofs(this);
-}
-inline uint64_t ByteRegion::discard_ofs() const {
-  return upb_byteregion_discardofs(this);
-}
-inline uint64_t ByteRegion::fetch_ofs() const {
-  return upb_byteregion_fetchofs(this);
-}
-inline uint64_t ByteRegion::end_ofs() const {
-  return upb_byteregion_endofs(this);
-}
-inline uint64_t ByteRegion::BytesAvailable(uint64_t offset) const {
-  return upb_byteregion_available(this, offset);
-}
-inline uint64_t ByteRegion::BytesRemaining(uint64_t offset) const {
-  return upb_byteregion_remaining(this, offset);
-}
-inline uint64_t ByteRegion::Length() const {
-  return upb_byteregion_len(this);
-}
-inline void ByteRegion::Reset(
-    const upb_byteregion *src, uint64_t ofs, uint64_t len) {
-  upb_byteregion_reset(this, src, ofs, len);
-}
-inline void ByteRegion::Release() {
-  upb_byteregion_release(this);
-}
-inline ByteRegion::ByteSuccess ByteRegion::Fetch() {
-  return upb_byteregion_fetch(this);
-}
-inline ByteRegion::ByteSuccess ByteRegion::FetchAll() {
-  return upb_byteregion_fetchall(this);
-}
-inline void ByteRegion::Discard(uint64_t ofs) {
-  upb_byteregion_discard(this, ofs);
-}
-inline void ByteRegion::Copy(uint64_t ofs, size_t len, char *dst) const {
-  upb_byteregion_copy(this, ofs, len, dst);
-}
-inline void ByteRegion::CopyAll(char *dst) const {
-  upb_byteregion_copyall(this, dst);
-}
-inline const char *ByteRegion::GetPtr(uint64_t ofs, size_t *len) const {
-  return upb_byteregion_getptr(this, ofs, len);
-}
-inline char *ByteRegion::StrDup() const {
-  return upb_byteregion_strdup(this);
-}
-template <typename T> void ByteRegion::AssignToString(T* str) {
-  uint64_t ofs = start_ofs();
-  size_t len;
-  const char *ptr = GetPtr(ofs, &len);
-  // Emperically calling reserve() here is counterproductive and slows down
-  // benchmarks.  If the parsing is happening in a tight loop that is reusing
-  // the string object, there is probably enough data reserved already and
-  // the reserve() call is extra overhead.
-  str->assign(ptr, len);
-  ofs += len;
-  while (ofs < end_ofs()) {
-    ptr = GetPtr(ofs, &len);
-    str->append(ptr, len);
-    ofs += len;
-  }
-}
-
-template <> inline ByteRegion* GetValue<ByteRegion*>(Value v) {
-  return static_cast<ByteRegion*>(upb_value_getbyteregion(v));
-}
-
-template <> inline Value MakeValue<ByteRegion*>(ByteRegion* v) {
-  return upb_value_byteregion(v);
+inline bool PutStringToBytestream(Sink* s, const char* buf, size_t n) {
+  return upb_bytestream_putstr(s, buf, n);
 }
 
-inline StringSource::StringSource() { upb_stringsrc_init(this); }
-template <typename T> StringSource::StringSource(const T& str) {
-  upb_stringsrc_init(this);
-  Reset(str);
-}
-inline StringSource::StringSource(const char *data, size_t len) {
-  upb_stringsrc_init(this);
-  Reset(data, len);
-}
-inline StringSource::~StringSource() {
-  upb_stringsrc_uninit(this);
-}
-inline void StringSource::Reset(const char* data, size_t len) {
-  upb_stringsrc_reset(this, data, len);
-}
-template <typename T> void StringSource::Reset(const T& str) {
-  upb_stringsrc_reset(this, str.c_str(), str.size());
-}
-inline ByteRegion* StringSource::AllBytes() {
-  return upb_stringsrc_allbytes(this);
-}
-inline upb_bytesrc* StringSource::ByteSource() {
-  return upb_stringsrc_bytesrc(this);
+template <class T> bool PutStringToBytestream(Sink* s, T str) {
+  return upb_bytestream_putstr(s, str.c_str(), str.size());
 }
 
 }  // namespace upb
 
 #endif
 
-#endif
+#endif  // UPB_BYTESTREAM_H_

upb/bytestream.proto

@@ -0,0 +1,14 @@
+//
+// upb - a minimalist implementation of protocol buffers.
+//
+// Copyright (c) 2013 Google Inc.  See LICENSE for details.
+// Author: Josh Haberman <jhaberman@gmail.com>
+//
+// This file contains a proto definition for use by handlers that consume a
+// simple byte stream (like traditional UNIX pipes).
+
+package upb;
+
+message ByteStream {
+  optional bytes bytes = 1;
+}

upb/bytestream.upb.c

@@ -0,0 +1,40 @@
+// This file was generated by upbc (the upb compiler).
+// Do not edit -- your changes will be discarded when the file is
+// regenerated.
+
+#include "upb/def.h"
+
+const upb_msgdef upb_bytestream_msgs[1];
+const upb_fielddef upb_bytestream_fields[1];
+const upb_enumdef upb_bytestream_enums[0];
+const upb_tabent upb_bytestream_strentries[4];
+const upb_tabent upb_bytestream_intentries[0];
+const _upb_value upb_bytestream_arrays[3];
+
+const upb_msgdef upb_bytestream_msgs[1] = {
+  UPB_MSGDEF_INIT("upb.ByteStream", UPB_INTTABLE_INIT(0, 0, 9, 0, NULL, &upb_bytestream_arrays[0], 3, 1), UPB_STRTABLE_INIT(1, 3, 9, 2, &upb_bytestream_strentries[0]), 3),
+};
+
+const upb_fielddef upb_bytestream_fields[1] = {
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BYTES, 0, false, "bytes", 1, &upb_bytestream_msgs[0], NULL, 0, UPB_VALUE_INIT_NONE),
+};
+
+const upb_enumdef upb_bytestream_enums[0] = {
+};
+
+const upb_tabent upb_bytestream_strentries[4] = {
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_STR("bytes"), UPB_VALUE_INIT_CONSTPTR(&upb_bytestream_fields[0]), NULL},
+};
+
+const upb_tabent upb_bytestream_intentries[0] = {
+};
+
+const _upb_value upb_bytestream_arrays[3] = {
+  UPB_ARRAY_EMPTYENT,
+  UPB_VALUE_INIT_CONSTPTR(&upb_bytestream_fields[0]),
+  UPB_ARRAY_EMPTYENT,
+};
+

upb/bytestream.upb.h

@@ -0,0 +1,37 @@
+// This file was generated by upbc (the upb compiler).
+// Do not edit -- your changes will be discarded when the file is
+// regenerated.
+
+#ifndef UPB_BYTESTREAM_UPB_H_
+#define UPB_BYTESTREAM_UPB_H_
+
+#include "upb/def.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+// Enums
+
+// Do not refer to these forward declarations; use the constants
+// below.
+extern const upb_msgdef upb_bytestream_msgs[1];
+extern const upb_fielddef upb_bytestream_fields[1];
+extern const upb_enumdef upb_bytestream_enums[0];
+
+// Constants for references to defs.
+// We hide these behind macros to decouple users from the
+// details of how we have statically defined them (ie. whether
+// each def has its own symbol or lives in an array of defs).
+#define UPB_BYTESTREAM &upb_bytestream_msgs[0]
+
+// Selector definitions.
+#define UPB_BYTESTREAM_BYTES_ENDSTR 2
+#define UPB_BYTESTREAM_BYTES_STRING 0
+#define UPB_BYTESTREAM_BYTES_STARTSTR 1
+
+#ifdef __cplusplus
+};  // extern "C"
+#endif
+
+#endif  // UPB_BYTESTREAM_UPB_H_

upb/def.c

@@ -9,9 +9,25 @@
 
 #include <stdlib.h>
 #include <string.h>
-#include "upb/bytestream.h"
+#include "upb/descriptor/descriptor.upb.h"
 #include "upb/handlers.h"
 
+typedef struct {
+  size_t len;
+  char str[1];  // Null-terminated string data follows.
+} str_t;
+
+static str_t *newstr(const char *data, size_t len) {
+  str_t *ret = malloc(sizeof(*ret) + len);
+  if (!ret) return NULL;
+  ret->len = len;
+  memcpy(ret->str, data, len);
+  ret->str[len] = '\0';
+  return ret;
+}
+
+static void freestr(str_t *s) { free(s); }
+
 // isalpha() etc. from <ctype.h> are locale-dependent, which we don't want.
 static bool upb_isbetween(char c, char low, char high) {
   return c >= low && c <= high;
@@ -113,6 +129,10 @@ static bool upb_validate_field(upb_fielddef *f, upb_status *s) {
     upb_status_seterrliteral(s, "fielddef must have name and number set");
     return false;
   }
+  if (!f->type_is_set_) {
+    upb_status_seterrliteral(s, "fielddef type was not initialized");
+    return false;
+  }
   if (upb_fielddef_hassubdef(f)) {
     if (f->subdef_is_symbolic) {
       upb_status_seterrf(s,
@@ -281,7 +301,7 @@ bool upb_enumdef_addval(upb_enumdef *e, const char *name, int32_t num,
     upb_status_seterrliteral(status, "out of memory");
     return false;
   }
-  if (!upb_inttable_lookup(&e->iton, num) &&
+  if (!upb_inttable_lookup(&e->iton, num, NULL) &&
       !upb_inttable_insert(&e->iton, num, upb_value_cstr(upb_strdup(name)))) {
     upb_status_seterrliteral(status, "out of memory");
     upb_strtable_remove(&e->ntoi, name, NULL);
@@ -310,15 +330,18 @@ void upb_enum_next(upb_enum_iter *iter) { upb_strtable_next(iter); }
 bool upb_enum_done(upb_enum_iter *iter) { return upb_strtable_done(iter); }
 
 bool upb_enumdef_ntoi(const upb_enumdef *def, const char *name, int32_t *num) {
-  const upb_value *v = upb_strtable_lookup(&def->ntoi, name);
-  if (!v) return false;
-  if (num) *num = upb_value_getint32(*v);
+  upb_value v;
+  if (!upb_strtable_lookup(&def->ntoi, name, &v)) {
+    return false;
+  }
+  if (num) *num = upb_value_getint32(v);
   return true;
 }
 
 const char *upb_enumdef_iton(const upb_enumdef *def, int32_t num) {
-  const upb_value *v = upb_inttable_lookup32(&def->iton, num);
-  return v ? upb_value_getcstr(*v) : NULL;
+  upb_value v;
+  return upb_inttable_lookup32(&def->iton, num, &v) ?
+      upb_value_getcstr(v) : NULL;
 }
 
 const char *upb_enum_iter_name(upb_enum_iter *iter) {
@@ -332,37 +355,11 @@ int32_t upb_enum_iter_number(upb_enum_iter *iter) {
 
 /* upb_fielddef ***************************************************************/
 
-#define alignof(t) offsetof(struct { char c; t x; }, x)
-#define TYPE_INFO(ctype, inmemory_type) \
-    {alignof(ctype), sizeof(ctype), UPB_CTYPE_ ## inmemory_type}
-
-const upb_typeinfo upb_types[UPB_NUM_TYPES] = {
-  TYPE_INFO(void*,     PTR),        // (unused)
-  TYPE_INFO(double,    DOUBLE),     // DOUBLE
-  TYPE_INFO(float,     FLOAT),      // FLOAT
-  TYPE_INFO(int64_t,   INT64),      // INT64
-  TYPE_INFO(uint64_t,  UINT64),     // UINT64
-  TYPE_INFO(int32_t,   INT32),      // INT32
-  TYPE_INFO(uint64_t,  UINT64),     // FIXED64
-  TYPE_INFO(uint32_t,  UINT32),     // FIXED32
-  TYPE_INFO(bool,      BOOL),       // BOOL
-  TYPE_INFO(void*,     BYTEREGION), // STRING
-  TYPE_INFO(void*,     PTR),        // GROUP
-  TYPE_INFO(void*,     PTR),        // MESSAGE
-  TYPE_INFO(void*,     BYTEREGION), // BYTES
-  TYPE_INFO(uint32_t,  UINT32),     // UINT32
-  TYPE_INFO(int32_t,   INT32),      // ENUM
-  TYPE_INFO(int32_t,   INT32),      // SFIXED32
-  TYPE_INFO(int64_t,   INT64),      // SFIXED64
-  TYPE_INFO(int32_t,   INT32),      // SINT32
-  TYPE_INFO(int64_t,   INT64),      // SINT64
-};
-
 static void upb_fielddef_init_default(upb_fielddef *f);
 
 static void upb_fielddef_uninit_default(upb_fielddef *f) {
-  if (f->default_is_string)
-    upb_byteregion_free(upb_value_getbyteregion(f->defaultval));
+  if (f->type_is_set_ && f->default_is_string)
+    freestr(upb_value_getptr(f->defaultval));
 }
 
 static void visitfield(const upb_refcounted *r, upb_refcounted_visit *visit,
@@ -396,16 +393,21 @@ upb_fielddef *upb_fielddef_new(const void *owner) {
   f->msgdef = NULL;
   f->sub.def = NULL;
   f->subdef_is_symbolic = false;
-  f->subdef_is_owned = false;
-  f->label_ = UPB_LABEL(OPTIONAL);
-
-  // These are initialized to be invalid; the user must set them explicitly.
-  // Could relax this later if it's convenient and non-confusing to have a
-  // defaults for them.
-  f->type_ = UPB_TYPE_NONE;
+  f->label_ = UPB_LABEL_OPTIONAL;
+  f->type_ = UPB_TYPE_INT32;
   f->number_ = 0;
-
-  upb_fielddef_init_default(f);
+  f->type_is_set_ = false;
+  f->tagdelim = false;
+
+  // For the moment we default this to UPB_INTFMT_VARIABLE, since it will work
+  // with all integer types and is in some since more "default" since the most
+  // normal-looking proto2 types int32/int64/uint32/uint64 use variable.
+  //
+  // Other options to consider:
+  // - there is no default; users must set this manually (like type).
+  // - default signed integers to UPB_INTFMT_ZIGZAG, since it's more likely to
+  //   be an optimal default for signed integers.
+  f->intfmt = UPB_INTFMT_VARIABLE;
   return f;
 }
 
@@ -417,11 +419,8 @@ upb_fielddef *upb_fielddef_dup(const upb_fielddef *f, const void *owner) {
   upb_fielddef_setnumber(newf, upb_fielddef_number(f));
   upb_fielddef_setname(newf, upb_fielddef_name(f));
   if (f->default_is_string) {
-    upb_byteregion *r = upb_value_getbyteregion(upb_fielddef_default(f));
-    size_t len;
-    const char *ptr = upb_byteregion_getptr(r, 0, &len);
-    assert(len == upb_byteregion_len(r));
-    upb_fielddef_setdefaultstr(newf, ptr, len);
+    str_t *s = upb_value_getptr(upb_fielddef_default(f));
+    upb_fielddef_setdefaultstr(newf, s->str, s->len);
   } else {
     upb_fielddef_setdefault(newf, upb_fielddef_default(f));
   }
@@ -468,7 +467,12 @@ void upb_fielddef_checkref(const upb_fielddef *f, const void *owner) {
   upb_def_checkref(upb_upcast(f), owner);
 }
 
+bool upb_fielddef_typeisset(const upb_fielddef *f) {
+  return f->type_is_set_;
+}
+
 upb_fieldtype_t upb_fielddef_type(const upb_fielddef *f) {
+  assert(f->type_is_set_);
   return f->type_;
 }
 
@@ -476,7 +480,17 @@ upb_label_t upb_fielddef_label(const upb_fielddef *f) {
   return f->label_;
 }
 
-uint32_t upb_fielddef_number(const upb_fielddef *f) { return f->number_; }
+upb_intfmt_t upb_fielddef_intfmt(const upb_fielddef *f) {
+  return f->intfmt;
+}
+
+bool upb_fielddef_istagdelim(const upb_fielddef *f) {
+  return f->tagdelim;
+}
+
+uint32_t upb_fielddef_number(const upb_fielddef *f) {
+  return f->number_;
+}
 
 const char *upb_fielddef_name(const upb_fielddef *f) {
   return upb_def_fullname(upb_upcast(f));
@@ -495,34 +509,37 @@ bool upb_fielddef_setname(upb_fielddef *f, const char *name) {
 }
 
 upb_value upb_fielddef_default(const upb_fielddef *f) {
+  assert(f->type_is_set_);
   return f->defaultval;
 }
 
+const char *upb_fielddef_defaultstr(const upb_fielddef *f, size_t *len) {
+  assert(f->type_is_set_);
+  if (f->default_is_string) {
+    str_t *str = upb_value_getptr(f->defaultval);
+    if (len) *len = str->len;
+    return str->str;
+  }
+  return NULL;
+}
+
 static void upb_fielddef_init_default(upb_fielddef *f) {
   f->default_is_string = false;
   switch (upb_fielddef_type(f)) {
-    case UPB_TYPE(DOUBLE): upb_value_setdouble(&f->defaultval, 0); break;
-    case UPB_TYPE(FLOAT): upb_value_setfloat(&f->defaultval, 0); break;
-    case UPB_TYPE(UINT64):
-    case UPB_TYPE(FIXED64): upb_value_setuint64(&f->defaultval, 0); break;
-    case UPB_TYPE(INT64):
-    case UPB_TYPE(SFIXED64):
-    case UPB_TYPE(SINT64): upb_value_setint64(&f->defaultval, 0); break;
-    case UPB_TYPE(ENUM):
-    case UPB_TYPE(INT32):
-    case UPB_TYPE(SINT32):
-    case UPB_TYPE(SFIXED32): upb_value_setint32(&f->defaultval, 0); break;
-    case UPB_TYPE(UINT32):
-    case UPB_TYPE(FIXED32): upb_value_setuint32(&f->defaultval, 0); break;
-    case UPB_TYPE(BOOL): upb_value_setbool(&f->defaultval, false); break;
-    case UPB_TYPE(STRING):
-    case UPB_TYPE(BYTES):
-        upb_value_setbyteregion(&f->defaultval, upb_byteregion_new(""));
-        f->default_is_string = true;
-        break;
-    case UPB_TYPE(GROUP):
-    case UPB_TYPE(MESSAGE): upb_value_setptr(&f->defaultval, NULL); break;
-    case UPB_TYPE_NONE: break;
+    case UPB_TYPE_DOUBLE: upb_value_setdouble(&f->defaultval, 0); break;
+    case UPB_TYPE_FLOAT: upb_value_setfloat(&f->defaultval, 0); break;
+    case UPB_TYPE_UINT64: upb_value_setuint64(&f->defaultval, 0); break;
+    case UPB_TYPE_INT64: upb_value_setint64(&f->defaultval, 0); break;
+    case UPB_TYPE_ENUM:
+    case UPB_TYPE_INT32: upb_value_setint32(&f->defaultval, 0); break;
+    case UPB_TYPE_UINT32: upb_value_setuint32(&f->defaultval, 0); break;
+    case UPB_TYPE_BOOL: upb_value_setbool(&f->defaultval, false); break;
+    case UPB_TYPE_STRING:
+    case UPB_TYPE_BYTES:
+      upb_value_setptr(&f->defaultval, newstr("", 0));
+      f->default_is_string = true;
+      break;
+    case UPB_TYPE_MESSAGE: upb_value_setptr(&f->defaultval, NULL); break;
   }
 }
 
@@ -554,89 +571,199 @@ bool upb_fielddef_settype(upb_fielddef *f, upb_fieldtype_t type) {
   assert(!upb_fielddef_isfrozen(f));
   upb_fielddef_uninit_default(f);
   f->type_ = type;
+  f->type_is_set_ = true;
   upb_fielddef_init_default(f);
   return true;
 }
 
+bool upb_fielddef_setdescriptortype(upb_fielddef *f, int type) {
+  assert(!upb_fielddef_isfrozen(f));
+  switch (type) {
+    case UPB_DESCRIPTOR_TYPE_DOUBLE:
+      upb_fielddef_settype(f, UPB_TYPE_DOUBLE);
+      break;
+    case UPB_DESCRIPTOR_TYPE_FLOAT:
+      upb_fielddef_settype(f, UPB_TYPE_FLOAT);
+      break;
+    case UPB_DESCRIPTOR_TYPE_INT64:
+    case UPB_DESCRIPTOR_TYPE_SFIXED64:
+    case UPB_DESCRIPTOR_TYPE_SINT64:
+      upb_fielddef_settype(f, UPB_TYPE_INT64);
+      break;
+    case UPB_DESCRIPTOR_TYPE_UINT64:
+    case UPB_DESCRIPTOR_TYPE_FIXED64:
+      upb_fielddef_settype(f, UPB_TYPE_UINT64);
+      break;
+    case UPB_DESCRIPTOR_TYPE_INT32:
+    case UPB_DESCRIPTOR_TYPE_SFIXED32:
+    case UPB_DESCRIPTOR_TYPE_SINT32:
+      upb_fielddef_settype(f, UPB_TYPE_INT32);
+      break;
+    case UPB_DESCRIPTOR_TYPE_UINT32:
+    case UPB_DESCRIPTOR_TYPE_FIXED32:
+      upb_fielddef_settype(f, UPB_TYPE_UINT32);
+      break;
+    case UPB_DESCRIPTOR_TYPE_BOOL:
+      upb_fielddef_settype(f, UPB_TYPE_BOOL);
+      break;
+    case UPB_DESCRIPTOR_TYPE_STRING:
+      upb_fielddef_settype(f, UPB_TYPE_STRING);
+      break;
+    case UPB_DESCRIPTOR_TYPE_BYTES:
+      upb_fielddef_settype(f, UPB_TYPE_BYTES);
+      break;
+    case UPB_DESCRIPTOR_TYPE_GROUP:
+    case UPB_DESCRIPTOR_TYPE_MESSAGE:
+      upb_fielddef_settype(f, UPB_TYPE_MESSAGE);
+      break;
+    case UPB_DESCRIPTOR_TYPE_ENUM:
+      upb_fielddef_settype(f, UPB_TYPE_ENUM);
+      break;
+    default:
+      return false;
+  }
+
+  if (type == UPB_DESCRIPTOR_TYPE_FIXED64 ||
+      type == UPB_DESCRIPTOR_TYPE_FIXED32 ||
+      type == UPB_DESCRIPTOR_TYPE_SFIXED64 ||
+      type == UPB_DESCRIPTOR_TYPE_SFIXED32) {
+    upb_fielddef_setintfmt(f, UPB_INTFMT_FIXED);
+  } else if (type == UPB_DESCRIPTOR_TYPE_SINT64 ||
+             type == UPB_DESCRIPTOR_TYPE_SINT32) {
+    upb_fielddef_setintfmt(f, UPB_INTFMT_ZIGZAG);
+  } else {
+    upb_fielddef_setintfmt(f, UPB_INTFMT_VARIABLE);
+  }
+
+  upb_fielddef_settagdelim(f, type == UPB_DESCRIPTOR_TYPE_GROUP);
+
+  return true;
+}
+
+upb_descriptortype_t upb_fielddef_descriptortype(const upb_fielddef *f) {
+  switch (upb_fielddef_type(f)) {
+    case UPB_TYPE_FLOAT:  return UPB_DESCRIPTOR_TYPE_FLOAT;
+    case UPB_TYPE_DOUBLE: return UPB_DESCRIPTOR_TYPE_DOUBLE;
+    case UPB_TYPE_BOOL:   return UPB_DESCRIPTOR_TYPE_BOOL;
+    case UPB_TYPE_STRING: return UPB_DESCRIPTOR_TYPE_STRING;
+    case UPB_TYPE_BYTES:  return UPB_DESCRIPTOR_TYPE_BYTES;
+    case UPB_TYPE_ENUM:   return UPB_DESCRIPTOR_TYPE_ENUM;
+    case UPB_TYPE_INT32:
+      switch (upb_fielddef_intfmt(f)) {
+        case UPB_INTFMT_VARIABLE: return UPB_DESCRIPTOR_TYPE_INT32;
+        case UPB_INTFMT_FIXED:    return UPB_DESCRIPTOR_TYPE_SFIXED32;
+        case UPB_INTFMT_ZIGZAG:   return UPB_DESCRIPTOR_TYPE_SINT32;
+      }
+    case UPB_TYPE_INT64:
+      switch (upb_fielddef_intfmt(f)) {
+        case UPB_INTFMT_VARIABLE: return UPB_DESCRIPTOR_TYPE_INT64;
+        case UPB_INTFMT_FIXED:    return UPB_DESCRIPTOR_TYPE_SFIXED64;
+        case UPB_INTFMT_ZIGZAG:   return UPB_DESCRIPTOR_TYPE_SINT64;
+      }
+    case UPB_TYPE_UINT32:
+      switch (upb_fielddef_intfmt(f)) {
+        case UPB_INTFMT_VARIABLE: return UPB_DESCRIPTOR_TYPE_UINT32;
+        case UPB_INTFMT_FIXED:    return UPB_DESCRIPTOR_TYPE_FIXED32;
+        case UPB_INTFMT_ZIGZAG:   return -1;
+      }
+    case UPB_TYPE_UINT64:
+      switch (upb_fielddef_intfmt(f)) {
+        case UPB_INTFMT_VARIABLE: return UPB_DESCRIPTOR_TYPE_UINT64;
+        case UPB_INTFMT_FIXED:    return UPB_DESCRIPTOR_TYPE_FIXED64;
+        case UPB_INTFMT_ZIGZAG:   return -1;
+      }
+    case UPB_TYPE_MESSAGE:
+      return upb_fielddef_istagdelim(f) ?
+          UPB_DESCRIPTOR_TYPE_GROUP : UPB_DESCRIPTOR_TYPE_MESSAGE;
+  }
+}
+
 bool upb_fielddef_setlabel(upb_fielddef *f, upb_label_t label) {
   assert(!upb_fielddef_isfrozen(f));
   f->label_ = label;
   return true;
 }
 
+bool upb_fielddef_setintfmt(upb_fielddef *f, upb_intfmt_t fmt) {
+  assert(!upb_fielddef_isfrozen(f));
+  f->intfmt = fmt;
+  return true;
+}
+
+bool upb_fielddef_settagdelim(upb_fielddef *f, bool tag_delim) {
+  assert(!upb_fielddef_isfrozen(f));
+  f->tagdelim = tag_delim;
+  return true;
+}
+
 void upb_fielddef_setdefault(upb_fielddef *f, upb_value value) {
+  assert(f->type_is_set_);
   assert(!upb_fielddef_isfrozen(f));
   assert(!upb_fielddef_isstring(f) && !upb_fielddef_issubmsg(f));
   if (f->default_is_string) {
-    upb_byteregion *bytes = upb_value_getbyteregion(f->defaultval);
-    assert(bytes);
-    upb_byteregion_free(bytes);
+    str_t *s = upb_value_getptr(f->defaultval);
+    assert(s);
+    freestr(s);
   }
   f->defaultval = value;
   f->default_is_string = false;
 }
 
 bool upb_fielddef_setdefaultstr(upb_fielddef *f, const void *str, size_t len) {
-  assert(upb_fielddef_isstring(f) || f->type_ == UPB_TYPE(ENUM));
-  if (f->type_ == UPB_TYPE(ENUM) && !upb_isident(str, len, false)) return false;
+  assert(upb_fielddef_isstring(f) || f->type_ == UPB_TYPE_ENUM);
+  if (f->type_ == UPB_TYPE_ENUM && !upb_isident(str, len, false))
+    return false;
 
   if (f->default_is_string) {
-    upb_byteregion *bytes = upb_value_getbyteregion(f->defaultval);
-    assert(bytes);
-    upb_byteregion_free(bytes);
+    str_t *s = upb_value_getptr(f->defaultval);
+    assert(s);
+    freestr(s);
   } else {
-    assert(f->type_ == UPB_TYPE(ENUM));
+    assert(f->type_ == UPB_TYPE_ENUM);
   }
 
-  upb_byteregion *r = upb_byteregion_newl(str, len);
-  upb_value_setbyteregion(&f->defaultval, r);
-  upb_bytesuccess_t ret = upb_byteregion_fetch(r);
-  UPB_ASSERT_VAR(ret, ret == (len == 0 ? UPB_BYTE_EOF : UPB_BYTE_OK));
-  assert(upb_byteregion_available(r, 0) == upb_byteregion_len(r));
+  str_t *s = newstr(str, len);
+  upb_value_setptr(&f->defaultval, s);
   f->default_is_string = true;
   return true;
 }
 
 void upb_fielddef_setdefaultcstr(upb_fielddef *f, const char *str) {
+  assert(f->type_is_set_);
   upb_fielddef_setdefaultstr(f, str, str ? strlen(str) : 0);
 }
 
 bool upb_fielddef_default_is_symbolic(const upb_fielddef *f) {
-  return f->default_is_string && f->type_ == UPB_TYPE_ENUM;
+  return f->type_is_set_ &&
+      f->default_is_string &&
+      f->type_ == UPB_TYPE_ENUM;
 }
 
 bool upb_fielddef_resolvedefault(upb_fielddef *f) {
   if (!upb_fielddef_default_is_symbolic(f)) return true;
 
-  upb_byteregion *bytes = upb_value_getbyteregion(f->defaultval);
+  str_t *s = upb_value_getptr(f->defaultval);
   const upb_enumdef *e = upb_downcast_enumdef(upb_fielddef_subdef(f));
-  assert(bytes);  // Points to either a real default or the empty string.
+  assert(s);  // Points to either a real default or the empty string.
   assert(e);
-  if (upb_byteregion_len(bytes) == 0) {
+  if (s->len == 0) {
     // The "default default" for an enum is the first defined value.
     upb_value_setint32(&f->defaultval, e->defaultval);
   } else {
-    size_t len;
     int32_t val = 0;
-    // ptr is guaranteed to be NULL-terminated because the byteregion was
-    // created with upb_byteregion_newl().
-    const char *ptr = upb_byteregion_getptr(
-        bytes, upb_byteregion_startofs(bytes), &len);
-    assert(len == upb_byteregion_len(bytes));  // Should all be in one chunk
-    if (!upb_enumdef_ntoi(e, ptr, &val)) {
+    if (!upb_enumdef_ntoi(e, s->str, &val))
       return false;
-    }
     upb_value_setint32(&f->defaultval, val);
   }
   f->default_is_string = false;
-  upb_byteregion_free(bytes);
+  freestr(s);
   return true;
 }
 
 static bool upb_subdef_typecheck(upb_fielddef *f, const upb_def *subdef) {
-  if (f->type_ == UPB_TYPE(MESSAGE) || f->type_ == UPB_TYPE(GROUP))
+  if (f->type_ == UPB_TYPE_MESSAGE)
     return upb_dyncast_msgdef(subdef) != NULL;
-  else if (f->type_ == UPB_TYPE(ENUM))
+  else if (f->type_ == UPB_TYPE_ENUM)
     return upb_dyncast_enumdef(subdef) != NULL;
   else {
     assert(false);
@@ -673,8 +800,7 @@ bool upb_fielddef_setsubdefname(upb_fielddef *f, const char *name) {
 }
 
 bool upb_fielddef_issubmsg(const upb_fielddef *f) {
-  return upb_fielddef_type(f) == UPB_TYPE_GROUP ||
-         upb_fielddef_type(f) == UPB_TYPE_MESSAGE;
+  return upb_fielddef_type(f) == UPB_TYPE_MESSAGE;
 }
 
 bool upb_fielddef_isstring(const upb_fielddef *f) {
@@ -691,7 +817,7 @@ bool upb_fielddef_isprimitive(const upb_fielddef *f) {
 }
 
 bool upb_fielddef_hassubdef(const upb_fielddef *f) {
-  return upb_fielddef_issubmsg(f) || upb_fielddef_type(f) == UPB_TYPE(ENUM);
+  return upb_fielddef_issubmsg(f) || upb_fielddef_type(f) == UPB_TYPE_ENUM;
 }
 
 
@@ -808,13 +934,15 @@ bool upb_msgdef_addfield(upb_msgdef *m, upb_fielddef *f,
 }
 
 const upb_fielddef *upb_msgdef_itof(const upb_msgdef *m, uint32_t i) {
-  const upb_value *val = upb_inttable_lookup32(&m->itof, i);
-  return val ? (const upb_fielddef*)upb_value_getptr(*val) : NULL;
+  upb_value val;
+  return upb_inttable_lookup32(&m->itof, i, &val) ?
+      upb_value_getptr(val) : NULL;
 }
 
 const upb_fielddef *upb_msgdef_ntof(const upb_msgdef *m, const char *name) {
-  const upb_value *val = upb_strtable_lookup(&m->ntof, name);
-  return val ? (upb_fielddef*)upb_value_getptr(*val) : NULL;
+  upb_value val;
+  return upb_strtable_lookup(&m->ntof, name, &val) ?
+      upb_value_getptr(val) : NULL;
 }
 
 upb_fielddef *upb_msgdef_itof_mutable(upb_msgdef *m, uint32_t i) {

upb/def.h

@@ -153,50 +153,59 @@ bool upb_def_freeze(upb_def *const*defs, int n, upb_status *status);
 
 /* upb::FieldDef **************************************************************/
 
-// We choose these to match descriptor.proto.  Clients may use UPB_TYPE() and
-// UPB_LABEL() instead of referencing these directly.
+// The types a field can have.  Note that this list is not identical to the
+// types defined in descriptor.proto, which gives INT32 and SINT32 separate
+// types (we distinguish the two with the "integer encoding" enum below).
 typedef enum {
-  UPB_TYPE_NONE     = -1,  // Internal-only, may be removed.
-  UPB_TYPE_DOUBLE   = 1,
-  UPB_TYPE_FLOAT    = 2,
-  UPB_TYPE_INT64    = 3,
-  UPB_TYPE_UINT64   = 4,
-  UPB_TYPE_INT32    = 5,
-  UPB_TYPE_FIXED64  = 6,
-  UPB_TYPE_FIXED32  = 7,
-  UPB_TYPE_BOOL     = 8,
-  UPB_TYPE_STRING   = 9,
-  UPB_TYPE_GROUP    = 10,
-  UPB_TYPE_MESSAGE  = 11,
-  UPB_TYPE_BYTES    = 12,
-  UPB_TYPE_UINT32   = 13,
-  UPB_TYPE_ENUM     = 14,
-  UPB_TYPE_SFIXED32 = 15,
-  UPB_TYPE_SFIXED64 = 16,
-  UPB_TYPE_SINT32   = 17,
-  UPB_TYPE_SINT64   = 18,
+  UPB_TYPE_FLOAT    = 1,
+  UPB_TYPE_DOUBLE   = 2,
+  UPB_TYPE_BOOL     = 3,
+  UPB_TYPE_STRING   = 4,
+  UPB_TYPE_BYTES    = 5,
+  UPB_TYPE_MESSAGE  = 6,
+  UPB_TYPE_ENUM     = 7,   // Enum values are int32.
+  UPB_TYPE_INT32    = 8,
+  UPB_TYPE_UINT32   = 9,
+  UPB_TYPE_INT64    = 10,
+  UPB_TYPE_UINT64   = 11,
 } upb_fieldtype_t;
 
-#define UPB_NUM_TYPES 19
-
+// The repeated-ness of each field; this matches descriptor.proto.
 typedef enum {
   UPB_LABEL_OPTIONAL = 1,
   UPB_LABEL_REQUIRED = 2,
   UPB_LABEL_REPEATED = 3,
 } upb_label_t;
 
-// These macros are provided for legacy reasons.
-#define UPB_TYPE(type) UPB_TYPE_ ## type
-#define UPB_LABEL(type) UPB_LABEL_ ## type
-
-// Info for a given field type.
-typedef struct {
-  uint8_t align;
-  uint8_t size;
-  uint8_t inmemory_type;    // For example, INT32, SINT32, and SFIXED32 -> INT32
-} upb_typeinfo;
+// How integers should be encoded in serializations that offer multiple
+// integer encoding methods.
+typedef enum {
+  UPB_INTFMT_VARIABLE = 1,
+  UPB_INTFMT_FIXED = 2,
+  UPB_INTFMT_ZIGZAG = 3,  // Only for signed types (INT32/INT64).
+} upb_intfmt_t;
 
-extern const upb_typeinfo upb_types[UPB_NUM_TYPES];
+// Descriptor types, as defined in descriptor.proto.
+typedef enum {
+  UPB_DESCRIPTOR_TYPE_DOUBLE   = 1,
+  UPB_DESCRIPTOR_TYPE_FLOAT    = 2,
+  UPB_DESCRIPTOR_TYPE_INT64    = 3,
+  UPB_DESCRIPTOR_TYPE_UINT64   = 4,
+  UPB_DESCRIPTOR_TYPE_INT32    = 5,
+  UPB_DESCRIPTOR_TYPE_FIXED64  = 6,
+  UPB_DESCRIPTOR_TYPE_FIXED32  = 7,
+  UPB_DESCRIPTOR_TYPE_BOOL     = 8,
+  UPB_DESCRIPTOR_TYPE_STRING   = 9,
+  UPB_DESCRIPTOR_TYPE_GROUP    = 10,
+  UPB_DESCRIPTOR_TYPE_MESSAGE  = 11,
+  UPB_DESCRIPTOR_TYPE_BYTES    = 12,
+  UPB_DESCRIPTOR_TYPE_UINT32   = 13,
+  UPB_DESCRIPTOR_TYPE_ENUM     = 14,
+  UPB_DESCRIPTOR_TYPE_SFIXED32 = 15,
+  UPB_DESCRIPTOR_TYPE_SFIXED64 = 16,
+  UPB_DESCRIPTOR_TYPE_SINT32   = 17,
+  UPB_DESCRIPTOR_TYPE_SINT64   = 18,
+} upb_descriptortype_t;
 
 #ifdef __cplusplus
 
@@ -207,6 +216,8 @@ class upb::FieldDef {
  public:
   typedef upb_fieldtype_t Type;
   typedef upb_label_t Label;
+  typedef upb_intfmt_t IntegerFormat;
+  typedef upb_descriptortype_t DescriptorType;
 
   // Returns NULL if memory allocation failed.
   static FieldDef* New(const void *owner);
@@ -235,17 +246,27 @@ class upb::FieldDef {
   bool set_full_name(const char *fullname);
   bool set_full_name(const std::string& fullname);
 
-  Type type() const;        // Return UPB_TYPE_NONE if uninitialized.
-  Label label() const;      // Defaults to UPB_LABEL_OPTIONAL.
-  uint32_t number() const;  // Returns 0 if uninitialized.
+  bool type_is_set() const;  // Whether set_type() has been called.
+  Type type() const;         // Requires that type_is_set() == true.
+  Label label() const;       // Defaults to UPB_LABEL_OPTIONAL.
+  uint32_t number() const;   // Returns 0 if uninitialized.
   const MessageDef* message_def() const;
 
-  // "number" and "name" must be set before the fielddef is added to a msgdef.
-  // For the moment we do not allow these to be set once the fielddef is added
-  // to a msgdef -- this could be relaxed in the future.
+  // Gets/sets the field's type according to the enum in descriptor.proto.
+  // This is not the same as UPB_TYPE_*, because it distinguishes between
+  // (for example) INT32 and SINT32, whereas our "type" enum does not.
+  // This return of descriptor_type() is a function of type(),
+  // integer_format(), and is_tag_delimited().  Likewise set_descriptor_type()
+  // sets all three appropriately.
+  DescriptorType descriptor_type() const;
+  bool set_descriptor_type(DescriptorType type);
+
+  // "number" and "name" must be set before the FieldDef is added to a
+  // MessageDef, and may not be set after that.  "type" must be set explicitly
+  // before the fielddef is finalized.
   bool set_number(uint32_t number);
-  bool set_type(upb_fieldtype_t type);
-  bool set_label(upb_label_t label);
+  bool set_type(Type type);
+  bool set_label(Label label);
 
   // These are the same as full_name()/set_full_name(), but since fielddefs
   // most often use simple, non-qualified names, we provide this accessor
@@ -255,15 +276,26 @@ class upb::FieldDef {
   bool set_name(const std::string& name);
   const char *name() const;
 
+  // Convenient field type tests.
   bool IsSubMessage() const;
   bool IsString() const;
   bool IsSequence() const;
   bool IsPrimitive() const;
 
-  // Returns the default value for this fielddef, which may either be something
-  // the client set explicitly or the "default default" (0 for numbers, empty
-  // for strings).  The field's type indicates the type of the returned value,
-  // except for enum fields that are still mutable.
+  // How integers are encoded.  Only meaningful for integer types.
+  // Defaults to UPB_INTFMT_VARIABLE, and is reset when "type" changes.
+  IntegerFormat integer_format() const;
+  bool set_integer_format(IntegerFormat format);
+
+  // Whether a submessage field is tag-delimited or not (if false, then
+  // length-delimited).  Only meaningful when type() == UPB_TYPE_MESSAGE.
+  bool is_tag_delimited() const;
+  bool set_tag_delimited(bool tag_delimited);
+
+  // Returns the non-string default value for this fielddef, which may either
+  // be something the client set explicitly or the "default default" (0 for
+  // numbers, empty for strings).  The field's type indicates the type of the
+  // returned value, except for enum fields that are still mutable.
   //
   // For enums the default can be set either numerically or symbolically -- the
   // upb_fielddef_default_is_symbolic() function below will indicate which it
@@ -273,6 +305,12 @@ class upb::FieldDef {
   // always have a default of type int32.
   Value default_value() const;
 
+  // Returns the NULL-terminated string default value for this field, or NULL
+  // if the default for this field is not a string.  The user may optionally
+  // pass "len" to retrieve the length of the default also (this would be
+  // required to get default values with embedded NULLs).
+  const char *GetDefaultString(size_t* len) const;
+
   // Sets default value for the field.  For numeric types, use
   // upb_fielddef_setdefault(), and "value" must match the type of the field.
   // For string/bytes types, use upb_fielddef_setdefaultstr().  Enum types may
@@ -337,6 +375,7 @@ class upb::FieldDef {
 struct upb_fielddef {
 #endif
   upb_def base;
+  upb_value defaultval;  // Only for non-repeated scalars and strings.
   const upb_msgdef *msgdef;
   union {
     const upb_def *def;  // If !subdef_is_symbolic.
@@ -344,23 +383,20 @@ struct upb_fielddef {
   } sub;  // The msgdef or enumdef for this field, if upb_hassubdef(f).
   bool subdef_is_symbolic;
   bool default_is_string;
-  bool subdef_is_owned;
+  bool type_is_set_;     // False until type is explicitly set.
+  upb_intfmt_t intfmt;
+  bool tagdelim;
   upb_fieldtype_t type_;
   upb_label_t label_;
   uint32_t number_;
-  upb_value defaultval;  // Only for non-repeated scalars and strings.
   uint32_t selector_base;  // Used to index into a upb::Handlers table.
 };
 
-// This will only work for static initialization because of the subdef_is_owned
-// initialization.  Theoretically the other _INIT() macros could possible work
-// for non-static initialization, but this has not been tested.
-#define UPB_FIELDDEF_INIT(label, type, name, num, msgdef, subdef, \
-                          selector_base, defaultval) \
-  {UPB_DEF_INIT(name, UPB_DEF_FIELD), msgdef, {subdef}, false, \
-   type == UPB_TYPE_STRING || type == UPB_TYPE_BYTES, \
-   false, /* subdef_is_owned: not used since fielddef is not freed. */ \
-   type, label, num, defaultval, selector_base}
+#define UPB_FIELDDEF_INIT(label, type, intfmt, tagdelim, name, num, \
+                          msgdef, subdef, selector_base, defaultval) \
+  {UPB_DEF_INIT(name, UPB_DEF_FIELD), defaultval, msgdef, {subdef}, \
+   false, type == UPB_TYPE_STRING || type == UPB_TYPE_BYTES, true, \
+   intfmt, tagdelim, type, label, num, selector_base}
 
 // Native C API.
 #ifdef __cplusplus
@@ -381,31 +417,40 @@ void upb_fielddef_checkref(const upb_fielddef *f, const void *owner);
 const char *upb_fielddef_fullname(const upb_fielddef *f);
 bool upb_fielddef_setfullname(upb_fielddef *f, const char *fullname);
 
+bool upb_fielddef_typeisset(const upb_fielddef *f);
 upb_fieldtype_t upb_fielddef_type(const upb_fielddef *f);
+upb_descriptortype_t upb_fielddef_descriptortype(const upb_fielddef *f);
 upb_label_t upb_fielddef_label(const upb_fielddef *f);
 uint32_t upb_fielddef_number(const upb_fielddef *f);
 const char *upb_fielddef_name(const upb_fielddef *f);
 const upb_msgdef *upb_fielddef_msgdef(const upb_fielddef *f);
 upb_msgdef *upb_fielddef_msgdef_mutable(upb_fielddef *f);
-bool upb_fielddef_settype(upb_fielddef *f, upb_fieldtype_t type);
-bool upb_fielddef_setlabel(upb_fielddef *f, upb_label_t label);
-bool upb_fielddef_setnumber(upb_fielddef *f, uint32_t number);
-bool upb_fielddef_setname(upb_fielddef *f, const char *name);
+upb_intfmt_t upb_fielddef_intfmt(const upb_fielddef *f);
+bool upb_fielddef_istagdelim(const upb_fielddef *f);
 bool upb_fielddef_issubmsg(const upb_fielddef *f);
 bool upb_fielddef_isstring(const upb_fielddef *f);
 bool upb_fielddef_isseq(const upb_fielddef *f);
 bool upb_fielddef_isprimitive(const upb_fielddef *f);
 upb_value upb_fielddef_default(const upb_fielddef *f);
+const char *upb_fielddef_defaultstr(const upb_fielddef *f, size_t *len);
+bool upb_fielddef_default_is_symbolic(const upb_fielddef *f);
+bool upb_fielddef_hassubdef(const upb_fielddef *f);
+const upb_def *upb_fielddef_subdef(const upb_fielddef *f);
+const char *upb_fielddef_subdefname(const upb_fielddef *f);
+
+bool upb_fielddef_settype(upb_fielddef *f, upb_fieldtype_t type);
+bool upb_fielddef_setdescriptortype(upb_fielddef *f, int type);
+bool upb_fielddef_setlabel(upb_fielddef *f, upb_label_t label);
+bool upb_fielddef_setnumber(upb_fielddef *f, uint32_t number);
+bool upb_fielddef_setname(upb_fielddef *f, const char *name);
+bool upb_fielddef_setintfmt(upb_fielddef *f, upb_intfmt_t fmt);
+bool upb_fielddef_settagdelim(upb_fielddef *f, bool tag_delim);
 void upb_fielddef_setdefault(upb_fielddef *f, upb_value value);
 bool upb_fielddef_setdefaultstr(upb_fielddef *f, const void *str, size_t len);
 void upb_fielddef_setdefaultcstr(upb_fielddef *f, const char *str);
-bool upb_fielddef_default_is_symbolic(const upb_fielddef *f);
 bool upb_fielddef_resolvedefault(upb_fielddef *f);
-bool upb_fielddef_hassubdef(const upb_fielddef *f);
 bool upb_fielddef_setsubdef(upb_fielddef *f, const upb_def *subdef);
 bool upb_fielddef_setsubdefname(upb_fielddef *f, const char *name);
-const upb_def *upb_fielddef_subdef(const upb_fielddef *f);
-const char *upb_fielddef_subdefname(const upb_fielddef *f);
 #ifdef __cplusplus
 }  // extern "C"
 #endif
@@ -453,9 +498,9 @@ class upb::MessageDef {
 
   // These return NULL if the field is not found.
   FieldDef* FindFieldByNumber(uint32_t number);
-  FieldDef* FieldFieldByName(const char *name);
+  FieldDef* FindFieldByName(const char *name);
   const FieldDef* FindFieldByNumber(uint32_t number) const;
-  const FieldDef* FieldFieldByName(const char *name) const;
+  const FieldDef* FindFieldByName(const char *name) const;
 
   // Returns a new msgdef that is a copy of the given msgdef (and a copy of all
   // the fields) but with any references to submessages broken and replaced
@@ -685,18 +730,18 @@ int32_t upb_enum_iter_number(upb_enum_iter *iter);
 // Downcasts, for when some wants to assert that a def is of a particular type.
 // These are only checked if we are building debug.
 #define UPB_DEF_CASTS(lower, upper) \
-  INLINE const upb_ ## lower *upb_dyncast_ ## lower(const upb_def *def) { \
+  UPB_INLINE const upb_ ## lower *upb_dyncast_ ## lower(const upb_def *def) { \
     if (upb_def_type(def) != UPB_DEF_ ## upper) return NULL; \
     return (upb_ ## lower*)def; \
   } \
-  INLINE const upb_ ## lower *upb_downcast_ ## lower(const upb_def *def) { \
+  UPB_INLINE const upb_ ## lower *upb_downcast_ ## lower(const upb_def *def) { \
     assert(upb_def_type(def) == UPB_DEF_ ## upper); \
     return (const upb_ ## lower*)def; \
   } \
-  INLINE upb_ ## lower *upb_dyncast_ ## lower ## _mutable(upb_def *def) { \
+  UPB_INLINE upb_ ## lower *upb_dyncast_ ## lower ## _mutable(upb_def *def) { \
     return (upb_ ## lower*)upb_dyncast_ ## lower(def); \
   } \
-  INLINE upb_ ## lower *upb_downcast_ ## lower ## _mutable(upb_def *def) { \
+  UPB_INLINE upb_ ## lower *upb_downcast_ ## lower ## _mutable(upb_def *def) { \
     return (upb_ ## lower*)upb_downcast_ ## lower(def); \
   }
 UPB_DEF_CASTS(msgdef, MSG);
@@ -706,7 +751,7 @@ UPB_DEF_CASTS(enumdef, ENUM);
 
 #ifdef __cplusplus
 
-INLINE const char *upb_safecstr(const std::string& str) {
+UPB_INLINE const char *upb_safecstr(const std::string& str) {
   assert(str.size() == std::strlen(str.c_str()));
   return str.c_str();
 }
@@ -793,9 +838,15 @@ inline bool FieldDef::set_full_name(const char *fullname) {
 inline bool FieldDef::set_full_name(const std::string& fullname) {
   return upb_fielddef_setfullname(this, upb_safecstr(fullname));
 }
+inline bool FieldDef::type_is_set() const {
+  return upb_fielddef_typeisset(this);
+}
 inline FieldDef::Type FieldDef::type() const {
   return upb_fielddef_type(this);
 }
+inline FieldDef::DescriptorType FieldDef::descriptor_type() const {
+  return upb_fielddef_descriptortype(this);
+}
 inline FieldDef::Label FieldDef::label() const {
   return upb_fielddef_label(this);
 }
@@ -820,6 +871,9 @@ inline bool FieldDef::set_name(const std::string& name) {
 inline bool FieldDef::set_type(upb_fieldtype_t type) {
   return upb_fielddef_settype(this, type);
 }
+inline bool FieldDef::set_descriptor_type(FieldDef::DescriptorType type) {
+  return upb_fielddef_setdescriptortype(this, type);
+}
 inline bool FieldDef::set_label(upb_label_t label) {
   return upb_fielddef_setlabel(this, label);
 }
@@ -835,6 +889,9 @@ inline bool FieldDef::IsSequence() const {
 inline Value FieldDef::default_value() const {
   return upb_fielddef_default(this);
 }
+inline const char *FieldDef::GetDefaultString(size_t* len) const {
+  return upb_fielddef_defaultstr(this, len);
+}
 inline void FieldDef::set_default_value(Value value) {
   upb_fielddef_setdefault(this, value);
 }
@@ -914,13 +971,13 @@ inline bool MessageDef::AddField(upb_fielddef *f, const void *ref_donor) {
 inline FieldDef* MessageDef::FindFieldByNumber(uint32_t number) {
   return upb_msgdef_itof_mutable(this, number);
 }
-inline FieldDef* MessageDef::FieldFieldByName(const char *name) {
+inline FieldDef* MessageDef::FindFieldByName(const char *name) {
   return upb_msgdef_ntof_mutable(this, name);
 }
 inline const FieldDef* MessageDef::FindFieldByNumber(uint32_t number) const {
   return upb_msgdef_itof(this, number);
 }
-inline const FieldDef* MessageDef::FieldFieldByName(const char *name) const {
+inline const FieldDef* MessageDef::FindFieldByName(const char *name) const {
   return upb_msgdef_ntof(this, name);
 }
 inline MessageDef* MessageDef::Dup(const void *owner) const {

upb/descriptor/descriptor.upb.c

@@ -9,105 +9,105 @@ const upb_fielddef google_protobuf_fields[73];
 const upb_enumdef google_protobuf_enums[4];
 const upb_tabent google_protobuf_strentries[192];
 const upb_tabent google_protobuf_intentries[66];
-const upb_value google_protobuf_arrays[97];
+const _upb_value google_protobuf_arrays[97];
 
 const upb_msgdef google_protobuf_msgs[20] = {
-  UPB_MSGDEF_INIT("google.protobuf.DescriptorProto", UPB_INTTABLE_INIT(2, 3, 9, 2, &google_protobuf_intentries[0], &google_protobuf_arrays[0], 6, 5), UPB_STRTABLE_INIT(7, 15, 9, 4, &google_protobuf_strentries[0]), 31),
+  UPB_MSGDEF_INIT("google.protobuf.DescriptorProto", UPB_INTTABLE_INIT(2, 3, 9, 2, &google_protobuf_intentries[0], &google_protobuf_arrays[0], 6, 5), UPB_STRTABLE_INIT(7, 15, 9, 4, &google_protobuf_strentries[0]), 25),
   UPB_MSGDEF_INIT("google.protobuf.DescriptorProto.ExtensionRange", UPB_INTTABLE_INIT(0, 0, 9, 0, NULL, &google_protobuf_arrays[6], 4, 2), UPB_STRTABLE_INIT(2, 3, 9, 2, &google_protobuf_strentries[16]), 2),
-  UPB_MSGDEF_INIT("google.protobuf.EnumDescriptorProto", UPB_INTTABLE_INIT(0, 0, 9, 0, NULL, &google_protobuf_arrays[10], 4, 3), UPB_STRTABLE_INIT(3, 3, 9, 2, &google_protobuf_strentries[20]), 11),
-  UPB_MSGDEF_INIT("google.protobuf.EnumOptions", UPB_INTTABLE_INIT(1, 1, 9, 1, &google_protobuf_intentries[4], &google_protobuf_arrays[14], 1, 0), UPB_STRTABLE_INIT(1, 3, 9, 2, &google_protobuf_strentries[24]), 5),
-  UPB_MSGDEF_INIT("google.protobuf.EnumValueDescriptorProto", UPB_INTTABLE_INIT(0, 0, 9, 0, NULL, &google_protobuf_arrays[15], 4, 3), UPB_STRTABLE_INIT(3, 3, 9, 2, &google_protobuf_strentries[28]), 7),
-  UPB_MSGDEF_INIT("google.protobuf.EnumValueOptions", UPB_INTTABLE_INIT(1, 1, 9, 1, &google_protobuf_intentries[6], &google_protobuf_arrays[19], 1, 0), UPB_STRTABLE_INIT(1, 3, 9, 2, &google_protobuf_strentries[32]), 5),
-  UPB_MSGDEF_INIT("google.protobuf.FieldDescriptorProto", UPB_INTTABLE_INIT(3, 3, 9, 2, &google_protobuf_intentries[8], &google_protobuf_arrays[20], 6, 5), UPB_STRTABLE_INIT(8, 15, 9, 4, &google_protobuf_strentries[36]), 18),
-  UPB_MSGDEF_INIT("google.protobuf.FieldOptions", UPB_INTTABLE_INIT(2, 3, 9, 2, &google_protobuf_intentries[12], &google_protobuf_arrays[26], 5, 3), UPB_STRTABLE_INIT(5, 7, 9, 3, &google_protobuf_strentries[52]), 11),
-  UPB_MSGDEF_INIT("google.protobuf.FileDescriptorProto", UPB_INTTABLE_INIT(4, 7, 9, 3, &google_protobuf_intentries[16], &google_protobuf_arrays[31], 6, 5), UPB_STRTABLE_INIT(9, 15, 9, 4, &google_protobuf_strentries[60]), 37),
-  UPB_MSGDEF_INIT("google.protobuf.FileDescriptorSet", UPB_INTTABLE_INIT(0, 0, 9, 0, NULL, &google_protobuf_arrays[37], 3, 1), UPB_STRTABLE_INIT(1, 3, 9, 2, &google_protobuf_strentries[76]), 5),
-  UPB_MSGDEF_INIT("google.protobuf.FileOptions", UPB_INTTABLE_INIT(8, 15, 9, 4, &google_protobuf_intentries[24], &google_protobuf_arrays[40], 6, 1), UPB_STRTABLE_INIT(9, 15, 9, 4, &google_protobuf_strentries[80]), 17),
-  UPB_MSGDEF_INIT("google.protobuf.MessageOptions", UPB_INTTABLE_INIT(1, 1, 9, 1, &google_protobuf_intentries[40], &google_protobuf_arrays[46], 4, 2), UPB_STRTABLE_INIT(3, 3, 9, 2, &google_protobuf_strentries[96]), 7),
-  UPB_MSGDEF_INIT("google.protobuf.MethodDescriptorProto", UPB_INTTABLE_INIT(0, 0, 9, 0, NULL, &google_protobuf_arrays[50], 5, 4), UPB_STRTABLE_INIT(4, 7, 9, 3, &google_protobuf_strentries[100]), 12),
-  UPB_MSGDEF_INIT("google.protobuf.MethodOptions", UPB_INTTABLE_INIT(1, 1, 9, 1, &google_protobuf_intentries[42], &google_protobuf_arrays[55], 1, 0), UPB_STRTABLE_INIT(1, 3, 9, 2, &google_protobuf_strentries[108]), 5),
-  UPB_MSGDEF_INIT("google.protobuf.ServiceDescriptorProto", UPB_INTTABLE_INIT(0, 0, 9, 0, NULL, &google_protobuf_arrays[56], 4, 3), UPB_STRTABLE_INIT(3, 3, 9, 2, &google_protobuf_strentries[112]), 11),
-  UPB_MSGDEF_INIT("google.protobuf.ServiceOptions", UPB_INTTABLE_INIT(1, 1, 9, 1, &google_protobuf_intentries[44], &google_protobuf_arrays[60], 1, 0), UPB_STRTABLE_INIT(1, 3, 9, 2, &google_protobuf_strentries[116]), 5),
-  UPB_MSGDEF_INIT("google.protobuf.SourceCodeInfo", UPB_INTTABLE_INIT(0, 0, 9, 0, NULL, &google_protobuf_arrays[61], 3, 1), UPB_STRTABLE_INIT(1, 3, 9, 2, &google_protobuf_strentries[120]), 5),
+  UPB_MSGDEF_INIT("google.protobuf.EnumDescriptorProto", UPB_INTTABLE_INIT(0, 0, 9, 0, NULL, &google_protobuf_arrays[10], 4, 3), UPB_STRTABLE_INIT(3, 3, 9, 2, &google_protobuf_strentries[20]), 9),
+  UPB_MSGDEF_INIT("google.protobuf.EnumOptions", UPB_INTTABLE_INIT(1, 1, 9, 1, &google_protobuf_intentries[4], &google_protobuf_arrays[14], 1, 0), UPB_STRTABLE_INIT(1, 3, 9, 2, &google_protobuf_strentries[24]), 4),
+  UPB_MSGDEF_INIT("google.protobuf.EnumValueDescriptorProto", UPB_INTTABLE_INIT(0, 0, 9, 0, NULL, &google_protobuf_arrays[15], 4, 3), UPB_STRTABLE_INIT(3, 3, 9, 2, &google_protobuf_strentries[28]), 6),
+  UPB_MSGDEF_INIT("google.protobuf.EnumValueOptions", UPB_INTTABLE_INIT(1, 1, 9, 1, &google_protobuf_intentries[6], &google_protobuf_arrays[19], 1, 0), UPB_STRTABLE_INIT(1, 3, 9, 2, &google_protobuf_strentries[32]), 4),
+  UPB_MSGDEF_INIT("google.protobuf.FieldDescriptorProto", UPB_INTTABLE_INIT(3, 3, 9, 2, &google_protobuf_intentries[8], &google_protobuf_arrays[20], 6, 5), UPB_STRTABLE_INIT(8, 15, 9, 4, &google_protobuf_strentries[36]), 17),
+  UPB_MSGDEF_INIT("google.protobuf.FieldOptions", UPB_INTTABLE_INIT(2, 3, 9, 2, &google_protobuf_intentries[12], &google_protobuf_arrays[26], 5, 3), UPB_STRTABLE_INIT(5, 7, 9, 3, &google_protobuf_strentries[52]), 10),
+  UPB_MSGDEF_INIT("google.protobuf.FileDescriptorProto", UPB_INTTABLE_INIT(4, 7, 9, 3, &google_protobuf_intentries[16], &google_protobuf_arrays[31], 6, 5), UPB_STRTABLE_INIT(9, 15, 9, 4, &google_protobuf_strentries[60]), 31),
+  UPB_MSGDEF_INIT("google.protobuf.FileDescriptorSet", UPB_INTTABLE_INIT(0, 0, 9, 0, NULL, &google_protobuf_arrays[37], 3, 1), UPB_STRTABLE_INIT(1, 3, 9, 2, &google_protobuf_strentries[76]), 4),
+  UPB_MSGDEF_INIT("google.protobuf.FileOptions", UPB_INTTABLE_INIT(8, 15, 9, 4, &google_protobuf_intentries[24], &google_protobuf_arrays[40], 6, 1), UPB_STRTABLE_INIT(9, 15, 9, 4, &google_protobuf_strentries[80]), 16),
+  UPB_MSGDEF_INIT("google.protobuf.MessageOptions", UPB_INTTABLE_INIT(1, 1, 9, 1, &google_protobuf_intentries[40], &google_protobuf_arrays[46], 4, 2), UPB_STRTABLE_INIT(3, 3, 9, 2, &google_protobuf_strentries[96]), 6),
+  UPB_MSGDEF_INIT("google.protobuf.MethodDescriptorProto", UPB_INTTABLE_INIT(0, 0, 9, 0, NULL, &google_protobuf_arrays[50], 5, 4), UPB_STRTABLE_INIT(4, 7, 9, 3, &google_protobuf_strentries[100]), 11),
+  UPB_MSGDEF_INIT("google.protobuf.MethodOptions", UPB_INTTABLE_INIT(1, 1, 9, 1, &google_protobuf_intentries[42], &google_protobuf_arrays[55], 1, 0), UPB_STRTABLE_INIT(1, 3, 9, 2, &google_protobuf_strentries[108]), 4),
+  UPB_MSGDEF_INIT("google.protobuf.ServiceDescriptorProto", UPB_INTTABLE_INIT(0, 0, 9, 0, NULL, &google_protobuf_arrays[56], 4, 3), UPB_STRTABLE_INIT(3, 3, 9, 2, &google_protobuf_strentries[112]), 9),
+  UPB_MSGDEF_INIT("google.protobuf.ServiceOptions", UPB_INTTABLE_INIT(1, 1, 9, 1, &google_protobuf_intentries[44], &google_protobuf_arrays[60], 1, 0), UPB_STRTABLE_INIT(1, 3, 9, 2, &google_protobuf_strentries[116]), 4),
+  UPB_MSGDEF_INIT("google.protobuf.SourceCodeInfo", UPB_INTTABLE_INIT(0, 0, 9, 0, NULL, &google_protobuf_arrays[61], 3, 1), UPB_STRTABLE_INIT(1, 3, 9, 2, &google_protobuf_strentries[120]), 4),
   UPB_MSGDEF_INIT("google.protobuf.SourceCodeInfo.Location", UPB_INTTABLE_INIT(0, 0, 9, 0, NULL, &google_protobuf_arrays[64], 4, 2), UPB_STRTABLE_INIT(2, 3, 9, 2, &google_protobuf_strentries[124]), 6),
-  UPB_MSGDEF_INIT("google.protobuf.UninterpretedOption", UPB_INTTABLE_INIT(3, 3, 9, 2, &google_protobuf_intentries[46], &google_protobuf_arrays[68], 6, 4), UPB_STRTABLE_INIT(7, 15, 9, 4, &google_protobuf_strentries[128]), 17),
+  UPB_MSGDEF_INIT("google.protobuf.UninterpretedOption", UPB_INTTABLE_INIT(3, 3, 9, 2, &google_protobuf_intentries[46], &google_protobuf_arrays[68], 6, 4), UPB_STRTABLE_INIT(7, 15, 9, 4, &google_protobuf_strentries[128]), 16),
   UPB_MSGDEF_INIT("google.protobuf.UninterpretedOption.NamePart", UPB_INTTABLE_INIT(0, 0, 9, 0, NULL, &google_protobuf_arrays[74], 4, 2), UPB_STRTABLE_INIT(2, 3, 9, 2, &google_protobuf_strentries[144]), 4),
 };
 
 const upb_fielddef google_protobuf_fields[73] = {
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, "aggregate_value", 8, &google_protobuf_msgs[18], NULL, 10, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, "cc_generic_services", 16, &google_protobuf_msgs[10], NULL, 3, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_ENUM, "ctype", 1, &google_protobuf_msgs[7], upb_upcast(&google_protobuf_enums[2]), 0, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, "default_value", 7, &google_protobuf_msgs[6], NULL, 15, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_STRING, "dependency", 3, &google_protobuf_msgs[8], NULL, 8, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, "deprecated", 3, &google_protobuf_msgs[7], NULL, 2, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_DOUBLE, "double_value", 6, &google_protobuf_msgs[18], NULL, 13, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_INT32, "end", 2, &google_protobuf_msgs[1], NULL, 1, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "enum_type", 4, &google_protobuf_msgs[0], upb_upcast(&google_protobuf_msgs[2]), 15, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "enum_type", 5, &google_protobuf_msgs[8], upb_upcast(&google_protobuf_msgs[2]), 18, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, "experimental_map_key", 9, &google_protobuf_msgs[7], NULL, 3, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, "extendee", 2, &google_protobuf_msgs[6], NULL, 3, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "extension", 7, &google_protobuf_msgs[8], upb_upcast(&google_protobuf_msgs[6]), 34, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "extension", 6, &google_protobuf_msgs[0], upb_upcast(&google_protobuf_msgs[6]), 25, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "extension_range", 5, &google_protobuf_msgs[0], upb_upcast(&google_protobuf_msgs[1]), 20, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "field", 2, &google_protobuf_msgs[0], upb_upcast(&google_protobuf_msgs[6]), 5, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "file", 1, &google_protobuf_msgs[9], upb_upcast(&google_protobuf_msgs[8]), 2, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, "identifier_value", 3, &google_protobuf_msgs[18], NULL, 5, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, "input_type", 2, &google_protobuf_msgs[12], NULL, 3, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REQUIRED, UPB_TYPE_BOOL, "is_extension", 2, &google_protobuf_msgs[19], NULL, 3, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, "java_generate_equals_and_hash", 20, &google_protobuf_msgs[10], NULL, 6, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, "java_generic_services", 17, &google_protobuf_msgs[10], NULL, 4, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, "java_multiple_files", 10, &google_protobuf_msgs[10], NULL, 16, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, "java_outer_classname", 8, &google_protobuf_msgs[10], NULL, 12, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, "java_package", 1, &google_protobuf_msgs[10], NULL, 0, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_ENUM, "label", 4, &google_protobuf_msgs[6], upb_upcast(&google_protobuf_enums[0]), 7, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "location", 1, &google_protobuf_msgs[16], upb_upcast(&google_protobuf_msgs[17]), 2, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, "message_set_wire_format", 1, &google_protobuf_msgs[11], NULL, 0, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "message_type", 4, &google_protobuf_msgs[8], upb_upcast(&google_protobuf_msgs[0]), 13, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "method", 2, &google_protobuf_msgs[14], upb_upcast(&google_protobuf_msgs[12]), 5, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, "name", 1, &google_protobuf_msgs[12], NULL, 0, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, "name", 1, &google_protobuf_msgs[4], NULL, 0, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, "name", 1, &google_protobuf_msgs[14], NULL, 0, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, "name", 1, &google_protobuf_msgs[2], NULL, 0, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, "name", 1, &google_protobuf_msgs[6], NULL, 0, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "name", 2, &google_protobuf_msgs[18], upb_upcast(&google_protobuf_msgs[19]), 2, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, "name", 1, &google_protobuf_msgs[0], NULL, 0, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, "name", 1, &google_protobuf_msgs[8], NULL, 0, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REQUIRED, UPB_TYPE_STRING, "name_part", 1, &google_protobuf_msgs[19], NULL, 0, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_INT64, "negative_int_value", 5, &google_protobuf_msgs[18], NULL, 9, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "nested_type", 3, &google_protobuf_msgs[0], upb_upcast(&google_protobuf_msgs[0]), 10, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, "no_standard_descriptor_accessor", 2, &google_protobuf_msgs[11], NULL, 1, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_INT32, "number", 2, &google_protobuf_msgs[4], NULL, 3, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_INT32, "number", 3, &google_protobuf_msgs[6], NULL, 6, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_ENUM, "optimize_for", 9, &google_protobuf_msgs[10], upb_upcast(&google_protobuf_enums[3]), 15, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, "options", 4, &google_protobuf_msgs[12], upb_upcast(&google_protobuf_msgs[13]), 9, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, "options", 3, &google_protobuf_msgs[14], upb_upcast(&google_protobuf_msgs[15]), 8, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, "options", 8, &google_protobuf_msgs[8], upb_upcast(&google_protobuf_msgs[10]), 21, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, "options", 3, &google_protobuf_msgs[2], upb_upcast(&google_protobuf_msgs[3]), 8, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, "options", 7, &google_protobuf_msgs[0], upb_upcast(&google_protobuf_msgs[11]), 28, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, "options", 8, &google_protobuf_msgs[6], upb_upcast(&google_protobuf_msgs[7]), 9, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, "options", 3, &google_protobuf_msgs[4], upb_upcast(&google_protobuf_msgs[5]), 4, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, "output_type", 3, &google_protobuf_msgs[12], NULL, 6, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, "package", 2, &google_protobuf_msgs[8], NULL, 3, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, "packed", 2, &google_protobuf_msgs[7], NULL, 1, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_INT32, "path", 1, &google_protobuf_msgs[17], NULL, 2, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_UINT64, "positive_int_value", 4, &google_protobuf_msgs[18], NULL, 8, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, "py_generic_services", 18, &google_protobuf_msgs[10], NULL, 5, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "service", 6, &google_protobuf_msgs[8], upb_upcast(&google_protobuf_msgs[14]), 29, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, "source_code_info", 9, &google_protobuf_msgs[8], upb_upcast(&google_protobuf_msgs[16]), 24, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_INT32, "span", 2, &google_protobuf_msgs[17], NULL, 5, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_INT32, "start", 1, &google_protobuf_msgs[1], NULL, 0, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BYTES, "string_value", 7, &google_protobuf_msgs[18], NULL, 14, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_ENUM, "type", 5, &google_protobuf_msgs[6], upb_upcast(&google_protobuf_enums[1]), 8, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, "type_name", 6, &google_protobuf_msgs[6], NULL, 12, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "uninterpreted_option", 999, &google_protobuf_msgs[15], upb_upcast(&google_protobuf_msgs[18]), 2, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "uninterpreted_option", 999, &google_protobuf_msgs[11], upb_upcast(&google_protobuf_msgs[18]), 4, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "uninterpreted_option", 999, &google_protobuf_msgs[13], upb_upcast(&google_protobuf_msgs[18]), 2, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "uninterpreted_option", 999, &google_protobuf_msgs[10], upb_upcast(&google_protobuf_msgs[18]), 9, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "uninterpreted_option", 999, &google_protobuf_msgs[7], upb_upcast(&google_protobuf_msgs[18]), 8, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "uninterpreted_option", 999, &google_protobuf_msgs[3], upb_upcast(&google_protobuf_msgs[18]), 2, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "uninterpreted_option", 999, &google_protobuf_msgs[5], upb_upcast(&google_protobuf_msgs[18]), 2, UPB_VALUE_INIT_NONE),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, "value", 2, &google_protobuf_msgs[2], upb_upcast(&google_protobuf_msgs[4]), 5, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, "aggregate_value", 8, &google_protobuf_msgs[18], NULL, 9, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, "cc_generic_services", 16, &google_protobuf_msgs[10], NULL, 3, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_ENUM, 0, false, "ctype", 1, &google_protobuf_msgs[7], upb_upcast(&google_protobuf_enums[2]), 0, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, "default_value", 7, &google_protobuf_msgs[6], NULL, 14, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_STRING, 0, false, "dependency", 3, &google_protobuf_msgs[8], NULL, 8, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, "deprecated", 3, &google_protobuf_msgs[7], NULL, 2, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_DOUBLE, 0, false, "double_value", 6, &google_protobuf_msgs[18], NULL, 12, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_INT32, UPB_INTFMT_VARIABLE, false, "end", 2, &google_protobuf_msgs[1], NULL, 1, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "enum_type", 4, &google_protobuf_msgs[0], upb_upcast(&google_protobuf_msgs[2]), 13, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "enum_type", 5, &google_protobuf_msgs[8], upb_upcast(&google_protobuf_msgs[2]), 17, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, "experimental_map_key", 9, &google_protobuf_msgs[7], NULL, 3, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, "extendee", 2, &google_protobuf_msgs[6], NULL, 3, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "extension", 7, &google_protobuf_msgs[8], upb_upcast(&google_protobuf_msgs[6]), 29, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "extension", 6, &google_protobuf_msgs[0], upb_upcast(&google_protobuf_msgs[6]), 21, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "extension_range", 5, &google_protobuf_msgs[0], upb_upcast(&google_protobuf_msgs[1]), 17, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "field", 2, &google_protobuf_msgs[0], upb_upcast(&google_protobuf_msgs[6]), 5, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "file", 1, &google_protobuf_msgs[9], upb_upcast(&google_protobuf_msgs[8]), 2, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, "identifier_value", 3, &google_protobuf_msgs[18], NULL, 4, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, "input_type", 2, &google_protobuf_msgs[12], NULL, 3, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REQUIRED, UPB_TYPE_BOOL, 0, false, "is_extension", 2, &google_protobuf_msgs[19], NULL, 3, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, "java_generate_equals_and_hash", 20, &google_protobuf_msgs[10], NULL, 6, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, "java_generic_services", 17, &google_protobuf_msgs[10], NULL, 4, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, "java_multiple_files", 10, &google_protobuf_msgs[10], NULL, 15, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, "java_outer_classname", 8, &google_protobuf_msgs[10], NULL, 11, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, "java_package", 1, &google_protobuf_msgs[10], NULL, 0, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_ENUM, 0, false, "label", 4, &google_protobuf_msgs[6], upb_upcast(&google_protobuf_enums[0]), 7, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "location", 1, &google_protobuf_msgs[16], upb_upcast(&google_protobuf_msgs[17]), 2, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, "message_set_wire_format", 1, &google_protobuf_msgs[11], NULL, 0, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "message_type", 4, &google_protobuf_msgs[8], upb_upcast(&google_protobuf_msgs[0]), 13, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "method", 2, &google_protobuf_msgs[14], upb_upcast(&google_protobuf_msgs[12]), 5, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, "name", 1, &google_protobuf_msgs[12], NULL, 0, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, "name", 1, &google_protobuf_msgs[4], NULL, 0, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, "name", 1, &google_protobuf_msgs[14], NULL, 0, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, "name", 1, &google_protobuf_msgs[2], NULL, 0, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, "name", 1, &google_protobuf_msgs[6], NULL, 0, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "name", 2, &google_protobuf_msgs[18], upb_upcast(&google_protobuf_msgs[19]), 2, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, "name", 1, &google_protobuf_msgs[0], NULL, 0, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, "name", 1, &google_protobuf_msgs[8], NULL, 0, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REQUIRED, UPB_TYPE_STRING, 0, false, "name_part", 1, &google_protobuf_msgs[19], NULL, 0, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_INT64, UPB_INTFMT_VARIABLE, false, "negative_int_value", 5, &google_protobuf_msgs[18], NULL, 8, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "nested_type", 3, &google_protobuf_msgs[0], upb_upcast(&google_protobuf_msgs[0]), 9, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, "no_standard_descriptor_accessor", 2, &google_protobuf_msgs[11], NULL, 1, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_INT32, UPB_INTFMT_VARIABLE, false, "number", 2, &google_protobuf_msgs[4], NULL, 3, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_INT32, UPB_INTFMT_VARIABLE, false, "number", 3, &google_protobuf_msgs[6], NULL, 6, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_ENUM, 0, false, "optimize_for", 9, &google_protobuf_msgs[10], upb_upcast(&google_protobuf_enums[3]), 14, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, "options", 4, &google_protobuf_msgs[12], upb_upcast(&google_protobuf_msgs[13]), 9, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, "options", 3, &google_protobuf_msgs[14], upb_upcast(&google_protobuf_msgs[15]), 7, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, "options", 8, &google_protobuf_msgs[8], upb_upcast(&google_protobuf_msgs[10]), 19, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, "options", 3, &google_protobuf_msgs[2], upb_upcast(&google_protobuf_msgs[3]), 7, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, "options", 7, &google_protobuf_msgs[0], upb_upcast(&google_protobuf_msgs[11]), 23, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, "options", 8, &google_protobuf_msgs[6], upb_upcast(&google_protobuf_msgs[7]), 9, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, "options", 3, &google_protobuf_msgs[4], upb_upcast(&google_protobuf_msgs[5]), 4, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, "output_type", 3, &google_protobuf_msgs[12], NULL, 6, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, "package", 2, &google_protobuf_msgs[8], NULL, 3, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, "packed", 2, &google_protobuf_msgs[7], NULL, 1, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_INT32, UPB_INTFMT_VARIABLE, false, "path", 1, &google_protobuf_msgs[17], NULL, 2, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_UINT64, UPB_INTFMT_VARIABLE, false, "positive_int_value", 4, &google_protobuf_msgs[18], NULL, 7, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, "py_generic_services", 18, &google_protobuf_msgs[10], NULL, 5, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "service", 6, &google_protobuf_msgs[8], upb_upcast(&google_protobuf_msgs[14]), 25, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, "source_code_info", 9, &google_protobuf_msgs[8], upb_upcast(&google_protobuf_msgs[16]), 21, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_INT32, UPB_INTFMT_VARIABLE, false, "span", 2, &google_protobuf_msgs[17], NULL, 5, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_INT32, UPB_INTFMT_VARIABLE, false, "start", 1, &google_protobuf_msgs[1], NULL, 0, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BYTES, 0, false, "string_value", 7, &google_protobuf_msgs[18], NULL, 13, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_ENUM, 0, false, "type", 5, &google_protobuf_msgs[6], upb_upcast(&google_protobuf_enums[1]), 8, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, "type_name", 6, &google_protobuf_msgs[6], NULL, 11, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "uninterpreted_option", 999, &google_protobuf_msgs[15], upb_upcast(&google_protobuf_msgs[18]), 2, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "uninterpreted_option", 999, &google_protobuf_msgs[11], upb_upcast(&google_protobuf_msgs[18]), 4, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "uninterpreted_option", 999, &google_protobuf_msgs[13], upb_upcast(&google_protobuf_msgs[18]), 2, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "uninterpreted_option", 999, &google_protobuf_msgs[10], upb_upcast(&google_protobuf_msgs[18]), 9, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "uninterpreted_option", 999, &google_protobuf_msgs[7], upb_upcast(&google_protobuf_msgs[18]), 8, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "uninterpreted_option", 999, &google_protobuf_msgs[3], upb_upcast(&google_protobuf_msgs[18]), 2, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "uninterpreted_option", 999, &google_protobuf_msgs[5], upb_upcast(&google_protobuf_msgs[18]), 2, UPB_VALUE_INIT_NONE),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, "value", 2, &google_protobuf_msgs[2], upb_upcast(&google_protobuf_msgs[4]), 5, UPB_VALUE_INIT_NONE),
 };
 
 const upb_enumdef google_protobuf_enums[4] = {
@@ -119,257 +119,257 @@ const upb_enumdef google_protobuf_enums[4] = {
 
 const upb_tabent google_protobuf_strentries[192] = {
   {UPB_TABKEY_STR("extension"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[13]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("name"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[36]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("field"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[15]), NULL},
   {UPB_TABKEY_STR("extension_range"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[14]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("nested_type"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[40]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("options"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[49]), NULL},
   {UPB_TABKEY_STR("enum_type"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[8]), &google_protobuf_strentries[14]},
   {UPB_TABKEY_STR("start"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[61]), NULL},
   {UPB_TABKEY_STR("end"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[7]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("value"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[72]), NULL},
   {UPB_TABKEY_STR("options"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[48]), NULL},
   {UPB_TABKEY_STR("name"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[33]), &google_protobuf_strentries[22]},
   {UPB_TABKEY_STR("uninterpreted_option"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[70]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("number"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[42]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("options"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[51]), NULL},
   {UPB_TABKEY_STR("name"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[31]), &google_protobuf_strentries[30]},
   {UPB_TABKEY_STR("uninterpreted_option"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[71]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("label"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[25]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("name"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[34]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("number"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[43]), &google_protobuf_strentries[49]},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("type_name"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[64]), NULL},
   {UPB_TABKEY_STR("extendee"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[11]), NULL},
   {UPB_TABKEY_STR("type"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[63]), &google_protobuf_strentries[48]},
   {UPB_TABKEY_STR("default_value"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[3]), NULL},
   {UPB_TABKEY_STR("options"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[50]), NULL},
   {UPB_TABKEY_STR("experimental_map_key"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[10]), &google_protobuf_strentries[58]},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("ctype"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[2]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("deprecated"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[5]), NULL},
   {UPB_TABKEY_STR("uninterpreted_option"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[69]), NULL},
   {UPB_TABKEY_STR("packed"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[54]), NULL},
   {UPB_TABKEY_STR("extension"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[12]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("name"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[37]), NULL},
   {UPB_TABKEY_STR("service"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[58]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("source_code_info"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[59]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("dependency"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[4]), NULL},
   {UPB_TABKEY_STR("message_type"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[28]), NULL},
   {UPB_TABKEY_STR("package"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[53]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("options"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[47]), NULL},
   {UPB_TABKEY_STR("enum_type"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[9]), &google_protobuf_strentries[74]},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("file"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[16]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("uninterpreted_option"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[68]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("cc_generic_services"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[1]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("java_multiple_files"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[22]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("java_generic_services"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[21]), &google_protobuf_strentries[94]},
   {UPB_TABKEY_STR("java_generate_equals_and_hash"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[20]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("java_package"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[24]), NULL},
   {UPB_TABKEY_STR("optimize_for"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[44]), NULL},
   {UPB_TABKEY_STR("py_generic_services"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[57]), NULL},
   {UPB_TABKEY_STR("java_outer_classname"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[23]), NULL},
   {UPB_TABKEY_STR("message_set_wire_format"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[27]), &google_protobuf_strentries[98]},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("uninterpreted_option"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[66]), NULL},
   {UPB_TABKEY_STR("no_standard_descriptor_accessor"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[41]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("name"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[30]), NULL},
   {UPB_TABKEY_STR("input_type"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[18]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("output_type"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[52]), NULL},
   {UPB_TABKEY_STR("options"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[45]), NULL},
   {UPB_TABKEY_STR("uninterpreted_option"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[67]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("options"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[46]), &google_protobuf_strentries[114]},
   {UPB_TABKEY_STR("method"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[29]), NULL},
   {UPB_TABKEY_STR("name"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[32]), &google_protobuf_strentries[113]},
   {UPB_TABKEY_STR("uninterpreted_option"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[65]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("location"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[26]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("span"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[60]), NULL},
   {UPB_TABKEY_STR("path"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[55]), &google_protobuf_strentries[126]},
   {UPB_TABKEY_STR("double_value"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[6]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("name"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[35]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("negative_int_value"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[39]), NULL},
   {UPB_TABKEY_STR("aggregate_value"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[0]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("positive_int_value"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[56]), NULL},
   {UPB_TABKEY_STR("identifier_value"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[17]), NULL},
   {UPB_TABKEY_STR("string_value"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[62]), &google_protobuf_strentries[142]},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("is_extension"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[19]), NULL},
   {UPB_TABKEY_STR("name_part"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[38]), NULL},
   {UPB_TABKEY_STR("LABEL_REQUIRED"), UPB_VALUE_INIT_INT32(2), &google_protobuf_strentries[150]},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("LABEL_REPEATED"), UPB_VALUE_INIT_INT32(3), NULL},
   {UPB_TABKEY_STR("LABEL_OPTIONAL"), UPB_VALUE_INIT_INT32(1), NULL},
   {UPB_TABKEY_STR("TYPE_FIXED64"), UPB_VALUE_INIT_INT32(6), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("TYPE_STRING"), UPB_VALUE_INIT_INT32(9), NULL},
   {UPB_TABKEY_STR("TYPE_FLOAT"), UPB_VALUE_INIT_INT32(2), &google_protobuf_strentries[181]},
   {UPB_TABKEY_STR("TYPE_DOUBLE"), UPB_VALUE_INIT_INT32(1), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("TYPE_INT32"), UPB_VALUE_INIT_INT32(5), NULL},
   {UPB_TABKEY_STR("TYPE_SFIXED32"), UPB_VALUE_INIT_INT32(15), NULL},
   {UPB_TABKEY_STR("TYPE_FIXED32"), UPB_VALUE_INIT_INT32(7), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("TYPE_MESSAGE"), UPB_VALUE_INIT_INT32(11), &google_protobuf_strentries[182]},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("TYPE_INT64"), UPB_VALUE_INIT_INT32(3), &google_protobuf_strentries[179]},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("TYPE_ENUM"), UPB_VALUE_INIT_INT32(14), NULL},
   {UPB_TABKEY_STR("TYPE_UINT32"), UPB_VALUE_INIT_INT32(13), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("TYPE_UINT64"), UPB_VALUE_INIT_INT32(4), &google_protobuf_strentries[178]},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("TYPE_SFIXED64"), UPB_VALUE_INIT_INT32(16), NULL},
   {UPB_TABKEY_STR("TYPE_BYTES"), UPB_VALUE_INIT_INT32(12), NULL},
   {UPB_TABKEY_STR("TYPE_SINT64"), UPB_VALUE_INIT_INT32(18), NULL},
   {UPB_TABKEY_STR("TYPE_BOOL"), UPB_VALUE_INIT_INT32(8), NULL},
   {UPB_TABKEY_STR("TYPE_GROUP"), UPB_VALUE_INIT_INT32(10), NULL},
   {UPB_TABKEY_STR("TYPE_SINT32"), UPB_VALUE_INIT_INT32(17), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("CORD"), UPB_VALUE_INIT_INT32(1), NULL},
   {UPB_TABKEY_STR("STRING"), UPB_VALUE_INIT_INT32(0), &google_protobuf_strentries[185]},
   {UPB_TABKEY_STR("STRING_PIECE"), UPB_VALUE_INIT_INT32(2), NULL},
   {UPB_TABKEY_STR("CODE_SIZE"), UPB_VALUE_INIT_INT32(2), NULL},
   {UPB_TABKEY_STR("SPEED"), UPB_VALUE_INIT_INT32(1), &google_protobuf_strentries[191]},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_STR("LITE_RUNTIME"), UPB_VALUE_INIT_INT32(3), NULL},
 };
 
 const upb_tabent google_protobuf_intentries[66] = {
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_NUM(6), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[13]), NULL},
   {UPB_TABKEY_NUM(7), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[49]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_NUM(999), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[70]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_NUM(999), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[71]), NULL},
   {UPB_TABKEY_NUM(8), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[50]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_NUM(6), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[64]), NULL},
   {UPB_TABKEY_NUM(7), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[3]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_NUM(9), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[10]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_NUM(999), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[69]), NULL},
   {UPB_TABKEY_NUM(8), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[47]), NULL},
   {UPB_TABKEY_NUM(9), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[59]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_NUM(6), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[58]), NULL},
   {UPB_TABKEY_NUM(7), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[12]), NULL},
   {UPB_TABKEY_NUM(16), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[1]), NULL},
   {UPB_TABKEY_NUM(17), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[21]), NULL},
   {UPB_TABKEY_NUM(18), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[57]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_NUM(20), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[20]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_NUM(999), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[68]), NULL},
   {UPB_TABKEY_NUM(8), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[23]), NULL},
   {UPB_TABKEY_NUM(9), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[44]), NULL},
   {UPB_TABKEY_NUM(10), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[22]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_NUM(999), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[66]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_NUM(999), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[67]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_NUM(999), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[65]), NULL},
   {UPB_TABKEY_NUM(8), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[0]), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_NUM(6), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[6]), NULL},
   {UPB_TABKEY_NUM(7), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[62]), NULL},
   {UPB_TABKEY_NUM(16), UPB_VALUE_INIT_CONSTPTR("TYPE_SFIXED64"), NULL},
   {UPB_TABKEY_NUM(17), UPB_VALUE_INIT_CONSTPTR("TYPE_SINT32"), NULL},
   {UPB_TABKEY_NUM(18), UPB_VALUE_INIT_CONSTPTR("TYPE_SINT64"), NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NONE, UPB_VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_NUM(7), UPB_VALUE_INIT_CONSTPTR("TYPE_FIXED32"), NULL},
   {UPB_TABKEY_NUM(8), UPB_VALUE_INIT_CONSTPTR("TYPE_BOOL"), NULL},
   {UPB_TABKEY_NUM(9), UPB_VALUE_INIT_CONSTPTR("TYPE_STRING"), NULL},
@@ -381,7 +381,7 @@ const upb_tabent google_protobuf_intentries[66] = {
   {UPB_TABKEY_NUM(15), UPB_VALUE_INIT_CONSTPTR("TYPE_SFIXED32"), NULL},
 };
 
-const upb_value google_protobuf_arrays[97] = {
+const _upb_value google_protobuf_arrays[97] = {
   UPB_ARRAY_EMPTYENT,
   UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[36]),
   UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[15]),

upb/descriptor/descriptor.upb.h

@@ -83,6 +83,200 @@ extern const upb_enumdef google_protobuf_enums[4];
 #define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION &google_protobuf_msgs[18]
 #define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_NAMEPART &google_protobuf_msgs[19]
 
+// Selector definitions.
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_AGGREGATE_VALUE_ENDSTR 11
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_AGGREGATE_VALUE_STRING 9
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_AGGREGATE_VALUE_STARTSTR 10
+#define GOOGLE_PROTOBUF_FILEOPTIONS_CC_GENERIC_SERVICES_BOOL 3
+#define GOOGLE_PROTOBUF_FIELDOPTIONS_CTYPE_INT32 0
+#define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_DEFAULT_VALUE_ENDSTR 16
+#define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_DEFAULT_VALUE_STRING 14
+#define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_DEFAULT_VALUE_STARTSTR 15
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_DEPENDENCY_ENDSTR 10
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_DEPENDENCY_ENDSEQ 7
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_DEPENDENCY_STRING 8
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_DEPENDENCY_STARTSTR 9
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_DEPENDENCY_STARTSEQ 6
+#define GOOGLE_PROTOBUF_FIELDOPTIONS_DEPRECATED_BOOL 2
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_DOUBLE_VALUE_DOUBLE 12
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSIONRANGE_END_INT32 1
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_ENUM_TYPE_STARTSUBMSG 13
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_ENUM_TYPE_ENDSEQ 12
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_ENUM_TYPE_STARTSEQ 11
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_ENUM_TYPE_ENDSUBMSG 14
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_ENUM_TYPE_STARTSUBMSG 17
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_ENUM_TYPE_ENDSEQ 16
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_ENUM_TYPE_STARTSEQ 15
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_ENUM_TYPE_ENDSUBMSG 18
+#define GOOGLE_PROTOBUF_FIELDOPTIONS_EXPERIMENTAL_MAP_KEY_ENDSTR 5
+#define GOOGLE_PROTOBUF_FIELDOPTIONS_EXPERIMENTAL_MAP_KEY_STRING 3
+#define GOOGLE_PROTOBUF_FIELDOPTIONS_EXPERIMENTAL_MAP_KEY_STARTSTR 4
+#define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_EXTENDEE_ENDSTR 5
+#define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_EXTENDEE_STRING 3
+#define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_EXTENDEE_STARTSTR 4
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_EXTENSION_STARTSUBMSG 29
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_EXTENSION_ENDSEQ 28
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_EXTENSION_STARTSEQ 27
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_EXTENSION_ENDSUBMSG 30
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_STARTSUBMSG 21
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_ENDSEQ 20
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_STARTSEQ 19
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_ENDSUBMSG 22
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_RANGE_STARTSUBMSG 17
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_RANGE_ENDSEQ 16
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_RANGE_STARTSEQ 15
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_RANGE_ENDSUBMSG 18
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_FIELD_STARTSUBMSG 5
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_FIELD_ENDSEQ 4
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_FIELD_STARTSEQ 3
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_FIELD_ENDSUBMSG 6
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORSET_FILE_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORSET_FILE_ENDSEQ 1
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORSET_FILE_STARTSEQ 0
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORSET_FILE_ENDSUBMSG 3
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_IDENTIFIER_VALUE_ENDSTR 6
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_IDENTIFIER_VALUE_STRING 4
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_IDENTIFIER_VALUE_STARTSTR 5
+#define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_INPUT_TYPE_ENDSTR 5
+#define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_INPUT_TYPE_STRING 3
+#define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_INPUT_TYPE_STARTSTR 4
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_NAMEPART_IS_EXTENSION_BOOL 3
+#define GOOGLE_PROTOBUF_FILEOPTIONS_JAVA_GENERATE_EQUALS_AND_HASH_BOOL 6
+#define GOOGLE_PROTOBUF_FILEOPTIONS_JAVA_GENERIC_SERVICES_BOOL 4
+#define GOOGLE_PROTOBUF_FILEOPTIONS_JAVA_MULTIPLE_FILES_BOOL 15
+#define GOOGLE_PROTOBUF_FILEOPTIONS_JAVA_OUTER_CLASSNAME_ENDSTR 13
+#define GOOGLE_PROTOBUF_FILEOPTIONS_JAVA_OUTER_CLASSNAME_STRING 11
+#define GOOGLE_PROTOBUF_FILEOPTIONS_JAVA_OUTER_CLASSNAME_STARTSTR 12
+#define GOOGLE_PROTOBUF_FILEOPTIONS_JAVA_PACKAGE_ENDSTR 2
+#define GOOGLE_PROTOBUF_FILEOPTIONS_JAVA_PACKAGE_STRING 0
+#define GOOGLE_PROTOBUF_FILEOPTIONS_JAVA_PACKAGE_STARTSTR 1
+#define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_LABEL_INT32 7
+#define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_ENDSEQ 1
+#define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_STARTSEQ 0
+#define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_ENDSUBMSG 3
+#define GOOGLE_PROTOBUF_MESSAGEOPTIONS_MESSAGE_SET_WIRE_FORMAT_BOOL 0
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_MESSAGE_TYPE_STARTSUBMSG 13
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_MESSAGE_TYPE_ENDSEQ 12
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_MESSAGE_TYPE_STARTSEQ 11
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_MESSAGE_TYPE_ENDSUBMSG 14
+#define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_METHOD_STARTSUBMSG 5
+#define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_METHOD_ENDSEQ 4
+#define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_METHOD_STARTSEQ 3
+#define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_METHOD_ENDSUBMSG 6
+#define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_NAME_ENDSTR 2
+#define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_NAME_STRING 0
+#define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_NAME_STARTSTR 1
+#define GOOGLE_PROTOBUF_ENUMVALUEDESCRIPTORPROTO_NAME_ENDSTR 2
+#define GOOGLE_PROTOBUF_ENUMVALUEDESCRIPTORPROTO_NAME_STRING 0
+#define GOOGLE_PROTOBUF_ENUMVALUEDESCRIPTORPROTO_NAME_STARTSTR 1
+#define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_NAME_ENDSTR 2
+#define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_NAME_STRING 0
+#define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_NAME_STARTSTR 1
+#define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_NAME_ENDSTR 2
+#define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_NAME_STRING 0
+#define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_NAME_STARTSTR 1
+#define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_NAME_ENDSTR 2
+#define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_NAME_STRING 0
+#define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_NAME_STARTSTR 1
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_NAME_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_NAME_ENDSEQ 1
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_NAME_STARTSEQ 0
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_NAME_ENDSUBMSG 3
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_NAME_ENDSTR 2
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_NAME_STRING 0
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_NAME_STARTSTR 1
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_NAME_ENDSTR 2
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_NAME_STRING 0
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_NAME_STARTSTR 1
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_NAMEPART_NAME_PART_ENDSTR 2
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_NAMEPART_NAME_PART_STRING 0
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_NAMEPART_NAME_PART_STARTSTR 1
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_NEGATIVE_INT_VALUE_INT64 8
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_NESTED_TYPE_STARTSUBMSG 9
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_NESTED_TYPE_ENDSEQ 8
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_NESTED_TYPE_STARTSEQ 7
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_NESTED_TYPE_ENDSUBMSG 10
+#define GOOGLE_PROTOBUF_MESSAGEOPTIONS_NO_STANDARD_DESCRIPTOR_ACCESSOR_BOOL 1
+#define GOOGLE_PROTOBUF_ENUMVALUEDESCRIPTORPROTO_NUMBER_INT32 3
+#define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_NUMBER_INT32 6
+#define GOOGLE_PROTOBUF_FILEOPTIONS_OPTIMIZE_FOR_INT32 14
+#define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_OPTIONS_STARTSUBMSG 9
+#define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_OPTIONS_ENDSUBMSG 10
+#define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_OPTIONS_STARTSUBMSG 7
+#define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_OPTIONS_ENDSUBMSG 8
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_OPTIONS_STARTSUBMSG 19
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_OPTIONS_ENDSUBMSG 20
+#define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_OPTIONS_STARTSUBMSG 7
+#define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_OPTIONS_ENDSUBMSG 8
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_OPTIONS_STARTSUBMSG 23
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_OPTIONS_ENDSUBMSG 24
+#define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_OPTIONS_STARTSUBMSG 9
+#define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_OPTIONS_ENDSUBMSG 10
+#define GOOGLE_PROTOBUF_ENUMVALUEDESCRIPTORPROTO_OPTIONS_STARTSUBMSG 4
+#define GOOGLE_PROTOBUF_ENUMVALUEDESCRIPTORPROTO_OPTIONS_ENDSUBMSG 5
+#define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_OUTPUT_TYPE_ENDSTR 8
+#define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_OUTPUT_TYPE_STRING 6
+#define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_OUTPUT_TYPE_STARTSTR 7
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_PACKAGE_ENDSTR 5
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_PACKAGE_STRING 3
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_PACKAGE_STARTSTR 4
+#define GOOGLE_PROTOBUF_FIELDOPTIONS_PACKED_BOOL 1
+#define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_PATH_ENDSEQ 1
+#define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_PATH_INT32 2
+#define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_PATH_STARTSEQ 0
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_POSITIVE_INT_VALUE_UINT64 7
+#define GOOGLE_PROTOBUF_FILEOPTIONS_PY_GENERIC_SERVICES_BOOL 5
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_SERVICE_STARTSUBMSG 25
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_SERVICE_ENDSEQ 24
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_SERVICE_STARTSEQ 23
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_SERVICE_ENDSUBMSG 26
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_SOURCE_CODE_INFO_STARTSUBMSG 21
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_SOURCE_CODE_INFO_ENDSUBMSG 22
+#define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_SPAN_ENDSEQ 4
+#define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_SPAN_INT32 5
+#define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_SPAN_STARTSEQ 3
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSIONRANGE_START_INT32 0
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_STRING_VALUE_ENDSTR 15
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_STRING_VALUE_STRING 13
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_STRING_VALUE_STARTSTR 14
+#define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_INT32 8
+#define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_NAME_ENDSTR 13
+#define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_NAME_STRING 11
+#define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_NAME_STARTSTR 12
+#define GOOGLE_PROTOBUF_SERVICEOPTIONS_UNINTERPRETED_OPTION_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_SERVICEOPTIONS_UNINTERPRETED_OPTION_ENDSEQ 1
+#define GOOGLE_PROTOBUF_SERVICEOPTIONS_UNINTERPRETED_OPTION_STARTSEQ 0
+#define GOOGLE_PROTOBUF_SERVICEOPTIONS_UNINTERPRETED_OPTION_ENDSUBMSG 3
+#define GOOGLE_PROTOBUF_MESSAGEOPTIONS_UNINTERPRETED_OPTION_STARTSUBMSG 4
+#define GOOGLE_PROTOBUF_MESSAGEOPTIONS_UNINTERPRETED_OPTION_ENDSEQ 3
+#define GOOGLE_PROTOBUF_MESSAGEOPTIONS_UNINTERPRETED_OPTION_STARTSEQ 2
+#define GOOGLE_PROTOBUF_MESSAGEOPTIONS_UNINTERPRETED_OPTION_ENDSUBMSG 5
+#define GOOGLE_PROTOBUF_METHODOPTIONS_UNINTERPRETED_OPTION_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_METHODOPTIONS_UNINTERPRETED_OPTION_ENDSEQ 1
+#define GOOGLE_PROTOBUF_METHODOPTIONS_UNINTERPRETED_OPTION_STARTSEQ 0
+#define GOOGLE_PROTOBUF_METHODOPTIONS_UNINTERPRETED_OPTION_ENDSUBMSG 3
+#define GOOGLE_PROTOBUF_FILEOPTIONS_UNINTERPRETED_OPTION_STARTSUBMSG 9
+#define GOOGLE_PROTOBUF_FILEOPTIONS_UNINTERPRETED_OPTION_ENDSEQ 8
+#define GOOGLE_PROTOBUF_FILEOPTIONS_UNINTERPRETED_OPTION_STARTSEQ 7
+#define GOOGLE_PROTOBUF_FILEOPTIONS_UNINTERPRETED_OPTION_ENDSUBMSG 10
+#define GOOGLE_PROTOBUF_FIELDOPTIONS_UNINTERPRETED_OPTION_STARTSUBMSG 8
+#define GOOGLE_PROTOBUF_FIELDOPTIONS_UNINTERPRETED_OPTION_ENDSEQ 7
+#define GOOGLE_PROTOBUF_FIELDOPTIONS_UNINTERPRETED_OPTION_STARTSEQ 6
+#define GOOGLE_PROTOBUF_FIELDOPTIONS_UNINTERPRETED_OPTION_ENDSUBMSG 9
+#define GOOGLE_PROTOBUF_ENUMOPTIONS_UNINTERPRETED_OPTION_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_ENUMOPTIONS_UNINTERPRETED_OPTION_ENDSEQ 1
+#define GOOGLE_PROTOBUF_ENUMOPTIONS_UNINTERPRETED_OPTION_STARTSEQ 0
+#define GOOGLE_PROTOBUF_ENUMOPTIONS_UNINTERPRETED_OPTION_ENDSUBMSG 3
+#define GOOGLE_PROTOBUF_ENUMVALUEOPTIONS_UNINTERPRETED_OPTION_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_ENUMVALUEOPTIONS_UNINTERPRETED_OPTION_ENDSEQ 1
+#define GOOGLE_PROTOBUF_ENUMVALUEOPTIONS_UNINTERPRETED_OPTION_STARTSEQ 0
+#define GOOGLE_PROTOBUF_ENUMVALUEOPTIONS_UNINTERPRETED_OPTION_ENDSUBMSG 3
+#define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_VALUE_STARTSUBMSG 5
+#define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_VALUE_ENDSEQ 4
+#define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_VALUE_STARTSEQ 3
+#define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_VALUE_ENDSUBMSG 6
+
 #ifdef __cplusplus
 };  // extern "C"
 #endif

upb/descriptor/reader.c

@@ -16,8 +16,8 @@
 #include <errno.h>
 #include <stdlib.h>
 #include <string.h>
-#include "upb/bytestream.h"
 #include "upb/def.h"
+#include "upb/sink.h"
 #include "upb/descriptor/descriptor.upb.h"
 
 static char *upb_strndup(const char *buf, size_t n) {
@@ -28,7 +28,8 @@ static char *upb_strndup(const char *buf, size_t n) {
   return ret;
 }
 
-// Returns a newly allocated string that joins input strings together, for example:
+// Returns a newly allocated string that joins input strings together, for
+// example:
 //   join("Foo.Bar", "Baz") -> "Foo.Bar.Baz"
 //   join("", "Baz") -> "Baz"
 // Caller owns a ref on the returned string.
@@ -45,9 +46,21 @@ static char *upb_join(const char *base, const char *name) {
   }
 }
 
+
+/* upb_deflist ****************************************************************/
+
+// upb_deflist is an internal-only dynamic array for storing a growing list of
+// upb_defs.
+typedef struct {
+  upb_def **defs;
+  size_t len;
+  size_t size;
+  bool owned;
+} upb_deflist;
+
 void upb_deflist_init(upb_deflist *l) {
-  l->size = 8;
-  l->defs = malloc(l->size * sizeof(void*));
+  l->size = 0;
+  l->defs = NULL;
   l->len = 0;
   l->owned = true;
 }
@@ -56,15 +69,19 @@ void upb_deflist_uninit(upb_deflist *l) {
   if (l->owned)
     for(size_t i = 0; i < l->len; i++)
       upb_def_unref(l->defs[i], &l->defs);
-  free(l->defs);
 }
 
-void upb_deflist_push(upb_deflist *l, upb_def *d) {
-  if(l->len == l->size) {
-    l->size *= 2;
-    l->defs = realloc(l->defs, l->size * sizeof(void*));
+bool upb_deflist_push(upb_deflist *l, upb_def *d, upb_pipeline *p) {
+  if(++l->len >= l->size) {
+    size_t new_size = UPB_MAX(l->size, 4);
+    new_size *= 2;
+    l->defs = upb_pipeline_realloc(
+        p, l->defs, l->size * sizeof(void*), new_size * sizeof(void*));
+    if (!l->defs) return false;
+    l->size = new_size;
   }
-  l->defs[l->len++] = d;
+  l->defs[l->len - 1] = d;
+  return true;
 }
 
 void upb_deflist_donaterefs(upb_deflist *l, void *owner) {
@@ -74,9 +91,6 @@ void upb_deflist_donaterefs(upb_deflist *l, void *owner) {
   l->owned = false;
 }
 
-
-/* upb_descreader  ************************************************************/
-
 static upb_def *upb_deflist_last(upb_deflist *l) {
   return l->defs[l->len-1];
 }
@@ -91,17 +105,67 @@ static void upb_deflist_qualify(upb_deflist *l, char *str, int32_t start) {
   }
 }
 
-void upb_descreader_init(upb_descreader *r) {
+
+/* upb_descreader  ************************************************************/
+
+// We keep a stack of all the messages scopes we are currently in, as well as
+// the top-level file scope.  This is necessary to correctly qualify the
+// definitions that are contained inside.  "name" tracks the name of the
+// message or package (a bare name -- not qualified by any enclosing scopes).
+typedef struct {
+  char *name;
+  // Index of the first def that is under this scope.  For msgdefs, the
+  // msgdef itself is at start-1.
+  int start;
+} upb_descreader_frame;
+
+struct upb_descreader {
+  upb_deflist defs;
+  upb_descreader_frame stack[UPB_MAX_TYPE_DEPTH];
+  int stack_len;
+
+  uint32_t number;
+  char *name;
+  bool saw_number;
+  bool saw_name;
+
+  char *default_string;
+
+  upb_fielddef *f;
+};
+
+void upb_descreader_init(void *r);
+void upb_descreader_uninit(void *r);
+
+const upb_frametype upb_descreader_frametype = {
+  sizeof(upb_descreader),
+  upb_descreader_init,
+  upb_descreader_uninit,
+  NULL,
+};
+
+const upb_frametype *upb_descreader_getframetype() {
+  return &upb_descreader_frametype;
+}
+
+// Registers handlers that will build the defs.  Pass the descreader as the
+// closure.
+const upb_handlers *upb_descreader_gethandlers(const void *owner);
+
+
+/* upb_descreader  ************************************************************/
+
+void upb_descreader_init(void *_r) {
+  upb_descreader *r = _r;
   upb_deflist_init(&r->defs);
-  upb_status_init(&r->status);
   r->stack_len = 0;
   r->name = NULL;
   r->default_string = NULL;
 }
 
-void upb_descreader_uninit(upb_descreader *r) {
+void upb_descreader_uninit(void *_r) {
+  upb_descreader *r = _r;
   free(r->name);
-  upb_status_uninit(&r->status);
   upb_deflist_uninit(&r->defs);
   free(r->default_string);
   while (r->stack_len > 0) {
@@ -149,37 +213,37 @@ void upb_descreader_setscopename(upb_descreader *r, char *str) {
 }
 
 // Handlers for google.protobuf.FileDescriptorProto.
-static bool file_startmsg(void *_r) {
-  upb_descreader *r = _r;
+static bool file_startmsg(const upb_sinkframe *frame) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
   upb_descreader_startcontainer(r);
   return true;
 }
 
-static void file_endmsg(void *_r, upb_status *status) {
+static void file_endmsg(const upb_sinkframe *frame, upb_status *status) {
   UPB_UNUSED(status);
-  upb_descreader *r = _r;
+  upb_descreader *r = upb_sinkframe_userdata(frame);
   upb_descreader_endcontainer(r);
 }
 
-static size_t file_onpackage(void *_r, void *fval, const char *buf, size_t n) {
-  UPB_UNUSED(fval);
-  upb_descreader *r = _r;
+static size_t file_onpackage(const upb_sinkframe *frame,
+                             const char *buf, size_t n) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
   // XXX: see comment at the top of the file.
   upb_descreader_setscopename(r, upb_strndup(buf, n));
   return n;
 }
 
 // Handlers for google.protobuf.EnumValueDescriptorProto.
-static bool enumval_startmsg(void *_r) {
-  upb_descreader *r = _r;
+static bool enumval_startmsg(const upb_sinkframe *frame) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
   r->saw_number = false;
   r->saw_name = false;
   return true;
 }
 
-static size_t enumval_onname(void *_r, void *fval, const char *buf, size_t n) {
-  UPB_UNUSED(fval);
-  upb_descreader *r = _r;
+static size_t enumval_onname(const upb_sinkframe *frame,
+                             const char *buf, size_t n) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
   // XXX: see comment at the top of the file.
   free(r->name);
   r->name = upb_strndup(buf, n);
@@ -187,16 +251,15 @@ static size_t enumval_onname(void *_r, void *fval, const char *buf, size_t n) {
   return n;
 }
 
-static bool enumval_onnumber(void *_r, void *fval, int32_t val) {
-  UPB_UNUSED(fval);
-  upb_descreader *r = _r;
+static bool enumval_onnumber(const upb_sinkframe *frame, int32_t val) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
   r->number = val;
   r->saw_number = true;
   return true;
 }
 
-static void enumval_endmsg(void *_r, upb_status *status) {
-  upb_descreader *r = _r;
+static void enumval_endmsg(const upb_sinkframe *frame, upb_status *status) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
   if(!r->saw_number || !r->saw_name) {
     upb_status_seterrliteral(status, "Enum value missing name or number.");
     return;
@@ -214,16 +277,17 @@ static void enumval_endmsg(void *_r, upb_status *status) {
 
 
 // Handlers for google.protobuf.EnumDescriptorProto.
-static bool enum_startmsg(void *_r) {
-  upb_descreader *r = _r;
-  upb_deflist_push(&r->defs, upb_upcast(upb_enumdef_new(&r->defs)));
+static bool enum_startmsg(const upb_sinkframe *frame) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
+  upb_pipeline *p = upb_sinkframe_pipeline(frame);
+  upb_deflist_push(&r->defs, upb_upcast(upb_enumdef_new(&r->defs)), p);
   return true;
 }
 
-static void enum_endmsg(void *_r, upb_status *status) {
-  upb_descreader *r = _r;
+static void enum_endmsg(const upb_sinkframe *frame, upb_status *status) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
   upb_enumdef *e = upb_downcast_enumdef_mutable(upb_descreader_last(r));
-  if (upb_def_fullname(upb_descreader_last((upb_descreader*)_r)) == NULL) {
+  if (upb_def_fullname(upb_descreader_last(r)) == NULL) {
     upb_status_seterrliteral(status, "Enum had no name.");
     return;
   }
@@ -233,9 +297,9 @@ static void enum_endmsg(void *_r, upb_status *status) {
   }
 }
 
-static size_t enum_onname(void *_r, void *fval, const char *buf, size_t n) {
-  UPB_UNUSED(fval);
-  upb_descreader *r = _r;
+static size_t enum_onname(const upb_sinkframe *frame,
+                          const char *buf, size_t n) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
   // XXX: see comment at the top of the file.
   char *fullname = upb_strndup(buf, n);
   upb_def_setfullname(upb_descreader_last(r), fullname);
@@ -244,8 +308,8 @@ static size_t enum_onname(void *_r, void *fval, const char *buf, size_t n) {
 }
 
 // Handlers for google.protobuf.FieldDescriptorProto
-static bool field_startmsg(void *_r) {
-  upb_descreader *r = _r;
+static bool field_startmsg(const upb_sinkframe *frame) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
   r->f = upb_fielddef_new(&r->defs);
   free(r->default_string);
   r->default_string = NULL;
@@ -258,27 +322,19 @@ static bool parse_default(char *str, upb_value *d, int type) {
   bool success = true;
   if (str) {
     switch(type) {
-      case UPB_TYPE(INT32):
-      case UPB_TYPE(SINT32):
-      case UPB_TYPE(SFIXED32): upb_value_setint32(d, 0); break;
-      case UPB_TYPE(INT64):
-      case UPB_TYPE(SINT64):
-      case UPB_TYPE(SFIXED64): upb_value_setint64(d, 0); break;
-      case UPB_TYPE(UINT32):
-      case UPB_TYPE(FIXED32): upb_value_setuint32(d, 0);
-      case UPB_TYPE(UINT64):
-      case UPB_TYPE(FIXED64): upb_value_setuint64(d, 0); break;
-      case UPB_TYPE(DOUBLE): upb_value_setdouble(d, 0); break;
-      case UPB_TYPE(FLOAT): upb_value_setfloat(d, 0); break;
-      case UPB_TYPE(BOOL): upb_value_setbool(d, false); break;
+      case UPB_TYPE_INT32: upb_value_setint32(d, 0); break;
+      case UPB_TYPE_INT64: upb_value_setint64(d, 0); break;
+      case UPB_TYPE_UINT32: upb_value_setuint32(d, 0);
+      case UPB_TYPE_UINT64: upb_value_setuint64(d, 0); break;
+      case UPB_TYPE_FLOAT: upb_value_setfloat(d, 0); break;
+      case UPB_TYPE_DOUBLE: upb_value_setdouble(d, 0); break;
+      case UPB_TYPE_BOOL: upb_value_setbool(d, false); break;
       default: abort();
     }
   } else {
     char *end;
     switch (type) {
-      case UPB_TYPE(INT32):
-      case UPB_TYPE(SINT32):
-      case UPB_TYPE(SFIXED32): {
+      case UPB_TYPE_INT32: {
         long val = strtol(str, &end, 0);
         if (val > INT32_MAX || val < INT32_MIN || errno == ERANGE || *end)
           success = false;
@@ -286,14 +342,11 @@ static bool parse_default(char *str, upb_value *d, int type) {
           upb_value_setint32(d, val);
         break;
       }
-      case UPB_TYPE(INT64):
-      case UPB_TYPE(SINT64):
-      case UPB_TYPE(SFIXED64):
+      case UPB_TYPE_INT64:
         upb_value_setint64(d, strtoll(str, &end, 0));
         if (errno == ERANGE || *end) success = false;
         break;
-      case UPB_TYPE(UINT32):
-      case UPB_TYPE(FIXED32): {
+      case UPB_TYPE_UINT32: {
         unsigned long val = strtoul(str, &end, 0);
         if (val > UINT32_MAX || errno == ERANGE || *end)
           success = false;
@@ -301,20 +354,19 @@ static bool parse_default(char *str, upb_value *d, int type) {
           upb_value_setuint32(d, val);
         break;
       }
-      case UPB_TYPE(UINT64):
-      case UPB_TYPE(FIXED64):
+      case UPB_TYPE_UINT64:
         upb_value_setuint64(d, strtoull(str, &end, 0));
         if (errno == ERANGE || *end) success = false;
         break;
-      case UPB_TYPE(DOUBLE):
+      case UPB_TYPE_DOUBLE:
         upb_value_setdouble(d, strtod(str, &end));
         if (errno == ERANGE || *end) success = false;
         break;
-      case UPB_TYPE(FLOAT):
+      case UPB_TYPE_FLOAT:
         upb_value_setfloat(d, strtof(str, &end));
         if (errno == ERANGE || *end) success = false;
         break;
-      case UPB_TYPE(BOOL): {
+      case UPB_TYPE_BOOL: {
         if (strcmp(str, "false") == 0)
           upb_value_setbool(d, false);
         else if (strcmp(str, "true") == 0)
@@ -328,8 +380,8 @@ static bool parse_default(char *str, upb_value *d, int type) {
   return success;
 }
 
-static void field_endmsg(void *_r, upb_status *status) {
-  upb_descreader *r = _r;
+static void field_endmsg(const upb_sinkframe *frame, upb_status *status) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
   upb_fielddef *f = r->f;
   // TODO: verify that all required fields were present.
   assert(upb_fielddef_number(f) != 0 && upb_fielddef_name(f) != NULL);
@@ -340,7 +392,7 @@ static void field_endmsg(void *_r, upb_status *status) {
       upb_status_seterrliteral(status, "Submessages cannot have defaults.");
       return;
     }
-    if (upb_fielddef_isstring(f) || upb_fielddef_type(f) == UPB_TYPE(ENUM)) {
+    if (upb_fielddef_isstring(f) || upb_fielddef_type(f) == UPB_TYPE_ENUM) {
       upb_fielddef_setdefaultcstr(f, r->default_string);
     } else {
       upb_value val;
@@ -356,30 +408,27 @@ static void field_endmsg(void *_r, upb_status *status) {
   }
 }
 
-static bool field_ontype(void *_r, void *fval, int32_t val) {
-  UPB_UNUSED(fval);
-  upb_descreader *r = _r;
-  upb_fielddef_settype(r->f, val);
+static bool field_ontype(const upb_sinkframe *frame, int32_t val) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
+  upb_fielddef_setdescriptortype(r->f, val);
   return true;
 }
 
-static bool field_onlabel(void *_r, void *fval, int32_t val) {
-  UPB_UNUSED(fval);
-  upb_descreader *r = _r;
+static bool field_onlabel(const upb_sinkframe *frame, int32_t val) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
   upb_fielddef_setlabel(r->f, val);
   return true;
 }
 
-static bool field_onnumber(void *_r, void *fval, int32_t val) {
-  UPB_UNUSED(fval);
-  upb_descreader *r = _r;
+static bool field_onnumber(const upb_sinkframe *frame, int32_t val) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
   upb_fielddef_setnumber(r->f, val);
   return true;
 }
 
-static size_t field_onname(void *_r, void *fval, const char *buf, size_t n) {
-  UPB_UNUSED(fval);
-  upb_descreader *r = _r;
+static size_t field_onname(const upb_sinkframe *frame,
+                           const char *buf, size_t n) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
   // XXX: see comment at the top of the file.
   char *name = upb_strndup(buf, n);
   upb_fielddef_setname(r->f, name);
@@ -387,10 +436,9 @@ static size_t field_onname(void *_r, void *fval, const char *buf, size_t n) {
   return n;
 }
 
-static size_t field_ontypename(void *_r, void *fval, const char *buf,
-                               size_t n) {
-  UPB_UNUSED(fval);
-  upb_descreader *r = _r;
+static size_t field_ontypename(const upb_sinkframe *frame,
+                               const char *buf, size_t n) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
   // XXX: see comment at the top of the file.
   char *name = upb_strndup(buf, n);
   upb_fielddef_setsubdefname(r->f, name);
@@ -398,10 +446,9 @@ static size_t field_ontypename(void *_r, void *fval, const char *buf,
   return n;
 }
 
-static size_t field_ondefaultval(void *_r, void *fval, const char *buf,
-                                 size_t n) {
-  UPB_UNUSED(fval);
-  upb_descreader *r = _r;
+static size_t field_ondefaultval(const upb_sinkframe *frame,
+                                 const char *buf, size_t n) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
   // Have to convert from string to the correct type, but we might not know the
   // type yet, so we save it as a string until the end of the field.
   // XXX: see comment at the top of the file.
@@ -411,15 +458,16 @@ static size_t field_ondefaultval(void *_r, void *fval, const char *buf,
 }
 
 // Handlers for google.protobuf.DescriptorProto (representing a message).
-static bool msg_startmsg(void *_r) {
-  upb_descreader *r = _r;
-  upb_deflist_push(&r->defs, upb_upcast(upb_msgdef_new(&r->defs)));
+static bool msg_startmsg(const upb_sinkframe *frame) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
+  upb_pipeline *p = upb_sinkframe_pipeline(frame);
+  upb_deflist_push(&r->defs, upb_upcast(upb_msgdef_new(&r->defs)), p);
   upb_descreader_startcontainer(r);
   return true;
 }
 
-static void msg_endmsg(void *_r, upb_status *status) {
-  upb_descreader *r = _r;
+static void msg_endmsg(const upb_sinkframe *frame, upb_status *status) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
   upb_msgdef *m = upb_descreader_top(r);
   if(!upb_def_fullname(upb_upcast(m))) {
     upb_status_seterrliteral(status, "Encountered message with no name.");
@@ -428,9 +476,9 @@ static void msg_endmsg(void *_r, upb_status *status) {
   upb_descreader_endcontainer(r);
 }
 
-static size_t msg_onname(void *_r, void *fval, const char *buf, size_t n) {
-  UPB_UNUSED(fval);
-  upb_descreader *r = _r;
+static size_t msg_onname(const upb_sinkframe *frame,
+                         const char *buf, size_t n) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
   upb_msgdef *m = upb_descreader_top(r);
   // XXX: see comment at the top of the file.
   char *name = upb_strndup(buf, n);
@@ -439,18 +487,16 @@ static size_t msg_onname(void *_r, void *fval, const char *buf, size_t n) {
   return n;
 }
 
-static bool msg_onendfield(void *_r, void *fval) {
-  UPB_UNUSED(fval);
-  upb_descreader *r = _r;
+static bool msg_onendfield(const upb_sinkframe *frame) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
   upb_msgdef *m = upb_descreader_top(r);
   upb_msgdef_addfield(m, r->f, &r->defs);
   r->f = NULL;
   return true;
 }
 
-static bool discardfield(void *_r, void *fval) {
-  UPB_UNUSED(fval);
-  upb_descreader *r = _r;
+static bool discardfield(const upb_sinkframe *frame) {
+  upb_descreader *r = upb_sinkframe_userdata(frame);
   // Discard extension field so we don't leak it.
   upb_fielddef_unref(r->f, &r->defs);
   r->f = NULL;
@@ -496,7 +542,8 @@ static void reghandlers(void *closure, upb_handlers *h) {
   }
 }
 
-const upb_handlers *upb_descreader_newhandlers(const void *owner) {
+const upb_handlers *upb_descreader_gethandlers(const void *owner) {
   return upb_handlers_newfrozen(
-      GOOGLE_PROTOBUF_FILEDESCRIPTORSET, owner, reghandlers, NULL);
+      GOOGLE_PROTOBUF_FILEDESCRIPTORSET, &upb_descreader_frametype,
+      owner, reghandlers, NULL);
 }

upb/descriptor/reader.h

@@ -4,9 +4,8 @@
  * Copyright (c) 2011 Google Inc.  See LICENSE for details.
  * Author: Josh Haberman <jhaberman@gmail.com>
  *
- * upb_descreader provides a set of sink handlers that will build defs from a
- * data source that uses the descriptor.proto schema (like a protobuf binary
- * descriptor).
+ * upb::descriptor::Reader provides a way of building upb::Defs from
+ * data in descriptor.proto format.
  */
 
 #ifndef UPB_DESCRIPTOR_H
@@ -15,69 +14,62 @@
 #include "upb/handlers.h"
 
 #ifdef __cplusplus
-extern "C" {
-#endif
+namespace upb {
+namespace descriptor {
 
-/* upb_deflist ****************************************************************/
-
-// upb_deflist is an internal-only dynamic array for storing a growing list of
-// upb_defs.
-typedef struct {
-  upb_def **defs;
-  size_t len;
-  size_t size;
-  bool owned;
-} upb_deflist;
-
-void upb_deflist_init(upb_deflist *l);
-void upb_deflist_uninit(upb_deflist *l);
-void upb_deflist_push(upb_deflist *l, upb_def *d);
-
-/* upb_descreader  ************************************************************/
-
-// We keep a stack of all the messages scopes we are currently in, as well as
-// the top-level file scope.  This is necessary to correctly qualify the
-// definitions that are contained inside.  "name" tracks the name of the
-// message or package (a bare name -- not qualified by any enclosing scopes).
-typedef struct {
-  char *name;
-  // Index of the first def that is under this scope.  For msgdefs, the
-  // msgdef itself is at start-1.
-  int start;
-} upb_descreader_frame;
-
-typedef struct {
-  upb_deflist defs;
-  upb_descreader_frame stack[UPB_MAX_TYPE_DEPTH];
-  int stack_len;
-  upb_status status;
-
-  uint32_t number;
-  char *name;
-  bool saw_number;
-  bool saw_name;
-
-  char *default_string;
-
-  upb_fielddef *f;
-} upb_descreader;
-
-void upb_descreader_init(upb_descreader *r);
-void upb_descreader_uninit(upb_descreader *r);
-
-// Registers handlers that will build the defs.  Pass the descreader as the
-// closure.
-const upb_handlers *upb_descreader_newhandlers(const void *owner);
+// Frame type that accumulates defs as they are being built from a descriptor
+// according to the descriptor.proto schema.
+class Reader;
 
 // Gets the array of defs that have been parsed and removes them from the
 // descreader.  Ownership of the defs is passed to the caller using the given
 // owner), but the ownership of the returned array is retained and is
 // invalidated by any other call into the descreader.  The defs will not have
 // been resolved, and are ready to be added to a symtab.
+inline upb::Def** GetDefs(Reader* r, void* owner, int* n);
+
+// Gets the handlers for reading a FileDescriptorSet, which builds defs and
+// accumulates them in a Reader object (which the handlers use as their
+// FrameType).
+inline const upb::Handlers* GetReaderHandlers(const void* owner);
+
+}  // namespace descriptor
+}  // namespace upb
+
+typedef upb::descriptor::Reader upb_descreader;
+
+extern "C" {
+#else
+struct upb_descreader;
+typedef struct upb_descreader upb_descreader;
+#endif
+
+// C API.
+const upb_frametype *upb_descreader_getframetype();
 upb_def **upb_descreader_getdefs(upb_descreader *r, void *owner, int *n);
+const upb_handlers *upb_descreader_gethandlers(const void *owner);
+
+
+// C++ implementation details. /////////////////////////////////////////////////
 
 #ifdef __cplusplus
-}  /* extern "C" */
+}  // extern "C"
+
+namespace upb {
+
+template<> inline const FrameType* GetFrameType<upb::descriptor::Reader>() {
+  return upb_descreader_getframetype();
+}
+
+namespace descriptor {
+inline upb::Def** GetDefs(Reader* r, void* owner, int* n) {
+  return upb_descreader_getdefs(r, owner, n);
+}
+inline const upb::Handlers* GetReaderHandlers(const void* owner) {
+  return upb_descreader_gethandlers(owner);
+}
+}  // namespace descriptor
+}  // namespace upb
 #endif
 
-#endif
+#endif  // UPB_DESCRIPTOR_H

upb/google/bridge.cc

@@ -47,11 +47,15 @@ class me::Defs {
       const upb::FieldDef* upb_f = i.field();
       const goog::FieldDescriptor* proto2_f =
           d->FindFieldByNumber(upb_f->number());
+      if (!proto2_f) {
+        proto2_f = d->file()->pool()->FindExtensionByNumber(d, upb_f->number());
+      }
+      assert(proto2_f);
       if (!upb::google::TrySetWriteHandlers(proto2_f, m, upb_f, h)
 #ifdef UPB_GOOGLE3
           && !upb::google::TrySetProto1WriteHandlers(proto2_f, m, upb_f, h)
 #endif
-         ) {
+          ) {
         // Unsupported reflection class.
         //
         // Should we fall back to using the public Reflection interface in this
@@ -62,7 +66,7 @@ class me::Defs {
     }
   }
 
-  static void StaticOnMessage(void *closure, upb::Handlers* handlers) {
+  static void StaticOnMessage(void* closure, upb::Handlers* handlers) {
     me::Defs* defs = static_cast<me::Defs*>(closure);
     defs->OnMessage(handlers);
   }
@@ -121,51 +125,53 @@ FieldDef* AddFieldDef(const goog::Message& m, const goog::FieldDescriptor* f,
   upb_f->set_number(f->number());
   upb_f->set_name(f->name());
   upb_f->set_label(static_cast<upb::FieldDef::Label>(f->label()));
-  upb_f->set_type(weak_prototype ?
-      UPB_TYPE_MESSAGE : static_cast<upb::FieldDef::Type>(f->type()));
+  upb_f->set_descriptor_type(
+      weak_prototype ? UPB_DESCRIPTOR_TYPE_MESSAGE :
+          static_cast<upb::FieldDef::DescriptorType>(f->type()));
 
   if (weak_prototype) {
     upb_f->set_subdef_name(weak_prototype->GetDescriptor()->full_name());
-  } else if (upb_f->IsSubMessage()) {
-    upb_f->set_subdef_name(f->message_type()->full_name());
-  } else if (upb_f->type() == UPB_TYPE(ENUM)) {
-    // We set the enum default numerically.
-    upb_f->set_default_value(
-        MakeValue(static_cast<int32_t>(f->default_value_enum()->number())));
-    upb_f->set_subdef_name(f->enum_type()->full_name());
   } else {
-    // Set field default for primitive types.  Need to switch on the upb type
-    // rather than the proto2 type, because upb_f->type() may have been changed
-    // from BYTES to MESSAGE for a weak field.
-    switch (upb_types[upb_f->type()].inmemory_type) {
-      case UPB_CTYPE_INT32:
+    switch (upb_f->type()) {
+      case UPB_TYPE_INT32:
         upb_f->set_default_value(MakeValue(f->default_value_int32()));
         break;
-      case UPB_CTYPE_INT64:
+      case UPB_TYPE_INT64:
         upb_f->set_default_value(
             MakeValue(static_cast<int64_t>(f->default_value_int64())));
         break;
-      case UPB_CTYPE_UINT32:
+      case UPB_TYPE_UINT32:
         upb_f->set_default_value(MakeValue(f->default_value_uint32()));
         break;
-      case UPB_CTYPE_UINT64:
+      case UPB_TYPE_UINT64:
         upb_f->set_default_value(
             MakeValue(static_cast<uint64_t>(f->default_value_uint64())));
         break;
-      case UPB_CTYPE_DOUBLE:
+      case UPB_TYPE_DOUBLE:
         upb_f->set_default_value(MakeValue(f->default_value_double()));
         break;
-      case UPB_CTYPE_FLOAT:
+      case UPB_TYPE_FLOAT:
         upb_f->set_default_value(MakeValue(f->default_value_float()));
         break;
-      case UPB_CTYPE_BOOL:
+      case UPB_TYPE_BOOL:
         upb_f->set_default_value(MakeValue(f->default_value_bool()));
         break;
-      case UPB_CTYPE_BYTEREGION:
+      case UPB_TYPE_STRING:
+      case UPB_TYPE_BYTES:
         upb_f->set_default_string(f->default_value_string());
         break;
+      case UPB_TYPE_MESSAGE:
+        upb_f->set_subdef_name(f->message_type()->full_name());
+        break;
+      case UPB_TYPE_ENUM:
+        // We set the enum default numerically.
+        upb_f->set_default_value(
+            MakeValue(static_cast<int32_t>(f->default_value_enum()->number())));
+        upb_f->set_subdef_name(f->enum_type()->full_name());
+        break;
     }
   }
+
   bool ok = md->AddField(upb_f, &upb_f);
   UPB_ASSERT_VAR(ok, ok);
 
@@ -174,8 +180,7 @@ FieldDef* AddFieldDef(const goog::Message& m, const goog::FieldDescriptor* f,
   } else if (f->cpp_type() == goog::FieldDescriptor::CPPTYPE_MESSAGE) {
     *subm = upb::google::GetFieldPrototype(m, f);
 #ifdef UPB_GOOGLE3
-    if (!*subm)
-      *subm = upb::google::GetProto1FieldPrototype(m, f);
+    if (!*subm) *subm = upb::google::GetProto1FieldPrototype(m, f);
 #endif
     assert(*subm);
   }
@@ -183,7 +188,7 @@ FieldDef* AddFieldDef(const goog::Message& m, const goog::FieldDescriptor* f,
   return upb_f;
 }
 
-upb::EnumDef* NewEnumDef(const goog::EnumDescriptor* desc, void *owner) {
+upb::EnumDef* NewEnumDef(const goog::EnumDescriptor* desc, const void* owner) {
   upb::EnumDef* e = upb::EnumDef::New(owner);
   e->set_full_name(desc->full_name());
   for (int i = 0; i < desc->value_count(); i++) {
@@ -194,25 +199,28 @@ upb::EnumDef* NewEnumDef(const goog::EnumDescriptor* desc, void *owner) {
   return e;
 }
 
-static upb::MessageDef* NewMessageDef(const goog::Message& m, void *owner,
+static upb::MessageDef* NewMessageDef(const goog::Message& m, const void* owner,
                                       me::Defs* defs) {
   upb::MessageDef* md = upb::MessageDef::New(owner);
+  const goog::Descriptor* d = m.GetDescriptor();
   md->set_full_name(m.GetDescriptor()->full_name());
 
   // Must do this before processing submessages to prevent infinite recursion.
   defs->AddMessage(&m, md);
 
-  const goog::Descriptor* d = m.GetDescriptor();
+  vector<const goog::FieldDescriptor*> fields;
+  d->file()->pool()->FindAllExtensions(d, &fields);
   for (int i = 0; i < d->field_count(); i++) {
-    const goog::FieldDescriptor* proto2_f = d->field(i);
+    fields.push_back(d->field(i));
+  }
 
+  for (int i = 0; i < fields.size(); i++) {
+    const goog::FieldDescriptor* proto2_f = fields[i];
+    assert(proto2_f);
 #ifdef UPB_GOOGLE3
     // Skip lazy fields for now since we can't properly handle them.
     if (proto2_f->options().lazy()) continue;
 #endif
-    // Extensions not supported yet.
-    if (proto2_f->is_extension()) continue;
-
     const goog::Message* subm_prototype;
     upb::FieldDef* f = AddFieldDef(m, proto2_f, md, &subm_prototype);
 
@@ -220,7 +228,7 @@ static upb::MessageDef* NewMessageDef(const goog::Message& m, void *owner,
 
     upb::Def* subdef = defs->FindSymbol(f->subdef_name());
     if (!subdef) {
-      if (f->type() == UPB_TYPE(ENUM)) {
+      if (f->type() == UPB_TYPE_ENUM) {
         subdef = NewEnumDef(proto2_f->enum_type(), owner)->Upcast();
         defs->AddSymbol(subdef->full_name(), subdef);
       } else {
@@ -231,11 +239,11 @@ static upb::MessageDef* NewMessageDef(const goog::Message& m, void *owner,
     }
     f->set_subdef(subdef);
   }
-
   return md;
 }
 
-const upb::Handlers* NewWriteHandlers(const goog::Message& m, void *owner) {
+const upb::Handlers* NewWriteHandlers(const goog::Message& m,
+                                      const void* owner) {
   me::Defs defs;
   const upb::MessageDef* md = NewMessageDef(m, owner, &defs);
 
@@ -245,8 +253,8 @@ const upb::Handlers* NewWriteHandlers(const goog::Message& m, void *owner) {
   bool success = Def::Freeze(defs_vec, &status);
   UPB_ASSERT_VAR(success, success);
 
-  const upb::Handlers* ret =
-      upb::Handlers::NewFrozen(md, owner, me::Defs::StaticOnMessage, &defs);
+  const upb::Handlers* ret = upb::Handlers::NewFrozen(
+      md, NULL, owner, me::Defs::StaticOnMessage, &defs);
 
   // Unref all defs, since they're now ref'd by the handlers.
   for (int i = 0; i < static_cast<int>(defs_vec.size()); i++) {

upb/google/bridge.h

@@ -66,9 +66,10 @@ namespace google {
 //
 // TODO(haberman): Add handler caching functionality so that we don't use
 // O(n^2) memory in the worst case when incrementally building handlers.
-const upb::Handlers* NewWriteHandlers(const proto2::Message& m, void *owner);
+const upb::Handlers* NewWriteHandlers(const proto2::Message& m,
+                                      const void* owner);
 const upb::Handlers* NewWriteHandlers(const ::google::protobuf::Message& m,
-                                      void *owner);
+                                      const void* owner);
 
 }  // namespace google
 }  // namespace upb

upb/google/cord.h

@@ -1,48 +0,0 @@
-//
-// upb - a minimalist implementation of protocol buffers.
-//
-// Copyright (c) 2011-2012 Google Inc.  See LICENSE for details.
-// Author: Josh Haberman <jhaberman@gmail.com>
-//
-// Functionality for interoperating with Cord.  Only needed inside Google.
-
-#ifndef UPB_GOOGLE_CORD_H
-#define UPB_GOOGLE_CORD_H
-
-#include "strings/cord.h"
-#include "upb/bytestream.h"
-
-namespace upb {
-
-namespace proto2_bridge_google3 { class FieldAccessor; }
-namespace proto2_bridge_opensource { class FieldAccessor; }
-
-namespace google {
-
-class P2R_Handlers;
-
-class CordSupport {
- private:
-  UPB_DISALLOW_POD_OPS(CordSupport);
-
-  inline static void AssignToCord(const upb::ByteRegion* r, Cord* cord) {
-    // TODO(haberman): ref source data if source is a cord.
-    cord->Clear();
-    uint64_t ofs = r->start_ofs();
-    while (ofs < r->end_ofs()) {
-      size_t len;
-      const char *buf = r->GetPtr(ofs, &len);
-      cord->Append(StringPiece(buf, len));
-      ofs += len;
-    }
-  }
-
-  friend class ::upb::proto2_bridge_google3::FieldAccessor;
-  friend class ::upb::proto2_bridge_opensource::FieldAccessor;
-  friend class P2R_Handlers;
-};
-
-}  // namespace google
-}  // namespace upb
-
-#endif  // UPB_GOOGLE_CORD_H

upb/google/proto1.cc

@@ -18,23 +18,14 @@
 
 #include "upb/google/proto1.h"
 
-// TODO(haberman): friend upb so that this isn't required.
-#define protected public
 #include "net/proto2/public/repeated_field.h"
-#undef private
-
-// TODO(haberman): friend upb so that this isn't required.
-#define private public
-#include "net/proto/proto2_reflection.h"
-#undef private
-
 #include "net/proto/internal_layout.h"
-#include "upb/bytestream.h"
+#include "net/proto/proto2_reflection.h"
 #include "upb/def.h"
-#include "upb/google/cord.h"
 #include "upb/handlers.h"
+#include "upb/sink.h"
 
-template<class T> static T* GetPointer(void *message, size_t offset) {
+template <class T> static T* GetPointer(void* message, size_t offset) {
   return reinterpret_cast<T*>(static_cast<char*>(message) + offset);
 }
 
@@ -47,31 +38,32 @@ class P2R_Handlers {
   // of the FieldDef that are necessary to read/write this field to a
   // proto2::Message.
   static bool TrySet(const proto2::FieldDescriptor* proto2_f,
-                     const proto2::Message& m,
-                     const upb::FieldDef* upb_f, upb::Handlers* h) {
+                     const proto2::Message& m, const upb::FieldDef* upb_f,
+                     upb::Handlers* h) {
     const proto2::Reflection* base_r = m.GetReflection();
     // See file comment re: dynamic_cast.
     const _pi::Proto2Reflection* r =
         dynamic_cast<const _pi::Proto2Reflection*>(base_r);
     if (!r) return false;
-    // Extensions not supported yet.
-    if (proto2_f->is_extension()) return false;
+    // Extensions don't exist in proto1.
+    assert(!proto2_f->is_extension());
+
+#define PRIMITIVE(name, type_name)                                             \
+  case _pi::CREP_REQUIRED_##name:                                              \
+  case _pi::CREP_OPTIONAL_##name:                                              \
+  case _pi::CREP_REPEATED_##name:                                              \
+    SetPrimitiveHandlers<type_name>(proto2_f, r, upb_f, h);                    \
+    return true;
 
     switch (r->GetFieldLayout(proto2_f)->crep) {
-#define PRIMITIVE(name, type_name) \
-      case _pi::CREP_REQUIRED_ ## name: \
-      case _pi::CREP_OPTIONAL_ ## name: \
-      case _pi::CREP_REPEATED_ ## name: \
-        SetPrimitiveHandlers<type_name>(proto2_f, r, upb_f, h); return true;
-      PRIMITIVE(DOUBLE,   double);
-      PRIMITIVE(FLOAT,    float);
-      PRIMITIVE(INT64,    int64_t);
-      PRIMITIVE(UINT64,   uint64_t);
-      PRIMITIVE(INT32,    int32_t);
-      PRIMITIVE(FIXED64,  uint64_t);
-      PRIMITIVE(FIXED32,  uint32_t);
-      PRIMITIVE(BOOL,     bool);
-#undef PRIMITIVE
+      PRIMITIVE(DOUBLE, double);
+      PRIMITIVE(FLOAT, float);
+      PRIMITIVE(INT64, int64_t);
+      PRIMITIVE(UINT64, uint64_t);
+      PRIMITIVE(INT32, int32_t);
+      PRIMITIVE(FIXED64, uint64_t);
+      PRIMITIVE(FIXED32, uint32_t);
+      PRIMITIVE(BOOL, bool);
       case _pi::CREP_REQUIRED_STRING:
       case _pi::CREP_OPTIONAL_STRING:
       case _pi::CREP_REPEATED_STRING:
@@ -102,16 +94,19 @@ class P2R_Handlers {
       case _pi::CREP_OPTIONAL_FOREIGN_WEAK_PROTO2:
         SetWeakMessageHandlers(proto2_f, m, r, upb_f, h);
         return true;
-      default: assert(false); return false;
+      default:
+        assert(false);
+        return false;
     }
   }
 
+#undef PRIMITIVE
+
   // If the field "f" in the message "m" is a weak field, returns the prototype
   // of the submessage (which may be a specific type or may be OpaqueMessage).
   // Otherwise returns NULL.
   static const proto2::Message* GetWeakPrototype(
-      const proto2::Message& m,
-      const proto2::FieldDescriptor* f) {
+      const proto2::Message& m, const proto2::FieldDescriptor* f) {
     // See file comment re: dynamic_cast.
     const _pi::Proto2Reflection* r =
         dynamic_cast<const _pi::Proto2Reflection*>(m.GetReflection());
@@ -132,8 +127,7 @@ class P2R_Handlers {
   // the submessage (which may be OpaqueMessage for a weak field that is not
   // linked in).  Otherwise returns NULL.
   static const proto2::Message* GetFieldPrototype(
-      const proto2::Message& m,
-      const proto2::FieldDescriptor* f) {
+      const proto2::Message& m, const proto2::FieldDescriptor* f) {
     // See file comment re: dynamic_cast.
     const proto2::Message* ret = GetWeakPrototype(m, f);
     if (ret) {
@@ -143,7 +137,7 @@ class P2R_Handlers {
       // factory.
       assert(f->cpp_type() == proto2::FieldDescriptor::CPPTYPE_MESSAGE);
       ret = proto2::MessageFactory::generated_factory()->GetPrototype(
-          f->message_type());
+                f->message_type());
       assert(ret);
       return ret;
     } else {
@@ -154,11 +148,9 @@ class P2R_Handlers {
  private:
   class FieldOffset {
    public:
-    FieldOffset(
-        const proto2::FieldDescriptor* f,
-        const _pi::Proto2Reflection* r)
-        : offset_(GetOffset(f, r)),
-          is_repeated_(f->is_repeated()) {
+    FieldOffset(const proto2::FieldDescriptor* f,
+                const _pi::Proto2Reflection* r)
+        : offset_(GetOffset(f, r)), is_repeated_(f->is_repeated()) {
       if (!is_repeated_) {
         int64_t hasbit = GetHasbit(f, r);
         hasbyte_ = hasbit / 8;
@@ -166,7 +158,7 @@ class P2R_Handlers {
       }
     }
 
-    template<class T> T* GetFieldPointer(void* message) const {
+    template <class T> T* GetFieldPointer(void* message) const {
       return GetPointer<T>(message, offset_);
     }
 
@@ -193,7 +185,6 @@ class P2R_Handlers {
     return selector;
   }
 
-
   static int16_t GetHasbit(const proto2::FieldDescriptor* f,
                            const _pi::Proto2Reflection* r) {
     assert(!f->is_repeated());
@@ -211,60 +202,60 @@ class P2R_Handlers {
       const proto2::FieldDescriptor* proto2_f, const _pi::Proto2Reflection* r,
       const upb::FieldDef* f, upb::Handlers* h) {
     assert(f->IsSequence());
-    h->SetStartSequenceHandler(
-        f, &PushOffset, new FieldOffset(proto2_f, r),
-        &upb::DeletePointer<FieldOffset>);
+    h->SetStartSequenceHandler(f, &PushOffset, new FieldOffset(proto2_f, r),
+                               &upb::DeletePointer<FieldOffset>);
   }
 
-  static void* PushOffset(void *m, void *fval) {
-    const FieldOffset* offset = static_cast<FieldOffset*>(fval);
-    return offset->GetFieldPointer<void>(m);
+  static void* PushOffset(const upb::SinkFrame* frame) {
+    const FieldOffset* offset =
+        static_cast<FieldOffset*>(frame->handler_data());
+    return offset->GetFieldPointer<void>(frame->userdata());
   }
 
   // Primitive Value (numeric, enum, bool) /////////////////////////////////////
 
-  template <typename T> static void SetPrimitiveHandlers(
-      const proto2::FieldDescriptor* proto2_f,
-      const _pi::Proto2Reflection* r,
-      const upb::FieldDef* f, upb::Handlers* h) {
+  template <typename T>
+  static void SetPrimitiveHandlers(const proto2::FieldDescriptor* proto2_f,
+                                   const _pi::Proto2Reflection* r,
+                                   const upb::FieldDef* f, upb::Handlers* h) {
     if (f->IsSequence()) {
       SetStartSequenceHandler(proto2_f, r, f, h);
       h->SetValueHandler<T>(f, &Append<T>, NULL, NULL);
     } else {
-      upb::SetStoreValueHandler<T>(
-          f, GetOffset(proto2_f, r), GetHasbit(proto2_f, r), h);
+      upb::SetStoreValueHandler<T>(f, GetOffset(proto2_f, r),
+                                   GetHasbit(proto2_f, r), h);
     }
   }
 
   template <typename T>
-  static bool Append(void *_r, void *fval, T val) {
-    UPB_UNUSED(fval);
+  static bool Append(const upb::SinkFrame* frame, T val) {
     // Proto1's ProtoArray class derives from proto2::RepeatedField.
-    proto2::RepeatedField<T>* r = static_cast<proto2::RepeatedField<T>*>(_r);
+    proto2::RepeatedField<T>* r =
+        static_cast<proto2::RepeatedField<T>*>(frame->userdata());
     r->Add(val);
     return true;
   }
 
   // String ////////////////////////////////////////////////////////////////////
 
-  static void SetStringHandlers(
-      const proto2::FieldDescriptor* proto2_f,
-      const _pi::Proto2Reflection* r,
-      const upb::FieldDef* f, upb::Handlers* h) {
+  static void SetStringHandlers(const proto2::FieldDescriptor* proto2_f,
+                                const _pi::Proto2Reflection* r,
+                                const upb::FieldDef* f, upb::Handlers* h) {
     h->SetStringHandler(f, &OnStringBuf, NULL, NULL);
     if (f->IsSequence()) {
       SetStartSequenceHandler(proto2_f, r, f, h);
       h->SetStartStringHandler(f, &StartRepeatedString, NULL, NULL);
     } else {
-      h->SetStartStringHandler(
-          f, &StartString, new FieldOffset(proto2_f, r),
-          &upb::DeletePointer<FieldOffset>);
+      h->SetStartStringHandler(f, &StartString, new FieldOffset(proto2_f, r),
+                               &upb::DeletePointer<FieldOffset>);
     }
   }
 
-  static void* StartString(void *m, void *fval, size_t size_hint) {
+  static void* StartString(const upb::SinkFrame* frame, size_t size_hint) {
     UPB_UNUSED(size_hint);
-    const FieldOffset* info = static_cast<const FieldOffset*>(fval);
+    void* m = frame->userdata();
+    const FieldOffset* info =
+        static_cast<const FieldOffset*>(frame->handler_data());
     info->SetHasbit(m);
     string* str = info->GetFieldPointer<string>(m);
     str->clear();
@@ -272,16 +263,18 @@ class P2R_Handlers {
     return str;
   }
 
-  static size_t OnStringBuf(void *_s, void *fval, const char *buf, size_t n) {
-    string* s = static_cast<string*>(_s);
+  static size_t OnStringBuf(const upb::SinkFrame* frame,
+                            const char* buf,
+                            size_t n) {
+    string* s = static_cast<string*>(frame->userdata());
     s->append(buf, n);
     return n;
   }
 
-  static void* StartRepeatedString(void *_r, void *fval, size_t size_hint) {
-    UPB_UNUSED(fval);
+  static void* StartRepeatedString(const upb::SinkFrame* frame,
+                                   size_t size_hint) {
     proto2::RepeatedPtrField<string>* r =
-        static_cast<proto2::RepeatedPtrField<string>*>(_r);
+        static_cast<proto2::RepeatedPtrField<string>*>(frame->userdata());
     string* str = r->Add();
     // reserve() here appears to hurt performance rather than help.
     return str;
@@ -290,22 +283,24 @@ class P2R_Handlers {
   // Out-of-line string ////////////////////////////////////////////////////////
 
   static void SetOutOfLineStringHandlers(
-      const proto2::FieldDescriptor* proto2_f,
-      const _pi::Proto2Reflection* r,
+      const proto2::FieldDescriptor* proto2_f, const _pi::Proto2Reflection* r,
       const upb::FieldDef* f, upb::Handlers* h) {
     // This type is only used for non-repeated string fields.
     assert(!f->IsSequence());
-    h->SetStartStringHandler(
-        f, &StartOutOfLineString, new FieldOffset(proto2_f, r),
-        &upb::DeletePointer<FieldOffset>);
+    h->SetStartStringHandler(f, &StartOutOfLineString,
+                             new FieldOffset(proto2_f, r),
+                             &upb::DeletePointer<FieldOffset>);
     h->SetStringHandler(f, &OnStringBuf, NULL, NULL);
   }
 
-  static void* StartOutOfLineString(void *m, void *fval, size_t size_hint) {
-    const FieldOffset* info = static_cast<const FieldOffset*>(fval);
+  static void* StartOutOfLineString(const upb::SinkFrame* frame,
+                                    size_t size_hint) {
+    const FieldOffset* info =
+        static_cast<const FieldOffset*>(frame->handler_data());
+    void* m = frame->userdata();
     info->SetHasbit(m);
-    string **str = info->GetFieldPointer<string*>(m);
-    if (*str == &::ProtocolMessage::___empty_internal_proto_string_)
+    string** str = info->GetFieldPointer<string*>(m);
+    if (*str == &::proto2::internal::GetEmptyString())
       *str = new string();
     (*str)->clear();
     // reserve() here appears to hurt performance rather than help.
@@ -314,43 +309,43 @@ class P2R_Handlers {
 
   // Cord //////////////////////////////////////////////////////////////////////
 
-  static void SetCordHandlers(
-      const proto2::FieldDescriptor* proto2_f,
-      const _pi::Proto2Reflection* r,
-      const upb::FieldDef* f, upb::Handlers* h) {
+  static void SetCordHandlers(const proto2::FieldDescriptor* proto2_f,
+                              const _pi::Proto2Reflection* r,
+                              const upb::FieldDef* f, upb::Handlers* h) {
     h->SetStringHandler(f, &OnCordBuf, NULL, NULL);
     if (f->IsSequence()) {
       SetStartSequenceHandler(proto2_f, r, f, h);
       h->SetStartStringHandler(f, &StartRepeatedCord, NULL, NULL);
     } else {
-      h->SetStartStringHandler(
-          f, &StartCord, new FieldOffset(proto2_f, r),
-          &upb::DeletePointer<FieldOffset*>);
+      h->SetStartStringHandler(f, &StartCord, new FieldOffset(proto2_f, r),
+                               &upb::DeletePointer<FieldOffset*>);
     }
   }
 
-  static void* StartCord(void *m, void *fval, size_t size_hint) {
+  static void* StartCord(const upb::SinkFrame* frame, size_t size_hint) {
     UPB_UNUSED(size_hint);
-    UPB_UNUSED(fval);
-    const FieldOffset* offset = static_cast<const FieldOffset*>(fval);
+    void* m = frame->userdata();
+    const FieldOffset* offset =
+        static_cast<const FieldOffset*>(frame->handler_data());
     offset->SetHasbit(m);
     Cord* field = offset->GetFieldPointer<Cord>(m);
     field->Clear();
     return field;
   }
 
-  static size_t OnCordBuf(void *_c, void *fval, const char *buf, size_t n) {
-    UPB_UNUSED(fval);
-    Cord* c = static_cast<Cord*>(_c);
+  static size_t OnCordBuf(const upb::SinkFrame* frame,
+                          const char* buf,
+                          size_t n) {
+    Cord* c = static_cast<Cord*>(frame->userdata());
     c->Append(StringPiece(buf, n));
     return true;
   }
 
-  static void* StartRepeatedCord(void *_r, void *fval, size_t size_hint) {
+  static void* StartRepeatedCord(const upb::SinkFrame* frame,
+                                 size_t size_hint) {
     UPB_UNUSED(size_hint);
-    UPB_UNUSED(fval);
     proto2::RepeatedField<Cord>* r =
-        static_cast<proto2::RepeatedField<Cord>*>(_r);
+        static_cast<proto2::RepeatedField<Cord>*>(frame->userdata());
     return r->Add();
   }
 
@@ -358,14 +353,12 @@ class P2R_Handlers {
 
   class SubMessageHandlerData : public FieldOffset {
    public:
-    SubMessageHandlerData(
-        const proto2::Message& prototype,
-        const proto2::FieldDescriptor* f,
-        const _pi::Proto2Reflection* r)
+    SubMessageHandlerData(const proto2::Message& prototype,
+                          const proto2::FieldDescriptor* f,
+                          const _pi::Proto2Reflection* r)
         : FieldOffset(f, r) {
       prototype_ = GetWeakPrototype(prototype, f);
-      if (!prototype_)
-        prototype_ = GetFieldPrototype(prototype, f);
+      if (!prototype_) prototype_ = GetFieldPrototype(prototype, f);
     }
 
     const proto2::Message* prototype() const { return prototype_; }
@@ -375,43 +368,40 @@ class P2R_Handlers {
   };
 
   static void SetStartSubMessageHandler(
-      const proto2::FieldDescriptor* proto2_f,
-      const proto2::Message& m,
-      const _pi::Proto2Reflection* r,
-      upb::Handlers::StartFieldHandler* handler,
+      const proto2::FieldDescriptor* proto2_f, const proto2::Message& m,
+      const _pi::Proto2Reflection* r, upb::Handlers::StartFieldHandler* handler,
       const upb::FieldDef* f, upb::Handlers* h) {
-    h->SetStartSubMessageHandler(
-        f, handler,
-        new SubMessageHandlerData(m, proto2_f, r),
-        &upb::DeletePointer<SubMessageHandlerData>);
+    h->SetStartSubMessageHandler(f, handler,
+                                 new SubMessageHandlerData(m, proto2_f, r),
+                                 &upb::DeletePointer<SubMessageHandlerData>);
   }
 
   static void SetRequiredMessageHandlers(
-      const proto2::FieldDescriptor* proto2_f,
-      const proto2::Message& m,
-      const _pi::Proto2Reflection* r,
-      const upb::FieldDef* f, upb::Handlers* h) {
+      const proto2::FieldDescriptor* proto2_f, const proto2::Message& m,
+      const _pi::Proto2Reflection* r, const upb::FieldDef* f,
+      upb::Handlers* h) {
     if (f->IsSequence()) {
       SetStartSequenceHandler(proto2_f, r, f, h);
       SetStartSubMessageHandler(proto2_f, m, r, &StartRepeatedSubMessage, f, h);
     } else {
-      h->SetStartSubMessageHandler(
-          f, &StartRequiredSubMessage, new FieldOffset(proto2_f, r),
-          &upb::DeletePointer<FieldOffset>);
+      h->SetStartSubMessageHandler(f, &StartRequiredSubMessage,
+                                   new FieldOffset(proto2_f, r),
+                                   &upb::DeletePointer<FieldOffset>);
     }
   }
 
-  static void* StartRequiredSubMessage(void *m, void *fval) {
-    const FieldOffset* offset = static_cast<FieldOffset*>(fval);
+  static void* StartRequiredSubMessage(const upb::SinkFrame* frame) {
+    const FieldOffset* offset =
+        static_cast<FieldOffset*>(frame->handler_data());
+    void* m = frame->userdata();
     offset->SetHasbit(m);
     return offset->GetFieldPointer<void>(m);
   }
 
-  static void SetMessageHandlers(
-      const proto2::FieldDescriptor* proto2_f,
-      const proto2::Message& m,
-      const _pi::Proto2Reflection* r,
-      const upb::FieldDef* f, upb::Handlers* h) {
+  static void SetMessageHandlers(const proto2::FieldDescriptor* proto2_f,
+                                 const proto2::Message& m,
+                                 const _pi::Proto2Reflection* r,
+                                 const upb::FieldDef* f, upb::Handlers* h) {
     if (f->IsSequence()) {
       SetStartSequenceHandler(proto2_f, r, f, h);
       SetStartSubMessageHandler(proto2_f, m, r, &StartRepeatedSubMessage, f, h);
@@ -420,11 +410,10 @@ class P2R_Handlers {
     }
   }
 
-  static void SetWeakMessageHandlers(
-      const proto2::FieldDescriptor* proto2_f,
-      const proto2::Message& m,
-      const _pi::Proto2Reflection* r,
-      const upb::FieldDef* f, upb::Handlers* h) {
+  static void SetWeakMessageHandlers(const proto2::FieldDescriptor* proto2_f,
+                                     const proto2::Message& m,
+                                     const _pi::Proto2Reflection* r,
+                                     const upb::FieldDef* f, upb::Handlers* h) {
     if (f->IsSequence()) {
       SetStartSequenceHandler(proto2_f, r, f, h);
       SetStartSubMessageHandler(proto2_f, m, r, &StartRepeatedSubMessage, f, h);
@@ -433,20 +422,22 @@ class P2R_Handlers {
     }
   }
 
-  static void* StartSubMessage(void *m, void *fval) {
+  static void* StartSubMessage(const upb::SinkFrame* frame) {
+    void* m = frame->userdata();
     const SubMessageHandlerData* info =
-        static_cast<const SubMessageHandlerData*>(fval);
+        static_cast<const SubMessageHandlerData*>(frame->handler_data());
     info->SetHasbit(m);
-    proto2::Message **subm = info->GetFieldPointer<proto2::Message*>(m);
+    proto2::Message** subm = info->GetFieldPointer<proto2::Message*>(m);
     if (*subm == info->prototype()) *subm = (*subm)->New();
     return *subm;
   }
 
-  static void* StartWeakSubMessage(void *m, void *fval) {
+  static void* StartWeakSubMessage(const upb::SinkFrame* frame) {
+    void* m = frame->userdata();
     const SubMessageHandlerData* info =
-        static_cast<const SubMessageHandlerData*>(fval);
+        static_cast<const SubMessageHandlerData*>(frame->handler_data());
     info->SetHasbit(m);
-    proto2::Message **subm = info->GetFieldPointer<proto2::Message*>(m);
+    proto2::Message** subm = info->GetFieldPointer<proto2::Message*>(m);
     if (*subm == NULL) {
       *subm = info->prototype()->New();
     }
@@ -459,19 +450,19 @@ class P2R_Handlers {
     // AddAllocated() calls this, but only if other objects are sitting
     // around waiting for reuse, which we will not do.
     static void Delete(Type* t) {
-      (void)t;
+      UPB_UNUSED(t);
       assert(false);
     }
   };
 
   // Closure is a RepeatedPtrField<SubMessageType>*, but we access it through
   // its base class RepeatedPtrFieldBase*.
-  static void* StartRepeatedSubMessage(void* _r, void *fval) {
+  static void* StartRepeatedSubMessage(const upb::SinkFrame* frame) {
     const SubMessageHandlerData* info =
-        static_cast<const SubMessageHandlerData*>(fval);
-    proto2::internal::RepeatedPtrFieldBase *r =
-        static_cast<proto2::internal::RepeatedPtrFieldBase*>(_r);
-    void *submsg = r->AddFromCleared<RepeatedMessageTypeHandler>();
+        static_cast<const SubMessageHandlerData*>(frame->handler_data());
+    proto2::internal::RepeatedPtrFieldBase* r =
+        static_cast<proto2::internal::RepeatedPtrFieldBase*>(frame->userdata());
+    void* submsg = r->AddFromCleared<RepeatedMessageTypeHandler>();
     if (!submsg) {
       submsg = info->prototype()->New();
       r->AddAllocated<RepeatedMessageTypeHandler>(submsg);
@@ -487,14 +478,12 @@ bool TrySetProto1WriteHandlers(const proto2::FieldDescriptor* proto2_f,
 }
 
 const proto2::Message* GetProto1WeakPrototype(
-    const proto2::Message& m,
-    const proto2::FieldDescriptor* f) {
+    const proto2::Message& m, const proto2::FieldDescriptor* f) {
   return P2R_Handlers::GetWeakPrototype(m, f);
 }
 
 const proto2::Message* GetProto1FieldPrototype(
-    const proto2::Message& m,
-    const proto2::FieldDescriptor* f) {
+    const proto2::Message& m, const proto2::FieldDescriptor* f) {
   return P2R_Handlers::GetFieldPrototype(m, f);
 }
 

upb/google/proto1.h

@@ -38,14 +38,12 @@ bool TrySetProto1WriteHandlers(const proto2::FieldDescriptor* proto2_f,
 // Returns a prototype for the given field in "m", if it is weak.  The returned
 // message could be the linked-in message type or OpaqueMessage, if the weak
 // message is *not* linked in.  Otherwise returns NULL.
-const proto2::Message* GetProto1WeakPrototype(
-    const proto2::Message& m,
-    const proto2::FieldDescriptor* f);
+const proto2::Message* GetProto1WeakPrototype(const proto2::Message& m,
+                                              const proto2::FieldDescriptor* f);
 
 // Returns a prototype for the given non-weak field in "m".
 const proto2::Message* GetProto1FieldPrototype(
-    const proto2::Message& m,
-    const proto2::FieldDescriptor* f);
+    const proto2::Message& m, const proto2::FieldDescriptor* f);
 
 }  // namespace google
 }  // namespace upb

upb/google/proto2.cc

@@ -15,42 +15,36 @@
 
 #include "upb/google/proto2.h"
 
-#include "upb/google/proto1.h"
-#include "upb/bytestream.h"
 #include "upb/def.h"
+#include "upb/google/proto1.h"
 #include "upb/handlers.h"
+#include "upb/sink.h"
 
 namespace upb {
-namespace proto2_bridge_google3 { class FieldAccessor; }
-namespace proto2_bridge_opensource { class FieldAccessor; }
+namespace google_google3 { class GMR_Handlers; }
+namespace google_opensource { class GMR_Handlers; }
 }  // namespace upb
 
 // BEGIN DOUBLE COMPILATION TRICKERY. //////////////////////////////////////////
 
 #ifdef UPB_GOOGLE3
 
-// TODO(haberman): friend upb so that this isn't required.
-#define protected public
-#include "net/proto2/public/repeated_field.h"
-#undef protected
-
-#define private public
-#include "net/proto2/public/generated_message_reflection.h"
-#undef private
-
 #include "net/proto2/proto/descriptor.pb.h"
 #include "net/proto2/public/descriptor.h"
+#include "net/proto2/public/extension_set.h"
+#include "net/proto2/public/generated_message_reflection.h"
 #include "net/proto2/public/lazy_field.h"
 #include "net/proto2/public/message.h"
+#include "net/proto2/public/repeated_field.h"
 #include "net/proto2/public/string_piece_field_support.h"
-#include "upb/google/cord.h"
 
 namespace goog = ::proto2;
-namespace me = ::upb::proto2_bridge_google3;
+namespace me = ::upb::google_google3;
 
 #else
 
-// TODO(haberman): friend upb so that this isn't required.
+// TODO(haberman): remove these once new versions of protobuf that "friend"
+// upb are pervasive in the wild.
 #define protected public
 #include "google/protobuf/repeated_field.h"
 #undef protected
@@ -61,10 +55,16 @@ namespace me = ::upb::proto2_bridge_google3;
 
 #include "google/protobuf/descriptor.h"
 #include "google/protobuf/descriptor.pb.h"
+#include "google/protobuf/extension_set.h"
 #include "google/protobuf/message.h"
 
 namespace goog = ::google::protobuf;
-namespace me = ::upb::proto2_bridge_opensource;
+namespace me = ::upb::google_opensource;
+
+using goog::int32;
+using goog::int64;
+using goog::uint32;
+using goog::uint64;
 
 #endif  // ifdef UPB_GOOGLE3
 
@@ -74,9 +74,13 @@ namespace me = ::upb::proto2_bridge_opensource;
 // an enum value for STRING.
 #define UPB_CTYPE_STRING 0
 
-template<class T> static T* GetPointer(void *message, size_t offset) {
+template <class T> static T* GetPointer(void* message, size_t offset) {
   return reinterpret_cast<T*>(static_cast<char*>(message) + offset);
 }
+template <class T>
+static const T* GetConstPointer(const void* message, size_t offset) {
+  return reinterpret_cast<const T*>(static_cast<const char*>(message) + offset);
+}
 
 // This class contains handlers that can write into a proto2 class whose
 // reflection class is GeneratedMessageReflection.  (Despite the name, even
@@ -86,57 +90,60 @@ template<class T> static T* GetPointer(void *message, size_t offset) {
 // internal interfaces that are not guaranteed to be stable.  This class will
 // need to be updated if any non-backward-compatible changes are made to
 // GeneratedMessageReflection.
-//
-// TODO(haberman): change class name?  In retrospect, "FieldAccessor" isn't the
-// best (something more specific like GeneratedMessageReflectionHandlers or
-// GMR_Handlers would be better) but we're depending on a "friend" declaration
-// in proto2 that already specifies "FieldAccessor."  No versions of proto2 have
-// been released that include the "friend FieldAccessor" declaration, so there's
-// still time to change this.  On the other hand, perhaps it's simpler to just
-// rely on "#define private public" since it may be a long time before new
-// versions of proto2 open source are pervasive enough that we can remove this
-// anyway.
-class me::FieldAccessor {
+class me::GMR_Handlers {
  public:
   // Returns true if we were able to set an accessor and any other properties
   // of the FieldDef that are necessary to read/write this field to a
   // proto2::Message.
   static bool TrySet(const goog::FieldDescriptor* proto2_f,
-                     const goog::Message& m,
-                     const upb::FieldDef* upb_f, upb::Handlers* h) {
+                     const goog::Message& m, const upb::FieldDef* upb_f,
+                     upb::Handlers* h) {
     const goog::Reflection* base_r = m.GetReflection();
     // See file comment re: dynamic_cast.
     const goog::internal::GeneratedMessageReflection* r =
         dynamic_cast<const goog::internal::GeneratedMessageReflection*>(base_r);
     if (!r) return false;
-    // Extensions not supported yet.
-    if (proto2_f->is_extension()) return false;
+
+#define PRIMITIVE_TYPE(cpptype, cident)                                        \
+case goog::FieldDescriptor::cpptype:                                           \
+  SetPrimitiveHandlers<cident>(proto2_f, r, upb_f, h);                         \
+  return true;
 
     switch (proto2_f->cpp_type()) {
-#define PRIMITIVE_TYPE(cpptype, cident) \
-      case goog::FieldDescriptor::cpptype: \
-        SetPrimitiveHandlers<cident>(proto2_f, r, upb_f, h); return true;
-      PRIMITIVE_TYPE(CPPTYPE_INT32,  int32_t);
-      PRIMITIVE_TYPE(CPPTYPE_INT64,  int64_t);
-      PRIMITIVE_TYPE(CPPTYPE_UINT32, uint32_t);
-      PRIMITIVE_TYPE(CPPTYPE_UINT64, uint64_t);
+      PRIMITIVE_TYPE(CPPTYPE_INT32, int32);
+      PRIMITIVE_TYPE(CPPTYPE_INT64, int64);
+      PRIMITIVE_TYPE(CPPTYPE_UINT32, uint32);
+      PRIMITIVE_TYPE(CPPTYPE_UINT64, uint64);
       PRIMITIVE_TYPE(CPPTYPE_DOUBLE, double);
-      PRIMITIVE_TYPE(CPPTYPE_FLOAT,  float);
-      PRIMITIVE_TYPE(CPPTYPE_BOOL,   bool);
-#undef PRIMITIVE_TYPE
+      PRIMITIVE_TYPE(CPPTYPE_FLOAT, float);
+      PRIMITIVE_TYPE(CPPTYPE_BOOL, bool);
       case goog::FieldDescriptor::CPPTYPE_ENUM:
-        SetEnumHandlers(proto2_f, r, upb_f, h);
+        if (proto2_f->is_extension()) {
+          SetEnumExtensionHandlers(proto2_f, r, upb_f, h);
+        } else {
+          SetEnumHandlers(proto2_f, r, upb_f, h);
+        }
         return true;
       case goog::FieldDescriptor::CPPTYPE_STRING: {
+        if (proto2_f->is_extension()) {
+#ifdef UPB_GOOGLE3
+          SetStringExtensionHandlers<string>(proto2_f, r, upb_f, h);
+#else
+          SetStringExtensionHandlers<std::string>(proto2_f, r, upb_f, h);
+#endif
+          return true;
+        }
+
         // Old versions of the open-source protobuf release erroneously default
         // to Cord even though that has never been supported in the open-source
         // release.
         int32_t ctype = proto2_f->options().has_ctype() ?
-            proto2_f->options().ctype() : UPB_CTYPE_STRING;
+                        proto2_f->options().ctype()
+                        : UPB_CTYPE_STRING;
         switch (ctype) {
 #ifdef UPB_GOOGLE3
           case goog::FieldOptions::STRING:
-            SetStringHandlers<string>(proto2_f, m, r, upb_f, h);
+            SetStringHandlers<string>(proto2_f, r, upb_f, h);
             return true;
           case goog::FieldOptions::CORD:
             SetCordHandlers(proto2_f, r, upb_f, h);
@@ -146,7 +153,7 @@ class me::FieldAccessor {
             return true;
 #else
           case UPB_CTYPE_STRING:
-            SetStringHandlers<std::string>(proto2_f, m, r, upb_f, h);
+            SetStringHandlers<std::string>(proto2_f, r, upb_f, h);
             return true;
 #endif
           default:
@@ -156,23 +163,25 @@ class me::FieldAccessor {
       case goog::FieldDescriptor::CPPTYPE_MESSAGE:
 #ifdef UPB_GOOGLE3
         if (proto2_f->options().lazy()) {
+          assert(false);
           return false;  // Not yet implemented.
-        } else {
-          SetSubMessageHandlers(proto2_f, m, r, upb_f, h);
+        }
+#endif
+        if (proto2_f->is_extension()) {
+          SetSubMessageExtensionHandlers(proto2_f, m, r, upb_f, h);
           return true;
         }
-#else
         SetSubMessageHandlers(proto2_f, m, r, upb_f, h);
         return true;
-#endif
       default:
         return false;
     }
   }
 
+#undef PRIMITIVE_TYPE
+
   static const goog::Message* GetFieldPrototype(
-      const goog::Message& m,
-      const goog::FieldDescriptor* f) {
+      const goog::Message& m, const goog::FieldDescriptor* f) {
     // We assume that all submessages (and extensions) will be constructed
     // using the same MessageFactory as this message.  This doesn't cover the
     // case of CodedInputStream::SetExtensionRegistry().
@@ -209,11 +218,9 @@ class me::FieldAccessor {
 
   class FieldOffset {
    public:
-    FieldOffset(
-        const goog::FieldDescriptor* f,
-        const goog::internal::GeneratedMessageReflection* r)
-        : offset_(GetOffset(f, r)),
-          is_repeated_(f->is_repeated()) {
+    FieldOffset(const goog::FieldDescriptor* f,
+                const goog::internal::GeneratedMessageReflection* r)
+        : offset_(GetOffset(f, r)), is_repeated_(f->is_repeated()) {
       if (!is_repeated_) {
         int64_t hasbit = GetHasbit(f, r);
         hasbyte_ = hasbit / 8;
@@ -221,7 +228,7 @@ class me::FieldAccessor {
       }
     }
 
-    template<class T> T* GetFieldPointer(void *message) const {
+    template <class T> T* GetFieldPointer(void* message) const {
       return GetPointer<T>(message, offset_);
     }
 
@@ -240,6 +247,29 @@ class me::FieldAccessor {
     int8_t mask_;
   };
 
+  class ExtensionFieldData {
+   public:
+    ExtensionFieldData(
+        const goog::FieldDescriptor* proto2_f,
+        const goog::internal::GeneratedMessageReflection* r)
+        : offset_(r->extensions_offset_),
+          number_(proto2_f->number()),
+          type_(proto2_f->type()) {
+    }
+
+    int number() const { return number_; }
+    goog::internal::FieldType type() const { return type_; }
+
+    goog::internal::ExtensionSet* GetExtensionSet(goog::Message* m) const {
+      return GetPointer<goog::internal::ExtensionSet>(m, offset_);
+    }
+
+   private:
+    const size_t offset_;
+    int number_;
+    goog::internal::FieldType type_;
+  };
+
   // StartSequence /////////////////////////////////////////////////////////////
 
   static void SetStartSequenceHandler(
@@ -247,14 +277,14 @@ class me::FieldAccessor {
       const goog::internal::GeneratedMessageReflection* r,
       const upb::FieldDef* f, upb::Handlers* h) {
     assert(f->IsSequence());
-    h->SetStartSequenceHandler(
-        f, &PushOffset, new FieldOffset(proto2_f, r),
-        &upb::DeletePointer<FieldOffset>);
+    h->SetStartSequenceHandler(f, &PushOffset, new FieldOffset(proto2_f, r),
+                               &upb::DeletePointer<FieldOffset>);
   }
 
-  static void* PushOffset(void *m, void *fval) {
-    const FieldOffset* offset = static_cast<FieldOffset*>(fval);
-    return offset->GetFieldPointer<void>(m);
+  static void* PushOffset(const upb::SinkFrame* frame) {
+    const FieldOffset* offset =
+        static_cast<FieldOffset*>(frame->handler_data());
+    return offset->GetFieldPointer<void>(frame->userdata());
   }
 
   // Primitive Value (numeric, bool) ///////////////////////////////////////////
@@ -262,38 +292,68 @@ class me::FieldAccessor {
   template <typename T> static void SetPrimitiveHandlers(
       const goog::FieldDescriptor* proto2_f,
       const goog::internal::GeneratedMessageReflection* r,
-      const upb::FieldDef* f,
-      upb::Handlers* h) {
-    if (f->IsSequence()) {
-      SetStartSequenceHandler(proto2_f, r, f, h);
-      h->SetValueHandler<T>(f, &AppendPrimitive<T>, NULL, NULL);
+      const upb::FieldDef* f, upb::Handlers* h) {
+    if (proto2_f->is_extension()) {
+      ExtensionFieldData* data = new ExtensionFieldData(proto2_f, r);
+      upb::Handlers::Free* free = &upb::DeletePointer<ExtensionFieldData>;
+      if (f->IsSequence()) {
+        h->SetValueHandler<T>(f, &AppendPrimitiveExtension<T>, data, free);
+      } else {
+        h->SetValueHandler<T>(f, &SetPrimitiveExtension<T>, data, free);
+      }
     } else {
-      upb::SetStoreValueHandler<T>(
-          f, GetOffset(proto2_f, r), GetHasbit(proto2_f, r), h);
+      if (f->IsSequence()) {
+        SetStartSequenceHandler(proto2_f, r, f, h);
+        h->SetValueHandler<T>(f, &AppendPrimitive<T>, NULL, NULL);
+      } else {
+        upb::SetStoreValueHandler<T>(f, GetOffset(proto2_f, r),
+                                     GetHasbit(proto2_f, r), h);
+      }
     }
   }
 
   template <typename T>
-  static bool AppendPrimitive(void *_r, void *fval, T val) {
-    UPB_UNUSED(fval);
-    goog::RepeatedField<T>* r = static_cast<goog::RepeatedField<T>*>(_r);
+  static bool AppendPrimitive(const upb::SinkFrame* frame, T val) {
+    goog::RepeatedField<T>* r =
+        static_cast<goog::RepeatedField<T>*>(frame->userdata());
     r->Add(val);
     return true;
   }
 
+  template <typename T>
+  static bool AppendPrimitiveExtension(const upb::SinkFrame* frame, T val) {
+    goog::Message* m = frame->GetUserdata<goog::Message>();
+    ExtensionFieldData* data =
+        static_cast<ExtensionFieldData*>(frame->handler_data());
+    goog::internal::ExtensionSet* set = data->GetExtensionSet(m);
+    // TODO(haberman): give an accurate value for "packed"
+    goog::internal::RepeatedPrimitiveTypeTraits<T>::Add(
+        data->number(), data->type(), true, val, set);
+    return true;
+  }
+
+  template <typename T>
+  static bool SetPrimitiveExtension(const upb::SinkFrame* frame, T val) {
+    goog::Message* m = frame->GetUserdata<goog::Message>();
+    ExtensionFieldData* data =
+        static_cast<ExtensionFieldData*>(frame->handler_data());
+    goog::internal::ExtensionSet* set = data->GetExtensionSet(m);
+    goog::internal::PrimitiveTypeTraits<T>::Set(data->number(), data->type(),
+                                                val, set);
+    return true;
+  }
+
   // Enum //////////////////////////////////////////////////////////////////////
 
   class EnumHandlerData : public FieldOffset {
    public:
-    EnumHandlerData(
-        const goog::FieldDescriptor* proto2_f,
-        const goog::internal::GeneratedMessageReflection* r,
-        const upb::FieldDef* f)
+    EnumHandlerData(const goog::FieldDescriptor* proto2_f,
+                    const goog::internal::GeneratedMessageReflection* r,
+                    const upb::FieldDef* f)
         : FieldOffset(proto2_f, r),
           field_number_(f->number()),
           unknown_fields_offset_(r->unknown_fields_offset_),
-          enum_(upb_downcast_enumdef(f->subdef())) {
-    }
+          enum_(upb_downcast_enumdef(f->subdef())) {}
 
     bool IsValidValue(int32_t val) const {
       return enum_->FindValueByNumber(val) != NULL;
@@ -314,21 +374,22 @@ class me::FieldAccessor {
   static void SetEnumHandlers(
       const goog::FieldDescriptor* proto2_f,
       const goog::internal::GeneratedMessageReflection* r,
-      const upb::FieldDef* f,
-      upb::Handlers* h) {
+      const upb::FieldDef* f, upb::Handlers* h) {
+    assert(!proto2_f->is_extension());
     EnumHandlerData* data = new EnumHandlerData(proto2_f, r, f);
     if (f->IsSequence()) {
-      h->SetInt32Handler(
-          f, &AppendEnum, data, &upb::DeletePointer<EnumHandlerData>);
+      h->SetInt32Handler(f, &AppendEnum, data,
+                         &upb::DeletePointer<EnumHandlerData>);
     } else {
-      h->SetInt32Handler(
-          f, &SetEnum, data, &upb::DeletePointer<EnumHandlerData>);
+      h->SetInt32Handler(f, &SetEnum, data,
+                         &upb::DeletePointer<EnumHandlerData>);
     }
   }
 
-  static bool SetEnum(void *_m, void *fval, int32_t val) {
-    goog::Message* m = static_cast<goog::Message*>(_m);
-    const EnumHandlerData* data = static_cast<const EnumHandlerData*>(fval);
+  static bool SetEnum(const upb::SinkFrame* frame, int32_t val) {
+    goog::Message* m = static_cast<goog::Message*>(frame->userdata());
+    const EnumHandlerData* data =
+        static_cast<const EnumHandlerData*>(frame->handler_data());
     if (data->IsValidValue(val)) {
       int32_t* message_val = data->GetFieldPointer<int32_t>(m);
       *message_val = val;
@@ -339,15 +400,16 @@ class me::FieldAccessor {
     return true;
   }
 
-  static bool AppendEnum(void *_m, void *fval, int32_t val) {
+  static bool AppendEnum(const upb::SinkFrame* frame, int32_t val) {
     // Closure is the enclosing message.  We can't use the RepeatedField<> as
     // the closure because we need to go back to the message for unrecognized
     // enum values, which go into the unknown field set.
-    goog::Message* m = static_cast<goog::Message*>(_m);
-    const EnumHandlerData* data = static_cast<const EnumHandlerData*>(fval);
+    goog::Message* m = static_cast<goog::Message*>(frame->userdata());
+    const EnumHandlerData* data =
+        static_cast<const EnumHandlerData*>(frame->handler_data());
     if (data->IsValidValue(val)) {
       goog::RepeatedField<int32_t>* r =
-          data->GetFieldPointer<goog::RepeatedField<int32_t> >(m);
+          data->GetFieldPointer<goog::RepeatedField<int32_t>>(m);
       r->Add(val);
     } else {
       data->mutable_unknown_fields(m)->AddVarint(data->field_number(), val);
@@ -355,26 +417,56 @@ class me::FieldAccessor {
     return true;
   }
 
+  // EnumExtension /////////////////////////////////////////////////////////////
+
+  static void SetEnumExtensionHandlers(
+      const goog::FieldDescriptor* proto2_f,
+      const goog::internal::GeneratedMessageReflection* r,
+      const upb::FieldDef* f, upb::Handlers* h) {
+    assert(proto2_f->is_extension());
+    ExtensionFieldData* data = new ExtensionFieldData(proto2_f, r);
+    if (f->IsSequence()) {
+      h->SetInt32Handler(f, &AppendEnumExtension, data,
+                         upb::DeletePointer<ExtensionFieldData>);
+    } else {
+      h->SetInt32Handler(f, &SetEnumExtension, data,
+                         upb::DeletePointer<ExtensionFieldData>);
+    }
+  }
+
+  static bool SetEnumExtension(const upb::SinkFrame* frame, int32_t val) {
+    goog::Message* m = frame->GetUserdata<goog::Message>();
+    const ExtensionFieldData* data =
+        static_cast<const ExtensionFieldData*>(frame->handler_data());
+    goog::internal::ExtensionSet* set = data->GetExtensionSet(m);
+    set->SetEnum(data->number(), data->type(), val, NULL);
+    return true;
+  }
+
+  static bool AppendEnumExtension(const upb::SinkFrame* frame, int32_t val) {
+    goog::Message* m = frame->GetUserdata<goog::Message>();
+    const ExtensionFieldData* data =
+        static_cast<const ExtensionFieldData*>(frame->handler_data());
+    goog::internal::ExtensionSet* set = data->GetExtensionSet(m);
+    // TODO(haberman): give an accurate value for "packed"
+    set->AddEnum(data->number(), data->type(), true, val, NULL);
+    return true;
+  }
+
   // String ////////////////////////////////////////////////////////////////////
 
   // For scalar (non-repeated) string fields.
-  template<class T>
-  class StringHandlerData : public FieldOffset {
+  template <class T> class StringHandlerData : public FieldOffset {
    public:
     StringHandlerData(const goog::FieldDescriptor* proto2_f,
-                      const goog::internal::GeneratedMessageReflection* r,
-                      const goog::Message& prototype)
-        : FieldOffset(proto2_f, r) {
-      // "prototype" isn't guaranteed to be empty, so we create a copy to get
-      // the default string instance.
-      goog::Message* empty = prototype.New();
-      prototype_ = &r->GetStringReference(*empty, proto2_f, NULL);
-      delete empty;
-    }
+                      const goog::internal::GeneratedMessageReflection* r)
+        : FieldOffset(proto2_f, r),
+          prototype_(*GetConstPointer<T*>(r->default_instance_,
+                                          GetOffset(proto2_f, r))) {}
 
     const T* prototype() const { return prototype_; }
 
-    T** GetStringPointer(void *message) const {
+    T** GetStringPointer(void* message) const {
       return GetFieldPointer<T*>(message);
     }
 
@@ -384,27 +476,28 @@ class me::FieldAccessor {
 
   template <typename T> static void SetStringHandlers(
       const goog::FieldDescriptor* proto2_f,
-      const goog::Message& m,
       const goog::internal::GeneratedMessageReflection* r,
       const upb::FieldDef* f,
       upb::Handlers* h) {
+    assert(!proto2_f->is_extension());
     h->SetStringHandler(f, &OnStringBuf<T>, NULL, NULL);
     if (f->IsSequence()) {
       SetStartSequenceHandler(proto2_f, r, f, h);
       h->SetStartStringHandler(f, &StartRepeatedString<T>, NULL, NULL);
     } else {
-      StringHandlerData<T>* data = new StringHandlerData<T>(proto2_f, r, m);
-      h->SetStartStringHandler(
-          f, &StartString<T>, data, &upb::DeletePointer<StringHandlerData<T> >);
+      StringHandlerData<T>* data = new StringHandlerData<T>(proto2_f, r);
+      h->SetStartStringHandler(f, &StartString<T>, data,
+                               &upb::DeletePointer<StringHandlerData<T>>);
     }
   }
 
   // This needs to be templated because google3 string is not std::string.
-  template <typename T> static void* StartString(
-      void *m, void *fval, size_t size_hint) {
+  template <typename T>
+  static void* StartString(const upb::SinkFrame* frame, size_t size_hint) {
     UPB_UNUSED(size_hint);
+    goog::Message* m = static_cast<goog::Message*>(frame->userdata());
     const StringHandlerData<T>* data =
-        static_cast<const StringHandlerData<T>*>(fval);
+        static_cast<const StringHandlerData<T>*>(frame->handler_data());
     T** str = data->GetStringPointer(m);
     data->SetHasbit(m);
     // If it points to the default instance, we must create a new instance.
@@ -414,37 +507,75 @@ class me::FieldAccessor {
     return *str;
   }
 
-  template <typename T> static size_t OnStringBuf(
-      void *_str, void *fval, const char *buf, size_t n) {
-    UPB_UNUSED(fval);
-    T* str = static_cast<T*>(_str);
+  template <typename T>
+  static size_t OnStringBuf(const upb::SinkFrame* frame,
+                            const char* buf, size_t n) {
+    T* str = static_cast<T*>(frame->userdata());
     str->append(buf, n);
     return n;
   }
 
-
   template <typename T>
-  static void* StartRepeatedString(void *_r, void *fval, size_t size_hint) {
+  static void* StartRepeatedString(const upb::SinkFrame* frame,
+                                   size_t size_hint) {
     UPB_UNUSED(size_hint);
-    UPB_UNUSED(fval);
-    goog::RepeatedPtrField<T>* r = static_cast<goog::RepeatedPtrField<T>*>(_r);
+    goog::RepeatedPtrField<T>* r =
+        static_cast<goog::RepeatedPtrField<T>*>(frame->userdata());
     T* str = r->Add();
     str->clear();
     // reserve() here appears to hurt performance rather than help.
     return str;
   }
 
+  // StringExtension ///////////////////////////////////////////////////////////
+
+  template <typename T>
+  static void SetStringExtensionHandlers(
+      const goog::FieldDescriptor* proto2_f,
+      const goog::internal::GeneratedMessageReflection* r,
+      const upb::FieldDef* f, upb::Handlers* h) {
+    assert(proto2_f->is_extension());
+    h->SetStringHandler(f, &OnStringBuf<T>, NULL, NULL);
+    ExtensionFieldData* data = new ExtensionFieldData(proto2_f, r);
+    if (f->IsSequence()) {
+      h->SetStartStringHandler(f, &StartRepeatedStringExtension, data,
+                               upb::DeletePointer<ExtensionFieldData>);
+    } else {
+      h->SetStartStringHandler(f, &StartStringExtension, data,
+                               upb::DeletePointer<ExtensionFieldData>);
+    }
+  }
+
+  // google3 string is not std::string, but we avoid needing to template
+  // because we do not actually have to declare the string type.
+  static void* StartStringExtension(const upb::SinkFrame* frame,
+                                    size_t size_hint) {
+    UPB_UNUSED(size_hint);
+    goog::Message* m = frame->GetUserdata<goog::Message>();
+    const ExtensionFieldData* data =
+        static_cast<const ExtensionFieldData*>(frame->handler_data());
+    goog::internal::ExtensionSet* set = data->GetExtensionSet(m);
+    return set->MutableString(data->number(), data->type(), NULL);
+  }
+
+  static void* StartRepeatedStringExtension(const upb::SinkFrame* frame,
+                                            size_t size_hint) {
+    UPB_UNUSED(size_hint);
+    goog::Message* m = frame->GetUserdata<goog::Message>();
+    const ExtensionFieldData* data =
+        static_cast<const ExtensionFieldData*>(frame->handler_data());
+    goog::internal::ExtensionSet* set = data->GetExtensionSet(m);
+    return set->AddString(data->number(), data->type(), NULL);
+  }
+
   // SubMessage ////////////////////////////////////////////////////////////////
 
   class SubMessageHandlerData : public FieldOffset {
    public:
-    SubMessageHandlerData(
-        const goog::FieldDescriptor* f,
-        const goog::internal::GeneratedMessageReflection* r,
-        const goog::Message* prototype)
-        : FieldOffset(f, r),
-          prototype_(prototype) {
-    }
+    SubMessageHandlerData(const goog::FieldDescriptor* f,
+                          const goog::internal::GeneratedMessageReflection* r,
+                          const goog::Message* prototype)
+        : FieldOffset(f, r), prototype_(prototype) {}
 
     const goog::Message* prototype() const { return prototype_; }
 
@@ -453,13 +584,12 @@ class me::FieldAccessor {
   };
 
   static void SetSubMessageHandlers(
-      const goog::FieldDescriptor* proto2_f,
-      const goog::Message& m,
+      const goog::FieldDescriptor* proto2_f, const goog::Message& m,
       const goog::internal::GeneratedMessageReflection* r,
-      const upb::FieldDef* f,
-      upb::Handlers* h) {
+      const upb::FieldDef* f, upb::Handlers* h) {
+    const goog::Message* field_prototype = GetFieldPrototype(m, proto2_f);
     SubMessageHandlerData* data =
-        new SubMessageHandlerData(proto2_f, r, GetFieldPrototype(m, proto2_f));
+        new SubMessageHandlerData(proto2_f, r, field_prototype);
     upb::Handlers::Free* free = &upb::DeletePointer<SubMessageHandlerData>;
     if (f->IsSequence()) {
       SetStartSequenceHandler(proto2_f, r, f, h);
@@ -469,11 +599,13 @@ class me::FieldAccessor {
     }
   }
 
-  static void* StartSubMessage(void *m, void *fval) {
+  static void* StartSubMessage(const upb::SinkFrame* frame) {
+    void* m = frame->userdata();
     const SubMessageHandlerData* data =
-        static_cast<const SubMessageHandlerData*>(fval);
+        static_cast<const SubMessageHandlerData*>(frame->handler_data());
     data->SetHasbit(m);
-    goog::Message **subm = data->GetFieldPointer<goog::Message*>(m);
+    goog::Message** subm =
+        data->GetFieldPointer<goog::Message*>(frame->userdata());
     if (*subm == NULL || *subm == data->prototype()) {
       *subm = data->prototype()->New();
     }
@@ -486,19 +618,19 @@ class me::FieldAccessor {
     // AddAllocated() calls this, but only if other objects are sitting
     // around waiting for reuse, which we will not do.
     static void Delete(Type* t) {
-      (void)t;
+      UPB_UNUSED(t);
       assert(false);
     }
   };
 
   // Closure is a RepeatedPtrField<SubMessageType>*, but we access it through
   // its base class RepeatedPtrFieldBase*.
-  static void* StartRepeatedSubMessage(void* _r, void *fval) {
+  static void* StartRepeatedSubMessage(const upb::SinkFrame* frame) {
     const SubMessageHandlerData* data =
-        static_cast<const SubMessageHandlerData*>(fval);
-    goog::internal::RepeatedPtrFieldBase *r =
-        static_cast<goog::internal::RepeatedPtrFieldBase*>(_r);
-    void *submsg = r->AddFromCleared<RepeatedMessageTypeHandler>();
+        static_cast<const SubMessageHandlerData*>(frame->handler_data());
+    goog::internal::RepeatedPtrFieldBase* r =
+        static_cast<goog::internal::RepeatedPtrFieldBase*>(frame->userdata());
+    void* submsg = r->AddFromCleared<RepeatedMessageTypeHandler>();
     if (!submsg) {
       submsg = data->prototype()->New();
       r->AddAllocated<RepeatedMessageTypeHandler>(submsg);
@@ -506,7 +638,63 @@ class me::FieldAccessor {
     return submsg;
   }
 
-  // TODO(haberman): handle Extensions, Unknown Fields.
+  // SubMessageExtension ///////////////////////////////////////////////////////
+
+  class SubMessageExtensionHandlerData : public ExtensionFieldData {
+   public:
+    SubMessageExtensionHandlerData(
+        const goog::FieldDescriptor* proto2_f,
+        const goog::internal::GeneratedMessageReflection* r,
+        const goog::Message* prototype)
+        : ExtensionFieldData(proto2_f, r),
+          prototype_(prototype) {
+    }
+
+    const goog::Message* prototype() const { return prototype_; }
+
+   private:
+    const goog::Message* const prototype_;
+  };
+
+  static void SetSubMessageExtensionHandlers(
+      const goog::FieldDescriptor* proto2_f,
+      const goog::Message& m,
+      const goog::internal::GeneratedMessageReflection* r,
+      const upb::FieldDef* f,
+      upb::Handlers* h) {
+    const goog::Message* field_prototype = GetFieldPrototype(m, proto2_f);
+    SubMessageExtensionHandlerData* data =
+        new SubMessageExtensionHandlerData(proto2_f, r, field_prototype);
+    upb::Handlers::Free* free = &upb::DeletePointer<SubMessageHandlerData>;
+    if (f->IsSequence()) {
+      h->SetStartSubMessageHandler(f, &StartRepeatedSubMessageExtension, data,
+                                   free);
+    } else {
+      h->SetStartSubMessageHandler(f, &StartSubMessageExtension, data, free);
+    }
+  }
+
+  static void* StartRepeatedSubMessageExtension(const upb::SinkFrame* frame) {
+    goog::Message* m = frame->GetUserdata<goog::Message>();
+    const SubMessageExtensionHandlerData* data =
+        static_cast<const SubMessageExtensionHandlerData*>(
+            frame->handler_data());
+    goog::internal::ExtensionSet* set = data->GetExtensionSet(m);
+    return set->AddMessage(data->number(), data->type(), *data->prototype(),
+                           NULL);
+  }
+
+  static void* StartSubMessageExtension(const upb::SinkFrame* frame) {
+    goog::Message* m = frame->GetUserdata<goog::Message>();
+    const SubMessageExtensionHandlerData* data =
+        static_cast<const SubMessageExtensionHandlerData*>(
+            frame->handler_data());
+    goog::internal::ExtensionSet* set = data->GetExtensionSet(m);
+    return set->MutableMessage(data->number(), data->type(), *data->prototype(),
+                               NULL);
+  }
+
+  // TODO(haberman): handle Unknown Fields.
 
 #ifdef UPB_GOOGLE3
   // Handlers for types/features only included in internal proto2 release:
@@ -519,38 +707,40 @@ class me::FieldAccessor {
       const proto2::FieldDescriptor* proto2_f,
       const proto2::internal::GeneratedMessageReflection* r,
       const upb::FieldDef* f, upb::Handlers* h) {
+    assert(!proto2_f->is_extension());
     h->SetStringHandler(f, &OnCordBuf, NULL, NULL);
     if (f->IsSequence()) {
       SetStartSequenceHandler(proto2_f, r, f, h);
       h->SetStartStringHandler(f, &StartRepeatedCord, NULL, NULL);
     } else {
-      h->SetStartStringHandler(
-          f, &StartCord, new FieldOffset(proto2_f, r),
-          &upb::DeletePointer<FieldOffset*>);
+      h->SetStartStringHandler(f, &StartCord, new FieldOffset(proto2_f, r),
+                               &upb::DeletePointer<FieldOffset*>);
     }
   }
 
-  static void* StartCord(void *m, void *fval, size_t size_hint) {
+  static void* StartCord(const upb::SinkFrame* frame, size_t size_hint) {
     UPB_UNUSED(size_hint);
-    const FieldOffset* offset = static_cast<const FieldOffset*>(fval);
+    void* m = frame->userdata();
+    const FieldOffset* offset =
+        static_cast<const FieldOffset*>(frame->handler_data());
     offset->SetHasbit(m);
     Cord* field = offset->GetFieldPointer<Cord>(m);
     field->Clear();
     return field;
   }
 
-  static size_t OnCordBuf(void *_c, void *fval, const char *buf, size_t n) {
-    UPB_UNUSED(fval);
-    Cord* c = static_cast<Cord*>(_c);
+  static size_t OnCordBuf(const upb::SinkFrame* frame,
+                          const char* buf, size_t n) {
+    Cord* c = static_cast<Cord*>(frame->userdata());
     c->Append(StringPiece(buf, n));
     return n;
   }
 
-  static void* StartRepeatedCord(void *_r, void *fval, size_t size_hint) {
+  static void* StartRepeatedCord(const upb::SinkFrame* frame,
+                                 size_t size_hint) {
     UPB_UNUSED(size_hint);
-    UPB_UNUSED(fval);
     proto2::RepeatedField<Cord>* r =
-        static_cast<proto2::RepeatedField<Cord>*>(_r);
+        static_cast<proto2::RepeatedField<Cord>*>(frame->userdata());
     return r->Add();
   }
 
@@ -560,27 +750,27 @@ class me::FieldAccessor {
       const proto2::FieldDescriptor* proto2_f,
       const proto2::internal::GeneratedMessageReflection* r,
       const upb::FieldDef* f, upb::Handlers* h) {
+    assert(!proto2_f->is_extension());
     h->SetStringHandler(f, &OnStringPieceBuf, NULL, NULL);
     if (f->IsSequence()) {
       SetStartSequenceHandler(proto2_f, r, f, h);
       h->SetStartStringHandler(f, &StartRepeatedStringPiece, NULL, NULL);
     } else {
-      h->SetStartStringHandler(
-          f, &StartStringPiece, new FieldOffset(proto2_f, r),
-          &upb::DeletePointer<FieldOffset*>);
+      h->SetStartStringHandler(f, &StartStringPiece,
+                               new FieldOffset(proto2_f, r),
+                               &upb::DeletePointer<FieldOffset*>);
     }
   }
 
-  static size_t OnStringPieceBuf(void *_f, void *fval,
-                                 const char *buf, size_t len) {
-    UPB_UNUSED(fval);
+  static size_t OnStringPieceBuf(const upb::SinkFrame* frame,
+                                 const char* buf, size_t len) {
     // TODO(haberman): alias if possible and enabled on the input stream.
     // TODO(haberman): add a method to StringPieceField that lets us avoid
     // this copy/malloc/free.
     proto2::internal::StringPieceField* field =
-        static_cast<proto2::internal::StringPieceField*>(_f);
+        static_cast<proto2::internal::StringPieceField*>(frame->userdata());
     size_t new_len = field->size() + len;
-    char *data = new char[new_len];
+    char* data = new char[new_len];
     memcpy(data, field->data(), field->size());
     memcpy(data + field->size(), buf, len);
     field->CopyFrom(StringPiece(data, new_len));
@@ -588,9 +778,12 @@ class me::FieldAccessor {
     return len;
   }
 
-  static void* StartStringPiece(void *m, void *fval, size_t size_hint) {
+  static void* StartStringPiece(const upb::SinkFrame* frame,
+                                size_t size_hint) {
     UPB_UNUSED(size_hint);
-    const FieldOffset* offset = static_cast<const FieldOffset*>(fval);
+    void* m = frame->userdata();
+    const FieldOffset* offset =
+        static_cast<const FieldOffset*>(frame->handler_data());
     offset->SetHasbit(m);
     proto2::internal::StringPieceField* field =
         offset->GetFieldPointer<proto2::internal::StringPieceField>(m);
@@ -598,13 +791,13 @@ class me::FieldAccessor {
     return field;
   }
 
-  static void* StartRepeatedStringPiece(void* _r, void *fval,
+  static void* StartRepeatedStringPiece(const upb::SinkFrame* frame,
                                         size_t size_hint) {
     UPB_UNUSED(size_hint);
-    UPB_UNUSED(fval);
-    typedef proto2::RepeatedPtrField<proto2::internal::StringPieceField>
-        RepeatedStringPiece;
-    RepeatedStringPiece* r = static_cast<RepeatedStringPiece*>(_r);
+    typedef proto2::RepeatedPtrField<
+        proto2::internal::StringPieceField> RepeatedStringPiece;
+    RepeatedStringPiece* r =
+        static_cast<RepeatedStringPiece*>(frame->userdata());
     proto2::internal::StringPieceField* field = r->Add();
     field->Clear();
     return field;
@@ -619,13 +812,12 @@ namespace google {
 bool TrySetWriteHandlers(const goog::FieldDescriptor* proto2_f,
                          const goog::Message& prototype,
                          const upb::FieldDef* upb_f, upb::Handlers* h) {
-  return me::FieldAccessor::TrySet(proto2_f, prototype, upb_f, h);
+  return me::GMR_Handlers::TrySet(proto2_f, prototype, upb_f, h);
 }
 
-const goog::Message* GetFieldPrototype(
-    const goog::Message& m,
-    const goog::FieldDescriptor* f) {
-  return me::FieldAccessor::GetFieldPrototype(m, f);
+const goog::Message* GetFieldPrototype(const goog::Message& m,
+                                       const goog::FieldDescriptor* f) {
+  return me::GMR_Handlers::GetFieldPrototype(m, f);
 }
 
 }  // namespace google

upb/google/proto2.h

@@ -49,9 +49,8 @@ bool TrySetWriteHandlers(const ::google::protobuf::FieldDescriptor* proto2_f,
 // Returns a prototype for the given field in "m", if it is weak.  The returned
 // message could be the linked-in message type or OpaqueMessage, if the weak
 // message is *not* linked in.  Otherwise returns NULL.
-const proto2::Message* GetFieldPrototype(
-    const proto2::Message& m,
-    const proto2::FieldDescriptor* f);
+const proto2::Message* GetFieldPrototype(const proto2::Message& m,
+                                         const proto2::FieldDescriptor* f);
 const ::google::protobuf::Message* GetFieldPrototype(
     const ::google::protobuf::Message& m,
     const ::google::protobuf::FieldDescriptor* f);

upb/handlers.c

@@ -10,36 +10,22 @@
 #include <stdlib.h>
 #include <string.h>
 
+#include "upb/sink.h"
+
 // Defined for the sole purpose of having a unique pointer value for
 // UPB_NO_CLOSURE.
 char _upb_noclosure;
 
-typedef struct {
-  upb_func *handler;
-
-  // Could put either or both of these in a separate table to save memory when
-  // they are sparse.
-  void *data;
-  upb_handlerfree *cleanup;
-
-  // TODO(haberman): this is wasteful; only the first "fieldhandler" of a
-  // submessage field needs this.  To reduce memory footprint we should either:
-  // - put the subhandlers in a separate "fieldhandler", stored as part of
-  //   a union with one of the above fields.
-  // - count selector offsets by individual pointers instead of by whole
-  //   fieldhandlers.
-  const upb_handlers *subhandlers;
-} fieldhandler;
-
-static const fieldhandler *getfh(
+static const upb_fieldhandler *getfh(
     const upb_handlers *h, upb_selector_t selector) {
   assert(selector < upb_handlers_msgdef(h)->selector_count);
-  fieldhandler* fhbase = (void*)&h->fh_base;
+  upb_fieldhandler* fhbase = (void*)&h->fh_base;
   return &fhbase[selector];
 }
 
-static fieldhandler *getfh_mutable(upb_handlers *h, upb_selector_t selector) {
-  return (fieldhandler*)getfh(h, selector);
+static upb_fieldhandler *getfh_mutable(upb_handlers *h,
+                                       upb_selector_t selector) {
+  return (upb_fieldhandler*)getfh(h, selector);
 }
 
 bool upb_handlers_isfrozen(const upb_handlers *h) {
@@ -52,28 +38,22 @@ uint32_t upb_handlers_selectorbaseoffset(const upb_fielddef *f) {
 
 uint32_t upb_handlers_selectorcount(const upb_fielddef *f) {
   uint32_t ret = 1;
-  if (upb_fielddef_isstring(f)) ret += 2;  // STARTSTR/ENDSTR
+  if (upb_fielddef_isstring(f)) ret += 2;  // [STARTSTR]/STRING/ENDSTR
   if (upb_fielddef_isseq(f)) ret += 2;  // STARTSEQ/ENDSEQ
-  if (upb_fielddef_issubmsg(f)) ret += 2;  // STARTSUBMSG/ENDSUBMSG
+  if (upb_fielddef_issubmsg(f)) ret += 1;  // [STARTSUBMSG]/ENDSUBMSG
   return ret;
 }
 
 upb_handlertype_t upb_handlers_getprimitivehandlertype(const upb_fielddef *f) {
   switch (upb_fielddef_type(f)) {
     case UPB_TYPE_INT32:
-    case UPB_TYPE_SINT32:
-    case UPB_TYPE_SFIXED32:
     case UPB_TYPE_ENUM:
       return UPB_HANDLER_INT32;
     case UPB_TYPE_INT64:
-    case UPB_TYPE_SINT64:
-    case UPB_TYPE_SFIXED64:
       return UPB_HANDLER_INT64;
     case UPB_TYPE_UINT32:
-    case UPB_TYPE_FIXED32:
       return UPB_HANDLER_UINT32;
     case UPB_TYPE_UINT64:
-    case UPB_TYPE_FIXED64:
       return UPB_HANDLER_UINT64;
     case UPB_TYPE_FLOAT:
       return UPB_HANDLER_FLOAT;
@@ -103,11 +83,11 @@ bool upb_getselector(
         return false;
       *s = f->selector_base;
       break;
-    case UPB_HANDLER_STARTSTR:
+    case UPB_HANDLER_STRING:
       if (!upb_fielddef_isstring(f)) return false;
       *s = f->selector_base;
       break;
-    case UPB_HANDLER_STRING:
+    case UPB_HANDLER_STARTSTR:
       if (!upb_fielddef_isstring(f)) return false;
       *s = f->selector_base + 1;
       break;
@@ -125,11 +105,11 @@ bool upb_getselector(
       break;
     case UPB_HANDLER_STARTSUBMSG:
       if (!upb_fielddef_issubmsg(f)) return false;
-      *s = f->selector_base + 1;
+      *s = f->selector_base;
       break;
     case UPB_HANDLER_ENDSUBMSG:
       if (!upb_fielddef_issubmsg(f)) return false;
-      *s = f->selector_base + 2;
+      *s = f->selector_base + 1;
       break;
   }
   assert(*s < upb_fielddef_msgdef(f)->selector_count);
@@ -157,7 +137,7 @@ static void do_cleanup(upb_handlers* h, const upb_fielddef *f,
                        upb_handlertype_t type) {
   upb_selector_t selector;
   if (!upb_getselector(f, type, &selector)) return;
-  fieldhandler *fh = getfh_mutable(h, selector);
+  upb_fieldhandler *fh = getfh_mutable(h, selector);
   if (fh->cleanup) fh->cleanup(fh->data);
   fh->cleanup = NULL;
   fh->data = NULL;
@@ -187,13 +167,15 @@ static void visithandlers(const upb_refcounted *r, upb_refcounted_visit *visit,
   }
 }
 
-upb_handlers *upb_handlers_new(const upb_msgdef *md, const void *owner) {
+upb_handlers *upb_handlers_new(const upb_msgdef *md, const upb_frametype *ft,
+                               const void *owner) {
   assert(upb_msgdef_isfrozen(md));
   static const struct upb_refcounted_vtbl vtbl = {visithandlers, freehandlers};
-  size_t fhandlers_size = sizeof(fieldhandler) * md->selector_count;
+  size_t fhandlers_size = sizeof(upb_fieldhandler) * md->selector_count;
   upb_handlers *h = calloc(sizeof(*h) - sizeof(void*) + fhandlers_size, 1);
   if (!h) return NULL;
   h->msg = md;
+  h->ft = ft;
   upb_msgdef_ref(h->msg, h);
   if (!upb_refcounted_init(upb_upcast(h), &vtbl, owner)) goto oom;
 
@@ -212,6 +194,10 @@ bool upb_handlers_freeze(upb_handlers *const*handlers, int n, upb_status *s) {
 
 const upb_msgdef *upb_handlers_msgdef(const upb_handlers *h) { return h->msg; }
 
+const upb_frametype *upb_handlers_frametype(const upb_handlers *h) {
+  return h->ft;
+}
+
 void upb_handlers_setstartmsg(upb_handlers *h, upb_startmsg_handler *handler) {
   assert(!upb_handlers_isfrozen(h));
   h->startmsg = handler;
@@ -232,13 +218,18 @@ upb_endmsg_handler *upb_handlers_getendmsg(const upb_handlers *h) {
 
 // For now we stuff the subhandlers pointer into the fieldhandlers*
 // corresponding to the UPB_HANDLER_STARTSUBMSG handler.
+static const upb_handlers **subhandlersptr_sel(upb_handlers *h,
+                                               upb_selector_t startsubmsg) {
+  return &getfh_mutable(h, startsubmsg)->subhandlers;
+}
+
 static const upb_handlers **subhandlersptr(upb_handlers *h,
                                            const upb_fielddef *f) {
   assert(upb_fielddef_issubmsg(f));
   upb_selector_t selector;
   bool ok = upb_getselector(f, UPB_HANDLER_STARTSUBMSG, &selector);
   UPB_ASSERT_VAR(ok, ok);
-  return &getfh_mutable(h, selector)->subhandlers;
+  return subhandlersptr_sel(h, selector);
 }
 
 bool upb_handlers_setsubhandlers(upb_handlers *h, const upb_fielddef *f,
@@ -263,6 +254,12 @@ const upb_handlers *upb_handlers_getsubhandlers(const upb_handlers *h,
   return *stored;
 }
 
+const upb_handlers *upb_handlers_getsubhandlers_sel(const upb_handlers *h,
+                                                    upb_selector_t sel) {
+  const upb_handlers **stored = subhandlersptr_sel((upb_handlers*)h, sel);
+  return *stored;
+}
+
 #define SETTER(name, handlerctype, handlertype) \
   bool upb_handlers_set ## name(upb_handlers *h, const upb_fielddef *f, \
                                 handlerctype val, void *data, \
@@ -273,7 +270,7 @@ const upb_handlers *upb_handlers_getsubhandlers(const upb_handlers *h,
     bool ok = upb_getselector(f, handlertype, &selector); \
     if (!ok) return false; \
     do_cleanup(h, f, handlertype); \
-    fieldhandler *fh = getfh_mutable(h, selector); \
+    upb_fieldhandler *fh = getfh_mutable(h, selector); \
     fh->handler = (upb_func*)val; \
     fh->data = (upb_func*)data; \
     fh->cleanup = (upb_func*)cleanup; \
@@ -294,6 +291,47 @@ SETTER(startseq,    upb_startfield_handler*,  UPB_HANDLER_STARTSEQ);
 SETTER(startsubmsg, upb_startfield_handler*,  UPB_HANDLER_STARTSUBMSG);
 SETTER(endsubmsg,   upb_endfield_handler*,    UPB_HANDLER_ENDSUBMSG);
 SETTER(endseq,      upb_endfield_handler*,    UPB_HANDLER_ENDSEQ);
+
+// Our current implementation of these "alt" functions is, according to the
+// letter of the standard, undefined behavior, because we store the
+// upb_int32_handler2* to memory and then read it back (and call it) as a
+// upb_int32_handler*.  Even though both function pointer types take 32-bit
+// integer arguments, they are still technically different types (because one
+// takes an "int" argument and one takes a "long" argument), and calling a
+// function through a pointer to an incompatible type is undefined behavior.
+//
+// I think it is exceedingly unlikely that "int" and "long" would ever have
+// incompatible calling conventions when both are known to be 32 bit signed
+// two's complement integers.  But if absolute standards-compliance is ever
+// required, either due to a practical problem with the undefined behavior or a
+// tool that notices the incongruity, we have an available option for being
+// perfectly standard-compliant; we can store a bool for every function pointer
+// indicating whether it is an "alt" pointer or not.  Then at the call site
+// (inside upb_sink) we can do:
+//
+//   if (is_alt) {
+//     upb_int32_handler2 *func = fp;
+//     func(...);
+//   } else {
+//     upb_int32_handler *func = fp;
+//     func(...);
+//   }
+//
+// We could do this now, but it adds complexity and wastes the memory to store
+// these useless bools.  The bools are useless because the compiler will almost
+// certainly optimize away this branch and elide the two calls into a single
+// call with the 32-bit parameter calling convention.
+
+#ifdef UPB_TWO_32BIT_TYPES
+SETTER(int32alt,       upb_int32_handler2*,       UPB_HANDLER_INT32);
+SETTER(uint32alt,      upb_uint32_handler2*,      UPB_HANDLER_UINT32);
+#endif
+
+#ifdef UPB_TWO_64BIT_TYPES
+SETTER(int64alt,       upb_int64_handler2*,       UPB_HANDLER_INT64);
+SETTER(uint64alt,      upb_uint64_handler2*,      UPB_HANDLER_UINT64);
+#endif
+
 #undef SETTER
 
 upb_func *upb_handlers_gethandler(const upb_handlers *h, upb_selector_t s) {
@@ -310,9 +348,10 @@ typedef struct {
   void *closure;
 } dfs_state;
 
-static upb_handlers *newformsg(const upb_msgdef *m, const void *owner,
+static upb_handlers *newformsg(const upb_msgdef *m, const upb_frametype *ft,
+                               const void *owner,
                                dfs_state *s) {
-  upb_handlers *h = upb_handlers_new(m, owner);
+  upb_handlers *h = upb_handlers_new(m, ft, owner);
   if (!h) return NULL;
   if (!upb_inttable_insertptr(&s->tab, m, upb_value_ptr(h))) goto oom;
 
@@ -326,11 +365,11 @@ static upb_handlers *newformsg(const upb_msgdef *m, const void *owner,
     if (!upb_fielddef_issubmsg(f)) continue;
 
     const upb_msgdef *subdef = upb_downcast_msgdef(upb_fielddef_subdef(f));
-    const upb_value *subm_ent = upb_inttable_lookupptr(&s->tab, subdef);
-    if (subm_ent) {
-      upb_handlers_setsubhandlers(h, f, upb_value_getptr(*subm_ent));
+    upb_value subm_ent;
+    if (upb_inttable_lookupptr(&s->tab, subdef, &subm_ent)) {
+      upb_handlers_setsubhandlers(h, f, upb_value_getptr(subm_ent));
     } else {
-      upb_handlers *sub_mh = newformsg(subdef, &sub_mh, s);
+      upb_handlers *sub_mh = newformsg(subdef, ft, &sub_mh, s);
       if (!sub_mh) goto oom;
       upb_handlers_setsubhandlers(h, f, sub_mh);
       upb_handlers_unref(sub_mh, &sub_mh);
@@ -344,6 +383,7 @@ oom:
 }
 
 const upb_handlers *upb_handlers_newfrozen(const upb_msgdef *m,
+                                           const upb_frametype *ft,
                                            const void *owner,
                                            upb_handlers_callback *callback,
                                            void *closure) {
@@ -352,7 +392,7 @@ const upb_handlers *upb_handlers_newfrozen(const upb_msgdef *m,
   state.closure = closure;
   if (!upb_inttable_init(&state.tab, UPB_CTYPE_PTR)) return NULL;
 
-  upb_handlers *ret = newformsg(m, owner, &state);
+  upb_handlers *ret = newformsg(m, ft, owner, &state);
   if (!ret) return NULL;
   upb_refcounted *r = upb_upcast(ret);
   upb_status status = UPB_STATUS_INIT;
@@ -365,10 +405,9 @@ const upb_handlers *upb_handlers_newfrozen(const upb_msgdef *m,
 }
 
 #define STDMSG_WRITER(type, ctype)                                            \
-  bool upb_stdmsg_set ## type (void *_m, void *fval, ctype val) {             \
-    assert(_m != NULL);                                                       \
-    const upb_stdmsg_fval *f = fval;                                          \
-    uint8_t *m = _m;                                                          \
+  bool upb_stdmsg_set ## type (const upb_sinkframe *frame, ctype val) {       \
+    const upb_stdmsg_fval *f = upb_sinkframe_handlerdata(frame);              \
+    uint8_t *m = upb_sinkframe_userdata(frame);                               \
     if (f->hasbit > 0)                                                        \
       *(uint8_t*)&m[f->hasbit / 8] |= 1 << (f->hasbit % 8);                   \
     *(ctype*)&m[f->offset] = val;                                             \

upb/handlers.h

@@ -25,11 +25,23 @@
 #include "upb/def.h"
 
 #ifdef __cplusplus
-namespace upb { class Handlers; }
+struct upb_frametype;
+namespace upb {
+typedef upb_frametype FrameType;
+class Handlers;
+class SinkFrame;
+}
+typedef upb::FrameType upb_frametype;
 typedef upb::Handlers upb_handlers;
+typedef upb::SinkFrame upb_sinkframe;
+UPB_INLINE void *upb_sinkframe_handlerdata(const upb_sinkframe* frame);
 #else
+struct upb_frametype;
 struct upb_handlers;
+struct upb_sinkframe;
+typedef struct upb_frametype upb_frametype;
 typedef struct upb_handlers upb_handlers;
+typedef struct upb_sinkframe upb_sinkframe;
 #endif
 
 // All the different types of handlers that can be registered.
@@ -61,7 +73,7 @@ extern char _upb_noclosure;
 
 // A selector refers to a specific field handler in the Handlers object
 // (for example: the STARTSUBMSG handler for field "field15").
-typedef uint32_t upb_selector_t;
+typedef int32_t upb_selector_t;
 
 #ifdef __cplusplus
 
@@ -78,24 +90,34 @@ class upb::Handlers {
   typedef upb_selector_t Selector;
   typedef upb_handlertype_t Type;
 
-  typedef bool   StartMessageHandler(void* closure);
-  typedef void   EndMessageHandler(void* closure, Status* status);
-  typedef void*  StartFieldHandler(void* closure, void* data);
-  typedef bool   EndFieldHandler(void *closure, void *data);
-  typedef void*  StartStringHandler(void *c, void *d, size_t size_hint);
-  typedef size_t StringHandler(void *c, void *d, const char *buf, size_t len);
+  typedef bool StartMessageHandler(const SinkFrame*);
+  typedef void EndMessageHandler(const SinkFrame*, Status* status);
+  typedef void* StartFieldHandler(const SinkFrame*);
+  typedef bool EndFieldHandler(const SinkFrame*);
+  typedef void* StartStringHandler(const SinkFrame* c, size_t size_hint);
+  typedef size_t StringHandler(const SinkFrame* c, const char* buf, size_t len);
 
   template <class T> struct Value {
-    typedef bool Handler(void* closure, void* data, T val);
+    typedef bool Handler(const SinkFrame*, T val);
   };
 
-  typedef Value<int32_t>::Handler     Int32Handler;
-  typedef Value<int64_t>::Handler     Int64Handler;
-  typedef Value<uint32_t>::Handler    Uint32Handler;
-  typedef Value<uint64_t>::Handler    Uint64Handler;
-  typedef Value<float>::Handler       FloatHandler;
-  typedef Value<double>::Handler      DoubleHandler;
-  typedef Value<bool>::Handler        BoolHandler;
+  typedef Value<upb_int32_t>::Handler     Int32Handler;
+  typedef Value<upb_int64_t>::Handler     Int64Handler;
+  typedef Value<upb_uint32_t>::Handler    UInt32Handler;
+  typedef Value<upb_uint64_t>::Handler    UInt64Handler;
+  typedef Value<float>::Handler           FloatHandler;
+  typedef Value<double>::Handler          DoubleHandler;
+  typedef Value<bool>::Handler            BoolHandler;
+
+#ifdef UPB_TWO_32BIT_TYPES
+  typedef Value<upb_int32alt_t>::Handler  Int32Handler2;
+  typedef Value<upb_uint32alt_t>::Handler UInt32Handler2;
+#endif
+
+#ifdef UPB_TWO_64BIT_TYPES
+  typedef Value<upb_int64alt_t>::Handler  Int64Handler2;
+  typedef Value<upb_uint64alt_t>::Handler UInt64Handler2;
+#endif
 
   // Any function pointer can be converted to this and converted back to its
   // correct type.
@@ -106,17 +128,23 @@ class upb::Handlers {
 
   typedef void HandlersCallback(void *closure, upb_handlers *h);
 
-  // Returns a new handlers object for the given frozen msgdef.  A single ref
-  // will belong to the given owner.
+  // Returns a new handlers object for the given frozen msgdef that will use
+  // the given FrameType as its top-level state (can be NULL, for now).  A
+  // single ref on the returned object will belong to the given owner.
   // Returns NULL if memory allocation failed.
-  static Handlers* New(const MessageDef* m, const void *owner);
+  static Handlers* New(const MessageDef* m,
+                       const FrameType* ft,
+                       const void *owner);
 
   // Convenience function for registering a graph of handlers that mirrors the
   // graph of msgdefs for some message.  For "m" and all its children a new set
   // of handlers will be created and the given callback will be invoked,
   // allowing the client to register handlers for this message.  Note that any
-  // subhandlers set by the callback will be overwritten.
-  static const Handlers* NewFrozen(const MessageDef *m, const void *owner,
+  // subhandlers set by the callback will be overwritten.  A single ref on the
+  // returned object will belong to the given owner.
+  static const Handlers* NewFrozen(const MessageDef *m,
+                                   const FrameType* ft,
+                                   const void *owner,
                                    HandlersCallback *callback, void *closure);
 
   // Functionality from upb::RefCounted.
@@ -126,6 +154,9 @@ class upb::Handlers {
   void DonateRef(const void *from, const void *to) const;
   void CheckRef(const void *owner) const;
 
+  // Top-level frame type.
+  const FrameType* frame_type() const;
+
   // Freezes the given set of handlers.  You may not freeze a handler without
   // also freezing any handlers they point to.  In the future we may want to
   // require that all fields of the submessage have had subhandlers set for
@@ -137,7 +168,7 @@ class upb::Handlers {
 
   // Sets the startmsg handler for the message, which is defined as follows:
   //
-  //   bool startmsg(void *closure) {
+  //   bool startmsg(const upb::SinkFrame* frame) {
   //     // Called when the message begins.  Returns true if processing should
   //     // continue.
   //     return true;
@@ -147,7 +178,7 @@ class upb::Handlers {
 
   // Sets the endmsg handler for the message, which is defined as follows:
   //
-  //   void endmsg(void *closure, upb_status *status) {
+  //   void endmsg(const upb::SinkFrame* frame, upb_status *status) {
   //     // Called when processing of this message ends, whether in success or
   //     // failure.  "status" indicates the final status of processing, and
   //     // can also be modified in-place to update the final status.
@@ -159,7 +190,7 @@ class upb::Handlers {
   // (this is for an int32 field; other field types will pass their native
   // C/C++ type for "val"):
   //
-  //   bool value(void *closure, void *d, int32_t val) {
+  //   bool value(const upb::SinkFrame *frame, upb_int32_t val) {
   //     // Called when the field's value is encountered.  "d" contains
   //     // whatever data was bound to this field when it was registered.
   //     // Returns true if processing should continue.
@@ -168,24 +199,72 @@ class upb::Handlers {
   //
   // The value type must exactly match f->type().
   // For example, SetInt32Handler() may only be used for fields of type
-  // UPB_TYPE_INT32, UPB_TYPE_SINT32, UPB_TYPE_SFIXED32, and UPB_TYPE_ENUM.
+  // UPB_TYPE_INT32 and UPB_TYPE_ENUM.
   //
   // "d" is the data that will be bound to this callback and passed to it.
   // If "fr" is non-NULL it will be run when the data is no longer needed.
   //
   // Returns "false" if "f" does not belong to this message or has the wrong
   // type for this handler.
+  //
+  // NOTE: the prototype above uses "upb_int32_t" and not "int32_t" from
+  // stdint.h.  For C++ any int32 typedef will work correctly thanks to
+  // function overloading on the function pointer type.  But in C things are
+  // more complicated; "int" and "long" could both be 32-bit types, but the
+  // two are incompatible with each other when it comes to function pointers.
+  // Since we don't know what the underlying type of int32_t is, we have to
+  // define our own which we *do* know the underlying type of.  The easiest
+  // and most portable choice is to define handlers in C with the upb_intXX_t
+  // types.
   bool SetInt32Handler (const FieldDef* f,  Int32Handler* h, void* d, Free* fr);
   bool SetInt64Handler (const FieldDef* f,  Int64Handler* h, void* d, Free* fr);
-  bool SetUint32Handler(const FieldDef* f, Uint32Handler* h, void* d, Free* fr);
-  bool SetUint64Handler(const FieldDef* f, Uint64Handler* h, void* d, Free* fr);
+  bool SetUInt32Handler(const FieldDef* f, UInt32Handler* h, void* d, Free* fr);
+  bool SetUInt64Handler(const FieldDef* f, UInt64Handler* h, void* d, Free* fr);
   bool SetFloatHandler (const FieldDef* f,  FloatHandler* h, void* d, Free* fr);
   bool SetDoubleHandler(const FieldDef* f, DoubleHandler* h, void* d, Free* fr);
   bool SetBoolHandler  (const FieldDef* f,   BoolHandler* h, void* d, Free* fr);
 
+  // Convenience versions that look up the field by name first.  These return
+  // false if no field with this name exists, or for any of the other reasons
+  // that the FieldDef* version returns false.
+  bool SetInt32Handler (const char *name,   Int32Handler* h, void* d, Free* fr);
+  bool SetInt64Handler (const char *name,   Int64Handler* h, void* d, Free* fr);
+  bool SetUInt32Handler(const char *name,  UInt32Handler* h, void* d, Free* fr);
+  bool SetUInt64Handler(const char *name,  UInt64Handler* h, void* d, Free* fr);
+  bool SetFloatHandler (const char *name,   FloatHandler* h, void* d, Free* fr);
+  bool SetDoubleHandler(const char *name,  DoubleHandler* h, void* d, Free* fr);
+  bool SetBoolHandler  (const char *name,    BoolHandler* h, void* d, Free* fr);
+
+  // On platforms where there are two 32-bit or 64-bit integer types, provide
+  // registration functions for both.  Function overloading should make this
+  // all transparent to the user.
+#ifdef UPB_TWO_32BIT_TYPES
+  bool SetInt32Handler (const FieldDef* f,  Int32Handler2* h, void* d, Free* x);
+  bool SetUInt32Handler(const FieldDef* f, UInt32Handler2* h, void* d, Free* x);
+  bool SetInt32Handler (const char *name,   Int32Handler2* h, void* d, Free* x);
+  bool SetUInt32Handler(const char *name,  UInt32Handler2* h, void* d, Free* x);
+#endif
+
+#ifdef UPB_TWO_64BIT_TYPES
+  bool SetInt64Handler (const FieldDef* f,  Int64Handler2* h, void* d, Free* x);
+  bool SetUInt64Handler(const FieldDef* f, UInt64Handler2* h, void* d, Free* x);
+  bool SetInt64Handler (const char *name,   Int64Handler2* h, void* d, Free* x);
+  bool SetUInt64Handler(const char *name,  UInt64Handler2* h, void* d, Free* x);
+#endif
+
+  // Like the above, but these are templated on the type of the value.  For
+  // example, templating on int64_t is equivalent to calling SetInt64Handler.
+  // Attempts to template on a type that does not map to a UPB_TYPE_* type
+  // (like int8_t, since protobufs have no 8-bit type) will get an "undefined
+  // function" compilation error.
+  template<class T> bool SetValueHandler(
+      const FieldDef* f, typename Value<T>::Handler* h, void* d, Free* fr);
+  template<class T> bool SetValueHandler(
+      const char* name, typename Value<T>::Handler* h, void* d, Free* fr);
+
   // Sets handlers for a string field, which are defined as follows:
   //
-  //   void* startstr(void *closure, void *data, size_t size_hint) {
+  //   void* startstr(const upb::SinkFrame *frame, size_t size_hint) {
   //     // Called when a string value begins.  The return value indicates the
   //     // closure for the string.  "size_hint" indicates the size of the
   //     // string if it is known, however if the string is length-delimited
@@ -200,7 +279,7 @@ class upb::Handlers {
   //     return closure;
   //   }
   //
-  //   size_t str(void *closure, void *data, const char *str, size_t len) {
+  //   size_t str(const upb::SinkFrame* frame, const char *str, size_t len) {
   //     // Called for each buffer of string data; the multiple physical buffers
   //     // are all part of the same logical string.  The return value indicates
   //     // how many bytes were consumed.  If this number is less than "len",
@@ -211,7 +290,7 @@ class upb::Handlers {
   //     return len;
   //   }
   //
-  //   bool endstr(void *closure, void *data) {
+  //   bool endstr(const upb::SinkFrame* frame) {
   //     // Called when a string value ends.
   //     return true;
   //   }
@@ -221,13 +300,18 @@ class upb::Handlers {
   bool SetEndStringHandler(const FieldDef* f, EndFieldHandler* h,
                            void* d, Free* fr);
 
-  // A setter that is templated on the type of the value.
-  template<class T> bool SetValueHandler(
-      const FieldDef* f, typename Value<T>::Handler* h, void* d, Free* fr);
+  // Convenience versions that look up the field by name first.  These return
+  // false if no field with this name exists, or for any of the other reasons
+  // that the FieldDef* version returns false.
+  bool SetStartStringHandler(const char* name, StartStringHandler* h,
+                             void* d, Free* fr);
+  bool SetStringHandler(const char* name, StringHandler* h, void* d, Free* fr);
+  bool SetEndStringHandler(const char* name, EndFieldHandler* h,
+                           void* d, Free* fr);
 
   // Sets the startseq handler, which is defined as follows:
   //
-  //   void *startseq(void *closure, void *data) {
+  //   void *startseq(const upb::SinkFrame* frame) {
   //     // Called when a sequence (repeated field) begins.  The returned
   //     // pointer indicates the closure for the sequence (or UPB_BREAK
   //     // to interrupt processing).
@@ -241,11 +325,13 @@ class upb::Handlers {
   // If "cleanup" is non-NULL it will be run when the data is no longer needed.
   bool SetStartSequenceHandler(const FieldDef* f, StartFieldHandler *handler,
                                void* data, Free* cleanup);
+  bool SetStartSequenceHandler(const char* name, StartFieldHandler *handler,
+                               void* data, Free* cleanup);
 
   // Sets the startsubmsg handler for the given field, which is defined as
   // follows:
   //
-  //   void *startsubmsg(void *closure, void *data) {
+  //   void *startsubmsg(const upb::SinkFrame *frame) {
   //     // Called when a submessage begins.  The returned pointer indicates the
   //     // closure for the sequence (or UPB_BREAK to interrupt processing).
   //     return closure;
@@ -258,11 +344,13 @@ class upb::Handlers {
   // submessage/group field.
   bool SetStartSubMessageHandler(const FieldDef* f, StartFieldHandler *handler,
                                  void* data, Free* cleanup);
+  bool SetStartSubMessageHandler(const char* name, StartFieldHandler *handler,
+                                 void* data, Free* cleanup);
 
   // Sets the endsubmsg handler for the given field, which is defined as
   // follows:
   //
-  //   bool endsubmsg(void *closure, void *data) {
+  //   bool endsubmsg(const upb::SinkFrame *frame) {
   //     // Called when a submessage ends.  Returns true to continue processing.
   //     return true;
   //   }
@@ -274,11 +362,13 @@ class upb::Handlers {
   // submessage/group field.
   bool SetEndSubMessageHandler(const FieldDef* f, EndFieldHandler *handler,
                                void* data, Free* cleanup);
+  bool SetEndSubMessageHandler(const char* name, EndFieldHandler *handler,
+                               void* data, Free* cleanup);
 
   // Starts the endsubseq handler for the given field, which is defined as
   // follows:
   //
-  //   bool endseq(void *closure, void *data) {
+  //   bool endseq(const upb::SinkFrame *frame) {
   //     // Called when a sequence ends.  Returns true continue processing.
   //     return true;
   //   }
@@ -290,16 +380,17 @@ class upb::Handlers {
   // repeated field.
   bool SetEndSequenceHandler(const FieldDef* f, EndFieldHandler *handler,
                              void* data, Free* cleanup);
+  bool SetEndSequenceHandler(const char* name, EndFieldHandler *handler,
+                             void* data, Free* cleanup);
 
   // Sets or gets the object that specifies handlers for the given field, which
   // must be a submessage or group.  Returns NULL if no handlers are set.
   bool SetSubHandlers(const FieldDef* f, const Handlers* sub);
   const Handlers* GetSubHandlers(const FieldDef* f) const;
 
-  // NOTE: The remaining functions in this class are mostly of interest to
-  // byte-code/JIT compilers (or upb internals); most users will not need them.
-  // These functions also require more care, since passing a selector that
-  // does not match the type of these handlers yields undefined behavior.
+  // Equivalent to GetSubHandlers, but takes the STARTSUBMSG selector for the
+  // field.
+  const Handlers* GetSubHandlers(Selector startsubmsg) const;
 
   // A selector refers to a specific field handler in the Handlers object
   // (for example: the STARTSUBMSG handler for field "field15").
@@ -309,6 +400,9 @@ class upb::Handlers {
   // contains this FieldDef.
   static bool GetSelector(const FieldDef* f, Type type, Selector* s);
 
+  // Given a START selector of any kind, returns the corresponding END selector.
+  static Selector GetEndSelector(Selector start_selector);
+
   // Returns the function pointer for this handler.  It is the client's
   // responsibility to cast to the correct function type before calling it.
   GenericFunction* GetHandler(Selector selector);
@@ -325,18 +419,19 @@ class upb::Handlers {
   //
   // const FieldDef* GetFieldDef(Selector selector);
   // static bool IsSequence(Selector selector);
-  // Selector GetEndSelector(Selector start_selector);
 
  private:
   UPB_DISALLOW_POD_OPS(Handlers);
+  friend void* ::upb_sinkframe_handlerdata(const upb_sinkframe* frame);
 
 #else
 struct upb_handlers {
 #endif
   upb_refcounted base;
   const upb_msgdef *msg;
-  bool (*startmsg)(void*);
-  void (*endmsg)(void*, upb_status*);
+  const upb_frametype *ft;
+  bool (*startmsg)(const upb_sinkframe*);
+  void (*endmsg)(const upb_sinkframe*, upb_status*);
   void *fh_base[1];  // Start of dynamically-sized field handler array.
 };
 
@@ -344,26 +439,40 @@ struct upb_handlers {
 #ifdef __cplusplus
 extern "C" {
 #endif
-typedef bool upb_startmsg_handler(void *c);
-typedef void upb_endmsg_handler(void *c, upb_status *status);
-typedef void* upb_startfield_handler(void *closure, void *d);
-typedef bool upb_endfield_handler(void *closure, void *d);
+typedef bool upb_startmsg_handler(const upb_sinkframe *frame);
+typedef void upb_endmsg_handler(const upb_sinkframe *frame, upb_status *status);
+typedef void* upb_startfield_handler(const upb_sinkframe *frame);
+typedef bool upb_endfield_handler(const upb_sinkframe *frame);
 typedef void upb_handlers_callback(void *closure, upb_handlers *h);
 typedef void upb_handlerfree(void *d);
 typedef void upb_func();
 
-typedef bool upb_int32_handler(void *c, void *d, int32_t val);
-typedef bool upb_int64_handler(void *c, void *d, int64_t val);
-typedef bool upb_uint32_handler(void *c, void *d, uint32_t val);
-typedef bool upb_uint64_handler(void *c, void *d, uint64_t val);
-typedef bool upb_float_handler(void *c, void *d, float val);
-typedef bool upb_double_handler(void *c, void *d, double val);
-typedef bool upb_bool_handler(void *c, void *d, bool val);
-typedef void* upb_startstr_handler(void *closure, void *d, size_t size_hint);
-typedef size_t upb_string_handler(void *c, void *d, const char *buf, size_t n);
-
-upb_handlers *upb_handlers_new(const upb_msgdef *m, const void *owner);
+typedef bool upb_int32_handler(const upb_sinkframe *f, upb_int32_t val);
+typedef bool upb_int64_handler(const upb_sinkframe *f, upb_int64_t val);
+typedef bool upb_uint32_handler(const upb_sinkframe *f, upb_uint32_t val);
+typedef bool upb_uint64_handler(const upb_sinkframe *f, upb_uint64_t val);
+typedef bool upb_float_handler(const upb_sinkframe *f, float val);
+typedef bool upb_double_handler(const upb_sinkframe *f, double val);
+typedef bool upb_bool_handler(const upb_sinkframe *f, bool val);
+typedef void* upb_startstr_handler(const upb_sinkframe *f, size_t size_hint);
+typedef size_t upb_string_handler(
+    const upb_sinkframe *f, const char *buf, size_t n);
+
+#ifdef UPB_TWO_32BIT_TYPES
+typedef bool upb_int32_handler2(const upb_sinkframe *f, upb_int32alt_t val);
+typedef bool upb_uint32_handler2(const upb_sinkframe *f, upb_uint32alt_t val);
+#endif
+
+#ifdef UPB_TWO_64BIT_TYPES
+typedef bool upb_int64_handler2(const upb_sinkframe *f, upb_int64alt_t val);
+typedef bool upb_uint64_handler2(const upb_sinkframe *f, upb_uint64alt_t val);
+#endif
+
+upb_handlers *upb_handlers_new(const upb_msgdef *m,
+                               const upb_frametype *ft,
+                               const void *owner);
 const upb_handlers *upb_handlers_newfrozen(const upb_msgdef *m,
+                                           const upb_frametype *ft,
                                            const void *owner,
                                            upb_handlers_callback *callback,
                                            void *closure);
@@ -378,6 +487,7 @@ void upb_handlers_checkref(const upb_handlers *h, const void *owner);
 
 bool upb_handlers_freeze(upb_handlers *const*handlers, int n, upb_status *s);
 const upb_msgdef *upb_handlers_msgdef(const upb_handlers *h);
+const upb_frametype *upb_handlers_frametype(const upb_handlers *h);
 void upb_handlers_setstartmsg(upb_handlers *h, upb_startmsg_handler *handler);
 upb_startmsg_handler *upb_handlers_getstartmsg(const upb_handlers *h);
 void upb_handlers_setendmsg(upb_handlers *h, upb_endmsg_handler *handler);
@@ -428,13 +538,36 @@ bool upb_handlers_setsubhandlers(
     upb_handlers *h, const upb_fielddef *f, const upb_handlers *sub);
 const upb_handlers *upb_handlers_getsubhandlers(
     const upb_handlers *h, const upb_fielddef *f);
+const upb_handlers *upb_handlers_getsubhandlers_sel(
+    const upb_handlers *h, upb_selector_t sel);
 upb_handlertype_t upb_handlers_getprimitivehandlertype(const upb_fielddef *f);
 bool upb_getselector(
     const upb_fielddef *f, upb_handlertype_t type, upb_selector_t *s);
+UPB_INLINE upb_selector_t upb_getendselector(upb_selector_t start) {
+  return start + 1;
+}
 upb_func *upb_handlers_gethandler(const upb_handlers *h, upb_selector_t s);
 void *upb_handlers_gethandlerdata(const upb_handlers *h, upb_selector_t s);
 size_t upb_gethandleroffset(upb_selector_t s);
 
+#ifdef UPB_TWO_32BIT_TYPES
+bool upb_handlers_setint32alt(
+    upb_handlers *h, const upb_fielddef *f, upb_int32_handler2 *handler,
+    void *d, upb_handlerfree *fr);
+bool upb_handlers_setuint32alt(
+    upb_handlers *h, const upb_fielddef *f, upb_uint32_handler2 *handler,
+    void *d, upb_handlerfree *fr);
+#endif
+
+#ifdef UPB_TWO_64BIT_TYPES
+bool upb_handlers_setint64alt(
+    upb_handlers *h, const upb_fielddef *f, upb_int64_handler2 *handler,
+    void *d, upb_handlerfree *fr);
+bool upb_handlers_setuint64alt(
+    upb_handlers *h, const upb_fielddef *f, upb_uint64_handler2 *handler,
+    void *d, upb_handlerfree *fr);
+#endif
+
 // Internal-only.
 uint32_t upb_handlers_selectorbaseoffset(const upb_fielddef *f);
 uint32_t upb_handlers_selectorcount(const upb_fielddef *f);
@@ -444,11 +577,12 @@ uint32_t upb_handlers_selectorcount(const upb_fielddef *f);
 
 // Convenience versions of the above that first look up the field by name.
 #define DEFINE_NAME_SETTER(slot, type) \
-  INLINE void upb_handlers_set ## slot ## _n( \
+  UPB_INLINE bool upb_handlers_set ## slot ## _n( \
       upb_handlers *h, const char *name, type val, \
       void *d, upb_handlerfree *fr) { \
-    upb_handlers_set ## slot(h, upb_msgdef_ntof( \
-        upb_handlers_msgdef(h), name), val, d, fr); \
+    const upb_fielddef *f = upb_msgdef_ntof(upb_handlers_msgdef(h), name); \
+    if (!f) return false; \
+    return upb_handlers_set ## slot(h, f, val, d, fr); \
   }
 DEFINE_NAME_SETTER(int32, upb_int32_handler*);
 DEFINE_NAME_SETTER(int64, upb_int64_handler*);
@@ -464,6 +598,17 @@ DEFINE_NAME_SETTER(startseq, upb_startfield_handler*);
 DEFINE_NAME_SETTER(startsubmsg, upb_startfield_handler*);
 DEFINE_NAME_SETTER(endsubmsg, upb_endfield_handler*);
 DEFINE_NAME_SETTER(endseq, upb_endfield_handler*);
+
+#ifdef UPB_TWO_32BIT_TYPES
+DEFINE_NAME_SETTER(int32alt, upb_int32_handler2*);
+DEFINE_NAME_SETTER(uint32alt, upb_uint32_handler2*);
+#endif
+
+#ifdef UPB_TWO_64BIT_TYPES
+DEFINE_NAME_SETTER(int64alt, upb_int64_handler2*);
+DEFINE_NAME_SETTER(uint64alt, upb_uint64_handler2*);
+#endif
+
 #undef DEFINE_NAME_SETTER
 
 // Value writers for every in-memory type: write the data to a known offset
@@ -488,13 +633,13 @@ typedef struct upb_stdmsg_fval {
 #ifdef __cplusplus
 extern "C" {
 #endif
-bool upb_stdmsg_setint32(void *c, void *d, int32_t val);
-bool upb_stdmsg_setint64(void *c, void *d, int64_t val);
-bool upb_stdmsg_setuint32(void *c, void *d, uint32_t val);
-bool upb_stdmsg_setuint64(void *c, void *d, uint64_t val);
-bool upb_stdmsg_setfloat(void *c, void *d, float val);
-bool upb_stdmsg_setdouble(void *c, void *d, double val);
-bool upb_stdmsg_setbool(void *c, void *d, bool val);
+bool upb_stdmsg_setint32(const upb_sinkframe *frame, int32_t val);
+bool upb_stdmsg_setint64(const upb_sinkframe *frame, int64_t val);
+bool upb_stdmsg_setuint32(const upb_sinkframe *frame, uint32_t val);
+bool upb_stdmsg_setuint64(const upb_sinkframe *frame, uint64_t val);
+bool upb_stdmsg_setfloat(const upb_sinkframe *frame, float val);
+bool upb_stdmsg_setdouble(const upb_sinkframe *frame, double val);
+bool upb_stdmsg_setbool(const upb_sinkframe *frame, bool val);
 #ifdef __cplusplus
 }  // extern "C"
 #endif
@@ -503,14 +648,18 @@ bool upb_stdmsg_setbool(void *c, void *d, bool val);
 
 namespace upb {
 
+// This function should be specialized by types that have a FrameType.
+template<class T> inline const FrameType* GetFrameType() { return NULL; }
+
 // C++ Wrappers
-inline Handlers* Handlers::New(const MessageDef* m, const void *owner) {
-  return upb_handlers_new(m, owner);
+inline Handlers* Handlers::New(const MessageDef* m, const FrameType* ft,
+                               const void *owner) {
+  return upb_handlers_new(m, ft, owner);
 }
 inline const Handlers* Handlers::NewFrozen(
-    const MessageDef *m, const void *owner,
+    const MessageDef *m, const FrameType* ft, const void *owner,
     upb_handlers_callback *callback, void *closure) {
-  return upb_handlers_newfrozen(m, owner, callback, closure);
+  return upb_handlers_newfrozen(m, ft, owner, callback, closure);
 }
 inline bool Handlers::IsFrozen() const {
   return upb_handlers_isfrozen(this);
@@ -530,6 +679,9 @@ inline void Handlers::CheckRef(const void *owner) const {
 inline bool Handlers::Freeze(Handlers*const* handlers, int n, Status* s) {
   return upb_handlers_freeze(handlers, n, s);
 }
+inline const FrameType* Handlers::frame_type() const {
+  return upb_handlers_frametype(this);
+}
 inline const MessageDef* Handlers::message_def() const {
   return upb_handlers_msgdef(this);
 }
@@ -551,13 +703,13 @@ inline bool Handlers::SetInt64Handler(
     void *d, Handlers::Free *fr) {
   return upb_handlers_setint64(this, f, handler, d, fr);
 }
-inline bool Handlers::SetUint32Handler(
-    const FieldDef *f, Handlers::Uint32Handler *handler,
+inline bool Handlers::SetUInt32Handler(
+    const FieldDef *f, Handlers::UInt32Handler *handler,
     void *d, Handlers::Free *fr) {
   return upb_handlers_setuint32(this, f, handler, d, fr);
 }
-inline bool Handlers::SetUint64Handler(
-    const FieldDef *f, Handlers::Uint64Handler *handler,
+inline bool Handlers::SetUInt64Handler(
+    const FieldDef *f, Handlers::UInt64Handler *handler,
     void *d, Handlers::Free *fr) {
   return upb_handlers_setuint64(this, f, handler, d, fr);
 }
@@ -615,6 +767,76 @@ inline bool Handlers::SetSubHandlers(
     const FieldDef* f, const Handlers* sub) {
   return upb_handlers_setsubhandlers(this, f, sub);
 }
+inline bool Handlers::SetInt32Handler(
+    const char* name, Handlers::Int32Handler *handler,
+    void *d, Handlers::Free *fr) {
+  return upb_handlers_setint32_n(this, name, handler, d, fr);
+}
+inline bool Handlers::SetInt64Handler(
+    const char* name, Handlers::Int64Handler *handler,
+    void *d, Handlers::Free *fr) {
+  return upb_handlers_setint64_n(this, name, handler, d, fr);
+}
+inline bool Handlers::SetUInt32Handler(
+    const char* name, Handlers::UInt32Handler *handler,
+    void *d, Handlers::Free *fr) {
+  return upb_handlers_setuint32_n(this, name, handler, d, fr);
+}
+inline bool Handlers::SetUInt64Handler(
+    const char* name, Handlers::UInt64Handler *handler,
+    void *d, Handlers::Free *fr) {
+  return upb_handlers_setuint64_n(this, name, handler, d, fr);
+}
+inline bool Handlers::SetFloatHandler(
+    const char* name, Handlers::FloatHandler *handler,
+    void *d, Handlers::Free *fr) {
+  return upb_handlers_setfloat_n(this, name, handler, d, fr);
+}
+inline bool Handlers::SetDoubleHandler(
+    const char* name, Handlers::DoubleHandler *handler,
+    void *d, Handlers::Free *fr) {
+  return upb_handlers_setdouble_n(this, name, handler, d, fr);
+}
+inline bool Handlers::SetBoolHandler(
+    const char* name, Handlers::BoolHandler *handler,
+    void *d, Handlers::Free *fr) {
+  return upb_handlers_setbool_n(this, name, handler, d, fr);
+}
+inline bool Handlers::SetStartStringHandler(
+    const char* name, Handlers::StartStringHandler* handler,
+    void* d, Handlers::Free* fr) {
+  return upb_handlers_setstartstr_n(this, name, handler, d, fr);
+}
+inline bool Handlers::SetEndStringHandler(
+    const char* name, Handlers::EndFieldHandler* handler,
+    void* d, Handlers::Free* fr) {
+  return upb_handlers_setendstr_n(this, name, handler, d, fr);
+}
+inline bool Handlers::SetStringHandler(
+    const char* name, Handlers::StringHandler *handler,
+    void *d, Handlers::Free *fr) {
+  return upb_handlers_setstring_n(this, name, handler, d, fr);
+}
+inline bool Handlers::SetStartSequenceHandler(
+    const char* name, Handlers::StartFieldHandler *handler,
+    void *d, Handlers::Free *fr) {
+  return upb_handlers_setstartseq_n(this, name, handler, d, fr);
+}
+inline bool Handlers::SetStartSubMessageHandler(
+    const char* name, Handlers::StartFieldHandler *handler,
+    void *d, Handlers::Free *fr) {
+  return upb_handlers_setstartsubmsg_n(this, name, handler, d, fr);
+}
+inline bool Handlers::SetEndSubMessageHandler(
+    const char* name, Handlers::EndFieldHandler *handler,
+    void *d, Handlers::Free *fr) {
+  return upb_handlers_setendsubmsg_n(this, name, handler, d, fr);
+}
+inline bool Handlers::SetEndSequenceHandler(
+    const char* name, Handlers::EndFieldHandler *handler,
+    void *d, Handlers::Free *fr) {
+  return upb_handlers_setendseq_n(this, name, handler, d, fr);
+}
 inline Handlers::StartMessageHandler *Handlers::GetStartMessageHandler() const {
   return upb_handlers_getstartmsg(this);
 }
@@ -625,10 +847,17 @@ inline const Handlers* Handlers::GetSubHandlers(
     const FieldDef* f) const {
   return upb_handlers_getsubhandlers(this, f);
 }
+inline const Handlers* Handlers::GetSubHandlers(
+    Handlers::Selector sel) const {
+  return upb_handlers_getsubhandlers_sel(this, sel);
+}
 inline bool Handlers::GetSelector(
     const FieldDef* f, Handlers::Type type, Handlers::Selector* s) {
   return upb_getselector(f, type, s);
 }
+inline Handlers::Selector Handlers::GetEndSelector(Handlers::Selector start) {
+  return upb_getendselector(start);
+}
 inline Handlers::GenericFunction* Handlers::GetHandler(
     Handlers::Selector selector) {
   return upb_handlers_gethandler(this, selector);
@@ -640,6 +869,52 @@ inline size_t Handlers::GetHandlerOffset(Handlers::Selector selector) {
   return upb_gethandleroffset(selector);
 }
 
+#ifdef UPB_TWO_32BIT_TYPES
+inline bool Handlers::SetInt32Handler(
+    const FieldDef *f, Handlers::Int32Handler2 *handler,
+    void *d, Handlers::Free *fr) {
+  return upb_handlers_setint32alt(this, f, handler, d, fr);
+}
+inline bool Handlers::SetUInt32Handler(
+    const FieldDef *f, Handlers::UInt32Handler2 *handler,
+    void *d, Handlers::Free *fr) {
+  return upb_handlers_setuint32alt(this, f, handler, d, fr);
+}
+inline bool Handlers::SetInt32Handler(
+    const char* name, Handlers::Int32Handler2 *handler,
+    void *d, Handlers::Free *fr) {
+  return upb_handlers_setint32alt_n(this, name, handler, d, fr);
+}
+inline bool Handlers::SetUInt32Handler(
+    const char* name, Handlers::UInt32Handler2 *handler,
+    void *d, Handlers::Free *fr) {
+  return upb_handlers_setuint32alt_n(this, name, handler, d, fr);
+}
+#endif
+
+#ifdef UPB_TWO_64BIT_TYPES
+inline bool Handlers::SetInt64Handler(
+    const FieldDef *f, Handlers::Int64Handler2 *handler,
+    void *d, Handlers::Free *fr) {
+  return upb_handlers_setint64alt(this, f, handler, d, fr);
+}
+inline bool Handlers::SetUInt64Handler(
+    const FieldDef *f, Handlers::UInt64Handler2 *handler,
+    void *d, Handlers::Free *fr) {
+  return upb_handlers_setuint64alt(this, f, handler, d, fr);
+}
+inline bool Handlers::SetInt64Handler(
+    const char* name, Handlers::Int64Handler2 *handler,
+    void *d, Handlers::Free *fr) {
+  return upb_handlers_setint64alt_n(this, name, handler, d, fr);
+}
+inline bool Handlers::SetUInt64Handler(
+    const char* name, Handlers::UInt64Handler2 *handler,
+    void *d, Handlers::Free *fr) {
+  return upb_handlers_setuint64alt_n(this, name, handler, d, fr);
+}
+#endif
+
 #define SET_VALUE_HANDLER(type, ctype) \
     template<> \
     inline bool Handlers::SetValueHandler<ctype>( \
@@ -647,14 +922,32 @@ inline size_t Handlers::GetHandlerOffset(Handlers::Selector selector) {
         typename Handlers::Value<ctype>::Handler* handler, \
         void* data, Handlers::Free* cleanup) { \
       return upb_handlers_set ## type(this, f, handler, data, cleanup); \
+    } \
+    template<> \
+    inline bool Handlers::SetValueHandler<ctype>( \
+        const char* f, \
+        typename Handlers::Value<ctype>::Handler* handler, \
+        void* data, Handlers::Free* cleanup) { \
+      return upb_handlers_set ## type ## _n(this, f, handler, data, cleanup); \
     }
 SET_VALUE_HANDLER(double, double);
 SET_VALUE_HANDLER(float, float);
-SET_VALUE_HANDLER(uint64, uint64_t);
-SET_VALUE_HANDLER(uint32, uint32_t);
-SET_VALUE_HANDLER(int64, int64_t);
-SET_VALUE_HANDLER(int32, int32_t);
+SET_VALUE_HANDLER(uint64, upb_uint64_t);
+SET_VALUE_HANDLER(uint32, upb_uint32_t);
+SET_VALUE_HANDLER(int64, upb_int64_t);
+SET_VALUE_HANDLER(int32, upb_int32_t);
 SET_VALUE_HANDLER(bool, bool);
+
+#ifdef UPB_TWO_32BIT_TYPES
+SET_VALUE_HANDLER(int32alt, upb_int32alt_t);
+SET_VALUE_HANDLER(uint32alt, upb_uint32alt_t);
+#endif
+
+#ifdef UPB_TWO_64BIT_TYPES
+SET_VALUE_HANDLER(int64alt, upb_int64alt_t);
+SET_VALUE_HANDLER(uint64alt, upb_uint64alt_t);
+#endif
+
 #undef SET_VALUE_HANDLER
 
 template <class T> void DeletePointer(void *p) { delete static_cast<T*>(p); }
@@ -665,25 +958,55 @@ void SetStoreValueHandler(
 
 // A handy templated function that will retrieve a value handler for a given
 // C++ type.
-#define SET_STORE_VALUE_HANDLER(type, ctype) \
+#define SET_STORE_VALUE_HANDLER(type, ctype, handlerctype) \
     template <> \
     inline void SetStoreValueHandler<ctype>(const FieldDef* f, size_t offset, \
                                             int32_t hasbit, Handlers* h) { \
-      h->SetValueHandler<ctype>( \
+      h->SetValueHandler<handlerctype>( \
           f, upb_stdmsg_set ## type, new upb_stdmsg_fval(offset, hasbit), \
           &upb::DeletePointer<upb_stdmsg_fval>); \
     }
 
-SET_STORE_VALUE_HANDLER(double, double);
-SET_STORE_VALUE_HANDLER(float, float);
-SET_STORE_VALUE_HANDLER(uint64, uint64_t);
-SET_STORE_VALUE_HANDLER(uint32, uint32_t);
-SET_STORE_VALUE_HANDLER(int64, int64_t);
-SET_STORE_VALUE_HANDLER(int32, int32_t);
-SET_STORE_VALUE_HANDLER(bool, bool);
-#undef GET_VALUE_HANDLER
+SET_STORE_VALUE_HANDLER(double, double, double);
+SET_STORE_VALUE_HANDLER(float, float, float);
+SET_STORE_VALUE_HANDLER(uint64, upb_uint64_t, uint64_t);
+SET_STORE_VALUE_HANDLER(uint32, upb_uint32_t, uint32_t);
+SET_STORE_VALUE_HANDLER(int64, upb_int64_t, int64_t);
+SET_STORE_VALUE_HANDLER(int32, upb_int32_t, int32_t);
+SET_STORE_VALUE_HANDLER(bool, bool, bool);
+
+#ifdef UPB_TWO_32BIT_TYPES
+SET_STORE_VALUE_HANDLER(int32, upb_int32alt_t, int32_t);
+SET_STORE_VALUE_HANDLER(uint32, upb_uint32alt_t, uint32_t);
+#endif
+
+#ifdef UPB_TWO_64BIT_TYPES
+SET_STORE_VALUE_HANDLER(int64, upb_int64alt_t, int64_t);
+SET_STORE_VALUE_HANDLER(uint64, upb_uint64alt_t, uint64_t);
+#endif
+
+#undef SET_STORE_VALUE_HANDLER
 
 }  // namespace upb
 #endif
 
+// Implementation detail, put in the header file only so
+// upb_sinkframe_handlerdata() can be inlined.
+typedef struct {
+  upb_func *handler;
+
+  // Could put either or both of these in a separate table to save memory when
+  // they are sparse.
+  void *data;
+  upb_handlerfree *cleanup;
+
+  // TODO(haberman): this is wasteful; only the first "fieldhandler" of a
+  // submessage field needs this.  To reduce memory footprint we should either:
+  // - put the subhandlers in a separate "fieldhandler", stored as part of
+  //   a union with one of the above fields.
+  // - count selector offsets by individual pointers instead of by whole
+  //   fieldhandlers.
+  const upb_handlers *subhandlers;
+} upb_fieldhandler;
+
 #endif

upb/pb/decoder.h

@@ -4,163 +4,96 @@
  * Copyright (c) 2009-2010 Google Inc.  See LICENSE for details.
  * Author: Josh Haberman <jhaberman@gmail.com>
  *
- * upb_decoder implements a high performance, streaming decoder for protobuf
- * data that works by getting its input data from a upb_byteregion and calling
- * into a upb_handlers.
+ * upb::Decoder implements a high performance, streaming decoder for protobuf
+ * data that works by parsing input data one buffer at a time and calling into
+ * a upb::Handlers.
  */
 
 #ifndef UPB_DECODER_H_
 #define UPB_DECODER_H_
 
-#include <setjmp.h>
-#include "upb/bytestream.h"
 #include "upb/sink.h"
 
 #ifdef __cplusplus
-extern "C" {
-#endif
+namespace upb {
+namespace pb {
 
-/* upb_decoderplan ************************************************************/
-
-// A decoderplan contains whatever data structures and generated (JIT-ted) code
-// are necessary to decode protobuf data of a specific type to a specific set
-// of handlers.  By generating the plan ahead of time, we avoid having to
-// redo this work every time we decode.
-//
-// A decoderplan is threadsafe, meaning that it can be used concurrently by
-// different upb_decoders in different threads.  However, the upb_decoders are
-// *not* thread-safe.
-struct _upb_decoderplan;
-typedef struct _upb_decoderplan upb_decoderplan;
-
-// TODO(haberman):
-// - add support for letting any message in the plan be at the top level.
-// - make this object a handlers instead (when bytesrc/bytesink are merged
-//   into handlers).
-// - add support for sharing code with previously-built plans/handlers.
-upb_decoderplan *upb_decoderplan_new(const upb_handlers *h, bool allowjit);
-void upb_decoderplan_unref(upb_decoderplan *p);
-
-// Returns true if the plan contains JIT-ted code.  This may not be the same as
-// the "allowjit" parameter to the constructor if support for JIT-ting was not
-// compiled in.
-bool upb_decoderplan_hasjitcode(upb_decoderplan *p);
-
-
-/* upb_decoder ****************************************************************/
-
-struct dasm_State;
-
-typedef struct {
-  const upb_fielddef *f;
-  uint64_t end_ofs;
-  uint32_t group_fieldnum;  // UINT32_MAX for non-groups.
-  bool is_sequence;   // frame represents seq or submsg? (f might be both).
-  bool is_packed;     // !upb_issubmsg(f) && end_ofs != UINT64_MAX
-                      // (strings aren't pushed).
-} upb_decoder_frame;
-
-typedef struct _upb_decoder {
-  upb_decoderplan *plan;
-  upb_byteregion  *input;          // Input data (serialized), not owned.
-  upb_status      status;          // Where we store errors that occur.
-
-  // Where we push parsed data.
-  // TODO(haberman): make this a pointer and make upb_decoder_resetinput() take
-  // one of these instead of a void*.
-  upb_sink        sink;
-
-  // Our internal stack.
-  upb_decoder_frame *top, *limit;
-  upb_decoder_frame stack[UPB_MAX_NESTING];
-
-  // Current input buffer and its stream offset.
-  const char *buf, *ptr, *end;
-  uint64_t bufstart_ofs;
-
-  // End of the delimited region, relative to ptr, or NULL if not in this buf.
-  const char *delim_end;
-  // True if the top stack frame represents a packed field.
-  bool top_is_packed;
-
-#ifdef UPB_USE_JIT_X64
-  // For JIT, which doesn't do bounds checks in the middle of parsing a field.
-  const char *jit_end, *effective_end;  // == MIN(jit_end, delim_end)
-
-  // Used momentarily by the generated code to store a value while a user
-  // function is called.
-  uint32_t tmp_len;
-#endif
-
-  // For exiting the decoder on error.
-  jmp_buf exitjmp;
-} upb_decoder;
-
-void upb_decoder_init(upb_decoder *d);
-void upb_decoder_uninit(upb_decoder *d);
-
-// Resets the plan that the decoder will parse from.  "msg_offset" indicates
-// which message from the plan will be used as the top-level message.
-//
-// This will also reset the decoder's input to be uninitialized --
-// upb_decoder_resetinput() must be called before parsing can occur.  The plan
-// must live until the decoder is destroyed or reset to a different plan.
-//
-// Must be called before upb_decoder_resetinput() or upb_decoder_decode().
-void upb_decoder_resetplan(upb_decoder *d, upb_decoderplan *p);
-
-// Resets the input of an already-allocated decoder.  This puts it in a state
-// where it has not seen any data, and expects the next data to be from the
-// beginning of a new protobuf.  Decoders must have their input reset before
-// they can be used.  A decoder can have its input reset multiple times.
-// "input" must live until the decoder is destroyed or has it input reset
-// again. "c" is the closure that will be passed to the handlers.
-//
-// Must be called before upb_decoder_decode().
-void upb_decoder_resetinput(upb_decoder *d, upb_byteregion *input, void *c);
-
-// Decodes serialized data (calling handlers as the data is parsed), returning
-// the success of the operation (call upb_decoder_status() for details).
-upb_success_t upb_decoder_decode(upb_decoder *d);
-
-INLINE const upb_status *upb_decoder_status(upb_decoder *d) {
-  return &d->status;
-}
+// Frame type that encapsulates decoder state.
+class Decoder;
 
-// Implementation details
+// Resets the sink of the Decoder.  This must be called at  least once before
+// the decoder can be used.  It may only be called with the decoder is in a
+// state where it was just created or reset.  The given sink must be from the
+// same pipeline as this decoder.
+inline bool ResetDecoderSink(Decoder* d, Sink* sink);
 
-struct _upb_decoderplan {
-  // The top-level handlers that this plan calls into.  We own a ref.
-  const upb_handlers *handlers;
+// Gets the handlers suitable for parsing protobuf data according to the given
+// destination handlers.  The protobuf schema to parse is taken from dest.
+inline const upb::Handlers *GetDecoderHandlers(const upb::Handlers *dest,
+                                               bool allowjit,
+                                               const void *owner);
 
-#ifdef UPB_USE_JIT_X64
-  // JIT-generated machine code (else NULL).
-  char *jit_code;
-  size_t jit_size;
-  char *debug_info;
+// Returns true if these handlers represent a upb::pb::Decoder.
+bool IsDecoder(const upb::Handlers *h);
 
-  // For storing upb_jitmsginfo, which contains per-msg runtime data needed
-  // by the JIT.
-  // Maps upb_handlers* -> upb_jitmsginfo.
-  upb_inttable msginfo;
+// Returns true if IsDecoder(h) and the given handlers have JIT code.
+inline bool HasJitCode(const upb::Handlers* h);
 
-  // The following members are used only while the JIT is being built.
+// Returns the destination handlers if IsDecoder(h), otherwise returns NULL.
+const upb::Handlers* GetDestHandlers(const upb::Handlers* h);
 
-  // This pointer is allocated by dasm_init() and freed by dasm_free().
-  struct dasm_State *dynasm;
+}  // namespace pb
+}  // namespace upb
 
-  // For storing pclabel bases while we are building the JIT.
-  // Maps (upb_handlers* or upb_fielddef*) -> int32 pclabel_base
-  upb_inttable pclabels;
+typedef upb::pb::Decoder upb_pbdecoder;
 
-  // This is not the same as len(pclabels) because the table only contains base
-  // offsets for each def, but each def can have many pclabels.
-  uint32_t pclabel_count;
+extern "C" {
+#else
+struct upb_pbdecoder;
+typedef struct upb_pbdecoder upb_pbdecoder;
 #endif
-};
+
+// C API.
+const upb_frametype *upb_pbdecoder_getframetype();
+bool upb_pbdecoder_resetsink(upb_pbdecoder *d, upb_sink *sink);
+const upb_handlers *upb_pbdecoder_gethandlers(const upb_handlers *dest,
+                                              bool allowjit,
+                                              const void *owner);
+bool upb_pbdecoder_isdecoder(const upb_handlers *h);
+bool upb_pbdecoder_hasjitcode(const upb_handlers *h);
+const upb_handlers *upb_pbdecoder_getdesthandlers(const upb_handlers *h);
+
+// C++ implementation details. /////////////////////////////////////////////////
 
 #ifdef __cplusplus
-}  /* extern "C" */
+}  // extern "C"
+
+namespace upb {
+
+template<> inline const FrameType* GetFrameType<upb::pb::Decoder>() {
+  return upb_pbdecoder_getframetype();
+}
+
+namespace pb {
+inline bool ResetDecoderSink(Decoder* r, Sink* sink) {
+  return upb_pbdecoder_resetsink(r, sink);
+}
+inline const upb::Handlers* GetDecoderHandlers(const upb::Handlers* dest,
+                                               bool allowjit,
+                                               const void* owner) {
+  return upb_pbdecoder_gethandlers(dest, allowjit, owner);
+}
+inline bool IsDecoder(const upb::Handlers* h) {
+  return upb_pbdecoder_isdecoder(h);
+}
+inline bool HasJitCode(const upb::Handlers* h) {
+  return upb_pbdecoder_hasjitcode(h);
+}
+inline const upb::Handlers* GetDestHandlers(const upb::Handlers* h) {
+  return upb_pbdecoder_getdesthandlers(h);
+}
+}  // namespace pb
+}  // namespace upb
 #endif
 
 #endif  /* UPB_DECODER_H_ */

upb/pb/decoder_x64.dasc

@@ -4,7 +4,7 @@
 |// Copyright (c) 2011 Google Inc.  See LICENSE for details.
 |// Author: Josh Haberman <jhaberman@gmail.com>
 |//
-|// JIT compiler for upb_decoder on x86.  Given a upb_decoderplan object (which
+|// JIT compiler for upb_pbdecoder on x86.  Given a decoderplan object (which
 |// contains an embedded set of upb_handlers), generates code specialized to
 |// parsing the specific message and calling specific handlers.
 |//
@@ -54,17 +54,19 @@ typedef struct {
   void *jit_func;
 } upb_jitmsginfo;
 
-static uint32_t upb_getpclabel(upb_decoderplan *plan, const void *obj, int n) {
-  const upb_value *v = upb_inttable_lookupptr(&plan->pclabels, obj);
-  assert(v);
-  return upb_value_getuint32(*v) + n;
+static uint32_t upb_getpclabel(decoderplan *plan, const void *obj, int n) {
+  upb_value v;
+  bool found = upb_inttable_lookupptr(&plan->pclabels, obj, &v);
+  UPB_ASSERT_VAR(found, found);
+  return upb_value_getuint32(v) + n;
 }
 
-static upb_jitmsginfo *upb_getmsginfo(upb_decoderplan *plan,
+static upb_jitmsginfo *upb_getmsginfo(decoderplan *plan,
                                       const upb_handlers *h) {
-  const upb_value *v = upb_inttable_lookupptr(&plan->msginfo, h);
-  assert(v);
-  return upb_value_getptr(*v);
+  upb_value v;
+  bool found = upb_inttable_lookupptr(&plan->msginfo, h, &v);
+  UPB_ASSERT_VAR(found, found);
+  return upb_value_getptr(v);
 }
 
 // To debug JIT-ted code with GDB we need to tell GDB about the JIT-ted code
@@ -109,7 +111,7 @@ void __attribute__((noinline)) __jit_debug_register_code() {
   __asm__ __volatile__("");
 }
 
-void upb_reg_jit_gdb(upb_decoderplan *plan) {
+void upb_reg_jit_gdb(decoderplan *plan) {
   // Create debug info.
   size_t elf_len = sizeof(upb_jit_debug_elf_file);
   plan->debug_info = malloc(elf_len);
@@ -135,7 +137,7 @@ void upb_reg_jit_gdb(upb_decoderplan *plan) {
 
 #else
 
-void upb_reg_jit_gdb(upb_decoderplan *plan) {
+void upb_reg_jit_gdb(decoderplan *plan) {
   (void)plan;
 }
 
@@ -154,10 +156,9 @@ static void upb_assert_notnull(void *addr) { assert(addr != NULL); (void)addr; }
 |// Calling conventions.  Note -- this will need to be changed for
 |// Windows, which uses a different calling convention!
 |.define ARG1_64,   rdi
-|.define ARG2_8,    sil
+|.define ARG2_8,    r6b  // DynASM's equivalent to "sil" -- low byte of esi.
 |.define ARG2_32,   esi
 |.define ARG2_64,   rsi
-|.define ARG3_8,    dl
 |.define ARG3_32,   edx
 |.define ARG3_64,   rdx
 |.define ARG4_64,   rcx
@@ -170,9 +171,10 @@ static void upb_assert_notnull(void *addr) { assert(addr != NULL); (void)addr; }
 |// conventions, but of course when calling to user callbacks we must.
 |.define PTR,       rbx  // Writing this to DECODER->ptr commits our progress.
 |.define CLOSURE,   r12
-|.type   SINKFRAME, upb_sink_frame, r13
-|.type   FRAME,     upb_decoder_frame, r14
-|.type   DECODER,   upb_decoder, r15
+|.type   SINKFRAME, upb_sinkframe, r13
+|.type   FRAME,     frame, r14
+|.type   DECODER,   upb_pbdecoder, r15
+|.type   SINK,      upb_sink
 |
 |.macro callp, addr
 || upb_assert_notnull(addr);
@@ -187,6 +189,21 @@ static void upb_assert_notnull(void *addr) { assert(addr != NULL); (void)addr; }
 || }
 |.endmacro
 |
+|.macro load_handler_data, h, f, type
+||{
+||  uintptr_t data = (uintptr_t)gethandlerdata(h, f, type);
+|| if (data > 0xffffffff) {
+|    mov64  rax, data
+|    mov    SINKFRAME->u.handler_data, rax
+|| } else if (data > 0x7fffffff) {
+|    mov    eax, data
+|    mov    SINKFRAME->u.handler_data, rax
+|| } else {
+|    mov    qword SINKFRAME->u.handler_data, data
+|| }
+|| }
+|.endmacro
+|
 |// Checkpoints our progress by writing PTR to DECODER, and
 |// checks for end-of-buffer.
 |.macro checkpoint, h
@@ -205,25 +222,33 @@ static void upb_assert_notnull(void *addr) { assert(addr != NULL); (void)addr; }
 |  jz    ->exit_jit
 |.endmacro
 |
-|// Decodes varint from [PTR + offset] -> ARG3.
-|// Saves new pointer as rax.
+|// Decodes varint into ARG2.
+|// Inputs:
+|//   - ecx: first 4 bytes of varint
+|//   - offset: offset from PTR where varint begins
+|// Outputs:
+|//   - ARG2: contains decoded varint
+|//   - rax: new PTR
 |.macro decode_loaded_varint, offset
 |  // Check for <=2 bytes inline, otherwise jump to 2-10 byte decoder.
 |  lea    rax, [PTR + offset + 1]
-|  mov    ARG3_32, ecx
-|  and    ARG3_32, 0x7f
+|  mov    ARG2_32, ecx
+|  and    ARG2_32, 0x7f
 |  test   cl, cl
 |  jns    >9
 |  lea    rax, [PTR + offset + 2]
-|  movzx  esi, ch
-|  and    esi, 0x7f
-|  shl    esi, 7
-|  or     ARG3_32, esi
+|  movzx  edx, ch
+|  and    edx, 0x7f
+|  shl    edx, 7
+|  or     ARG2_32, edx
 |  test   cx, cx
 |  jns    >9
 |  mov    ARG1_64, rax
-|  mov    ARG2_32, ARG3_32
+|// XXX: I don't think this handles 64-bit values correctly.
+|// Test with UINT64_MAX
 |  callp  upb_vdecode_max8_fast
+|// rax return from function will contain new pointer
+|  mov    ARG2_64, rdx
 |  check_ptr_ret  // Check for unterminated, >10-byte varint.
 |9:
 |.endmacro
@@ -234,17 +259,22 @@ static void upb_assert_notnull(void *addr) { assert(addr != NULL); (void)addr; }
 |  mov    PTR, rax
 |.endmacro
 |
-|// Decode the tag -> edx.
+|// Table-based field dispatch.
+|// Inputs:
+|//   - ecx: first 4 bytes of tag
+|// Outputs:
+|//   - edx: field number
+|//   - esi: wire type
 |// Could specialize this by avoiding the value masking: could just key the
 |// table on the raw (length-masked) varint to save 3-4 cycles of latency.
 |// Currently only support tables where all entries are in the array part.
 |.macro dyndispatch_, h
 |=>upb_getpclabel(plan, h, DYNDISPATCH):
 |  decode_loaded_varint, 0
-|  mov  ecx, edx
+|  mov  ecx, esi
 |  shr  ecx, 3
-|  and  edx, 0x7   // Note: this value is used in the FIELD pclabel below.
-|  cmp  edx, UPB_WIRE_TYPE_END_GROUP
+|  and  esi, 0x7   // Note: this value is used in the FIELD pclabel below.
+|  cmp  esi, UPB_WIRE_TYPE_END_GROUP
 |  je   >1
 || upb_jitmsginfo *mi = upb_getmsginfo(plan, h);
 |  cmp  ecx, mi->max_field_number  // Bounds-check the field.
@@ -278,10 +308,31 @@ static void upb_assert_notnull(void *addr) { assert(addr != NULL); (void)addr; }
 |  .endmacro
 |.endif
 |
-|// Push a stack frame (not the CPU stack, the upb_decoder stack).
-|.macro pushframe, h, field, end_offset_, endtype
+|.macro pushsinkframe, handlers, field, endtype
+|  mov   rax, DECODER->sink
+|  mov   dword SINKFRAME->u.selector, getselector(field, endtype)
+|  lea   rcx, [SINKFRAME + sizeof(upb_sinkframe)]  // rcx for short addressing
+|  cmp   rcx, SINK:rax->limit
+|  jae   ->exit_jit  // Frame stack overflow.
+|  mov64 r9, (uintptr_t)handlers
+|  mov   SINKFRAME:rcx->h, r9
+|  mov   SINKFRAME:rcx->closure, CLOSURE
+|  mov   SINK:rax->top_, rcx
+|  mov   SINKFRAME:rcx->sink_, rax
+|  mov   SINKFRAME, rcx
+|.endmacro
+|
+|.macro popsinkframe
+|  sub   SINKFRAME, sizeof(upb_sinkframe)
+|  mov   rax, DECODER->sink
+|  mov   SINK:rax->top_, SINKFRAME
+|  mov   CLOSURE, SINKFRAME->closure
+|.endmacro
+|
+|// Push a stack frame (not the CPU stack, the upb_pbdecoder stack).
+|.macro pushframe, handlers, field, end_offset_, endtype
 |// Decoder Frame.
-|  lea   rax, [FRAME + sizeof(upb_decoder_frame)]  // rax for short addressing
+|  lea   rax, [FRAME + sizeof(frame)]  // rax for short addressing
 |  cmp   rax, DECODER->limit
 |  jae   ->exit_jit  // Frame stack overflow.
 |  mov64 r10, (uintptr_t)field
@@ -289,36 +340,21 @@ static void upb_assert_notnull(void *addr) { assert(addr != NULL); (void)addr; }
 |  mov   qword FRAME:rax->end_ofs, end_offset_
 |  mov   byte FRAME:rax->is_sequence, (endtype == UPB_HANDLER_ENDSEQ)
 |  mov   byte FRAME:rax->is_packed, 0
-|| if (upb_fielddef_type(field) == UPB_TYPE_GROUP &&
-||     endtype == UPB_HANDLER_ENDSUBMSG) {
+|| if (upb_fielddef_istagdelim(field) && endtype == UPB_HANDLER_ENDSUBMSG) {
 |    mov dword FRAME:rax->group_fieldnum, upb_fielddef_number(field)
 || } else {
 |    mov dword FRAME:rax->group_fieldnum, 0xffffffff
 || }
 |  mov   DECODER->top, rax
 |  mov   FRAME, rax
-|// Sink Frame.
-|  lea   rcx, [SINKFRAME + sizeof(upb_sink_frame)]  // rcx for short addressing
-|  cmp   rcx, DECODER->sink.limit
-|  jae   ->exit_jit  // Frame stack overflow.
-|  mov   dword SINKFRAME:rcx->end, getselector(field, endtype)
-|| if (upb_fielddef_issubmsg(field)) {
-|    mov64 r9, (uintptr_t)upb_handlers_getsubhandlers(h, field)
-|| } else {
-|    mov64 r9, (uintptr_t)h
-|| }
-|  mov   SINKFRAME:rcx->h, r9
-|  mov   DECODER->sink.top, rcx
-|  mov   SINKFRAME, rcx
+|  pushsinkframe handlers, field, endtype
 |.endmacro
 |
 |.macro popframe
-|  sub   FRAME, sizeof(upb_decoder_frame)
+|  sub   FRAME, sizeof(frame)
 |  mov   DECODER->top, FRAME
-|  sub   SINKFRAME, sizeof(upb_sink_frame)
-|  mov   DECODER->sink.top, SINKFRAME
+|  popsinkframe
 |  setmsgend
-|  mov   CLOSURE, SINKFRAME->closure
 |.endmacro
 |
 |.macro setmsgend
@@ -369,14 +405,6 @@ static void upb_assert_notnull(void *addr) { assert(addr != NULL); (void)addr; }
 #include <stdlib.h>
 #include "upb/pb/varint.h"
 
-static upb_selector_t getselector(const upb_fielddef *f,
-                                  upb_handlertype_t type) {
-  upb_selector_t selector;
-  bool ok = upb_getselector(f, type, &selector);
-  UPB_ASSERT_VAR(ok, ok);
-  return selector;
-}
-
 static upb_func *gethandler(const upb_handlers *h, const upb_fielddef *f,
                             upb_handlertype_t type) {
   return upb_handlers_gethandler(h, getselector(f, type));
@@ -387,73 +415,74 @@ static uintptr_t gethandlerdata(const upb_handlers *h, const upb_fielddef *f,
   return (uintptr_t)upb_handlers_gethandlerdata(h, getselector(f, type));
 }
 
-// Decodes the next val into ARG3, advances PTR.
-static void upb_decoderplan_jit_decodefield(upb_decoderplan *plan,
-                                            uint8_t type, size_t tag_size,
+// Decodes the next val into ARG2, advances PTR.
+static void upb_decoderplan_jit_decodefield(decoderplan *plan,
+                                            size_t tag_size,
                                             const upb_handlers *h,
                                             const upb_fielddef *f) {
   // Decode the value into arg 3 for the callback.
-  switch (type) {
-    case UPB_TYPE(DOUBLE):
+  switch (upb_fielddef_descriptortype(f)) {
+    case UPB_DESCRIPTOR_TYPE_DOUBLE:
       |  movsd  XMMARG1, qword [PTR + tag_size]
       |  add    PTR, 8 + tag_size
       break;
 
-    case UPB_TYPE(FIXED64):
-    case UPB_TYPE(SFIXED64):
-      |  mov  ARG3_64, qword [PTR + tag_size]
+    case UPB_DESCRIPTOR_TYPE_FIXED64:
+    case UPB_DESCRIPTOR_TYPE_SFIXED64:
+      |  mov  ARG2_64, qword [PTR + tag_size]
       |  add  PTR, 8 + tag_size
       break;
 
-    case UPB_TYPE(FLOAT):
+    case UPB_DESCRIPTOR_TYPE_FLOAT:
       |  movss  XMMARG1, dword [PTR + tag_size]
       |  add    PTR, 4 + tag_size
       break;
 
-    case UPB_TYPE(FIXED32):
-    case UPB_TYPE(SFIXED32):
-      |  mov  ARG3_32, dword [PTR + tag_size]
+    case UPB_DESCRIPTOR_TYPE_FIXED32:
+    case UPB_DESCRIPTOR_TYPE_SFIXED32:
+      |  mov  ARG2_32, dword [PTR + tag_size]
       |  add  PTR, 4 + tag_size
       break;
 
-    case UPB_TYPE(BOOL):
+    case UPB_DESCRIPTOR_TYPE_BOOL:
       // Can't assume it's one byte long, because bool must be wire-compatible
       // with all of the varint integer types.
       |  decode_varint  tag_size
-      |  test  ARG3_64, ARG3_64
-      |  setne ARG3_8   // Other bytes left with val, should be ok.
+      |  test  ARG2_64, ARG2_64
+      |  setne al
+      |  movzx ARG2_32, al
       break;
 
-    case UPB_TYPE(INT64):
-    case UPB_TYPE(UINT64):
-    case UPB_TYPE(INT32):
-    case UPB_TYPE(UINT32):
-    case UPB_TYPE(ENUM):
+    case UPB_DESCRIPTOR_TYPE_INT64:
+    case UPB_DESCRIPTOR_TYPE_UINT64:
+    case UPB_DESCRIPTOR_TYPE_INT32:
+    case UPB_DESCRIPTOR_TYPE_UINT32:
+    case UPB_DESCRIPTOR_TYPE_ENUM:
       |  decode_varint  tag_size
       break;
 
-    case UPB_TYPE(SINT64):
+    case UPB_DESCRIPTOR_TYPE_SINT64:
       // 64-bit zig-zag decoding.
       |  decode_varint  tag_size
-      |  mov  rax, ARG3_64
-      |  shr  ARG3_64, 1
+      |  mov  rax, ARG2_64
+      |  shr  ARG2_64, 1
       |  and  rax, 1
       |  neg  rax
-      |  xor  ARG3_64, rax
+      |  xor  ARG2_64, rax
       break;
 
-    case UPB_TYPE(SINT32):
+    case UPB_DESCRIPTOR_TYPE_SINT32:
       // 32-bit zig-zag decoding.
       |  decode_varint  tag_size
-      |  mov  eax, ARG3_32
-      |  shr  ARG3_32, 1
+      |  mov  eax, ARG2_32
+      |  shr  ARG2_32, 1
       |  and  eax, 1
       |  neg  eax
-      |  xor  ARG3_32, eax
+      |  xor  ARG2_32, eax
       break;
 
-    case UPB_TYPE(STRING):
-    case UPB_TYPE(BYTES): {
+    case UPB_DESCRIPTOR_TYPE_STRING:
+    case UPB_DESCRIPTOR_TYPE_BYTES: {
       // We only handle the case where the entire string is in our current
       // buf, which sidesteps any security problems.  The C path has more
       // robust checks.
@@ -461,39 +490,46 @@ static void upb_decoderplan_jit_decodefield(upb_decoderplan *plan,
       |  decode_loaded_varint tag_size
       |  mov  rdi, DECODER->end
       |  sub  rdi, rax
-      |  cmp  ARG3_64, rdi  // if (len > d->end - str)
+      |  cmp  ARG2_64, rdi  // if (len > d->end - str)
       |  ja   ->exit_jit    // Can't deliver, whole string not in buf.
       |  mov  PTR, rax
 
       upb_func *handler = gethandler(h, f, UPB_HANDLER_STARTSTR);
       if (handler) {
-        |  mov  DECODER->tmp_len, ARG3_64
-        |  mov  ARG1_64, CLOSURE
-        |  mov64 ARG2_64, gethandlerdata(h, f, UPB_HANDLER_STARTSTR)
+        |  mov  DECODER->tmp_len, ARG2_32
+        |  mov  ARG1_64, SINKFRAME
+        |  load_handler_data h, f, UPB_HANDLER_STARTSTR
         |  callp handler
         |  check_ptr_ret
-        |  mov  ARG1_64, rax   // sub-closure
-        |  mov  ARG4_64, DECODER->tmp_len
+        |  mov  CLOSURE, rax
+        |  mov  ARG3_32, DECODER->tmp_len
       } else {
-        |  mov  ARG1_64, CLOSURE
-        |  mov  ARG4_64, ARG3_64
+        |  mov  ARG3_64, ARG2_64
       }
 
       handler = gethandler(h, f, UPB_HANDLER_STRING);
       if (handler) {
-        |  mov64 ARG2_64, gethandlerdata(h, f, UPB_HANDLER_STRING)
-        |  mov   ARG3_64, PTR
+        // TODO: push a real frame so we can resume into the string.
+        // (but maybe do this only if the string breaks).
+        |  pushsinkframe h, f, UPB_HANDLER_ENDSTR
+
+        // size_t str(const upb_sinkframe *frame, const char *buf, size_t len)
+        |  mov   ARG1_64, SINKFRAME
+        |  load_handler_data h, f, UPB_HANDLER_STRING
+        |  mov   ARG2_64, PTR
         |  callp handler
         // TODO: properly handle returns other than "n" (the whole string).
         |  add   PTR, rax
+        |  popsinkframe
       } else {
-        |  add   PTR, ARG4_64
+        |  add   PTR, ARG3_64
       }
 
       handler = gethandler(h, f, UPB_HANDLER_ENDSTR);
       if (handler) {
-        |  mov    ARG1_64, CLOSURE
-        |  mov64  ARG2_64, gethandlerdata(h, f, UPB_HANDLER_ENDSTR)
+        // bool endstr(const upb_sinkframe *frame);
+        |  mov    ARG1_64, SINKFRAME
+        |  load_handler_data h, f, UPB_HANDLER_ENDSTR
         |  callp  handler
         |  check_bool_ret
       }
@@ -501,10 +537,10 @@ static void upb_decoderplan_jit_decodefield(upb_decoderplan *plan,
     }
 
     // Will dispatch callbacks and call submessage in a second.
-    case UPB_TYPE(MESSAGE):
+    case UPB_DESCRIPTOR_TYPE_MESSAGE:
       |  decode_varint  tag_size
       break;
-    case UPB_TYPE(GROUP):
+    case UPB_DESCRIPTOR_TYPE_GROUP:
       |  add  PTR, tag_size
       break;
 
@@ -512,52 +548,58 @@ static void upb_decoderplan_jit_decodefield(upb_decoderplan *plan,
   }
 }
 
-static void upb_decoderplan_jit_callcb(upb_decoderplan *plan,
+static void upb_decoderplan_jit_callcb(decoderplan *plan,
                                        const upb_handlers *h,
                                        const upb_fielddef *f) {
   // Call callbacks.  Specializing the append accessors didn't yield a speed
   // increase in benchmarks.
   if (upb_fielddef_issubmsg(f)) {
-    if (upb_fielddef_type(f) == UPB_TYPE(MESSAGE)) {
-      |   mov   rsi, PTR
-      |   sub   rsi, DECODER->buf
-      |   add   rsi, ARG3_64   // = (d->ptr - d->buf) + delim_len
-    } else {
-      assert(upb_fielddef_type(f) == UPB_TYPE(GROUP));
-      |   mov   rsi, UPB_NONDELIMITED
-    }
-    |  pushframe  h, f, rsi, UPB_HANDLER_ENDSUBMSG
-
     // Call startsubmsg handler (if any).
     upb_func *startsubmsg = gethandler(h, f, UPB_HANDLER_STARTSUBMSG);
     if (startsubmsg) {
-      // upb_sflow_t startsubmsg(void *closure, upb_value fval)
-      |  mov   ARG1_64, CLOSURE
-      |  mov64 ARG2_64, gethandlerdata(h, f, UPB_HANDLER_STARTSUBMSG);
+      // upb_sflow_t startsubmsg(const upb_sinkframe *frame)
+      |  mov DECODER->tmp_len, ARG2_32
+      |  mov ARG1_64, SINKFRAME
+      |  load_handler_data h, f, UPB_HANDLER_STARTSUBMSG
       |  callp startsubmsg
       |  check_ptr_ret
       |  mov  CLOSURE, rax
     }
-    |  mov   qword SINKFRAME->closure, CLOSURE
 
-    // TODO: have to decide what to do with NULLs subhandlers (or whether to
-    // disallow them and require a full handlers tree to match the def tree).
     const upb_handlers *sub_h = upb_handlers_getsubhandlers(h, f);
-    assert(sub_h);
-    |  call  =>upb_getpclabel(plan, sub_h, STARTMSG)
-    |  popframe
+    if (sub_h) {
+      if (upb_fielddef_istagdelim(f)) {
+        |  mov   rdx, UPB_NONDELIMITED
+      } else {
+        |  mov   esi, DECODER->tmp_len
+        |  mov   rdx, PTR
+        |  sub   rdx, DECODER->buf
+        |  add   rdx, DECODER->bufstart_ofs
+        |  add   rdx, rsi   // = d->bufstart_ofs + (d->ptr - d->buf) + delim_len
+      }
+      |  pushframe  sub_h, f, rdx, UPB_HANDLER_ENDSUBMSG
+      |  call  =>upb_getpclabel(plan, sub_h, STARTMSG)
+      |  popframe
+    } else {
+      if (upb_fielddef_istagdelim(f)) {
+        // Groups with no handlers not supported yet.
+        assert(false);
+      } else {
+        |  mov   esi, DECODER->tmp_len
+        |  add   PTR, rsi
+      }
+    }
 
     // Call endsubmsg handler (if any).
     upb_func *endsubmsg = gethandler(h, f, UPB_HANDLER_ENDSUBMSG);
     if (endsubmsg) {
       // upb_flow_t endsubmsg(void *closure, upb_value fval);
-      |  mov   ARG1_64, CLOSURE
-      |  mov64 ARG2_64, gethandlerdata(h, f, UPB_HANDLER_ENDSUBMSG);
+      |  mov   ARG1_64, SINKFRAME
+      |  load_handler_data h, f, UPB_HANDLER_ENDSUBMSG
       |  callp endsubmsg
       |  check_bool_ret
     }
   } else if (!upb_fielddef_isstring(f)) {
-    |  mov ARG1_64, CLOSURE
     upb_handlertype_t handlertype = upb_handlers_getprimitivehandlertype(f);
     upb_func *handler = gethandler(h, f, handlertype);
     const upb_stdmsg_fval *fv = (void*)gethandlerdata(h, f, handlertype);
@@ -565,24 +607,25 @@ static void upb_decoderplan_jit_callcb(upb_decoderplan *plan,
     // Can't switch() on function pointers.
     if (handler == (void*)&upb_stdmsg_setint64 ||
         handler == (void*)&upb_stdmsg_setuint64) {
-      |  mov   [ARG1_64 + fv->offset], ARG3_64
+      |  mov   [CLOSURE + fv->offset], ARG2_64
       |  sethas CLOSURE, fv->hasbit
     } else if (handler == (void*)&upb_stdmsg_setdouble) {
-      |  movsd  qword [ARG1_64 + fv->offset], XMMARG1
+      |  movsd  qword [CLOSURE + fv->offset], XMMARG1
       |  sethas CLOSURE, fv->hasbit
     } else if (handler == (void*)&upb_stdmsg_setint32 ||
                handler == (void*)&upb_stdmsg_setuint32) {
-      |  mov   [ARG1_64 + fv->offset], ARG3_32
+      |  mov   [CLOSURE + fv->offset], ARG2_32
       |  sethas CLOSURE, fv->hasbit
     } else if (handler == (void*)&upb_stdmsg_setfloat) {
-      |  movss  dword [ARG1_64 + fv->offset], XMMARG1
+      |  movss  dword [CLOSURE + fv->offset], XMMARG1
       |  sethas CLOSURE, fv->hasbit
     } else if (handler == (void*)&upb_stdmsg_setbool) {
-      |  mov   [ARG1_64 + fv->offset], ARG3_8
+      |  mov   [CLOSURE + fv->offset], ARG2_8
       |  sethas CLOSURE, fv->hasbit
     } else if (handler) {
-      // Load closure and fval into arg registers.
-      |  mov64  ARG2_64, gethandlerdata(h, f, handlertype);
+      // bool value(const upb_sinkframe* frame, ctype val)
+      |  mov    ARG1_64, SINKFRAME
+      |  load_handler_data h, f, handlertype
       |  callp  handler
       |  check_bool_ret
     }
@@ -591,15 +634,27 @@ static void upb_decoderplan_jit_callcb(upb_decoderplan *plan,
 
 static uint64_t upb_get_encoded_tag(const upb_fielddef *f) {
   uint32_t tag = (upb_fielddef_number(f) << 3) |
-      upb_decoder_types[upb_fielddef_type(f)].native_wire_type;
+      upb_decoder_types[upb_fielddef_descriptortype(f)].native_wire_type;
   uint64_t encoded_tag = upb_vencode32(tag);
   // No tag should be greater than 5 bytes.
   assert(encoded_tag <= 0xffffffffff);
   return encoded_tag;
 }
 
+static void upb_decoderplan_jit_endseq(decoderplan *plan,
+                                       const upb_handlers *h,
+                                       const upb_fielddef *f) {
+  |  popframe
+  upb_func *endseq = gethandler(h, f, UPB_HANDLER_ENDSEQ);
+  if (endseq) {
+    |  mov   ARG1_64, SINKFRAME
+    |  load_handler_data h, f, UPB_HANDLER_ENDSEQ
+    |  callp endseq
+  }
+}
+
 // PTR should point to the beginning of the tag.
-static void upb_decoderplan_jit_field(upb_decoderplan *plan,
+static void upb_decoderplan_jit_field(decoderplan *plan,
                                       const upb_handlers *h,
                                       const upb_fielddef *f,
                                       const upb_fielddef *next_f) {
@@ -608,45 +663,51 @@ static void upb_decoderplan_jit_field(upb_decoderplan *plan,
   int tag_size = upb_value_size(tag);
 
   // PC-label for the dispatch table.
-  // We check the wire type (which must be loaded in edx) because the
+  // We check the wire type (which must be loaded in edi) because the
   // table is keyed on field number, not type.
   |=>upb_getpclabel(plan, f, FIELD):
-  |  cmp  edx, (tag & 0x7)
+  |  cmp  esi, (tag & 0x7)
   |  jne  ->exit_jit     // In the future: could be an unknown field or packed.
   |=>upb_getpclabel(plan, f, FIELD_NO_TYPECHECK):
   if (upb_fielddef_isseq(f)) {
-    |  mov   rsi, FRAME->end_ofs
-    |  pushframe  h, f, rsi, UPB_HANDLER_ENDSEQ
     upb_func *startseq = gethandler(h, f, UPB_HANDLER_STARTSEQ);
     if (startseq) {
-      |  mov    ARG1_64, CLOSURE
-      |  mov64  ARG2_64, gethandlerdata(h, f, UPB_HANDLER_STARTSEQ);
-      |  callp  startseq
+      |  mov   ARG1_64, SINKFRAME
+      |  load_handler_data h, f, UPB_HANDLER_STARTSEQ
+      |  callp startseq
       |  check_ptr_ret
-      |  mov    CLOSURE, rax
+      |  mov   CLOSURE, rax
     }
-    |  mov   qword SINKFRAME->closure, CLOSURE
+    |  mov   rsi, FRAME->end_ofs
+    |  pushframe  h, f, rsi, UPB_HANDLER_ENDSEQ
   }
 
   |1:  // Label for repeating this field.
 
-  upb_decoderplan_jit_decodefield(plan, upb_fielddef_type(f), tag_size, h, f);
+  upb_decoderplan_jit_decodefield(plan, tag_size, h, f);
   upb_decoderplan_jit_callcb(plan, h, f);
 
-  // Epilogue: load next tag, check for repeated field.
-  |  checkpoint  h
-  |  mov         rcx, qword [PTR]
+  // This is kind of gross; future redesign should take into account how to
+  // make this work nicely.  The difficult part is that the sequence can be
+  // broken either by end-of-message or by seeing a different field; in both
+  // cases we need to call the endseq handler, but what we do after that
+  // depends on which case triggered the end-of-sequence.
+  |  mov   DECODER->ptr, PTR
+  |  cmp   PTR, DECODER->jit_end
+  |  jae   ->exit_jit
+  |  cmp   PTR, DECODER->effective_end
+  |  jb    >2
+  if (upb_fielddef_isseq(f)) {
+    upb_decoderplan_jit_endseq(plan, h, f);
+  }
+  |  jmp  =>upb_getpclabel(plan, h, ENDOFMSG)
+  |2:
+  |  mov   rcx, qword [PTR]
   if (upb_fielddef_isseq(f)) {
     |  checktag  tag
     |  je  <1
-    upb_func *endseq = gethandler(h, f, UPB_HANDLER_ENDSEQ);
-    if (endseq) {
-      |  mov   ARG1_64, CLOSURE
-      |  mov64  ARG2_64, gethandlerdata(h, f, UPB_HANDLER_ENDSEQ);
-      |  callp endseq
-    }
-    |  popframe
-    // Load next tag again (popframe clobbered it).
+    upb_decoderplan_jit_endseq(plan, h, f);
+    // Load next tag again (popframe/endseq clobbered it).
     |  mov         rcx, qword [PTR]
   }
 
@@ -663,22 +724,22 @@ static int upb_compare_uint32(const void *a, const void *b) {
   return *(uint32_t*)a - *(uint32_t*)b;
 }
 
-static void upb_decoderplan_jit_msg(upb_decoderplan *plan,
+static void upb_decoderplan_jit_msg(decoderplan *plan,
                                     const upb_handlers *h) {
   |=>upb_getpclabel(plan, h, AFTER_STARTMSG):
-  // There was a call to get here, so we need to align the stack.
-  |  sub  rsp, 8
+  |  push  rbp
+  |  mov   rbp, rsp
   |  jmp  >1
 
   |=>upb_getpclabel(plan, h, STARTMSG):
-  // There was a call to get here, so we need to align the stack.
-  |  sub  rsp, 8
+  |  push  rbp
+  |  mov   rbp, rsp
 
   // Call startmsg handler (if any):
   upb_startmsg_handler *startmsg = upb_handlers_getstartmsg(h);
   if (startmsg) {
     // upb_flow_t startmsg(void *closure);
-    |  mov   ARG1_64, SINKFRAME->closure
+    |  mov   ARG1_64, SINKFRAME
     |  callp startmsg
     |  check_bool_ret
   }
@@ -731,17 +792,18 @@ static void upb_decoderplan_jit_msg(upb_decoderplan *plan,
   upb_endmsg_handler *endmsg = upb_handlers_getendmsg(h);
   if (endmsg) {
     // void endmsg(void *closure, upb_status *status) {
-    |  mov   ARG1_64, SINKFRAME->closure
-    |  lea   ARG2_64, DECODER->sink.status
+    |  mov   ARG1_64, SINKFRAME
+    |  mov   ARG2_64, DECODER->sink
+    |  mov   ARG2_64, SINK:ARG2_64->pipeline_
+    |  add   ARG2_64, offsetof(upb_pipeline, status_)
     |  callp endmsg
   }
 
-  // Counter previous alignment.
-  |  add  rsp, 8
+  |  leave
   |  ret
 }
 
-static void upb_decoderplan_jit(upb_decoderplan *plan) {
+static void upb_decoderplan_jit(decoderplan *plan) {
   // The JIT prologue/epilogue trampoline that is generated in this function
   // does not depend on the handlers, so it will never vary.  Ideally we would
   // put it in an object file and just link it into upb so we could have only a
@@ -763,18 +825,18 @@ static void upb_decoderplan_jit(upb_decoderplan *plan) {
   // Align stack.
   |  sub   rsp, 8
   |  mov   DECODER, ARG1_64
+  |  mov   DECODER->saved_rbp, rbp
   |  mov   FRAME, DECODER:ARG1_64->top
-  |  mov   SINKFRAME, DECODER:ARG1_64->sink.top
+  |  mov   rax, DECODER:ARG1_64->sink
+  |  mov   SINKFRAME, SINK:rax->top_
   |  mov   CLOSURE, SINKFRAME->closure
   |  mov   PTR, DECODER->ptr
 
   // TODO: push return addresses for re-entry (will be necessary for multiple
   // buffer support).
   |  call  ARG2_64
-
   |->exit_jit:
-  // Restore stack pointer to where it was before any "call" instructions
-  // inside our generated code.
+  |  mov   rbp, DECODER->saved_rbp
   |  lea   rsp, [rbp - 48]
   // Counter previous alignment.
   |  add   rsp, 8
@@ -794,10 +856,10 @@ static void upb_decoderplan_jit(upb_decoderplan *plan) {
   }
 }
 
-static void upb_decoderplan_jit_assignpclabels(upb_decoderplan *plan,
+static void upb_decoderplan_jit_assignpclabels(decoderplan *plan,
                                                const upb_handlers *h) {
   // Limit the DFS.
-  if (upb_inttable_lookupptr(&plan->pclabels, h)) return;
+  if (upb_inttable_lookupptr(&plan->pclabels, h, NULL)) return;
 
   upb_inttable_insertptr(&plan->pclabels, h,
                          upb_value_uint32(plan->pclabel_count));
@@ -832,14 +894,14 @@ static void upb_decoderplan_jit_assignpclabels(upb_decoderplan *plan,
   info->tablearray = malloc((info->max_field_number + 1) * sizeof(void*));
 }
 
-static void upb_decoderplan_makejit(upb_decoderplan *plan) {
+static void upb_decoderplan_makejit(decoderplan *plan) {
   upb_inttable_init(&plan->msginfo, UPB_CTYPE_PTR);
   plan->debug_info = NULL;
 
   // Assign pclabels.
   plan->pclabel_count = 0;
   upb_inttable_init(&plan->pclabels, UPB_CTYPE_UINT32);
-  upb_decoderplan_jit_assignpclabels(plan, plan->handlers);
+  upb_decoderplan_jit_assignpclabels(plan, plan->dest_handlers);
 
   void **globals = malloc(UPB_JIT_GLOBAL__MAX * sizeof(*globals));
   dasm_init(plan, 1);
@@ -867,7 +929,7 @@ static void upb_decoderplan_makejit(upb_decoderplan *plan) {
     const upb_handlers *h = (const upb_handlers*)upb_inttable_iter_key(&i);
     upb_jitmsginfo *mi = upb_getmsginfo(plan, h);
     // We jump to after the startmsg handler since it is called before entering
-    // the JIT (either by upb_decoder or by a previous call to the JIT).
+    // the JIT (either by upb_pbdecoder or by a previous call to the JIT).
     mi->jit_func = plan->jit_code +
         dasm_getpclabel(plan, upb_getpclabel(plan, h, AFTER_STARTMSG));
     for (uint32_t j = 0; j <= mi->max_field_number; j++) {
@@ -899,7 +961,7 @@ static void upb_decoderplan_makejit(upb_decoderplan *plan) {
 #endif
 }
 
-static void upb_decoderplan_freejit(upb_decoderplan *plan) {
+static void upb_decoderplan_freejit(decoderplan *plan) {
   upb_inttable_iter i;
   upb_inttable_begin(&i, &plan->msginfo);
   for(; !upb_inttable_done(&i); upb_inttable_next(&i)) {
@@ -913,9 +975,10 @@ static void upb_decoderplan_freejit(upb_decoderplan *plan) {
   // TODO: unregister
 }
 
-static void upb_decoder_enterjit(upb_decoder *d) {
-  if (d->plan->jit_code &&
-      d->sink.top == d->sink.stack &&
+static void upb_decoder_enterjit(upb_pbdecoder *d, decoderplan *plan) {
+  if (plan->jit_code &&
+      d->top == d->stack &&
+      d->sink->top_ == d->sink->stack &&
       d->ptr && d->ptr < d->jit_end) {
 #ifndef NDEBUG
     register uint64_t rbx asm ("rbx") = 11;
@@ -926,8 +989,8 @@ static void upb_decoder_enterjit(upb_decoder *d) {
 #endif
     // Decodes as many fields as possible, updating d->ptr appropriately,
     // before falling through to the slow(er) path.
-    void (*upb_jit_decode)(upb_decoder *d, void*) = (void*)d->plan->jit_code;
-    upb_jitmsginfo *mi = upb_getmsginfo(d->plan, d->plan->handlers);
+    void (*upb_jit_decode)(upb_pbdecoder *d, void*) = (void*)plan->jit_code;
+    upb_jitmsginfo *mi = upb_getmsginfo(plan, plan->dest_handlers);
     assert(mi);
     upb_jit_decode(d, mi->jit_func);
     assert(d->ptr <= d->end);

upb/pb/glue.c

@@ -16,33 +16,37 @@
 
 upb_def **upb_load_defs_from_descriptor(const char *str, size_t len, int *n,
                                         void *owner, upb_status *status) {
-  upb_stringsrc strsrc;
-  upb_stringsrc_init(&strsrc);
-  upb_stringsrc_reset(&strsrc, str, len);
+  // Create handlers.
+  const upb_handlers *reader_h = upb_descreader_gethandlers(&reader_h);
+  const upb_handlers *decoder_h =
+      upb_pbdecoder_gethandlers(reader_h, false, &decoder_h);
 
-  const upb_handlers *h = upb_descreader_newhandlers(&h);
-  upb_decoderplan *p = upb_decoderplan_new(h, false);
-  upb_decoder d;
-  upb_decoder_init(&d);
-  upb_handlers_unref(h, &h);
-  upb_descreader r;
-  upb_descreader_init(&r);
-  upb_decoder_resetplan(&d, p);
-  upb_decoder_resetinput(&d, upb_stringsrc_allbytes(&strsrc), &r);
+  // Create pipeline.
+  upb_pipeline pipeline;
+  upb_pipeline_init(&pipeline, NULL, 0, upb_realloc, NULL);
+  upb_pipeline_donateref(&pipeline, reader_h, &reader_h);
+  upb_pipeline_donateref(&pipeline, decoder_h, &decoder_h);
 
-  upb_success_t ret = upb_decoder_decode(&d);
-  if (status) upb_status_copy(status, upb_decoder_status(&d));
-  upb_stringsrc_uninit(&strsrc);
-  upb_decoder_uninit(&d);
-  upb_decoderplan_unref(p);
-  if (ret != UPB_OK) {
-    upb_descreader_uninit(&r);
+  // Create sinks.
+  upb_sink *reader_sink = upb_pipeline_newsink(&pipeline, reader_h);
+  upb_sink *decoder_sink = upb_pipeline_newsink(&pipeline, decoder_h);
+  upb_pbdecoder *d = upb_sinkframe_userdata(upb_sink_base(decoder_sink));
+  upb_pbdecoder_resetsink(d, reader_sink);
+
+  // Push input data.
+  bool ok = upb_bytestream_putstr(decoder_sink, str, len);
+
+  if (status) upb_status_copy(status, upb_pipeline_status(&pipeline));
+  if (!ok) {
+    upb_pipeline_uninit(&pipeline);
     return NULL;
   }
-  upb_def **defs = upb_descreader_getdefs(&r, owner, n);
+
+  upb_descreader *r = upb_sinkframe_userdata(upb_sink_base(reader_sink));
+  upb_def **defs = upb_descreader_getdefs(r, owner, n);
   upb_def **defscopy = malloc(sizeof(upb_def*) * (*n));
   memcpy(defscopy, defs, sizeof(upb_def*) * (*n));
-  upb_descreader_uninit(&r);
+  upb_pipeline_uninit(&pipeline);
 
   return defscopy;
 }

upb/pb/textprinter.c

@@ -14,8 +14,9 @@
 #include <stdlib.h>
 #include <string.h>
 
+#include "upb/sink.h"
+
 struct _upb_textprinter {
-  upb_bytesink *sink;
   int indent_depth;
   bool single_line;
   upb_status status;
@@ -24,18 +25,17 @@ struct _upb_textprinter {
 #define CHECK(x) if ((x) < 0) goto err;
 
 static int indent(upb_textprinter *p) {
+  int i;
   if (!p->single_line)
-    CHECK(upb_bytesink_putrepeated(p->sink, ' ', p->indent_depth*2));
+    for (i = 0; i < p->indent_depth * 2; i++)
+      putchar(' ');
   return 0;
-err:
   return -1;
 }
 
 static int endfield(upb_textprinter *p) {
-  CHECK(upb_bytesink_putc(p->sink, p->single_line ? ' ' : '\n'));
+  putchar(p->single_line ? ' ' : '\n');
   return 0;
-err:
-  return -1;
 }
 
 static int putescaped(upb_textprinter *p, const char *buf, size_t len,
@@ -51,7 +51,7 @@ static int putescaped(upb_textprinter *p, const char *buf, size_t len,
 
   for (; buf < end; buf++) {
     if (dstend - dst < 4) {
-      CHECK(upb_bytesink_write(p->sink, dstbuf, dst - dstbuf));
+      fwrite(dstbuf, dst - dstbuf, 1, stdout);
       dst = dstbuf;
     }
 
@@ -79,24 +79,35 @@ static int putescaped(upb_textprinter *p, const char *buf, size_t len,
     last_hex_escape = is_hex_escape;
   }
   // Flush remaining data.
-  CHECK(upb_bytesink_write(p->sink, dst, dst - dstbuf));
+  fwrite(dst, dst - dstbuf, 1, stdout);
   return 0;
-err:
-  return -1;
 }
 
 #define TYPE(name, ctype, fmt) \
-  static bool put ## name(void *_p, void *fval, ctype val) {                 \
-    upb_textprinter *p = _p;                                                 \
-    const upb_fielddef *f = fval;                                            \
-    CHECK(indent(p));                                                        \
-    CHECK(upb_bytesink_writestr(p->sink, upb_fielddef_name(f)));             \
-    CHECK(upb_bytesink_writestr(p->sink, ": "));                             \
-    CHECK(upb_bytesink_printf(p->sink, fmt, val));                           \
-    CHECK(endfield(p));                                                      \
-    return true;                                                             \
-  err:                                                                       \
-    return false;                                                            \
+  static bool put ## name(const upb_sinkframe *frame, ctype val) {             \
+    upb_textprinter *p = upb_sinkframe_userdata(frame);                        \
+    const upb_fielddef *f = upb_sinkframe_handlerdata(frame);                  \
+    CHECK(indent(p));                                                          \
+    puts(upb_fielddef_name(f));                                                \
+    puts(": ");                                                                \
+    printf(fmt, val);                                                          \
+    CHECK(endfield(p));                                                        \
+    return true;                                                               \
+  err:                                                                         \
+    return false;                                                              \
+}
+
+static bool putbool(const upb_sinkframe *frame, bool val) {
+  upb_textprinter *p = upb_sinkframe_userdata(frame);
+  const upb_fielddef *f = upb_sinkframe_handlerdata(frame);
+  CHECK(indent(p));
+  puts(upb_fielddef_name(f));
+  puts(": ");
+  puts(val ? "true" : "false");
+  CHECK(endfield(p));
+  return true;
+err:
+  return false;
 }
 
 #define STRINGIFY_HELPER(x) #x
@@ -108,72 +119,61 @@ TYPE(uint32, uint32_t, "%" PRIu32);
 TYPE(uint64, uint64_t, "%" PRIu64)
 TYPE(float,  float,    "%." STRINGIFY_MACROVAL(FLT_DIG) "g")
 TYPE(double, double,   "%." STRINGIFY_MACROVAL(DBL_DIG) "g")
-TYPE(bool,   bool,     "%hhu");
 
 // Output a symbolic value from the enum if found, else just print as int32.
-static bool putenum(void *_p, void *fval, int32_t val) {
-
-  upb_textprinter *p = _p;
-  const upb_fielddef *f = fval;
+static bool putenum(const upb_sinkframe *frame, int32_t val) {
+  const upb_fielddef *f = upb_sinkframe_handlerdata(frame);
   const upb_enumdef *enum_def = upb_downcast_enumdef(upb_fielddef_subdef(f));
   const char *label = upb_enumdef_iton(enum_def, val);
   if (label) {
-    CHECK(upb_bytesink_writestr(p->sink, label));
+    puts(label);
   } else {
-    CHECK(putint32(_p, fval, val));
+    CHECK(putint32(frame, val));
   }
   return true;
 err:
   return false;
 }
 
-static void *startstr(void *_p, void *fval, size_t size_hint) {
+static void *startstr(const upb_sinkframe *frame, size_t size_hint) {
   UPB_UNUSED(size_hint);
-  UPB_UNUSED(fval);
-  upb_textprinter *p = _p;
-  CHECK(upb_bytesink_putc(p->sink, '"'));
+  upb_textprinter *p = upb_sinkframe_userdata(frame);
+  putchar('"');
   return p;
-err:
-  return UPB_BREAK;
 }
 
-static bool endstr(void *_p, void *fval) {
-  UPB_UNUSED(fval);
-  upb_textprinter *p = _p;
-  CHECK(upb_bytesink_putc(p->sink, '"'));
+static bool endstr(const upb_sinkframe *frame) {
+  putchar('"');
   return true;
-err:
-  return false;
 }
 
-static size_t putstr(void *_p, void *fval, const char *buf, size_t len) {
-  upb_textprinter *p = _p;
-  const upb_fielddef *f = fval;
-  CHECK(putescaped(p, buf, len, upb_fielddef_type(f) == UPB_TYPE(STRING)));
+static size_t putstr(const upb_sinkframe *frame, const char *buf, size_t len) {
+  upb_textprinter *p = upb_sinkframe_userdata(frame);
+  const upb_fielddef *f = upb_sinkframe_handlerdata(frame);
+  CHECK(putescaped(p, buf, len, upb_fielddef_type(f) == UPB_TYPE_STRING));
   return len;
 err:
   return 0;
 }
 
-static void *startsubmsg(void *_p, void *fval) {
-  upb_textprinter *p = _p;
-  const upb_fielddef *f = fval;
+static void *startsubmsg(const upb_sinkframe *frame) {
+  upb_textprinter *p = upb_sinkframe_userdata(frame);
+  const upb_fielddef *f = upb_sinkframe_handlerdata(frame);
   CHECK(indent(p));
-  CHECK(upb_bytesink_printf(p->sink, "%s {", upb_fielddef_name(f)));
+  printf("%s {", upb_fielddef_name(f));
   if (!p->single_line)
-    CHECK(upb_bytesink_putc(p->sink, '\n'));
+    putchar('\n');
   p->indent_depth++;
-  return _p;
+  return p;
 err:
   return UPB_BREAK;
 }
 
-static bool endsubmsg(void *_p, void *fval) {
-  UPB_UNUSED(fval);
-  upb_textprinter *p = _p;
+static bool endsubmsg(const upb_sinkframe *frame) {
+  upb_textprinter *p = upb_sinkframe_userdata(frame);
   p->indent_depth--;
   CHECK(indent(p));
-  CHECK(upb_bytesink_putc(p->sink, '}'));
+  putchar('}');
   CHECK(endfield(p));
   return true;
 err:
@@ -187,9 +187,7 @@ upb_textprinter *upb_textprinter_new() {
 
 void upb_textprinter_free(upb_textprinter *p) { free(p); }
 
-void upb_textprinter_reset(upb_textprinter *p, upb_bytesink *sink,
-                           bool single_line) {
-  p->sink = sink;
+void upb_textprinter_reset(upb_textprinter *p, bool single_line) {
   p->single_line = single_line;
   p->indent_depth = 0;
 }
@@ -202,21 +200,15 @@ static void onmreg(void *c, upb_handlers *h) {
     upb_fielddef *f = upb_msg_iter_field(&i);
     switch (upb_fielddef_type(f)) {
       case UPB_TYPE_INT32:
-      case UPB_TYPE_SINT32:
-      case UPB_TYPE_SFIXED32:
         upb_handlers_setint32(h, f, putint32, f, NULL);
         break;
-      case UPB_TYPE_SINT64:
-      case UPB_TYPE_SFIXED64:
       case UPB_TYPE_INT64:
         upb_handlers_setint64(h, f, putint64, f, NULL);
         break;
       case UPB_TYPE_UINT32:
-      case UPB_TYPE_FIXED32:
         upb_handlers_setuint32(h, f, putuint32, f, NULL);
         break;
       case UPB_TYPE_UINT64:
-      case UPB_TYPE_FIXED64:
         upb_handlers_setuint64(h, f, putuint64, f, NULL);
         break;
       case UPB_TYPE_FLOAT:
@@ -234,7 +226,6 @@ static void onmreg(void *c, upb_handlers *h) {
         upb_handlers_setstring(h, f, putstr, f, NULL);
         upb_handlers_setendstr(h, f, endstr, f, NULL);
         break;
-      case UPB_TYPE_GROUP:
       case UPB_TYPE_MESSAGE:
         upb_handlers_setstartsubmsg(h, f, &startsubmsg, f, NULL);
         upb_handlers_setendsubmsg(h, f, &endsubmsg, f, NULL);
@@ -250,5 +241,5 @@ static void onmreg(void *c, upb_handlers *h) {
 
 const upb_handlers *upb_textprinter_newhandlers(const void *owner,
                                                 const upb_msgdef *m) {
-  return upb_handlers_newfrozen(m, owner, &onmreg, NULL);
+  return upb_handlers_newfrozen(m, NULL, owner, &onmreg, NULL);
 }

upb/pb/textprinter.h

@@ -8,7 +8,6 @@
 #ifndef UPB_TEXT_H_
 #define UPB_TEXT_H_
 
-#include "upb/bytestream.h"
 #include "upb/handlers.h"
 
 #ifdef __cplusplus
@@ -20,8 +19,7 @@ typedef struct _upb_textprinter upb_textprinter;
 
 upb_textprinter *upb_textprinter_new();
 void upb_textprinter_free(upb_textprinter *p);
-void upb_textprinter_reset(upb_textprinter *p, upb_bytesink *sink,
-                           bool single_line);
+void upb_textprinter_reset(upb_textprinter *p, bool single_line);
 const upb_handlers *upb_textprinter_newhandlers(const void *owner,
                                                 const upb_msgdef *m);
 

upb/pb/varint.h

@@ -36,10 +36,14 @@ typedef enum {
 
 /* Zig-zag encoding/decoding **************************************************/
 
-INLINE int32_t upb_zzdec_32(uint32_t n) { return (n >> 1) ^ -(int32_t)(n & 1); }
-INLINE int64_t upb_zzdec_64(uint64_t n) { return (n >> 1) ^ -(int64_t)(n & 1); }
-INLINE uint32_t upb_zzenc_32(int32_t n) { return (n << 1) ^ (n >> 31); }
-INLINE uint64_t upb_zzenc_64(int64_t n) { return (n << 1) ^ (n >> 63); }
+UPB_INLINE int32_t upb_zzdec_32(uint32_t n) {
+  return (n >> 1) ^ -(int32_t)(n & 1);
+}
+UPB_INLINE int64_t upb_zzdec_64(uint64_t n) {
+  return (n >> 1) ^ -(int64_t)(n & 1);
+}
+UPB_INLINE uint32_t upb_zzenc_32(int32_t n) { return (n << 1) ^ (n >> 31); }
+UPB_INLINE uint64_t upb_zzenc_64(int64_t n) { return (n << 1) ^ (n >> 63); }
 
 /* Decoding *******************************************************************/
 
@@ -65,7 +69,7 @@ upb_decoderet upb_vdecode_max8_massimino(upb_decoderet r);
 // up to 10 bytes, so it must not be used unless there are at least ten bytes
 // left in the buffer!
 #define UPB_VARINT_DECODER_CHECK2(name, decode_max8_function)                  \
-INLINE upb_decoderet upb_vdecode_check2_ ## name(const char *_p) {             \
+UPB_INLINE upb_decoderet upb_vdecode_check2_ ## name(const char *_p) {         \
   uint8_t *p = (uint8_t*)_p;                                                   \
   if ((*p & 0x80) == 0) { upb_decoderet r = {_p + 1, *p & 0x7fU}; return r; }  \
   upb_decoderet r = {_p + 2, (*p & 0x7fU) | ((*(p + 1) & 0x7fU) << 7)};        \
@@ -81,21 +85,21 @@ UPB_VARINT_DECODER_CHECK2(massimino, upb_vdecode_max8_massimino);
 
 // Our canonical functions for decoding varints, based on the currently
 // favored best-performing implementations.
-INLINE upb_decoderet upb_vdecode_fast(const char *p) {
+UPB_INLINE upb_decoderet upb_vdecode_fast(const char *p) {
   if (sizeof(long) == 8)
     return upb_vdecode_check2_massimino(p);
   else
     return upb_vdecode_check2_branch32(p);
 }
 
-INLINE upb_decoderet upb_vdecode_max8_fast(upb_decoderet r) {
+UPB_INLINE upb_decoderet upb_vdecode_max8_fast(upb_decoderet r) {
   return upb_vdecode_max8_massimino(r);
 }
 
 
 /* Encoding *******************************************************************/
 
-INLINE int upb_value_size(uint64_t val) {
+UPB_INLINE int upb_value_size(uint64_t val) {
 #ifdef __GNUC__
   int high_bit = 63 - __builtin_clzll(val);  // 0-based, undef if val == 0.
 #else
@@ -110,7 +114,7 @@ INLINE int upb_value_size(uint64_t val) {
 // bytes long), returning how many bytes were used.
 //
 // TODO: benchmark and optimize if necessary.
-INLINE size_t upb_vencode64(uint64_t val, char *buf) {
+UPB_INLINE size_t upb_vencode64(uint64_t val, char *buf) {
   if (val == 0) { buf[0] = 0; return 1; }
   size_t i = 0;
   while (val) {
@@ -123,7 +127,7 @@ INLINE size_t upb_vencode64(uint64_t val, char *buf) {
 }
 
 // Encodes a 32-bit varint, *not* sign-extended.
-INLINE uint64_t upb_vencode32(uint32_t val) {
+UPB_INLINE uint64_t upb_vencode32(uint32_t val) {
   char buf[UPB_PB_VARINT_MAX_LEN];
   size_t bytes = upb_vencode64(val, buf);
   uint64_t ret = 0;

upb/refcounted.c

@@ -136,8 +136,8 @@ static const void *unobfuscate_v(upb_value x) {
 static upb_inttable reftracks = UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR);
 
 static upb_inttable *trygettab(const void *p) {
-  const upb_value *v = upb_inttable_lookupptr(&reftracks, p);
-  return v ? upb_value_getptr(*v) : NULL;
+  upb_value v;
+  return upb_inttable_lookupptr(&reftracks, p, &v) ? upb_value_getptr(v) : NULL;
 }
 
 // Gets or creates the tracking table for the given owner.
@@ -155,9 +155,9 @@ static upb_inttable *gettab(const void *p) {
 static void track(const upb_refcounted *r, const void *owner, bool ref2) {
   upb_lock();
   upb_inttable *refs = gettab(owner);
-  const upb_value *v = upb_inttable_lookup(refs, obfuscate(r));
-  if (v) {
-    trackedref *ref = (trackedref*)unobfuscate_v(*v);
+  upb_value v;
+  if (upb_inttable_lookup(refs, obfuscate(r), &v)) {
+    trackedref *ref = (trackedref*)unobfuscate_v(v);
     // Since we allow multiple ref2's for the same to/from pair without
     // allocating separate memory for each one, we lose the fine-grained
     // tracking behavior we get with regular refs.  Since ref2s only happen
@@ -177,10 +177,11 @@ static void track(const upb_refcounted *r, const void *owner, bool ref2) {
 static void untrack(const upb_refcounted *r, const void *owner, bool ref2) {
   upb_lock();
   upb_inttable *refs = gettab(owner);
-  const upb_value *v = upb_inttable_lookup(refs, obfuscate(r));
+  upb_value v;
+  bool found = upb_inttable_lookup(refs, obfuscate(r), &v);
   // This assert will fail if an owner attempts to release a ref it didn't have.
-  assert(v);
-  trackedref *ref = (trackedref*)unobfuscate_v(*v);
+  UPB_ASSERT_VAR(found, found);
+  trackedref *ref = (trackedref*)unobfuscate_v(v);
   assert(ref->is_ref2 == ref2);
   if (--ref->count == 0) {
     free(ref);
@@ -197,9 +198,10 @@ static void untrack(const upb_refcounted *r, const void *owner, bool ref2) {
 static void checkref(const upb_refcounted *r, const void *owner, bool ref2) {
   upb_lock();
   upb_inttable *refs = gettab(owner);
-  const upb_value *v = upb_inttable_lookup(refs, obfuscate(r));
-  assert(v);
-  trackedref *ref = (trackedref*)unobfuscate_v(*v);
+  upb_value v;
+  bool found = upb_inttable_lookup(refs, obfuscate(r), &v);
+  UPB_ASSERT_VAR(found, found);
+  trackedref *ref = (trackedref*)unobfuscate_v(v);
   assert(ref->obj == r);
   assert(ref->is_ref2 == ref2);
   upb_unlock();
@@ -339,14 +341,16 @@ UPB_NORETURN static void oom(tarjan *t) {
 }
 
 uint64_t trygetattr(const tarjan *t, const upb_refcounted *r) {
-  const upb_value *v = upb_inttable_lookupptr(&t->objattr, r);
-  return v ? upb_value_getuint64(*v) : 0;
+  upb_value v;
+  return upb_inttable_lookupptr(&t->objattr, r, &v) ?
+      upb_value_getuint64(v) : 0;
 }
 
 uint64_t getattr(const tarjan *t, const upb_refcounted *r) {
-  const upb_value *v = upb_inttable_lookupptr(&t->objattr, r);
-  assert(v);
-  return upb_value_getuint64(*v);
+  upb_value v;
+  bool found = upb_inttable_lookupptr(&t->objattr, r, &v);
+  UPB_ASSERT_VAR(found, found);
+  return upb_value_getuint64(v);
 }
 
 void setattr(tarjan *t, const upb_refcounted *r, uint64_t attr) {
@@ -420,9 +424,10 @@ static void set_lowlink(tarjan *t, const upb_refcounted *r, uint32_t lowlink) {
 uint32_t *group(tarjan *t, upb_refcounted *r) {
   assert(color(t, r) == WHITE);
   uint64_t groupnum = getattr(t, r) >> 8;
-  const upb_value *v = upb_inttable_lookup(&t->groups, groupnum);
-  assert(v);
-  return upb_value_getptr(*v);
+  upb_value v;
+  bool found = upb_inttable_lookup(&t->groups, groupnum, &v);
+  UPB_ASSERT_VAR(found, found);
+  return upb_value_getptr(v);
 }
 
 // If the group leader for this object's group has not previously been set,
@@ -430,10 +435,11 @@ uint32_t *group(tarjan *t, upb_refcounted *r) {
 static upb_refcounted *groupleader(tarjan *t, upb_refcounted *r) {
   assert(color(t, r) == WHITE);
   uint64_t leader_slot = (getattr(t, r) >> 8) + 1;
-  const upb_value *v = upb_inttable_lookup(&t->groups, leader_slot);
-  assert(v);
-  if (upb_value_getptr(*v)) {
-    return upb_value_getptr(*v);
+  upb_value v;
+  bool found = upb_inttable_lookup(&t->groups, leader_slot, &v);
+  UPB_ASSERT_VAR(found, found);
+  if (upb_value_getptr(v)) {
+    return upb_value_getptr(v);
   } else {
     upb_inttable_remove(&t->groups, leader_slot, NULL);
     upb_inttable_insert(&t->groups, leader_slot, upb_value_ptr(r));

upb/sink.c

@@ -7,61 +7,258 @@
 
 #include "upb/sink.h"
 
+#include <stdlib.h>
+#include <string.h>
+
+static void upb_sink_init(upb_sink *s, const upb_handlers *h, upb_pipeline *p);
+static void upb_sink_resetobj(void *obj);
+static const upb_frametype upb_sink_frametype;
+
 static bool chkstack(upb_sink *s) {
-  if (s->top + 1 >= s->limit) {
-    upb_status_seterrliteral(&s->status, "Nesting too deep.");
+  if (s->top_ + 1 >= s->limit) {
+    upb_status_seterrliteral(&s->pipeline_->status_, "Nesting too deep.");
     return false;
   } else {
     return true;
   }
 }
 
-static upb_selector_t getselector(const upb_fielddef *f,
-                                  upb_handlertype_t type) {
-  upb_selector_t selector;
-  bool ok = upb_getselector(f, type, &selector);
-  UPB_ASSERT_VAR(ok, ok);
-  return selector;
+#define alignof(type) offsetof (struct { char c; type member; }, member)
+
+typedef union { double u; void *p; long l; } maxalign_t;
+static const size_t maxalign = alignof(maxalign_t);
+
+static void *align_up(void *p) {
+  if (!p) return NULL;
+  uintptr_t val = (uintptr_t)p;
+  uintptr_t aligned =
+      val % maxalign == 0 ? val : val + maxalign - (val % maxalign);
+  return (void*)aligned;
+}
+
+void *upb_realloc(void *ud, void *ptr, size_t size) {
+  UPB_UNUSED(ud);
+  return realloc(ptr, size);
+}
+
+
+/* upb_pipeline ***************************************************************/
+
+// For the moment we get fixed-size blocks of this size, but we could change
+// this strategy if necessary.
+#define BLOCK_SIZE 8192
+
+struct region {
+  struct region *prev;
+  maxalign_t data[1];  // Region data follows.
+};
+
+size_t regionsize(size_t usable_size) {
+  return sizeof(struct region) - sizeof(maxalign_t) + usable_size;
+}
+
+struct obj {
+  struct obj *prev;
+  const upb_frametype *ft;
+  maxalign_t data;  // Region data follows.
+};
+
+size_t objsize(size_t memsize) {
+  return sizeof(struct obj) - sizeof(maxalign_t) + memsize;
+}
+
+void upb_pipeline_init(upb_pipeline *p, void *initial_mem, size_t initial_size,
+                       void *(*realloc)(void *ud, void *ptr, size_t bytes),
+                       void *ud) {
+  p->realloc = realloc;
+  p->ud = ud;
+  p->bump_top = initial_mem;
+  p->bump_limit = initial_mem ? initial_mem + initial_size : NULL;
+  p->region_head = NULL;
+  p->obj_head = NULL;
+  p->last_alloc = NULL;
+  upb_status_init(&p->status_);
+}
+
+void upb_pipeline_uninit(upb_pipeline *p) {
+  for (struct obj *o = p->obj_head; o; o = o->prev) {
+    if (o->ft->uninit)
+      o->ft->uninit(&o->data);
+  }
+
+  for (struct region *r = p->region_head; r; ) {
+    struct region *prev = r->prev;
+    p->realloc(p->ud, r, 0);
+    r = prev;
+  }
+  upb_status_uninit(&p->status_);
+}
+
+void *upb_pipeline_alloc(upb_pipeline *p, size_t bytes) {
+  void *mem = align_up(p->bump_top);
+  if (!mem || mem > p->bump_limit || p->bump_limit - mem < bytes) {
+    size_t size = regionsize(UPB_MAX(BLOCK_SIZE, bytes));
+    struct region *r;
+    if (!p->realloc || !(r = p->realloc(p->ud, NULL, size))) {
+      return NULL;
+    }
+    r->prev = p->region_head;
+    p->region_head = r;
+    p->bump_limit = (char*)r + size;
+    mem = &r->data[0];
+    assert(p->bump_limit > mem);
+    assert(p->bump_limit - mem >= bytes);
+  }
+  p->bump_top = mem + bytes;
+  p->last_alloc = mem;
+  return mem;
+}
+
+void *upb_pipeline_realloc(upb_pipeline *p, void *ptr,
+                           size_t oldsize, size_t bytes) {
+  if (ptr && ptr == p->last_alloc &&
+      p->bump_limit - ptr >= bytes) {
+    p->bump_top = ptr + bytes;
+    return ptr;
+  } else {
+    void *mem = upb_pipeline_alloc(p, bytes);
+    memcpy(mem, ptr, oldsize);
+    return mem;
+  }
+}
+
+void *upb_pipeline_allocobj(upb_pipeline *p, const upb_frametype *ft) {
+  struct obj *obj = upb_pipeline_alloc(p, objsize(ft->size));
+  if (!obj) return NULL;
+
+  obj->prev = p->obj_head;
+  obj->ft = ft;
+  p->obj_head = obj;
+  if (ft->init) ft->init(&obj->data);
+  return &obj->data;
+}
+
+void upb_pipeline_reset(upb_pipeline *p) {
+  upb_status_clear(&p->status_);
+  for (struct obj *o = p->obj_head; o; o = o->prev) {
+    if (o->ft->reset)
+      o->ft->reset(&o->data);
+  }
+}
+
+upb_sink *upb_pipeline_newsink(upb_pipeline *p, const upb_handlers *handlers) {
+  upb_sink *s = upb_pipeline_allocobj(p, &upb_sink_frametype);
+  upb_sink_init(s, handlers, p);
+  return s;
+}
+
+const upb_status *upb_pipeline_status(const upb_pipeline *p) {
+  return &p->status_;
+}
+
+typedef struct {
+  const upb_handlers *h;
+} handlersref_t;
+
+static void freehandlersref(void *r) {
+  handlersref_t *ref = r;
+  upb_handlers_unref(ref->h, &ref->h);
+}
+
+static const upb_frametype handlersref_frametype = {
+  sizeof(handlersref_t),
+  NULL,
+  freehandlersref,
+  NULL,
+};
+
+void upb_pipeline_donateref(
+    upb_pipeline *p, const upb_handlers *h, const void *owner) {
+  handlersref_t *ref = upb_pipeline_allocobj(p, &handlersref_frametype);
+  upb_handlers_donateref(h, owner, &ref->h);
+  ref->h = h;
 }
 
-void upb_sink_init(upb_sink *s, const upb_handlers *h) {
+
+/* upb_sinkframe **************************************************************/
+
+int upb_sinkframe_depth(const upb_sinkframe* frame) {
+  return frame - frame->sink_->stack;
+}
+
+const upb_handlers* upb_sinkframe_handlers(const upb_sinkframe* frame) {
+  return frame->h;
+}
+
+upb_pipeline *upb_sinkframe_pipeline(const upb_sinkframe* frame) {
+  return frame->sink_->pipeline_;
+}
+
+
+/* upb_sink *******************************************************************/
+
+static const upb_frametype upb_sink_frametype = {
+  sizeof(upb_sink),
+  NULL,
+  NULL,
+  upb_sink_resetobj,
+};
+
+void upb_sink_reset(upb_sink *s, void *closure) {
+  s->top_ = s->stack;
+  s->top_->closure = closure;
+}
+
+static void upb_sink_resetobj(void *obj) {
+  upb_sink *s = obj;
+  s->top_ = s->stack;
+}
+
+static void upb_sink_init(upb_sink *s, const upb_handlers *h, upb_pipeline *p) {
+  s->pipeline_ = p;
   s->limit = &s->stack[UPB_MAX_NESTING];
-  s->top = NULL;
   s->stack[0].h = h;
-  upb_status_init(&s->status);
+  s->top_ = s->stack;
+  if (h->ft) {
+    s->stack[0].closure = upb_pipeline_allocobj(p, h->ft);
+  }
 }
 
-void upb_sink_reset(upb_sink *s, void *closure) {
-  s->top = s->stack;
-  s->top->closure = closure;
+const upb_sinkframe *upb_sink_top(const upb_sink *s) {
+  return s->top_;
+}
+
+const upb_sinkframe *upb_sink_base(const upb_sink *s) {
+  return s->stack;
 }
 
-void upb_sink_uninit(upb_sink *s) {
-  upb_status_uninit(&s->status);
+upb_pipeline *upb_sink_pipeline(const upb_sink *s) {
+  return s->pipeline_;
 }
 
 bool upb_sink_startmsg(upb_sink *s) {
-  const upb_handlers *h = s->top->h;
+  const upb_handlers *h = s->top_->h;
   upb_startmsg_handler *startmsg = upb_handlers_getstartmsg(h);
-  return startmsg ? startmsg(s->top->closure) : true;
+  return startmsg ? startmsg(s->top_) : true;
 }
 
-void upb_sink_endmsg(upb_sink *s, upb_status *status) {
-  UPB_UNUSED(status);
-  assert(s->top == s->stack);
-  upb_endmsg_handler *endmsg = upb_handlers_getendmsg(s->top->h);
-  if (endmsg) endmsg(s->top->closure, &s->status);
+void upb_sink_endmsg(upb_sink *s) {
+  assert(s->top_ == s->stack);
+  upb_endmsg_handler *endmsg = upb_handlers_getendmsg(s->top_->h);
+  if (endmsg) {
+    endmsg(s->top_, &s->pipeline_->status_);
+  }
 }
 
 #define PUTVAL(type, ctype, htype) \
-  bool upb_sink_put ## type(upb_sink *s, const upb_fielddef *f, ctype val) { \
-    upb_selector_t selector; \
-    if (!upb_getselector(f, UPB_HANDLER_ ## htype, &selector)) return false; \
+  bool upb_sink_put ## type(upb_sink *s, upb_selector_t sel, ctype val) { \
+    const upb_handlers *h = s->top_->h; \
     upb_ ## type ## _handler *handler = (upb_ ## type ## _handler*) \
-        upb_handlers_gethandler(s->top->h, selector); \
+        upb_handlers_gethandler(h, sel); \
     if (handler) { \
-      void *data = upb_handlers_gethandlerdata(s->top->h, selector); \
-      if (!handler(s->top->closure, data, val)) return false; \
+      s->top_->u.handler_data = upb_handlers_gethandlerdata(h, sel); \
+      bool ok = handler(s->top_, val); \
+      if (!ok) return false; \
     } \
     return true; \
   }
@@ -75,131 +272,153 @@ PUTVAL(double, double,          DOUBLE);
 PUTVAL(bool,   bool,            BOOL);
 #undef PUTVAL
 
-size_t upb_sink_putstring(upb_sink *s, const upb_fielddef *f,
+size_t upb_sink_putstring(upb_sink *s, upb_selector_t sel,
                           const char *buf, size_t n) {
-  upb_selector_t selector;
-  if (!upb_getselector(f, UPB_HANDLER_STRING, &selector)) return false;
-  upb_string_handler *handler = (upb_string_handler*)
-      upb_handlers_gethandler(s->top->h, selector);
+  const upb_handlers *h = s->top_->h;
+  upb_string_handler *handler =
+      (upb_string_handler*)upb_handlers_gethandler(h, sel);
+
   if (handler) {
-    void *data = upb_handlers_gethandlerdata(s->top->h, selector); \
-    return handler(s->top->closure, data, buf, n);
+    s->top_->u.handler_data = upb_handlers_gethandlerdata(h, sel);;
+    n = handler(s->top_, buf, n);
   }
+
   return n;
 }
 
-bool upb_sink_startseq(upb_sink *s, const upb_fielddef *f) {
-  assert(upb_fielddef_isseq(f));
+bool upb_sink_startseq(upb_sink *s, upb_selector_t sel) {
   if (!chkstack(s)) return false;
 
-  void *subc = s->top->closure;
-  const upb_handlers *h = s->top->h;
-  upb_selector_t selector;
-  if (!upb_getselector(f, UPB_HANDLER_STARTSEQ, &selector)) return false;
+  void *subc = s->top_->closure;
+  const upb_handlers *h = s->top_->h;
   upb_startfield_handler *startseq =
-      (upb_startfield_handler*)upb_handlers_gethandler(h, selector);
+      (upb_startfield_handler*)upb_handlers_gethandler(h, sel);
+
   if (startseq) {
-    subc = startseq(s->top->closure, upb_handlers_gethandlerdata(h, selector));
-    if (!subc) return false;
+    s->top_->u.handler_data = upb_handlers_gethandlerdata(h, sel);
+    subc = startseq(s->top_);
+    if (subc == UPB_BREAK) {
+      return false;
+    }
   }
 
-  ++s->top;
-  s->top->end = getselector(f, UPB_HANDLER_ENDSEQ);
-  s->top->h = h;
-  s->top->closure = subc;
+  s->top_->u.selector = upb_getendselector(sel);
+  ++s->top_;
+  s->top_->h = h;
+  s->top_->closure = subc;
+  s->top_->sink_ = s;
   return true;
 }
 
-bool upb_sink_endseq(upb_sink *s, const upb_fielddef *f) {
-  upb_selector_t selector = s->top->end;
-  assert(selector == getselector(f, UPB_HANDLER_ENDSEQ));
-  --s->top;
+bool upb_sink_endseq(upb_sink *s, upb_selector_t sel) {
+  --s->top_;
+  assert(sel == s->top_->u.selector);
 
-  const upb_handlers *h = s->top->h;
+  const upb_handlers *h = s->top_->h;
   upb_endfield_handler *endseq =
-      (upb_endfield_handler*)upb_handlers_gethandler(h, selector);
-  return endseq ?
-      endseq(s->top->closure, upb_handlers_gethandlerdata(h, selector)) :
-      true;
+      (upb_endfield_handler*)upb_handlers_gethandler(h, sel);
+
+  if (endseq) {
+    bool ok = endseq(s->top_);
+    if (!ok) {
+      ++s->top_;
+      return false;
+    }
+  }
+
+  return true;
 }
 
-bool upb_sink_startstr(upb_sink *s, const upb_fielddef *f, size_t size_hint) {
-  assert(upb_fielddef_isstring(f));
+bool upb_sink_startstr(upb_sink *s, upb_selector_t sel, size_t size_hint) {
   if (!chkstack(s)) return false;
 
-  void *subc = s->top->closure;
-  const upb_handlers *h = s->top->h;
-  upb_selector_t selector;
-  if (!upb_getselector(f, UPB_HANDLER_STARTSTR, &selector)) return false;
+  void *subc = s->top_->closure;
+  const upb_handlers *h = s->top_->h;
   upb_startstr_handler *startstr =
-      (upb_startstr_handler*)upb_handlers_gethandler(h, selector);
+      (upb_startstr_handler*)upb_handlers_gethandler(h, sel);
+
   if (startstr) {
-    subc = startstr(
-        s->top->closure, upb_handlers_gethandlerdata(h, selector), size_hint);
-    if (!subc) return false;
+    s->top_->u.handler_data = upb_handlers_gethandlerdata(h, sel);
+    subc = startstr(s->top_, size_hint);
+    if (subc == UPB_BREAK) {
+      return false;
+    }
   }
 
-  ++s->top;
-  s->top->end = getselector(f, UPB_HANDLER_ENDSTR);
-  s->top->h = h;
-  s->top->closure = subc;
+  s->top_->u.selector = upb_getendselector(sel);
+  ++s->top_;
+  s->top_->h = h;
+  s->top_->closure = subc;
+  s->top_->sink_ = s;
   return true;
 }
 
-bool upb_sink_endstr(upb_sink *s, const upb_fielddef *f) {
-  upb_selector_t selector = s->top->end;
-  assert(selector == getselector(f, UPB_HANDLER_ENDSTR));
-  --s->top;
-
-  const upb_handlers *h = s->top->h;
+bool upb_sink_endstr(upb_sink *s, upb_selector_t sel) {
+  --s->top_;
+  assert(sel == s->top_->u.selector);
+  const upb_handlers *h = s->top_->h;
   upb_endfield_handler *endstr =
-      (upb_endfield_handler*)upb_handlers_gethandler(h, selector);
-  return endstr ?
-      endstr(s->top->closure, upb_handlers_gethandlerdata(h, selector)) :
-      true;
+      (upb_endfield_handler*)upb_handlers_gethandler(h, sel);
+
+  if (endstr) {
+    bool ok = endstr(s->top_);
+    if (!ok) {
+      ++s->top_;
+      return false;
+    }
+  }
+
+  return true;
 }
 
-bool upb_sink_startsubmsg(upb_sink *s, const upb_fielddef *f) {
-  assert(upb_fielddef_issubmsg(f));
+bool upb_sink_startsubmsg(upb_sink *s, upb_selector_t sel) {
   if (!chkstack(s)) return false;
 
-  const upb_handlers *h = s->top->h;
-  upb_selector_t selector;
-  if (!upb_getselector(f, UPB_HANDLER_STARTSUBMSG, &selector)) return false;
+  void *subc = s->top_->closure;
+  const upb_handlers *h = s->top_->h;
   upb_startfield_handler *startsubmsg =
-      (upb_startfield_handler*)upb_handlers_gethandler(h, selector);
-  void *subc = s->top->closure;
+      (upb_startfield_handler*)upb_handlers_gethandler(h, sel);
 
   if (startsubmsg) {
-    void *data = upb_handlers_gethandlerdata(h, selector);
-    subc = startsubmsg(s->top->closure, data);
-    if (!subc) return false;
+    s->top_->u.handler_data = upb_handlers_gethandlerdata(h, sel);
+    subc = startsubmsg(s->top_);
+    if (subc == UPB_BREAK) {
+      return false;
+    }
   }
 
-  ++s->top;
-  s->top->end = getselector(f, UPB_HANDLER_ENDSUBMSG);
-  s->top->h = upb_handlers_getsubhandlers(h, f);
-  s->top->closure = subc;
+  s->top_->u.selector= upb_getendselector(sel);
+  ++s->top_;
+  s->top_->h = upb_handlers_getsubhandlers_sel(h, sel);
+  // TODO: should add support for submessages without any handlers
+  assert(s->top_->h);
+  s->top_->closure = subc;
+  s->top_->sink_ = s;
   upb_sink_startmsg(s);
   return true;
 }
 
-bool upb_sink_endsubmsg(upb_sink *s, const upb_fielddef *f) {
-  upb_selector_t selector = s->top->end;
-  assert(selector == getselector(f, UPB_HANDLER_ENDSUBMSG));
+bool upb_sink_endsubmsg(upb_sink *s, upb_selector_t sel) {
+  upb_endmsg_handler *endmsg = upb_handlers_getendmsg(s->top_->h);
+  if (endmsg) endmsg(s->top_, &s->pipeline_->status_);
+  --s->top_;
 
-  upb_endmsg_handler *endmsg = upb_handlers_getendmsg(s->top->h);
-  if (endmsg) endmsg(s->top->closure, &s->status);
-  --s->top;
+  assert(sel == s->top_->u.selector);
+  const upb_handlers *h = s->top_->h;
+  upb_endfield_handler *endsubmsg =
+      (upb_endfield_handler*)upb_handlers_gethandler(h, sel);
 
-  const upb_handlers *h = s->top->h;
-  upb_endfield_handler *endfield =
-      (upb_endfield_handler*)upb_handlers_gethandler(h, selector);
-  return endfield ?
-      endfield(s->top->closure, upb_handlers_gethandlerdata(h, selector)) :
-      true;
+  if (endsubmsg) {
+    bool ok = endsubmsg(s->top_);
+    if (!ok) {
+      ++s->top_;
+      return false;
+    }
+  }
+
+  return true;
 }
 
 const upb_handlers *upb_sink_tophandlers(upb_sink *s) {
-  return s->top->h;
+  return s->top_->h;
 }

upb/sink.h

@@ -23,60 +23,451 @@
 #include "upb/handlers.h"
 
 #ifdef __cplusplus
-extern "C" {
+namespace upb {
+class Pipeline;
+class Sink;
+template <int size> class SeededPipeline;
+}
+typedef upb::Pipeline upb_pipeline;
+typedef upb::Sink upb_sink;
+UPB_INLINE upb_sink* upb_sinkframe_sink(const upb_sinkframe* frame);
+UPB_INLINE void* upb_sinkframe_userdata(const upb_sinkframe* frame);
+UPB_INLINE void* upb_sinkframe_handlerdata(const upb_sinkframe* frame);
+#else
+struct upb_pipeline;
+struct upb_sink;
+typedef struct upb_pipeline upb_pipeline;
+typedef struct upb_sink upb_sink;
+#endif
+
+struct upb_frametype {
+  size_t size;
+  void (*init)(void* obj);
+  void (*uninit)(void* obj);
+  void (*reset)(void* obj);
+};
+
+#ifdef __cplusplus
+
+// A upb::Pipeline is a set of sinks that can send data to each other.  The
+// pipeline object also contains an arena allocator that the sinks and their
+// associated processing state can use for fast memory allocation.  This makes
+// pipelines very fast to construct and destroy, especially if the arena is
+// supplied with an initial block of memory.  If this initial block of memory
+// is from the C stack and is large enough, then actual heap allocation can be
+// avoided entirely which significantly reduces overhead in some cases.
+//
+// All sinks and processing state are automatically freed when the pipeline is
+// destroyed, so Free() is not necessary or possible.  Allocated objects can
+// optionally specify a Reset() callback that will be called when whenever the
+// pipeline is Reset() or destroyed.  This can be used to free any outside
+// resources the object is holding.
+//
+// Pipelines (and sinks/objects allocated from them) are not thread-safe!
+class upb::Pipeline {
+ public:
+  // Initializes the pipeline's arena with the given initial memory that will
+  // be used before allocating memory using the given allocation function.
+  // The "ud" pointer will be passed as the first parameter to the realloc
+  // callback, and can be used to pass user-specific state.
+  Pipeline(void *initial_mem, size_t initial_size,
+        void *(*realloc)(void *ud, void *ptr, size_t size), void *ud);
+  ~Pipeline();
+
+  // Returns a newly-allocated Sink for the given handlers.  The sink is will
+  // live as long as the pipeline does.  Caller retains ownership of the
+  // handlers object, which must outlive the pipeline.
+  //
+  // TODO(haberman): add an option for the sink to take a ref, so the handlers
+  // don't have to outlive?  This would be simpler but imposes a minimum cost.
+  // Taking an atomic ref is not *so* bad in the single-threaded case, but this
+  // can degrade heavily under contention, so we need a way to avoid it in
+  // cases where this overhead would be significant and the caller can easily
+  // guarantee the outlive semantics.
+  Sink* NewSink(const Handlers* handlers);
+
+  // Accepts a ref donated from the given owner.  Will unref the Handlers when
+  // the Pipeline is destroyed.
+  void DonateRef(const Handlers* h, const void* owner);
+
+  // The current error status for the pipeline.
+  const upb::Status& status() const;
+
+  // Calls "reset" on all Sinks and resettable state objects in the arena, and
+  // resets the error status.  Useful for resetting processing state so new
+  // input can be accepted.
+  void Reset();
+
+  // Allocates/reallocates memory of the given size, or returns NULL if no
+  // memory is available.  It is not necessary (or possible) to manually free
+  // the memory obtained from these functions.
+  void* Alloc(size_t size);
+  void* Realloc(void* ptr, size_t old_size, size_t size);
+
+  // Allocates an object with the given FrameType.  Note that this object may
+  // *not* be resized with Realloc().
+  void* AllocObject(const FrameType* type);
+
+ private:
+#else
+struct upb_pipeline {
+#endif
+  void *(*realloc)(void *ud, void *ptr, size_t size);
+  void *ud;
+  void *bump_top;    // Current alloc offset, either from initial or dyn region.
+  void *bump_limit;  // Limit of current alloc block.
+  void *obj_head;    // Linked list of objects with "reset" functions.
+  void *region_head; // Linked list of dyn regions we got from user's realloc().
+  void *last_alloc;
+  upb_status status_;
+};
+
+#ifdef __cplusplus
+
+// For convenience, a template for a pipeline with an array of initial memory.
+template <int initial_size>
+class upb::SeededPipeline : public upb::Pipeline {
+ public:
+  SeededPipeline(void *(*realloc)(void *ud, void *ptr, size_t size), void *ud)
+      : Pipeline(mem_, initial_size, realloc, ud) {
+  }
+
+ private:
+  char mem_[initial_size];
+};
+
+class upb::SinkFrame {
+ public:
+  // Returns the sink that this frame belongs to.
+  Sink* sink() const;
+
+  // Returns the pipeline that this sink and frame belong to.
+  Pipeline* pipeline() const;
+
+  // The depth of this frame (counts all kind of frames (sequence, submessage,
+  // and string frames).
+  int depth() const;
+
+  // The Handlers object for this frame.
+  const Handlers* handlers() const;
+
+  // Returns the user data that is bound to this sink frame (as returned
+  // by the Start{SubMessage,String,Sequence} handler, or passed to
+  // Sink::Reset()).
+  void* userdata() const;
+
+  // A templated version of userdata() that type-checks the templated return
+  // type.
+  //
+  // TODO(haberman): this isn't truly robust until sequence and string frames
+  // have distinct FrameTypes in the Handlers.
+  template<class T>
+  T* GetUserdata() const {
+#ifdef NDEBUG
+    return static_cast<T*>(userdata());
+#else
+    const FrameType* type = handlers()->frame_type();
+    if (!type || type == GetFrameType<T>()) {
+      return static_cast<T*>(userdata());
+    } else {
+      assert(false);
+      return NULL;
+    }
 #endif
+  }
 
+  // Returns the data that was bound to the currently-executing callback in the
+  // Handlers object.  If not currently in a handler, the results are undefined.
+  void* handler_data() const;
 
-/* upb_sink *******************************************************************/
+ private:
+  UPB_DISALLOW_POD_OPS(SinkFrame);
+  friend class upb::Sink;
+  friend upb_sink* ::upb_sinkframe_sink(const upb_sinkframe* frame);
+  friend void* ::upb_sinkframe_userdata(const upb_sinkframe* frame);
+  friend void* ::upb_sinkframe_handlerdata(const upb_sinkframe* frame);
 
-typedef struct {
-  upb_selector_t end;  // From the enclosing message (unused at top-level).
+#else
+struct upb_sinkframe {
+#endif
+  upb_sink *sink_;
   const upb_handlers *h;
   void *closure;
-} upb_sink_frame;
 
-typedef struct {
-  upb_sink_frame *top, *limit;
-  upb_sink_frame stack[UPB_MAX_NESTING];
-  upb_status status;
-} upb_sink;
+  union {
+    // For the top frame (sink->top), the handler_data for the
+    // currently-executing callback, otherwise undefined.
+    // TODO(haberman): have a special pointer value to indicate "not in a
+    // callback"; this will be a way to enforce non-reentrancy of a sink.
+    void *handler_data;
 
-// Caller retains ownership of the handlers object.
-void upb_sink_init(upb_sink *s, const upb_handlers *h);
+    // For other frames, the END* callback that will run when the subframe is
+    // popped (for example, for a "sequence" frame the frame above it will be a
+    // UPB_HANDLER_ENDSEQ handler).  But this is only necessary for assertion
+    // checking inside upb_sink and can be omitted if the sink has only one
+    // caller.
+    // TODO(haberman): have a mechanism for ensuring that a sink only has one
+    // caller.
+    upb_selector_t selector;
+  } u;
+};
 
-// Resets the state of the sink so that it is ready to accept new input.
-// Any state from previously received data is discarded.  "Closure" will be
-// used as the top-level closure.
-void upb_sink_reset(upb_sink *s, void *closure);
+#ifdef __cplusplus
+
+// A upb::Sink is an object that binds a upb::Handlers object to some runtime
+// state.  It is the object that can actually call a set of handlers.
+//
+// Unlike upb::Def and upb::Handlers, upb::Sink is never frozen, immutable, or
+// thread-safe.  You can create as many of them as you want, but each one may
+// only be used in a single thread at a time.
+//
+// If we compare with class-based OOP, a you can think of a upb::Def as an
+// abstract base class, a upb::Handlers as a concrete derived class, and a
+// upb::Sink as an object (class instance).
+//
+// Each upb::Sink lives in exactly one pipeline.
+class upb::Sink {
+ public:
 
-void upb_sink_uninit(upb_sink *s);
+  // Resets the state of the sink so that it is ready to accept new input.
+  // Any state from previously received data is discarded.  "Closure" will be
+  // used as the top-level closure.
+  void Reset(void *closure);
 
-// Returns the handlers at the top of the stack.
-const upb_handlers *upb_sink_tophandlers(upb_sink *s);
+  // Returns the top-most and base (lowest) frame of the stack, respectively.
+  const SinkFrame* top() const;
+  const SinkFrame* base() const;
 
-// Functions for pushing data into the sink.
-// These return false if processing should stop (either due to error or just
-// to suspend).
+  // Returns the pipeline that this sink comes from.
+  Pipeline* pipeline() const;
+
+  // Functions for pushing data into the sink.
+  //
+  // These return false if processing should stop (either due to error or just
+  // to suspend).
+  //
+  // These may not be called from within one of the same sink's handlers (in
+  // other words, handlers are not re-entrant).
+
+  // Should be called at the start and end of processing.
+  bool StartMessage();
+  void EndMessage();
+
+  // Putting of individual values.  These work for both repeated and
+  // non-repeated fields, but for repeated fields you must wrap them in
+  // calls to StartSequence()/EndSequence().
+  bool PutInt32(Handlers::Selector s, int32_t val);
+  bool PutInt64(Handlers::Selector s, int64_t val);
+  bool PutUInt32(Handlers::Selector s, uint32_t val);
+  bool PutUInt64(Handlers::Selector s, uint64_t val);
+  bool PutFloat(Handlers::Selector s, float val);
+  bool PutDouble(Handlers::Selector s, double val);
+  bool PutBool(Handlers::Selector s, bool val);
+
+  // Putting of string/bytes values.  Each string can consist of zero or more
+  // non-contiguous buffers of data.
+  bool StartString(Handlers::Selector s, size_t size_hint);
+  size_t PutStringBuffer(Handlers::Selector s, const char *buf, size_t len);
+  bool EndString(Handlers::Selector s);
+
+  // For submessage fields.
+  bool StartSubMessage(Handlers::Selector s);
+  bool EndSubMessage(Handlers::Selector s);
+
+  // For repeated fields of any type, the sequence of values must be wrapped in
+  // these calls.
+  bool StartSequence(Handlers::Selector s);
+  bool EndSequence(Handlers::Selector s);
+
+ private:
+  UPB_DISALLOW_POD_OPS(Sink);
+#else
+struct upb_sink {
+#endif
+  upb_pipeline *pipeline_;
+  upb_sinkframe *top_, *limit;
+  upb_sinkframe stack[UPB_MAX_NESTING];
+};
+
+// C API.
+UPB_INLINE upb_sink *upb_sinkframe_sink(const upb_sinkframe* frame) {
+  return frame->sink_;
+}
+
+UPB_INLINE void *upb_sinkframe_userdata(const upb_sinkframe* frame) {
+  return frame->closure;
+}
+
+UPB_INLINE void *upb_sinkframe_handlerdata(const upb_sinkframe* frame) {
+  return frame->u.handler_data;
+}
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+void *upb_realloc(void *ud, void *ptr, size_t size);
+void upb_pipeline_init(upb_pipeline *p, void *initial_mem, size_t initial_size,
+                       void *(*realloc)(void *ud, void *ptr, size_t size),
+                       void *ud);
+void upb_pipeline_uninit(upb_pipeline *p);
+void *upb_pipeline_alloc(upb_pipeline *p, size_t size);
+void *upb_pipeline_realloc(
+    upb_pipeline *p, void *ptr, size_t old_size, size_t size);
+void *upb_pipeline_allocobj(upb_pipeline *p, const upb_frametype *type);
+void upb_pipeline_reset(upb_pipeline *p);
+void upb_pipeline_donateref(
+    upb_pipeline *p, const upb_handlers *h, const void *owner);
+upb_sink *upb_pipeline_newsink(upb_pipeline *p, const upb_handlers *h);
+const upb_status *upb_pipeline_status(const upb_pipeline *p);
+
+int upb_sinkframe_depth(const upb_sinkframe* frame);
+const upb_handlers* upb_sinkframe_handlers(const upb_sinkframe* frame);
+upb_pipeline* upb_sinkframe_pipeline(const upb_sinkframe* frame);
+
+void upb_sink_reset(upb_sink *s, void *closure);
+upb_pipeline *upb_sink_pipeline(const upb_sink *s);
+const upb_sinkframe *upb_sink_top(const upb_sink *s);
+const upb_sinkframe *upb_sink_base(const upb_sink *s);
 bool upb_sink_startmsg(upb_sink *s);
-void upb_sink_endmsg(upb_sink *s, upb_status *status);
-bool upb_sink_putint32(upb_sink *s, const upb_fielddef *f, int32_t val);
-bool upb_sink_putint64(upb_sink *s, const upb_fielddef *f, int64_t val);
-bool upb_sink_putuint32(upb_sink *s, const upb_fielddef *f, uint32_t val);
-bool upb_sink_putuint64(upb_sink *s, const upb_fielddef *f, uint64_t val);
-bool upb_sink_putfloat(upb_sink *s, const upb_fielddef *f, float val);
-bool upb_sink_putdouble(upb_sink *s, const upb_fielddef *f, double val);
-bool upb_sink_putbool(upb_sink *s, const upb_fielddef *f, bool val);
-bool upb_sink_startstr(upb_sink *s, const upb_fielddef *f, size_t size_hint);
-size_t upb_sink_putstring(upb_sink *s, const upb_fielddef *f, const char *buf,
+void upb_sink_endmsg(upb_sink *s);
+bool upb_sink_putint32(upb_sink *s, upb_selector_t sel, int32_t val);
+bool upb_sink_putint64(upb_sink *s, upb_selector_t sel, int64_t val);
+bool upb_sink_putuint32(upb_sink *s, upb_selector_t sel, uint32_t val);
+bool upb_sink_putuint64(upb_sink *s, upb_selector_t sel, uint64_t val);
+bool upb_sink_putfloat(upb_sink *s, upb_selector_t sel, float val);
+bool upb_sink_putdouble(upb_sink *s, upb_selector_t sel, double val);
+bool upb_sink_putbool(upb_sink *s, upb_selector_t sel, bool val);
+bool upb_sink_startstr(upb_sink *s, upb_selector_t sel, size_t size_hint);
+size_t upb_sink_putstring(upb_sink *s, upb_selector_t sel, const char *buf,
                           size_t len);
-bool upb_sink_endstr(upb_sink *s, const upb_fielddef *f);
-bool upb_sink_startsubmsg(upb_sink *s, const upb_fielddef *f);
-bool upb_sink_endsubmsg(upb_sink *s, const upb_fielddef *f);
-bool upb_sink_startseq(upb_sink *s, const upb_fielddef *f);
-bool upb_sink_endseq(upb_sink *s, const upb_fielddef *f);
+bool upb_sink_endstr(upb_sink *s, upb_selector_t sel);
+bool upb_sink_startsubmsg(upb_sink *s, upb_selector_t sel);
+bool upb_sink_endsubmsg(upb_sink *s, upb_selector_t sel);
+bool upb_sink_startseq(upb_sink *s, upb_selector_t sel);
+bool upb_sink_endseq(upb_sink *s, upb_selector_t sel);
 
 #ifdef __cplusplus
 }  /* extern "C" */
 #endif
 
+#ifdef __cplusplus
+
+namespace upb {
+
+
+inline Pipeline::Pipeline(void *initial_mem, size_t initial_size,
+                    void *(*realloc)(void *ud, void *ptr, size_t size),
+                    void *ud) {
+  upb_pipeline_init(this, initial_mem, initial_size, realloc, ud);
+}
+inline Pipeline::~Pipeline() {
+  upb_pipeline_uninit(this);
+}
+inline void* Pipeline::Alloc(size_t size) {
+  return upb_pipeline_alloc(this, size);
+}
+inline void* Pipeline::Realloc(void* ptr, size_t old_size, size_t size) {
+  return upb_pipeline_realloc(this, ptr, old_size, size);
+}
+inline void* Pipeline::AllocObject(const upb::FrameType* type) {
+  return upb_pipeline_allocobj(this, type);
+}
+inline void Pipeline::Reset() {
+  upb_pipeline_reset(this);
+}
+inline const upb::Status& Pipeline::status() const {
+  return *upb_pipeline_status(this);
+}
+inline Sink* Pipeline::NewSink(const upb::Handlers* handlers) {
+  return upb_pipeline_newsink(this, handlers);
+}
+inline void Pipeline::DonateRef(const upb::Handlers* h, const void *owner) {
+  return upb_pipeline_donateref(this, h, owner);
+}
+
+inline Sink* SinkFrame::sink() const {
+  return upb_sinkframe_sink(this);
+}
+inline Pipeline* SinkFrame::pipeline() const {
+  return upb_sinkframe_pipeline(this);
+}
+inline void* SinkFrame::userdata() const {
+  return upb_sinkframe_userdata(this);
+}
+inline void* SinkFrame::handler_data() const {
+  return upb_sinkframe_handlerdata(this);
+}
+inline int SinkFrame::depth() const {
+  return upb_sinkframe_depth(this);
+}
+inline const Handlers* SinkFrame::handlers() const {
+  return upb_sinkframe_handlers(this);
+}
+
+inline void Sink::Reset(void *closure) {
+  upb_sink_reset(this, closure);
+}
+inline Pipeline* Sink::pipeline() const {
+  return upb_sink_pipeline(this);
+}
+inline const SinkFrame* Sink::top() const {
+  return upb_sink_top(this);
+}
+inline const SinkFrame* Sink::base() const {
+  return upb_sink_base(this);
+}
+inline bool Sink::StartMessage() {
+  return upb_sink_startmsg(this);
+}
+inline void Sink::EndMessage() {
+  upb_sink_endmsg(this);
+}
+inline bool Sink::PutInt32(Handlers::Selector sel, int32_t val) {
+  return upb_sink_putint32(this, sel, val);
+}
+inline bool Sink::PutInt64(Handlers::Selector sel, int64_t val) {
+  return upb_sink_putint64(this, sel, val);
+}
+inline bool Sink::PutUInt32(Handlers::Selector sel, uint32_t val) {
+  return upb_sink_putuint32(this, sel, val);
+}
+inline bool Sink::PutUInt64(Handlers::Selector sel, uint64_t val) {
+  return upb_sink_putuint64(this, sel, val);
+}
+inline bool Sink::PutFloat(Handlers::Selector sel, float val) {
+  return upb_sink_putfloat(this, sel, val);
+}
+inline bool Sink::PutDouble(Handlers::Selector sel, double val) {
+  return upb_sink_putdouble(this, sel, val);
+}
+inline bool Sink::PutBool(Handlers::Selector sel, bool val) {
+  return upb_sink_putbool(this, sel, val);
+}
+inline bool Sink::StartString(Handlers::Selector sel, size_t size_hint) {
+  return upb_sink_startstr(this, sel, size_hint);
+}
+inline size_t Sink::PutStringBuffer(Handlers::Selector sel, const char *buf,
+                                  size_t len) {
+  return upb_sink_putstring(this, sel, buf, len);
+}
+inline bool Sink::EndString(Handlers::Selector sel) {
+  return upb_sink_endstr(this, sel);
+}
+inline bool Sink::StartSubMessage(Handlers::Selector sel) {
+  return upb_sink_startsubmsg(this, sel);
+}
+inline bool Sink::EndSubMessage(Handlers::Selector sel) {
+  return upb_sink_endsubmsg(this, sel);
+}
+inline bool Sink::StartSequence(Handlers::Selector sel) {
+  return upb_sink_startseq(this, sel);
+}
+inline bool Sink::EndSequence(Handlers::Selector sel) {
+  return upb_sink_endseq(this, sel);
+}
+
+}  // namespace upb
+#endif
+
 #endif

upb/symtab.c

@@ -10,8 +10,6 @@
 #include <stdlib.h>
 #include <string.h>
 
-#include "upb/bytestream.h"
-
 bool upb_symtab_isfrozen(const upb_symtab *s) {
   return upb_refcounted_isfrozen(upb_upcast(s));
 }
@@ -77,16 +75,18 @@ const upb_def **upb_symtab_getdefs(const upb_symtab *s, upb_deftype_t type,
 
 const upb_def *upb_symtab_lookup(const upb_symtab *s, const char *sym,
                                  const void *owner) {
-  const upb_value *v = upb_strtable_lookup(&s->symtab, sym);
-  upb_def *ret = v ? upb_value_getptr(*v) : NULL;
+  upb_value v;
+  upb_def *ret = upb_strtable_lookup(&s->symtab, sym, &v) ?
+      upb_value_getptr(v) : NULL;
   if (ret) upb_def_ref(ret, owner);
   return ret;
 }
 
 const upb_msgdef *upb_symtab_lookupmsg(const upb_symtab *s, const char *sym,
                                        const void *owner) {
-  const upb_value *v = upb_strtable_lookup(&s->symtab, sym);
-  upb_def *def = v ? upb_value_getptr(*v) : NULL;
+  upb_value v;
+  upb_def *def = upb_strtable_lookup(&s->symtab, sym, &v) ?
+      upb_value_getptr(v) : NULL;
   upb_msgdef *ret = NULL;
   if(def && def->type == UPB_DEF_MSG) {
     ret = upb_downcast_msgdef_mutable(def);
@@ -103,8 +103,8 @@ static upb_def *upb_resolvename(const upb_strtable *t,
   if(sym[0] == UPB_SYMBOL_SEPARATOR) {
     // Symbols starting with '.' are absolute, so we do a single lookup.
     // Slice to omit the leading '.'
-    const upb_value *v = upb_strtable_lookup(t, sym + 1);
-    return v ? upb_value_getptr(*v) : NULL;
+    upb_value v;
+    return upb_strtable_lookup(t, sym + 1, &v) ? upb_value_getptr(v) : NULL;
   } else {
     // Remove components from base until we find an entry or run out.
     // TODO: This branch is totally broken, but currently not used.
@@ -134,8 +134,9 @@ static bool upb_resolve_dfs(const upb_def *def, upb_strtable *addtab,
                             upb_status *s) {
   // Memoize results of this function for efficiency (since we're traversing a
   // DAG this is not needed to limit the depth of the search).
-  const upb_value *v = upb_inttable_lookup(seen, (uintptr_t)def);
-  if (v) return upb_value_getbool(*v);
+  upb_value v;
+  if (upb_inttable_lookup(seen, (uintptr_t)def, &v))
+    return upb_value_getbool(v);
 
   // Visit submessages for all messages in the SCC.
   bool need_dup = false;
@@ -143,10 +144,10 @@ static bool upb_resolve_dfs(const upb_def *def, upb_strtable *addtab,
   do {
     assert(upb_def_isfrozen(def));
     if (def->type == UPB_DEF_FIELD) continue;
-    const upb_value *v = upb_strtable_lookup(addtab, upb_def_fullname(def));
-    if (v) {
+    upb_value v;
+    if (upb_strtable_lookup(addtab, upb_def_fullname(def), &v)) {
       // Because we memoize we should not visit a node after we have dup'd it.
-      assert(((upb_def*)upb_value_getptr(*v))->came_from_user);
+      assert(((upb_def*)upb_value_getptr(v))->came_from_user);
       need_dup = true;
     }
     const upb_msgdef *m = upb_dyncast_msgdef(def);
@@ -169,7 +170,7 @@ static bool upb_resolve_dfs(const upb_def *def, upb_strtable *addtab,
     do {
       if (def->type == UPB_DEF_FIELD) continue;
       const char *name = upb_def_fullname(def);
-      if (upb_strtable_lookup(addtab, name) == NULL) {
+      if (!upb_strtable_lookup(addtab, name, NULL)) {
         upb_def *newdef = upb_def_dup(def, new_owner);
         if (!newdef) goto oom;
         newdef->came_from_user = false;
@@ -210,7 +211,7 @@ bool upb_symtab_add(upb_symtab *s, upb_def *const*defs, int n, void *ref_donor,
           status, "Anonymous defs cannot be added to a symtab");
       goto err;
     }
-    if (upb_strtable_lookup(&addtab, fullname) != NULL) {
+    if (upb_strtable_lookup(&addtab, fullname, NULL)) {
       upb_status_seterrf(status, "Conflicting defs named '%s'", fullname);
       goto err;
     }
@@ -263,10 +264,8 @@ bool upb_symtab_add(upb_symtab *s, upb_def *const*defs, int n, void *ref_donor,
       }
 
       if (!upb_fielddef_resolvedefault(f)) {
-        upb_byteregion *r = upb_value_getbyteregion(upb_fielddef_default(f));
-        size_t len;
-        const char *ptr = upb_byteregion_getptr(r, 0, &len);
-        upb_status_seterrf(status, "couldn't resolve enum default '%s'", ptr);
+        upb_status_seterrf(status, "couldn't resolve enum default '%s'",
+                           upb_fielddef_defaultstr(f, NULL));
         goto err;
       }
     }

upb/table.c

@@ -38,22 +38,22 @@ char *upb_strdup(const char *s) {
   return p;
 }
 
-static upb_tabkey upb_strkey(const char *str) {
+static upb_tabkey strkey(const char *str) {
   upb_tabkey k;
   k.str = (char*)str;
   return k;
 }
 
-typedef const upb_tabent *upb_hashfunc_t(const upb_table *t, upb_tabkey key);
-typedef bool upb_eqlfunc_t(upb_tabkey k1, upb_tabkey k2);
+typedef const upb_tabent *hashfunc_t(const upb_table *t, upb_tabkey key);
+typedef bool eqlfunc_t(upb_tabkey k1, upb_tabkey k2);
 
 /* Base table (shared code) ***************************************************/
 
-static bool upb_table_isfull(upb_table *t) {
+static bool isfull(upb_table *t) {
   return (double)(t->count + 1) / upb_table_size(t) > MAX_LOAD;
 }
 
-static bool upb_table_init(upb_table *t, upb_ctype_t type, uint8_t size_lg2) {
+static bool init(upb_table *t, upb_ctype_t type, uint8_t size_lg2) {
   t->count = 0;
   t->type = type;
   t->size_lg2 = size_lg2;
@@ -69,29 +69,41 @@ static bool upb_table_init(upb_table *t, upb_ctype_t type, uint8_t size_lg2) {
   return true;
 }
 
-static void upb_table_uninit(upb_table *t) { free((void*)t->entries); }
+static void uninit(upb_table *t) { free((void*)t->entries); }
 
-static upb_tabent *upb_table_emptyent(upb_table *t) {
+static upb_tabent *emptyent(upb_table *t) {
   upb_tabent *e = (upb_tabent*)t->entries + upb_table_size(t);
   while (1) { if (upb_tabent_isempty(--e)) return e; assert(e > t->entries); }
 }
 
-static const upb_value *upb_table_lookup(const upb_table *t, upb_tabkey key,
-                                         upb_hashfunc_t *hash,
-                                         upb_eqlfunc_t *eql) {
+static const upb_tabent *findentry(const upb_table *t, upb_tabkey key,
+                                   hashfunc_t *hash, eqlfunc_t *eql) {
   if (t->size_lg2 == 0) return NULL;
   const upb_tabent *e = hash(t, key);
   if (upb_tabent_isempty(e)) return NULL;
   while (1) {
-    if (eql(e->key, key)) return &e->val;
+    if (eql(e->key, key)) return e;
     if ((e = e->next) == NULL) return NULL;
   }
 }
 
+static bool lookup(const upb_table *t, upb_tabkey key, upb_value *v,
+                   hashfunc_t *hash, eqlfunc_t *eql) {
+  const upb_tabent *e = findentry(t, key, hash, eql);
+  if (e) {
+    if (v) {
+      _upb_value_setval(v, e->val, t->type);
+    }
+    return true;
+  } else {
+    return false;
+  }
+}
+
 // The given key must not already exist in the table.
-static void upb_table_insert(upb_table *t, upb_tabkey key, upb_value val,
-                             upb_hashfunc_t *hash, upb_eqlfunc_t *eql) {
-  assert(upb_table_lookup(t, key, hash, eql) == NULL);
+static void insert(upb_table *t, upb_tabkey key, upb_value val,
+                   hashfunc_t *hash, eqlfunc_t *eql) {
+  assert(findentry(t, key, hash, eql) == NULL);
   assert(val.type == t->type);
   t->count++;
   upb_tabent *mainpos_e = (upb_tabent*)hash(t, key);
@@ -101,7 +113,7 @@ static void upb_table_insert(upb_table *t, upb_tabkey key, upb_value val,
     our_e->next = NULL;
   } else {
     // Collision.
-    upb_tabent *new_e = upb_table_emptyent(t);
+    upb_tabent *new_e = emptyent(t);
     // Head of collider's chain.
     upb_tabent *chain = (upb_tabent*)hash(t, mainpos_e->key);
     if (chain == mainpos_e) {
@@ -125,26 +137,27 @@ static void upb_table_insert(upb_table *t, upb_tabkey key, upb_value val,
     }
   }
   our_e->key = key;
-  our_e->val = val;
-  assert(upb_table_lookup(t, key, hash, eql) == &our_e->val);
+  our_e->val = val.val;
+  assert(findentry(t, key, hash, eql) == our_e);
 }
 
-static bool upb_table_remove(upb_table *t, upb_tabkey key, upb_value *val,
-                             upb_tabkey *removed,
-                             upb_hashfunc_t *hash, upb_eqlfunc_t *eql) {
+static bool rm(upb_table *t, upb_tabkey key, upb_value *val,
+               upb_tabkey *removed, hashfunc_t *hash, eqlfunc_t *eql) {
   upb_tabent *chain = (upb_tabent*)hash(t, key);
   if (upb_tabent_isempty(chain)) return false;
   if (eql(chain->key, key)) {
     // Element to remove is at the head of its chain.
     t->count--;
-    if (val) *val = chain->val;
+    if (val) {
+      _upb_value_setval(val, chain->val, t->type);
+    }
     if (chain->next) {
       upb_tabent *move = (upb_tabent*)chain->next;
       *chain = *move;
-      *removed = move->key;
+      if (removed) *removed = move->key;
       move->key.num = 0;  // Make the slot empty.
     } else {
-      *removed = chain->key;
+      if (removed) *removed = chain->key;
       chain->key.num = 0;  // Make the slot empty.
     }
     return true;
@@ -154,11 +167,13 @@ static bool upb_table_remove(upb_table *t, upb_tabkey key, upb_value *val,
       chain = (upb_tabent*)chain->next;
     if (chain->next) {
       // Found element to remove.
-      if (val) *val = chain->next->val;
-      upb_tabent *remove = (upb_tabent*)chain->next;
-      *removed = remove->key;
-      remove->key.num = 0;
-      chain->next = remove->next;
+      if (val) {
+        _upb_value_setval(val, chain->next->val, t->type);
+      }
+      upb_tabent *rm = (upb_tabent*)chain->next;
+      if (removed) *removed = rm->key;
+      rm->key.num = 0;
+      chain->next = rm->next;
       t->count--;
       return true;
     } else {
@@ -167,8 +182,7 @@ static bool upb_table_remove(upb_table *t, upb_tabkey key, upb_value *val,
   }
 }
 
-static const upb_tabent *upb_table_next(const upb_table *t,
-                                        const upb_tabent *e) {
+static const upb_tabent *next(const upb_table *t, const upb_tabent *e) {
   const upb_tabent *end = t->entries + upb_table_size(t);
   do { if (++e == end) return NULL; } while(e->key.num == 0);
   return e;
@@ -176,8 +190,8 @@ static const upb_tabent *upb_table_next(const upb_table *t,
 
 // TODO: is calculating t->entries - 1 undefined behavior?  If so find a better
 // solution.
-static const upb_tabent *upb_table_begin(const upb_table *t) {
-  return upb_table_next(t, t->entries - 1);
+static const upb_tabent *begin(const upb_table *t) {
+  return next(t, t->entries - 1);
 }
 
 
@@ -185,30 +199,30 @@ static const upb_tabent *upb_table_begin(const upb_table *t) {
 
 // A simple "subclass" of upb_table that only adds a hash function for strings.
 
-static const upb_tabent *upb_strhash(const upb_table *t, upb_tabkey key) {
+static const upb_tabent *strhash(const upb_table *t, upb_tabkey key) {
   // Could avoid the strlen() by using a hash function that terminates on NULL.
   return t->entries + (MurmurHash2(key.str, strlen(key.str), 0) & t->mask);
 }
 
-static bool upb_streql(upb_tabkey k1, upb_tabkey k2) {
+static bool streql(upb_tabkey k1, upb_tabkey k2) {
   return strcmp(k1.str, k2.str) == 0;
 }
 
 bool upb_strtable_init(upb_strtable *t, upb_ctype_t type) {
-  return upb_table_init(&t->t, type, 2);
+  return init(&t->t, type, 2);
 }
 
 void upb_strtable_uninit(upb_strtable *t) {
   for (size_t i = 0; i < upb_table_size(&t->t); i++)
     free((void*)t->t.entries[i].key.str);
-  upb_table_uninit(&t->t);
+  uninit(&t->t);
 }
 
 bool upb_strtable_insert(upb_strtable *t, const char *k, upb_value v) {
-  if (upb_table_isfull(&t->t)) {
+  if (isfull(&t->t)) {
     // Need to resize.  New table of double the size, add old elements to it.
     upb_strtable new_table;
-    if (!upb_table_init(&new_table.t, t->t.type, t->t.size_lg2 + 1))
+    if (!init(&new_table.t, t->t.type, t->t.size_lg2 + 1))
       return false;
     upb_strtable_iter i;
     upb_strtable_begin(&i, t);
@@ -220,29 +234,31 @@ bool upb_strtable_insert(upb_strtable *t, const char *k, upb_value v) {
     *t = new_table;
   }
   if ((k = upb_strdup(k)) == NULL) return false;
-  upb_table_insert(&t->t, upb_strkey(k), v, &upb_strhash, &upb_streql);
+  insert(&t->t, strkey(k), v, &strhash, &streql);
   return true;
 }
 
-const upb_value *upb_strtable_lookup(const upb_strtable *t, const char *key) {
-  return upb_table_lookup(&t->t, upb_strkey(key), &upb_strhash, &upb_streql);
+bool upb_strtable_lookup(const upb_strtable *t, const char *key, upb_value *v) {
+  return lookup(&t->t, strkey(key), v, &strhash, &streql);
 }
 
 bool upb_strtable_remove(upb_strtable *t, const char *key, upb_value *val) {
-  upb_tabkey removed;
-  bool found = upb_table_remove(
-      &t->t, upb_strkey(key), val, &removed, &upb_strhash, &upb_streql);
-  if (found) free((void*)removed.str);
-  return found;
+  upb_tabkey tabkey;
+  if (rm(&t->t, strkey(key), val, &tabkey, &strhash, &streql)) {
+    free((void*)tabkey.str);
+    return true;
+  } else {
+    return false;
+  }
 }
 
 void upb_strtable_begin(upb_strtable_iter *i, const upb_strtable *t) {
   i->t = t;
-  i->e = upb_table_begin(&t->t);
+  i->e = begin(&t->t);
 }
 
 void upb_strtable_next(upb_strtable_iter *i) {
-  i->e = upb_table_next(&i->t->t, i->e);
+  i->e = next(&i->t->t, i->e);
 }
 
 
@@ -251,7 +267,7 @@ void upb_strtable_next(upb_strtable_iter *i) {
 // For inttables we use a hybrid structure where small keys are kept in an
 // array and large keys are put in the hash table.
 
-static bool upb_inteql(upb_tabkey k1, upb_tabkey k2) {
+static bool inteql(upb_tabkey k1, upb_tabkey k2) {
   return k1.num == k2.num;
 }
 
@@ -259,9 +275,23 @@ size_t upb_inttable_count(const upb_inttable *t) {
   return t->t.count + t->array_count;
 }
 
+static void check(upb_inttable *t) {
+  UPB_UNUSED(t);
+#if defined(UPB_DEBUG_TABLE) && !defined(NDEBUG)
+  // This check is very expensive (makes inserts/deletes O(N)).
+  size_t count = 0;
+  upb_inttable_iter i;
+  upb_inttable_begin(&i, t);
+  for(; !upb_inttable_done(&i); upb_inttable_next(&i), count++) {
+    assert(upb_inttable_lookup(t, upb_inttable_iter_key(&i), NULL));
+  }
+  assert(count == upb_inttable_count(t));
+#endif
+}
+
 bool upb_inttable_sizedinit(upb_inttable *t, upb_ctype_t type,
                             size_t asize, int hsize_lg2) {
-  if (!upb_table_init(&t->t, type, hsize_lg2)) return false;
+  if (!init(&t->t, type, hsize_lg2)) return false;
   // Always make the array part at least 1 long, so that we know key 0
   // won't be in the hash part, which simplifies things.
   t->array_size = UPB_MAX(1, asize);
@@ -269,10 +299,11 @@ bool upb_inttable_sizedinit(upb_inttable *t, upb_ctype_t type,
   size_t array_bytes = t->array_size * sizeof(upb_value);
   t->array = malloc(array_bytes);
   if (!t->array) {
-    upb_table_uninit(&t->t);
+    uninit(&t->t);
     return false;
   }
   memset((void*)t->array, 0xff, array_bytes);
+  check(t);
   return true;
 }
 
@@ -281,58 +312,50 @@ bool upb_inttable_init(upb_inttable *t, upb_ctype_t type) {
 }
 
 void upb_inttable_uninit(upb_inttable *t) {
-  upb_table_uninit(&t->t);
+  uninit(&t->t);
   free((void*)t->array);
 }
 
-static void upb_inttable_check(upb_inttable *t) {
-  UPB_UNUSED(t);
-#if defined(UPB_DEBUG_TABLE) && !defined(NDEBUG)
-  // This check is very expensive (makes inserts/deletes O(N)).
-  size_t count = 0;
-  upb_inttable_iter i;
-  upb_inttable_begin(&i, t);
-  for(; !upb_inttable_done(&i); upb_inttable_next(&i), count++) {
-    const upb_value *v = upb_inttable_lookup(t, upb_inttable_iter_key(&i));
-    assert(v);
-  }
-  assert(count == upb_inttable_count(t));
-#endif
-}
-
 bool upb_inttable_insert(upb_inttable *t, uintptr_t key, upb_value val) {
-  assert(upb_arrhas(val));
+  assert(upb_arrhas(val.val));
   if (key < t->array_size) {
     assert(!upb_arrhas(t->array[key]));
     t->array_count++;
-    ((upb_value*)t->array)[key] = val;
+    ((_upb_value*)t->array)[key] = val.val;
   } else {
-    if (upb_table_isfull(&t->t)) {
+    if (isfull(&t->t)) {
       // Need to resize the hash part, but we re-use the array part.
       upb_table new_table;
-      if (!upb_table_init(&new_table, t->t.type, t->t.size_lg2 + 1))
+      if (!init(&new_table, t->t.type, t->t.size_lg2 + 1))
         return false;
       const upb_tabent *e;
-      for (e = upb_table_begin(&t->t); e; e = upb_table_next(&t->t, e))
-        upb_table_insert(&new_table, e->key, e->val, &upb_inthash, &upb_inteql);
+      for (e = begin(&t->t); e; e = next(&t->t, e)) {
+        upb_value v;
+        _upb_value_setval(&v, e->val, t->t.type);
+        insert(&new_table, e->key, v, &upb_inthash, &inteql);
+      }
 
       assert(t->t.count == new_table.count);
 
-      upb_table_uninit(&t->t);
+      uninit(&t->t);
       t->t = new_table;
     }
-    upb_table_insert(&t->t, upb_intkey(key), val, &upb_inthash, &upb_inteql);
+    insert(&t->t, upb_intkey(key), val, &upb_inthash, &inteql);
   }
-  upb_inttable_check(t);
+  check(t);
   return true;
 }
 
-const upb_value *upb_inttable_lookup(const upb_inttable *t, uintptr_t key) {
+bool upb_inttable_lookup(const upb_inttable *t, uintptr_t key, upb_value *v) {
   if (key < t->array_size) {
-    const upb_value *v = &t->array[key];
-    return upb_arrhas(*v) ? v : NULL;
+    bool ret = upb_arrhas(t->array[key]);
+    if (ret && v) {
+      _upb_value_setval(v, t->array[key], t->t.type);
+    }
+    return ret;
+  } else {
+    return lookup(&t->t, upb_intkey(key), v, &upb_inthash, &inteql);
   }
-  return upb_table_lookup(&t->t, upb_intkey(key), &upb_inthash, &upb_inteql);
 }
 
 bool upb_inttable_remove(upb_inttable *t, uintptr_t key, upb_value *val) {
@@ -340,7 +363,9 @@ bool upb_inttable_remove(upb_inttable *t, uintptr_t key, upb_value *val) {
   if (key < t->array_size) {
     if (upb_arrhas(t->array[key])) {
       t->array_count--;
-      if (val) *val = t->array[key];
+      if (val) {
+        _upb_value_setval(val, t->array[key], t->t.type);
+      }
       ((upb_value*)t->array)[key] = upb_value_uint64(-1);
       success = true;
     } else {
@@ -348,10 +373,9 @@ bool upb_inttable_remove(upb_inttable *t, uintptr_t key, upb_value *val) {
     }
   } else {
     upb_tabkey removed;
-    success = upb_table_remove(
-        &t->t, upb_intkey(key), val, &removed, &upb_inthash, &upb_inteql);
+    success = rm(&t->t, upb_intkey(key), val, &removed, &upb_inthash, &inteql);
   }
-  upb_inttable_check(t);
+  check(t);
   return success;
 }
 
@@ -370,9 +394,9 @@ bool upb_inttable_insertptr(upb_inttable *t, const void *key, upb_value val) {
   return upb_inttable_insert(t, (uintptr_t)key, val);
 }
 
-const upb_value *upb_inttable_lookupptr(const upb_inttable *t,
-                                        const void *key) {
-  return upb_inttable_lookup(t, (uintptr_t)key);
+bool upb_inttable_lookupptr(const upb_inttable *t, const void *key,
+                            upb_value *v) {
+  return upb_inttable_lookup(t, (uintptr_t)key, v);
 }
 
 bool upb_inttable_removeptr(upb_inttable *t, const void *key, upb_value *val) {
@@ -426,7 +450,7 @@ void upb_inttable_next(upb_inttable_iter *iter) {
     iter->array_part = false;
     iter->ptr.ent = t->t.entries - 1;
   }
-  iter->ptr.ent = upb_table_next(&t->t, iter->ptr.ent);
+  iter->ptr.ent = next(&t->t, iter->ptr.ent);
 }
 
 #ifdef UPB_UNALIGNED_READS_OK

upb/table.h

@@ -47,11 +47,7 @@ typedef union {
 
 typedef struct _upb_tabent {
   upb_tabkey key;
-  // Storing a upb_value here wastes a bit of memory in debug mode because
-  // we are storing the type for each value even though we enforce that all
-  // values are the same.  But since this only affects debug mode, we don't
-  // worry too much about it.  The same applies to upb_inttable.array below.
-  upb_value val;
+  _upb_value val;
   // Internal chaining.  This is const so we can create static initializers for
   // tables.  We cast away const sometimes, but *only* when the containing
   // upb_table is known to be non-const.  This requires a bit of care, but
@@ -76,7 +72,7 @@ typedef struct {
 
 typedef struct {
   upb_table t;             // For entries that don't fit in the array part.
-  const upb_value *array;  // Array part of the table.
+  const _upb_value *array;  // Array part of the table.
   size_t array_size;       // Array part size.
   size_t array_count;      // Array part number of elements.
 } upb_inttable;
@@ -89,7 +85,7 @@ typedef struct {
 
 #define UPB_ARRAY_EMPTYENT UPB_VALUE_INIT_INT64(-1)
 
-INLINE size_t upb_table_size(const upb_table *t) {
+UPB_INLINE size_t upb_table_size(const upb_table *t) {
   if (t->size_lg2 == 0)
     return 0;
   else
@@ -97,12 +93,22 @@ INLINE size_t upb_table_size(const upb_table *t) {
 }
 
 // Internal-only functions, in .h file only out of necessity.
-INLINE bool upb_tabent_isempty(const upb_tabent *e) { return e->key.num == 0; }
-INLINE upb_tabkey upb_intkey(uintptr_t key) { upb_tabkey k = {key}; return k; }
-INLINE const upb_tabent *upb_inthash(const upb_table *t, upb_tabkey key) {
+UPB_INLINE bool upb_tabent_isempty(const upb_tabent *e) {
+  return e->key.num == 0;
+}
+
+UPB_INLINE upb_tabkey upb_intkey(uintptr_t key) {
+  upb_tabkey k = {key}; return k;
+}
+
+UPB_INLINE const upb_tabent *upb_inthash(const upb_table *t, upb_tabkey key) {
   return t->entries + ((uint32_t)key.num & t->mask);
 }
-INLINE bool upb_arrhas(upb_value v) { return v.val.uint64 != (uint64_t)-1; }
+
+UPB_INLINE bool upb_arrhas(_upb_value v) {
+  return v.uint64 != (uint64_t)-1;
+}
+
 uint32_t MurmurHash2(const void *key, size_t len, uint32_t seed);
 
 // Initialize and uninitialize a table, respectively.  If memory allocation
@@ -114,7 +120,9 @@ void upb_strtable_uninit(upb_strtable *table);
 
 // Returns the number of values in the table.
 size_t upb_inttable_count(const upb_inttable *t);
-INLINE size_t upb_strtable_count(const upb_strtable *t) { return t->t.count; }
+UPB_INLINE size_t upb_strtable_count(const upb_strtable *t) {
+  return t->t.count;
+}
 
 // Inserts the given key into the hashtable with the given value.  The key must
 // not already exist in the hash table.  For string tables, the key must be
@@ -129,8 +137,8 @@ bool upb_strtable_insert(upb_strtable *t, const char *key, upb_value val);
 // Looks up key in this table, returning a pointer to the table's internal copy
 // of the user's inserted data, or NULL if this key is not in the table.  The
 // returned pointer is invalidated by inserts.
-const upb_value *upb_inttable_lookup(const upb_inttable *t, uintptr_t key);
-const upb_value *upb_strtable_lookup(const upb_strtable *t, const char *key);
+bool upb_inttable_lookup(const upb_inttable *t, uintptr_t key, upb_value *v);
+bool upb_strtable_lookup(const upb_strtable *t, const char *key, upb_value *v);
 
 // Removes an item from the table.  Returns true if the remove was successful,
 // and stores the removed item in *val if non-NULL.
@@ -145,7 +153,8 @@ upb_value upb_inttable_pop(upb_inttable *t);
 // Convenience routines for inttables with pointer keys.
 bool upb_inttable_insertptr(upb_inttable *t, const void *key, upb_value val);
 bool upb_inttable_removeptr(upb_inttable *t, const void *key, upb_value *val);
-const upb_value *upb_inttable_lookupptr(const upb_inttable *t, const void *key);
+bool upb_inttable_lookupptr(
+    const upb_inttable *t, const void *key, upb_value *val);
 
 // Optimizes the table for the current set of entries, for both memory use and
 // lookup time.  Client should call this after all entries have been inserted;
@@ -153,17 +162,26 @@ const upb_value *upb_inttable_lookupptr(const upb_inttable *t, const void *key);
 void upb_inttable_compact(upb_inttable *t);
 
 // A special-case inlinable version of the lookup routine for 32-bit integers.
-INLINE const upb_value *upb_inttable_lookup32(const upb_inttable *t,
-                                              uint32_t key) {
+UPB_INLINE bool upb_inttable_lookup32(const upb_inttable *t, uint32_t key,
+                                      upb_value *v) {
   if (key < t->array_size) {
-    const upb_value *v = &t->array[key];
-    return upb_arrhas(*v) ? v : NULL;
-  }
-  const upb_tabent *e;
-  if (t->t.entries == NULL) return NULL;
-  for (e = upb_inthash(&t->t, upb_intkey(key)); true; e = e->next) {
-    if ((uint32_t)e->key.num == key) return &e->val;
-    if (e->next == NULL) return NULL;
+    _upb_value arrval = t->array[key];
+    if (upb_arrhas(arrval)) {
+      _upb_value_setval(v, arrval, t->t.type);
+      return true;
+    } else {
+      return false;
+    }
+  } else {
+    const upb_tabent *e;
+    if (t->t.entries == NULL) return NULL;
+    for (e = upb_inthash(&t->t, upb_intkey(key)); true; e = e->next) {
+      if ((uint32_t)e->key.num == key) {
+        _upb_value_setval(v, e->val, t->t.type);
+        return true;
+      }
+      if (e->next == NULL) return false;
+    }
   }
 }
 
@@ -185,12 +203,12 @@ typedef struct {
 
 void upb_strtable_begin(upb_strtable_iter *i, const upb_strtable *t);
 void upb_strtable_next(upb_strtable_iter *i);
-INLINE bool upb_strtable_done(upb_strtable_iter *i) { return i->e == NULL; }
-INLINE const char *upb_strtable_iter_key(upb_strtable_iter *i) {
+UPB_INLINE bool upb_strtable_done(upb_strtable_iter *i) { return i->e == NULL; }
+UPB_INLINE const char *upb_strtable_iter_key(upb_strtable_iter *i) {
   return i->e->key.str;
 }
-INLINE upb_value upb_strtable_iter_value(upb_strtable_iter *i) {
-  return i->e->val;
+UPB_INLINE upb_value upb_strtable_iter_value(upb_strtable_iter *i) {
+  return _upb_value_val(i->e->val, i->t->t.type);
 }
 
 
@@ -208,7 +226,7 @@ typedef struct {
   const upb_inttable *t;
   union {
     const upb_tabent *ent;  // For hash iteration.
-    const upb_value *val;   // For array iteration.
+    const _upb_value *val;   // For array iteration.
   } ptr;
   uintptr_t arrkey;
   bool array_part;
@@ -216,14 +234,15 @@ typedef struct {
 
 void upb_inttable_begin(upb_inttable_iter *i, const upb_inttable *t);
 void upb_inttable_next(upb_inttable_iter *i);
-INLINE bool upb_inttable_done(upb_inttable_iter *i) {
+UPB_INLINE bool upb_inttable_done(upb_inttable_iter *i) {
   return i->ptr.ent == NULL;
 }
-INLINE uintptr_t upb_inttable_iter_key(upb_inttable_iter *i) {
+UPB_INLINE uintptr_t upb_inttable_iter_key(upb_inttable_iter *i) {
   return i->array_part ? i->arrkey : i->ptr.ent->key.num;
 }
-INLINE upb_value upb_inttable_iter_value(upb_inttable_iter *i) {
-  return i->array_part ? *i->ptr.val : i->ptr.ent->val;
+UPB_INLINE upb_value upb_inttable_iter_value(upb_inttable_iter *i) {
+  return _upb_value_val(
+      i->array_part ? *i->ptr.val : i->ptr.ent->val, i->t->t.type);
 }
 
 #ifdef __cplusplus

upb/upb.h

@@ -14,6 +14,7 @@
 #define UPB_H_
 
 #include <assert.h>
+#include <limits.h>
 #include <stdarg.h>
 #include <stdbool.h>
 #include <stddef.h>
@@ -24,8 +25,10 @@ extern "C" {
 #endif
 
 // inline if possible, emit standalone code if required.
-#ifndef INLINE
-#define INLINE static inline
+#ifdef __cplusplus
+#define UPB_INLINE inline
+#else
+#define UPB_INLINE static inline
 #endif
 
 #if __STDC_VERSION__ >= 199901L
@@ -50,20 +53,107 @@ extern "C" {
   void operator=(const class_name&);
 #endif
 
+#if defined(__clang__) && defined(LANG_CXX11) && defined(__has_warning)
+#if __has_feature(cxx_attributes) && __has_warning("-Wimplicit-fallthrough")
+#define UPB_FALLTHROUGH_INTENDED [[clang::fallthrough]]
+#endif
+#endif
+
+#ifndef UPB_FALLTHROUGH_INTENDED
+#define UPB_FALLTHROUGH_INTENDED do { } while (0)
+#endif
+
 #ifdef __GNUC__
 #define UPB_NORETURN __attribute__((__noreturn__))
 #else
 #define UPB_NORETURN
 #endif
 
-#ifndef UINT16_MAX
-#define UINT16_MAX 0xffff
+// Type detection and typedefs for integer types.
+//
+// We unfortunately cannot just use stdint.h types in all cases, because some
+// platforms have more than one 32-bit type (or 64-bit type).  For example, on
+// x86-64, both "long" and "long long" are 64-bit types, but they are
+// unfortunately incompatible with each other despite being the same size.
+// Since the types are incompatible, functions pointers between them are
+// incompatible also, which leads to trouble since handlers are declared in
+// terms of function pointers.  Since we don't know which of these types
+// stdint.h will use (and we have no way of inspecting the typedefs, either at
+// preprocessing or compilation time), we are forced to declare our own
+// typedefs that we *do* know the real underlying type of.
+//
+// If any platform existed where there three integer types were the same size,
+// this would have to become more complicated.  For example, short, int, and
+// long could all be 32-bits.  Even more diabolically, short, int, long, and
+// long long could all be 64 bits and still be standard-compliant.  However,
+// few platforms are this strange, and it's unlikely that upb will be used on
+// the strangest ones.
+//
+// For more information, see:
+// http://blog.reverberate.org/2013/03/cc-gripe-1-integer-types.html
+
+// Can't count on stdint.h limits like INT32_MAX, because in C++ these are
+// only defined when __STDC_LIMIT_MACROS are defined before the *first* include
+// of stdint.h.  We can't guarantee that someone else didn't include these first
+// without defining __STDC_LIMIT_MACROS.
+#define UPB_INT32_MAX 0x7fffffffLL
+#define UPB_INT32_MIN (-UPB_INT32_MAX - 1)
+#define UPB_INT64_MAX 0x7fffffffffffffffLL
+#define UPB_INT64_MIN (-UPB_INT64_MAX - 1)
+
+#if INT_MAX == UPB_INT32_MAX && INT_MIN == UPB_INT32_MIN
+#define UPB_INT_IS_32BITS 1
+#endif
+
+#if LONG_MAX == UPB_INT32_MAX && LONG_MIN == UPB_INT32_MIN
+#define UPB_LONG_IS_32BITS 1
+#endif
+
+#if LONG_MAX == UPB_INT64_MAX && LONG_MIN == UPB_INT64_MIN
+#define UPB_LONG_IS_64BITS 1
 #endif
 
-#ifndef UINT32_MAX
-#define UINT32_MAX 0xffffffff
+#if LLONG_MAX == UPB_INT64_MAX && LLONG_MIN == UPB_INT64_MIN
+#define UPB_LLONG_IS_64BITS 1
 #endif
 
+#if UPB_INT_IS_32BITS
+typedef int upb_int32_t;
+typedef unsigned int upb_uint32_t;
+#define UPB_INT32_CTYPE i
+
+#if UPB_LONG_IS_32BITS
+#define UPB_TWO_32BIT_TYPES 1
+typedef long upb_int32alt_t;
+typedef unsigned long upb_uint32alt_t;
+#define UPB_INT32_CTYPE2 l
+#endif  // UPB_LONG_IS_32BITS
+
+#elif UPB_LONG_IS_32BITS  // && !UPB_INT_IS_32BITS
+typedef long upb_int32_t;
+typedef unsigned long upb_uint32_t;
+#define UPB_INT32_CTYPE l
+#endif  // UPB_INT_IS_32BITS
+
+
+#if UPB_LONG_IS_64BITS
+typedef long upb_int64_t;
+typedef unsigned long upb_uint64_t;
+#define UPB_INT64_CTYPE l
+
+#if UPB_LLONG_IS_64BITS
+#define UPB_TWO_64BIT_TYPES 1
+typedef long long upb_int64alt_t;
+typedef unsigned long long upb_uint64alt_t;
+#define UPB_INT64_CTYPE2 ll
+#endif  // UPB_LLONG_IS_64BITS
+
+#elif UPB_LLONG_IS_64BITS  // && !UPB_LONG_IS_64BITS
+typedef long long upb_int64_t;
+typedef unsigned long long upb_uint64_t;
+#define UPB_INT64_CTYPE ll
+#endif  // UPB_LONG_IS_64BITS
+
 #define UPB_MAX(x, y) ((x) > (y) ? (x) : (y))
 #define UPB_MIN(x, y) ((x) < (y) ? (x) : (y))
 
@@ -147,8 +237,8 @@ class upb::Status {
   Status();
   ~Status();
 
-  bool ok();
-  bool eof();
+  bool ok() const;
+  bool eof() const;
 
   const char *GetString() const;
   void SetEof();
@@ -225,31 +315,23 @@ typedef enum {
   UPB_CTYPE_FIELDDEF = 11,
 } upb_ctype_t;
 
-#ifdef __cplusplus
-namespace upb { class ByteRegion; }
-typedef upb::ByteRegion upb_byteregion;
-#else
-struct upb_byteregion;
-typedef struct upb_byteregion upb_byteregion;
-#endif
+typedef union {
+  uint64_t uint64;
+  int32_t int32;
+  int64_t int64;
+  uint32_t uint32;
+  double _double;
+  float _float;
+  bool _bool;
+  char *cstr;
+  void *ptr;
+  const void *constptr;
+} _upb_value;
 
 // A single .proto value.  The owner must have an out-of-band way of knowing
 // the type, so that it knows which union member to use.
 typedef struct {
-  union {
-    uint64_t uint64;
-    int32_t int32;
-    int64_t int64;
-    uint32_t uint32;
-    double _double;
-    float _float;
-    bool _bool;
-    char *cstr;
-    void *ptr;
-    const void *constptr;
-    upb_byteregion *byteregion;
-  } val;
-
+  _upb_value val;
 #ifndef NDEBUG
   // In debug mode we carry the value type around also so we can check accesses
   // to be sure the right member is being read.
@@ -258,30 +340,44 @@ typedef struct {
 } upb_value;
 
 #ifdef UPB_C99
-#define UPB_VAL_INIT(v, member) {.member = v}
+#define UPB_VALUE_INIT(v, member) {.member = v}
 #endif
 // TODO(haberman): C++
+//
+//
+#define UPB__VALUE_INIT_NONE      UPB_VALUE_INIT(NULL, ptr)
 
 #ifdef NDEBUG
 #define SET_TYPE(dest, val)
-#define UPB_VALUE_INIT(v, member, type) {UPB_VAL_INIT(v, member)}
+#define UPB_VALUE_INIT_NONE      {UPB__VALUE_INIT_NONE}
 #else
 #define SET_TYPE(dest, val) dest = val
-#define UPB_VALUE_INIT(v, member, type) {UPB_VAL_INIT(v, member), type}
+// Non-existent type, all reads will fail.
+#define UPB_VALUE_INIT_NONE      {UPB__VALUE_INIT_NONE, -1}
 #endif
 
-#define UPB_VALUE_INIT_INT32(v)  UPB_VALUE_INIT(v, int32,   UPB_CTYPE_INT32)
-#define UPB_VALUE_INIT_INT64(v)  UPB_VALUE_INIT(v, int64,   UPB_CTYPE_INT64)
-#define UPB_VALUE_INIT_UINT32(v) UPB_VALUE_INIT(v, uint32,  UPB_CTYPE_UINT32)
-#define UPB_VALUE_INIT_UINT64(v) UPB_VALUE_INIT(v, uint64,  UPB_CTYPE_UINT64)
-#define UPB_VALUE_INIT_DOUBLE(v) UPB_VALUE_INIT(v, _double, UPB_CTYPE_DOUBLE)
-#define UPB_VALUE_INIT_FLOAT(v)  UPB_VALUE_INIT(v, _float,  UPB_CTYPE_FLOAT)
-#define UPB_VALUE_INIT_BOOL(v)   UPB_VALUE_INIT(v, _bool,   UPB_CTYPE_BOOL)
-#define UPB_VALUE_INIT_CSTR(v)   UPB_VALUE_INIT(v, cstr,    UPB_CTYPE_CSTR)
-#define UPB_VALUE_INIT_PTR(v)    UPB_VALUE_INIT(v, ptr,     UPB_CTYPE_PTR)
-#define UPB_VALUE_INIT_CONSTPTR(v) UPB_VALUE_INIT(v, constptr, UPB_CTYPE_PTR)
-// Non-existent type, all reads will fail.
-#define UPB_VALUE_INIT_NONE      UPB_VALUE_INIT(NULL, ptr, -1)
+#define UPB_VALUE_INIT_INT32(v)  UPB_VALUE_INIT(v, int32)
+#define UPB_VALUE_INIT_INT64(v)  UPB_VALUE_INIT(v, int64)
+#define UPB_VALUE_INIT_UINT32(v) UPB_VALUE_INIT(v, uint32)
+#define UPB_VALUE_INIT_UINT64(v) UPB_VALUE_INIT(v, uint64)
+#define UPB_VALUE_INIT_DOUBLE(v) UPB_VALUE_INIT(v, _double)
+#define UPB_VALUE_INIT_FLOAT(v)  UPB_VALUE_INIT(v, _float)
+#define UPB_VALUE_INIT_BOOL(v)   UPB_VALUE_INIT(v, _bool)
+#define UPB_VALUE_INIT_CSTR(v)   UPB_VALUE_INIT(v, cstr)
+#define UPB_VALUE_INIT_PTR(v)    UPB_VALUE_INIT(v, ptr)
+#define UPB_VALUE_INIT_CONSTPTR(v) UPB_VALUE_INIT(v, constptr)
+
+UPB_INLINE void _upb_value_setval(upb_value *v, _upb_value val,
+                                  upb_ctype_t type) {
+  v->val = val;
+  SET_TYPE(v->type, type);
+}
+
+UPB_INLINE upb_value _upb_value_val(_upb_value val, upb_ctype_t type) {
+  upb_value ret;
+  _upb_value_setval(&ret, val, type);
+  return ret;
+}
 
 // For each value type, define the following set of functions:
 //
@@ -293,12 +389,12 @@ typedef struct {
 // upb_value upb_value_int32(int32_t val);
 
 #define WRITERS(name, membername, ctype, proto_type) \
-  INLINE void upb_value_set ## name(upb_value *val, ctype cval) { \
+  UPB_INLINE void upb_value_set ## name(upb_value *val, ctype cval) { \
     val->val.uint64 = 0; \
     SET_TYPE(val->type, proto_type); \
     val->val.membername = cval; \
   } \
-  INLINE upb_value upb_value_ ## name(ctype val) { \
+  UPB_INLINE upb_value upb_value_ ## name(ctype val) { \
     upb_value ret; \
     upb_value_set ## name(&ret, val); \
     return ret; \
@@ -307,17 +403,17 @@ typedef struct {
 #define ALL(name, membername, ctype, proto_type) \
   /* Can't reuse WRITERS() here unfortunately because "bool" is a macro \
    * that expands to _Bool, so it ends up defining eg. upb_value_set_Bool */ \
-  INLINE void upb_value_set ## name(upb_value *val, ctype cval) { \
+  UPB_INLINE void upb_value_set ## name(upb_value *val, ctype cval) { \
     val->val.uint64 = 0; \
     SET_TYPE(val->type, proto_type); \
     val->val.membername = cval; \
   } \
-  INLINE upb_value upb_value_ ## name(ctype val) { \
+  UPB_INLINE upb_value upb_value_ ## name(ctype val) { \
     upb_value ret; \
     upb_value_set ## name(&ret, val); \
     return ret; \
   } \
-  INLINE ctype upb_value_get ## name(upb_value val) { \
+  UPB_INLINE ctype upb_value_get ## name(upb_value val) { \
     assert(val.type == proto_type); \
     return val.val.membername; \
   }
@@ -329,7 +425,6 @@ ALL(uint64, uint64,  uint64_t, UPB_CTYPE_UINT64);
 ALL(bool,   _bool,   bool,     UPB_CTYPE_BOOL);
 ALL(cstr,   cstr,    char*,    UPB_CTYPE_CSTR);
 ALL(ptr,    ptr,     void*,    UPB_CTYPE_PTR);
-ALL(byteregion, byteregion, upb_byteregion*, UPB_CTYPE_BYTEREGION);
 
 #ifdef __KERNEL__
 // Linux kernel modules are compiled without SSE and therefore are incapable
@@ -387,8 +482,8 @@ template <typename T> inline Value MakePtrValue(T* v) {
 // C++ Wrappers
 inline Status::Status() { upb_status_init(this); }
 inline Status::~Status() { upb_status_uninit(this); }
-inline bool Status::ok() { return upb_ok(this); }
-inline bool Status::eof() { return upb_eof(this); }
+inline bool Status::ok() const { return upb_ok(this); }
+inline bool Status::eof() const { return upb_eof(this); }
 inline const char *Status::GetString() const { return upb_status_getstr(this); }
 inline void Status::SetEof() { upb_status_seteof(this); }
 inline void Status::SetErrorLiteral(const char* msg) {