Sync from Google development.

Makefile

@@ -47,7 +47,7 @@ endif
 CC=gcc
 CXX=g++
 CFLAGS=-std=gnu99
-CXXFLAGS=-Ibindings/cpp
+CXXFLAGS=-std=c++11
 INCLUDE=-Itests -I.
 CPPFLAGS=$(INCLUDE) -Wall -Wextra -Wno-sign-compare $(USER_CFLAGS)
 LDLIBS=-lpthread upb/libupb.a
@@ -220,10 +220,7 @@ SIMPLE_TESTS= \
 
 SIMPLE_CXX_TESTS= \
   tests/test_cpp \
-
-  # The build process for this test is complicated and hasn't been
-  # ported to the open-source Makefile yet.
-  # tests/test_decoder \
+  tests/pb/test_decoder
 
 VARIADIC_TESTS= \
   tests/t.test_vs_proto2.googlemessage1 \

tests/pb/test_decoder.cc

@@ -37,22 +37,25 @@
 #include <string.h>
 
 #include "tests/upb_test.h"
-#include "third_party/upb/tests/pb/test_decoder_schema.upb.h"
 #include "upb/handlers.h"
 #include "upb/pb/decoder.h"
 #include "upb/pb/varint.int.h"
 #include "upb/upb.h"
 
 #undef PRINT_FAILURE
-#define PRINT_FAILURE(expr) \
-  fprintf(stderr, "Assertion failed: %s:%d\n", __FILE__, __LINE__); \
-  fprintf(stderr, "expr: %s\n", #expr); \
-  if (testhash) { \
-    fprintf(stderr, "assertion failed running test %x.  " \
-                    "Run with the arg %x to run only this test.\n", \
-                    testhash, testhash); \
-    fprintf(stderr, "Failed at %02.2f%% through tests.\n", \
-                    (float)completed * 100 / total); \
+#define PRINT_FAILURE(expr)                                           \
+  fprintf(stderr, "Assertion failed: %s:%d\n", __FILE__, __LINE__);   \
+  fprintf(stderr, "expr: %s\n", #expr);                               \
+  if (testhash) {                                                     \
+    fprintf(stderr, "assertion failed running test %x.\n", testhash); \
+    if (!filter_hash) {                                               \
+      fprintf(stderr,                                                 \
+              "Run with the arg %x to run only this test. "           \
+              "(This will also turn on extra debugging output)\n",    \
+              testhash);                                              \
+    }                                                                 \
+    fprintf(stderr, "Failed at %02.2f%% through tests.\n",            \
+            (float)completed * 100 / total);                          \
   }
 
 uint32_t filter_hash = 0;
@@ -90,106 +93,101 @@ static const upb_decoder_typeinfo upb_decoder_types[] = {
   {UPB_WIRE_TYPE_VARINT,      true},   // SINT64
 };
 
+#ifndef USE_GOOGLE
+using std::string;
+#endif
 
-class buffer {
- public:
-  buffer(const void *data, size_t len) : len_(0) { append(data, len); }
-  explicit buffer(const char *data) : len_(0) { append(data); }
-  explicit buffer(size_t len) : len_(len) { memset(buf_, 0, len); }
-  buffer(const buffer& buf) : len_(0) { append(buf); }
-  buffer() : len_(0) {}
-
-  void append(const void *data, size_t len) {
-    ASSERT_NOCOUNT(len + len_ < sizeof(buf_));
-    memcpy(buf_ + len_, data, len);
-    len_ += len;
-    buf_[len_] = NULL;
-  }
-
-  void append(const buffer& buf) {
-    append(buf.buf_, buf.len_);
-  }
-
-  void append(const char *str) {
-    append(str, strlen(str));
-  }
-
-  void vappendf(const char *fmt, va_list args) {
-    size_t avail = sizeof(buf_) - len_;
-    size_t size = vsnprintf(buf_ + len_, avail, fmt, args);
-    ASSERT_NOCOUNT(avail > size);
-    len_ += size;
-  }
+void vappendf(string* str, const char *format, va_list args) {
+  va_list copy;
+  va_copy(copy, args);
 
-  void appendf(const char *fmt, ...) {
-    va_list args;
-    va_start(args, fmt);
-    vappendf(fmt, args);
-    va_end(args);
+  int count = vsnprintf(NULL, 0, format, args);
+  if (count >= 0)
+  {
+    assert(count < 32768);
+    char *buffer = new char[count + 1];
+    assert(buffer);
+    count = vsnprintf(buffer, count + 1, format, copy);
+    assert(count >= 0);
+    str->append(buffer, count);
+    delete [] buffer;
   }
+  va_end(copy);
+}
 
-  void assign(const buffer& buf) {
-    clear();
-    append(buf);
-  }
+void appendf(string* str, const char *fmt, ...) {
+  va_list args;
+  va_start(args, fmt);
+  vappendf(str, fmt, args);
+  va_end(args);
+}
 
-  bool eql(const buffer& other) const {
-    return len_ == other.len_ && memcmp(buf_, other.buf_, len_) == 0;
+void PrintBinary(const string& str) {
+  for (size_t i = 0; i < str.size(); i++) {
+    if (isprint(str[i])) {
+      fprintf(stderr, "%c", str[i]);
+    } else {
+      fprintf(stderr, "\\x%02x", str[i]);
+    }
   }
-
-  void clear() { len_ = 0; }
-  size_t len() const { return len_; }
-  const char *buf() const { return buf_; }
-
- private:
-  // Has to be big enough for the largest string used in the test.
-  char buf_[32768];
-  size_t len_;
-};
-
+}
 
 /* Routines for building arbitrary protos *************************************/
 
-const buffer empty;
-
-buffer cat(const buffer& a, const buffer& b,
-           const buffer& c = empty,
-           const buffer& d = empty,
-           const buffer& e = empty,
-           const buffer& f = empty) {
-  buffer ret;
+const string empty;
+
+string cat(const string& a, const string& b,
+           const string& c = empty,
+           const string& d = empty,
+           const string& e = empty,
+           const string& f = empty,
+           const string& g = empty,
+           const string& h = empty,
+           const string& i = empty,
+           const string& j = empty,
+           const string& k = empty,
+           const string& l = empty) {
+  string ret;
+  ret.reserve(a.size() + b.size() + c.size() + d.size() + e.size() + f.size() +
+              g.size() + h.size() + i.size() + j.size() + k.size() + l.size());
   ret.append(a);
   ret.append(b);
   ret.append(c);
   ret.append(d);
   ret.append(e);
   ret.append(f);
+  ret.append(g);
+  ret.append(h);
+  ret.append(i);
+  ret.append(j);
+  ret.append(k);
+  ret.append(l);
   return ret;
 }
 
-buffer varint(uint64_t x) {
+string varint(uint64_t x) {
   char buf[UPB_PB_VARINT_MAX_LEN];
   size_t len = upb_vencode64(x, buf);
-  return buffer(buf, len);
+  return string(buf, len);
 }
 
 // TODO: proper byte-swapping for big-endian machines.
-buffer fixed32(void *data) { return buffer(data, 4); }
-buffer fixed64(void *data) { return buffer(data, 8); }
-
-buffer delim(const buffer& buf) { return cat(varint(buf.len()), buf); }
-buffer uint32(uint32_t u32) { return fixed32(&u32); }
-buffer uint64(uint64_t u64) { return fixed64(&u64); }
-buffer flt(float f) { return fixed32(&f); }
-buffer dbl(double d) { return fixed64(&d); }
-buffer zz32(int32_t x) { return varint(upb_zzenc_32(x)); }
-buffer zz64(int64_t x) { return varint(upb_zzenc_64(x)); }
-
-buffer tag(uint32_t fieldnum, char wire_type) {
+string fixed32(void *data) { return string(static_cast<char*>(data), 4); }
+string fixed64(void *data) { return string(static_cast<char*>(data), 8); }
+
+string delim(const string& buf) { return cat(varint(buf.size()), buf); }
+string uint32(uint32_t u32) { return fixed32(&u32); }
+string uint64(uint64_t u64) { return fixed64(&u64); }
+string flt(float f) { return fixed32(&f); }
+string dbl(double d) { return fixed64(&d); }
+string zz32(int32_t x) { return varint(upb_zzenc_32(x)); }
+string zz64(int64_t x) { return varint(upb_zzenc_64(x)); }
+
+string tag(uint32_t fieldnum, char wire_type) {
   return varint((fieldnum << 3) | wire_type);
 }
 
-buffer submsg(uint32_t fn, const buffer& buf) {
+string submsg(uint32_t fn, const string& buf) {
   return cat( tag(fn, UPB_WIRE_TYPE_DELIMITED), delim(buf) );
 }
 
@@ -203,18 +201,17 @@ buffer submsg(uint32_t fn, const buffer& buf) {
 // handled.
 
 int closures[UPB_DECODER_MAX_NESTING];
-buffer output;
+string output;
 
-void indentbuf(buffer *buf, int depth) {
-  for (int i = 0; i < depth; i++)
-    buf->append("  ", 2);
+void indentbuf(string *buf, int depth) {
+  buf->append(2 * depth, ' ');
 }
 
-#define NUMERIC_VALUE_HANDLER(member, ctype, fmt) \
-  bool value_ ## member(int* depth, const uint32_t* num, ctype val) {          \
-    indentbuf(&output, *depth);                                                \
-    output.appendf("%" PRIu32 ":%" fmt "\n", *num, val);                       \
-    return true;                                                               \
+#define NUMERIC_VALUE_HANDLER(member, ctype, fmt)                   \
+  bool value_##member(int* depth, const uint32_t* num, ctype val) { \
+    indentbuf(&output, *depth);                                     \
+    appendf(&output, "%" PRIu32 ":%" fmt "\n", *num, val);          \
+    return true;                                                    \
   }
 
 NUMERIC_VALUE_HANDLER(uint32, uint32_t, PRIu32)
@@ -226,19 +223,20 @@ NUMERIC_VALUE_HANDLER(double, double,   "g")
 
 bool value_bool(int* depth, const uint32_t* num, bool val) {
   indentbuf(&output, *depth);
-  output.appendf("%" PRIu32 ":%s\n", *num, val ? "true" : "false");
+  appendf(&output, "%" PRIu32 ":%s\n", *num, val ? "true" : "false");
   return true;
 }
 
 int* startstr(int* depth, const uint32_t* num, size_t size_hint) {
   indentbuf(&output, *depth);
-  output.appendf("%" PRIu32 ":(%zu)\"", *num, size_hint);
+  appendf(&output, "%" PRIu32 ":(%zu)\"", *num, size_hint);
   return depth + 1;
 }
 
 size_t value_string(int* depth, const uint32_t* num, const char* buf,
                     size_t n, const upb::BufferHandle* handle) {
   UPB_UNUSED(num);
+  UPB_UNUSED(depth);
   output.append(buf, n);
   ASSERT(handle == &global_handle);
   return n;
@@ -253,7 +251,7 @@ bool endstr(int* depth, const uint32_t* num) {
 
 int* startsubmsg(int* depth, const uint32_t* num) {
   indentbuf(&output, *depth);
-  output.appendf("%" PRIu32 ":{\n", *num);
+  appendf(&output, "%" PRIu32 ":{\n", *num);
   return depth + 1;
 }
 
@@ -266,7 +264,7 @@ bool endsubmsg(int* depth, const uint32_t* num) {
 
 int* startseq(int* depth, const uint32_t* num) {
   indentbuf(&output, *depth);
-  output.appendf("%" PRIu32 ":[\n", *num);
+  appendf(&output, "%" PRIu32 ":[\n", *num);
   return depth + 1;
 }
 
@@ -284,6 +282,7 @@ bool startmsg(int* depth) {
 }
 
 bool endmsg(int* depth, upb_status* status) {
+  UPB_UNUSED(status);
   indentbuf(&output, *depth);
   output.append(">\n");
   return true;
@@ -298,7 +297,7 @@ template<class T, bool F(int*, const uint32_t*, T)>
 void doreg(upb_handlers *h, uint32_t num) {
   const upb_fielddef *f = upb_msgdef_itof(upb_handlers_msgdef(h), num);
   ASSERT(f);
-  ASSERT(h->SetValueHandler<T>(f, UpbBindT(F, new uint32_t(num))));
+  ASSERT(h->SetValueHandler<T>(f, UpbBindT(&F, new uint32_t(num))));
   if (f->IsSequence()) {
     ASSERT(h->SetStartSequenceHandler(f, UpbBind(startseq, new uint32_t(num))));
     ASSERT(h->SetEndSequenceHandler(f, UpbBind(endseq, new uint32_t(num))));
@@ -347,9 +346,93 @@ void reg_str(upb_handlers *h, uint32_t num) {
   ASSERT(h->SetStringHandler(f, UpbBind(value_string, new uint32_t(num))));
 }
 
+void AddField(upb_descriptortype_t descriptor_type, const std::string& name,
+              uint32_t fn, bool repeated, upb::MessageDef* md) {
+  // TODO: Fluent interface?  ie.
+  //   ASSERT(md->AddField(upb::BuildFieldDef()
+  //       .SetName("f_message")
+  //       .SetNumber(UPB_DESCRIPTOR_TYPE_MESSAGE)
+  //       .SetDescriptorType(UPB_DESCRIPTOR_TYPE_MESSAGE)
+  //       .SetMessageSubdef(md.get())));
+  upb::reffed_ptr<upb::FieldDef> f = upb::FieldDef::New();
+  ASSERT(f->set_name(name, NULL));
+  ASSERT(f->set_number(fn, NULL));
+  f->set_label(repeated ? UPB_LABEL_REPEATED : UPB_LABEL_OPTIONAL);
+  f->set_descriptor_type(descriptor_type);
+  ASSERT(md->AddField(f.get(), NULL));
+}
+
+void AddFieldsForType(upb_descriptortype_t descriptor_type,
+                      const char* basename, upb::MessageDef* md) {
+  const upb_descriptortype_t t = descriptor_type;
+  AddField(t, std::string("f_") + basename, t, false, md);
+  AddField(t, std::string("r_") + basename, rep_fn(t), true, md);
+}
+
+upb::reffed_ptr<const upb::MessageDef> NewMessageDef() {
+  upb::reffed_ptr<upb::MessageDef> md = upb::MessageDef::New();
+
+  md->set_full_name("DecoderTest", NULL);
+
+  AddFieldsForType(UPB_DESCRIPTOR_TYPE_DOUBLE, "double", md.get());
+  AddFieldsForType(UPB_DESCRIPTOR_TYPE_FLOAT, "float", md.get());
+  AddFieldsForType(UPB_DESCRIPTOR_TYPE_INT64, "int64", md.get());
+  AddFieldsForType(UPB_DESCRIPTOR_TYPE_UINT64, "uint64", md.get());
+  AddFieldsForType(UPB_DESCRIPTOR_TYPE_INT32, "int32", md.get());
+  AddFieldsForType(UPB_DESCRIPTOR_TYPE_FIXED64, "fixed64", md.get());
+  AddFieldsForType(UPB_DESCRIPTOR_TYPE_FIXED32, "fixed32", md.get());
+  AddFieldsForType(UPB_DESCRIPTOR_TYPE_BOOL, "bool", md.get());
+  AddFieldsForType(UPB_DESCRIPTOR_TYPE_STRING, "string", md.get());
+  AddFieldsForType(UPB_DESCRIPTOR_TYPE_BYTES, "bytes", md.get());
+  AddFieldsForType(UPB_DESCRIPTOR_TYPE_UINT32, "uint32", md.get());
+  AddFieldsForType(UPB_DESCRIPTOR_TYPE_SFIXED32, "sfixed32", md.get());
+  AddFieldsForType(UPB_DESCRIPTOR_TYPE_SFIXED64, "sfixed64", md.get());
+  AddFieldsForType(UPB_DESCRIPTOR_TYPE_SINT32, "sint32", md.get());
+  AddFieldsForType(UPB_DESCRIPTOR_TYPE_SINT64, "sint64", md.get());
+
+  AddField(UPB_DESCRIPTOR_TYPE_STRING, "nop_field", 40, false, md.get());
+
+  upb::reffed_ptr<upb::FieldDef> f = upb::FieldDef::New();
+  ASSERT(f->set_name("f_message", NULL));
+  ASSERT(f->set_number(UPB_DESCRIPTOR_TYPE_MESSAGE, NULL));
+  f->set_descriptor_type(UPB_DESCRIPTOR_TYPE_MESSAGE);
+  ASSERT(f->set_message_subdef(md.get(), NULL));
+  ASSERT(md->AddField(f.get(), NULL));
+
+  f = upb::FieldDef::New();
+  ASSERT(f->set_name("r_message", NULL));
+  ASSERT(f->set_number(rep_fn(UPB_DESCRIPTOR_TYPE_MESSAGE), NULL));
+  f->set_label(UPB_LABEL_REPEATED);
+  f->set_descriptor_type(UPB_DESCRIPTOR_TYPE_MESSAGE);
+  ASSERT(f->set_message_subdef(md.get(), NULL));
+  ASSERT(md->AddField(f.get(), NULL));
+
+  upb::reffed_ptr<upb::EnumDef> e = upb::EnumDef::New();
+  ASSERT(e->AddValue("FOO", 1, NULL));
+  ASSERT(e->Freeze(NULL));
+
+  f = upb::FieldDef::New();
+  ASSERT(f->set_name("f_enum", NULL));
+  ASSERT(f->set_number(UPB_DESCRIPTOR_TYPE_ENUM, NULL));
+  f->set_descriptor_type(UPB_DESCRIPTOR_TYPE_ENUM);
+  ASSERT(f->set_enum_subdef(e.get(), NULL));
+  ASSERT(md->AddField(f.get(), NULL));
+
+  f = upb::FieldDef::New();
+  ASSERT(f->set_name("r_enum", NULL));
+  ASSERT(f->set_number(rep_fn(UPB_DESCRIPTOR_TYPE_ENUM), NULL));
+  f->set_label(UPB_LABEL_REPEATED);
+  f->set_descriptor_type(UPB_DESCRIPTOR_TYPE_ENUM);
+  ASSERT(f->set_enum_subdef(e.get(), NULL));
+  ASSERT(md->AddField(f.get(), NULL));
+
+  ASSERT(md->Freeze(NULL));
+
+  return md;
+}
+
 upb::reffed_ptr<const upb::Handlers> NewHandlers() {
-  upb::reffed_ptr<upb::Handlers> h(
-      upb::Handlers::New(UPB_TEST_DECODER_DECODERTEST));
+  upb::reffed_ptr<upb::Handlers> h(upb::Handlers::New(NewMessageDef().get()));
 
   h->SetStartMessageHandler(UpbMakeHandler(startmsg));
   h->SetEndMessageHandler(UpbMakeHandler(endmsg));
@@ -395,11 +478,11 @@ upb::reffed_ptr<const upb::Handlers> NewHandlers() {
 const upb::Handlers *global_handlers;
 const upb::pb::DecoderMethod *global_method;
 
-uint32_t Hash(const buffer& proto, const buffer* expected_output, size_t seam1,
+uint32_t Hash(const string& proto, const string* expected_output, size_t seam1,
               size_t seam2) {
-  uint32_t hash = MurmurHash2(proto.buf(), proto.len(), 0);
+  uint32_t hash = MurmurHash2(proto.c_str(), proto.size(), 0);
   if (expected_output)
-    hash = MurmurHash2(expected_output->buf(), expected_output->len(), hash);
+    hash = MurmurHash2(expected_output->c_str(), expected_output->size(), hash);
   hash = MurmurHash2(&seam1, sizeof(seam1), hash);
   hash = MurmurHash2(&seam2, sizeof(seam2), hash);
   return hash;
@@ -410,8 +493,38 @@ bool parse(upb::BytesSink* s, void* subc, const char* buf, size_t start,
   start = UPB_MAX(start, *ofs);
   if (start <= end) {
     size_t len = end - start;
+    if (filter_hash) {
+      fprintf(stderr, "Calling parse(%zu) for bytes %zu-%zu of the input\n",
+              len, start, end);
+    }
     size_t parsed = s->PutBuffer(subc, buf + start, len, &global_handle);
+    if (filter_hash) {
+      if (parsed == len) {
+        fprintf(stderr,
+                "parse(%zu) = %zu, complete byte count indicates success\n",
+                len, len);
+      } else if (parsed > len) {
+        fprintf(stderr,
+                "parse(%zu) = %zu, long byte count indicates success and skip"
+                "of the next %zu bytes\n",
+                len, parsed, parsed - len);
+      } else {
+        fprintf(stderr,
+                "parse(%zu) = %zu, short byte count indicates failure; "
+                "last %zu bytes were not consumed\n",
+                len, parsed, len - parsed);
+      }
+    }
     if (status->ok() != (parsed >= len)) {
+      if (status->ok()) {
+        fprintf(stderr,
+                "Error: decode function returned short byte count but set no "
+                "error status\n");
+      } else {
+        fprintf(stderr,
+                "Error: decode function returned complete byte count but set "
+                "error status\n");
+      }
       fprintf(stderr, "Status: %s, parsed=%zu, len=%zu\n",
               status->error_message(), parsed, len);
       ASSERT(false);
@@ -424,13 +537,13 @@ bool parse(upb::BytesSink* s, void* subc, const char* buf, size_t start,
 }
 
 #define LINE(x) x "\n"
-void run_decoder(const buffer& proto, const buffer* expected_output) {
+void run_decoder(const string& proto, const string* expected_output) {
   upb::Status status;
   upb::pb::Decoder decoder(global_method, &status);
   upb::Sink sink(global_handlers, &closures[0]);
   decoder.ResetOutput(&sink);
-  for (size_t i = 0; i < proto.len(); i++) {
-    for (size_t j = i; j < UPB_MIN(proto.len(), i + 5); j++) {
+  for (size_t i = 0; i < proto.size(); i++) {
+    for (size_t j = i; j < UPB_MIN(proto.size(), i + 5); j++) {
       testhash = Hash(proto, expected_output, i, j);
       if (filter_hash && testhash != filter_hash) continue;
       if (!count_only) {
@@ -440,27 +553,50 @@ void run_decoder(const buffer& proto, const buffer* expected_output) {
         size_t ofs = 0;
         upb::BytesSink* input = decoder.input();
         void *sub;
+
+        if (filter_hash) {
+          fprintf(stderr, "RUNNING TEST CASE, hash=%x\n", testhash);
+          fprintf(stderr, "JIT on: %s\n",
+                  global_method->is_native() ? "true" : "false");
+          fprintf(stderr, "Input (len=%zu): ", proto.size());
+          PrintBinary(proto);
+          fprintf(stderr, "\n");
+          if (expected_output) {
+            fprintf(stderr, "Expected output: %s\n", expected_output->c_str());
+          } else {
+            fprintf(stderr, "Expected to FAIL\n");
+          }
+          fprintf(stderr, "Calling start()\n");
+        }
+
         bool ok =
-            input->Start(proto.len(), &sub) &&
-            parse(input, sub, proto.buf(), 0, i, &ofs, &status) &&
-            parse(input, sub, proto.buf(), i, j, &ofs, &status) &&
-            parse(input, sub, proto.buf(), j, proto.len(), &ofs, &status) &&
-            ofs == proto.len() &&
-            input->End();
+            input->Start(proto.size(), &sub) &&
+            parse(input, sub, proto.c_str(), 0, i, &ofs, &status) &&
+            parse(input, sub, proto.c_str(), i, j, &ofs, &status) &&
+            parse(input, sub, proto.c_str(), j, proto.size(), &ofs, &status) &&
+            ofs == proto.size();
+
+        if (ok) {
+          if (filter_hash) {
+            fprintf(stderr, "calling end()\n");
+          }
+          ok = input->End();
+        }
+
         if (expected_output) {
-          if (!output.eql(*expected_output)) {
+          if (output != *expected_output) {
             fprintf(stderr, "Text mismatch: '%s' vs '%s'\n",
-                    output.buf(), expected_output->buf());
+                    output.c_str(), expected_output->c_str());
           }
           if (!ok) {
             fprintf(stderr, "Failed: %s\n", status.error_message());
           }
           ASSERT(ok);
-          ASSERT(output.eql(*expected_output));
+          ASSERT(output == *expected_output);
         } else {
           if (ok) {
             fprintf(stderr, "Didn't expect ok result, but got output: '%s'\n",
-                    output.buf());
+                    output.c_str());
           }
           ASSERT(!ok);
         }
@@ -471,15 +607,15 @@ void run_decoder(const buffer& proto, const buffer* expected_output) {
   testhash = 0;
 }
 
-const static buffer thirty_byte_nop = buffer(cat(
-    tag(NOP_FIELD, UPB_WIRE_TYPE_DELIMITED), delim(buffer(30)) ));
+const static string thirty_byte_nop = cat(
+    tag(NOP_FIELD, UPB_WIRE_TYPE_DELIMITED), delim(string(30, 'X')) );
 
-void assert_successful_parse(const buffer& proto,
+void assert_successful_parse(const string& proto,
                              const char *expected_fmt, ...) {
-  buffer expected_text;
+  string expected_text;
   va_list args;
   va_start(args, expected_fmt);
-  expected_text.vappendf(expected_fmt, args);
+  vappendf(&expected_text, expected_fmt, args);
   va_end(args);
   // To test both middle-of-buffer and end-of-buffer code paths,
   // repeat once with no-op padding data at the end of buffer.
@@ -487,11 +623,11 @@ void assert_successful_parse(const buffer& proto,
   run_decoder(cat( proto, thirty_byte_nop ), &expected_text);
 }
 
-void assert_does_not_parse_at_eof(const buffer& proto) {
+void assert_does_not_parse_at_eof(const string& proto) {
   run_decoder(proto, NULL);
 }
 
-void assert_does_not_parse(const buffer& proto) {
+void assert_does_not_parse(const string& proto) {
   // Test that the error is caught both at end-of-buffer and middle-of-buffer.
   assert_does_not_parse_at_eof(proto);
   assert_does_not_parse_at_eof(cat( proto, thirty_byte_nop ));
@@ -502,19 +638,19 @@ void assert_does_not_parse(const buffer& proto) {
 
 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
-    buffer("abcdefg"),  // UPB_WIRE_TYPE_64BIT
-    buffer("\x80"),     // UPB_WIRE_TYPE_DELIMITED (partial length)
-    buffer(),           // UPB_WIRE_TYPE_START_GROUP (no value required)
-    buffer(),           // UPB_WIRE_TYPE_END_GROUP (no value required)
-    buffer("abc")       // UPB_WIRE_TYPE_32BIT
+  static const string incompletes[6] = {
+    string("\x80"),     // UPB_WIRE_TYPE_VARINT
+    string("abcdefg"),  // UPB_WIRE_TYPE_64BIT
+    string("\x80"),     // UPB_WIRE_TYPE_DELIMITED (partial length)
+    string(),           // UPB_WIRE_TYPE_START_GROUP (no value required)
+    string(),           // UPB_WIRE_TYPE_END_GROUP (no value required)
+    string("abc")       // UPB_WIRE_TYPE_32BIT
   };
 
   uint32_t fieldnum = type;
   uint32_t rep_fieldnum = rep_fn(type);
   int wire_type = upb_decoder_types[type].native_wire_type;
-  const buffer& incomplete = incompletes[wire_type];
+  const string& incomplete = incompletes[wire_type];
 
   // EOF before a known non-repeated value.
   assert_does_not_parse_at_eof(tag(fieldnum, wire_type));
@@ -552,19 +688,19 @@ void test_premature_eof_for_type(upb_descriptortype_t type) {
 
     if (type == UPB_DESCRIPTOR_TYPE_MESSAGE) {
       // Submessage ends in the middle of a value.
-      buffer incomplete_submsg =
+      string incomplete_submsg =
           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),
-               varint(incomplete_submsg.len()),
+               varint(incomplete_submsg.size()),
                incomplete_submsg ));
     }
   } else {
     // Packed region ends in the middle of a value.
     assert_does_not_parse(
         cat( tag(rep_fieldnum, UPB_WIRE_TYPE_DELIMITED),
-             varint(incomplete.len()),
+             varint(incomplete.size()),
              incomplete ));
 
     // EOF in the middle of packed region.
@@ -576,7 +712,7 @@ void test_premature_eof_for_type(upb_descriptortype_t type) {
 // "33" and "66" are just two random values that all numeric types can
 // represent.
 void test_valid_data_for_type(upb_descriptortype_t type,
-                              const buffer& enc33, const buffer& enc66) {
+                              const string& enc33, const string& enc66) {
   uint32_t fieldnum = type;
   uint32_t rep_fieldnum = rep_fn(type);
   int wire_type = upb_decoder_types[type].native_wire_type;
@@ -614,7 +750,7 @@ void test_valid_data_for_type(upb_descriptortype_t type,
 }
 
 void test_valid_data_for_signed_type(upb_descriptortype_t type,
-                                     const buffer& enc33, const buffer& enc66) {
+                                     const string& enc33, const string& enc66) {
   uint32_t fieldnum = type;
   uint32_t rep_fieldnum = rep_fn(type);
   int wire_type = upb_decoder_types[type].native_wire_type;
@@ -672,15 +808,14 @@ void test_invalid() {
   test_premature_eof_for_type(UPB_DESCRIPTOR_TYPE_SINT64);
 
   // EOF inside a tag's varint.
-  assert_does_not_parse_at_eof( buffer("\x80") );
+  assert_does_not_parse_at_eof( string("\x80") );
 
   // EOF inside a known group.
   // TODO(haberman): add group to decoder test schema.
   //assert_does_not_parse_at_eof( tag(4, UPB_WIRE_TYPE_START_GROUP) );
 
   // EOF inside an unknown group.
-  // TODO(haberman): unknown groups not supported yet.
-  //assert_does_not_parse_at_eof( tag(UNKNOWN_FIELD, UPB_WIRE_TYPE_START_GROUP) );
+  assert_does_not_parse_at_eof( tag(UNKNOWN_FIELD, UPB_WIRE_TYPE_START_GROUP) );
 
   // End group that we are not currently in.
   assert_does_not_parse( tag(4, UPB_WIRE_TYPE_END_GROUP) );
@@ -701,7 +836,7 @@ void test_invalid() {
             tag(UPB_DESCRIPTOR_TYPE_GROUP, UPB_WIRE_TYPE_END_GROUP)));
 
   // Test exceeding the resource limit of stack depth.
-  buffer buf;
+  string buf;
   for (int i = 0; i <= UPB_DECODER_MAX_NESTING; i++) {
     buf.assign(submsg(UPB_DESCRIPTOR_TYPE_MESSAGE, buf));
   }
@@ -710,11 +845,16 @@ void test_invalid() {
 
 void test_valid() {
   // Empty protobuf.
-  assert_successful_parse(buffer(""), "<\n>\n");
+  assert_successful_parse(string(""), "<\n>\n");
 
   // Empty protobuf where we never call PutString between
   // StartString/EndString.
-  {
+
+  // Randomly generated hash for this test, hope it doesn't conflict with others
+  // by chance.
+  const uint32_t emptyhash = 0x5709be8e;
+  if (!filter_hash || filter_hash == testhash) {
+    testhash = emptyhash;
     upb::Status status;
     upb::pb::Decoder decoder(global_method, &status);
     upb::Sink sink(global_handlers, &closures[0]);
@@ -723,7 +863,7 @@ void test_valid() {
     bool ok = upb::BufferSource::PutBuffer("", 0, decoder.input());
     ASSERT(ok);
     ASSERT(status.ok());
-    ASSERT(output.eql(buffer("<\n>\n")));
+    ASSERT(output == string("<\n>\n"));
   }
 
   test_valid_data_for_signed_type(UPB_DESCRIPTOR_TYPE_DOUBLE,
@@ -770,7 +910,7 @@ void test_valid() {
       cat( tag(12345, UPB_WIRE_TYPE_64BIT), uint64(2345678) ),
       "<\n>\n");
   assert_successful_parse(
-      submsg(12345, buffer("                ")),
+      submsg(12345, string("                ")),
       "<\n>\n");
 
   assert_successful_parse(
@@ -818,7 +958,7 @@ void test_valid() {
   // Submessage tests.
   uint32_t msg_fn = UPB_DESCRIPTOR_TYPE_MESSAGE;
   assert_successful_parse(
-      submsg(msg_fn, submsg(msg_fn, submsg(msg_fn, buffer()))),
+      submsg(msg_fn, submsg(msg_fn, submsg(msg_fn, string()))),
       LINE("<")
       LINE("%u:{")
       LINE("  <")
@@ -836,7 +976,7 @@ void test_valid() {
 
   uint32_t repm_fn = rep_fn(UPB_DESCRIPTOR_TYPE_MESSAGE);
   assert_successful_parse(
-      submsg(repm_fn, submsg(repm_fn, buffer())),
+      submsg(repm_fn, submsg(repm_fn, string())),
       LINE("<")
       LINE("%u:[")
       LINE("  %u:{")
@@ -852,16 +992,59 @@ void test_valid() {
       LINE("]")
       LINE(">"), repm_fn, repm_fn, repm_fn, repm_fn);
 
+  // Test unknown group.
+  uint32_t unknown_group_fn = 12321;
+  assert_successful_parse(
+      cat( tag(unknown_group_fn, UPB_WIRE_TYPE_START_GROUP),
+           tag(unknown_group_fn, UPB_WIRE_TYPE_END_GROUP) ),
+      LINE("<")
+      LINE(">")
+  );
+
+  // Test some unknown fields inside an unknown group.
+  const string unknown_group_with_data =
+      cat(
+          tag(unknown_group_fn, UPB_WIRE_TYPE_START_GROUP),
+          tag(12345, UPB_WIRE_TYPE_VARINT), varint(2345678),
+          tag(123456789, UPB_WIRE_TYPE_32BIT), uint32(2345678),
+          tag(123477, UPB_WIRE_TYPE_64BIT), uint64(2345678),
+          tag(123, UPB_WIRE_TYPE_DELIMITED), varint(0),
+          tag(unknown_group_fn, UPB_WIRE_TYPE_END_GROUP)
+         );
+
+  // Nested unknown group with data.
+  assert_successful_parse(
+      cat(
+           tag(unknown_group_fn, UPB_WIRE_TYPE_START_GROUP),
+           unknown_group_with_data,
+           tag(unknown_group_fn, UPB_WIRE_TYPE_END_GROUP),
+           tag(UPB_DESCRIPTOR_TYPE_INT32, UPB_WIRE_TYPE_VARINT), varint(1)
+         ),
+      LINE("<")
+      LINE("%u:1")
+      LINE(">"),
+      UPB_DESCRIPTOR_TYPE_INT32
+  );
+
+  assert_successful_parse(
+      cat( tag(unknown_group_fn, UPB_WIRE_TYPE_START_GROUP),
+           tag(unknown_group_fn + 1, UPB_WIRE_TYPE_START_GROUP),
+           tag(unknown_group_fn + 1, UPB_WIRE_TYPE_END_GROUP),
+           tag(unknown_group_fn, UPB_WIRE_TYPE_END_GROUP) ),
+      LINE("<")
+      LINE(">")
+  );
+
   // Staying within the stack limit should work properly.
-  buffer buf;
-  buffer textbuf;
+  string buf;
+  string textbuf;
   int total = UPB_DECODER_MAX_NESTING - 1;
   for (int i = 0; i < total; i++) {
     buf.assign(submsg(UPB_DESCRIPTOR_TYPE_MESSAGE, buf));
     indentbuf(&textbuf, i);
     textbuf.append("<\n");
     indentbuf(&textbuf, i);
-    textbuf.appendf("%u:{\n", UPB_DESCRIPTOR_TYPE_MESSAGE);
+    appendf(&textbuf, "%u:{\n", UPB_DESCRIPTOR_TYPE_MESSAGE);
   }
   indentbuf(&textbuf, total);
   textbuf.append("<\n");
@@ -873,7 +1056,7 @@ void test_valid() {
     indentbuf(&textbuf, total - i - 1);
     textbuf.append(">\n");
   }
-  assert_successful_parse(buf, "%s", textbuf.buf());
+  assert_successful_parse(buf, "%s", textbuf.c_str());
 }
 
 void run_tests() {
@@ -885,14 +1068,17 @@ upb::reffed_ptr<const upb::pb::DecoderMethod> NewMethod(
     const upb::Handlers* dest_handlers, bool allow_jit) {
   upb::pb::CodeCache cache;
   cache.set_allow_jit(allow_jit);
-  return cache.GetDecoderMethodForDestHandlers(dest_handlers);
+  return cache.GetDecoderMethod(upb::pb::DecoderMethodOptions(dest_handlers));
 }
 
 void test_emptyhandlers(bool allowjit) {
   // Create an empty handlers to make sure that the decoder can handle empty
   // messages.
-  upb::reffed_ptr<upb::Handlers> h(
-      upb::Handlers::New(UPB_TEST_DECODER_EMPTYMESSAGE));
+  upb::reffed_ptr<upb::MessageDef> md = upb::MessageDef::New();
+  ASSERT(md->set_full_name("Empty", NULL));
+  ASSERT(md->Freeze(NULL));
+
+  upb::reffed_ptr<upb::Handlers> h(upb::Handlers::New(md.get()));
   bool ok = h->Freeze(NULL);
   ASSERT(ok);
   NewMethod(h.get(), allowjit);

tests/pb/test_decoder_schema.proto

@@ -1,64 +0,0 @@
-//
-// upb - a minimalist implementation of protocol buffers.
-//
-// Copyright (c) 2012 Google Inc.  See LICENSE for details.
-// Author: Josh Haberman <jhaberman@gmail.com>
-//
-// Schema used in test_decoder.cc.  It contains two fields (one optional
-// and one repeated) for each type.
-
-package upb.test_decoder;
-
-message M {
-  optional M m = 1;
-}
-
-enum E {
-  FOO = 1;
-}
-
-message EmptyMessage {}
-
-message DecoderTest {
-  optional double   f_double    = 1;
-  optional float    f_float     = 2;
-  optional int64    f_int64     = 3;
-  optional uint64   f_uint64    = 4;
-  optional int32    f_int32     = 5;
-  optional fixed64  f_fixed64   = 6;
-  optional fixed32  f_fixed32   = 7;
-  optional bool     f_bool      = 8;
-  optional string   f_string    = 9;
-  optional bytes    f_bytes     = 12;
-  optional uint32   f_uint32    = 13;
-  optional sfixed32 f_sfixed32  = 15;
-  optional sfixed64 f_sfixed64  = 16;
-  optional sint32   f_sint32    = 17;
-  optional sint64   f_sint64    = 18;
-
-  optional DecoderTest f_message  = 11;
-  optional E f_enum     = 14;
-
-
-  repeated double   r_double    = 536869912;
-  repeated float    r_float     = 536869913;
-  repeated int64    r_int64     = 536869914;
-  repeated uint64   r_uint64    = 536869915;
-  repeated int32    r_int32     = 536869916;
-  repeated fixed64  r_fixed64   = 536869917;
-  repeated fixed32  r_fixed32   = 536869918;
-  repeated bool     r_bool      = 536869919;
-  repeated string   r_string    = 536869920;
-  repeated bytes    r_bytes     = 536869923;
-  repeated uint32   r_uint32    = 536869924;
-  repeated sfixed32 r_sfixed32  = 536869926;
-  repeated sfixed64 r_sfixed64  = 536869927;
-  repeated sint32   r_sint32    = 536869928;
-  repeated sint64   r_sint64    = 536869929;
-
-  repeated DecoderTest r_message  = 536869922;
-  repeated E r_enum     = 536869925;
-
-  // To allow arbitrary padding.
-  optional string nop_field = 40;
-}

tests/test_cpp.cc

@@ -12,10 +12,7 @@
 
 #include <iostream>
 #include <set>
-
-#ifdef UPB_CXX11
 #include <type_traits>
-#endif
 
 #include "upb/def.h"
 #include "upb/descriptor/reader.h"

tests/test_vs_proto2.cc

@@ -45,6 +45,15 @@ void compare_metadata(const google::protobuf::Descriptor* d,
   }
 }
 
+void print_diff(const google::protobuf::Message& msg1,
+                const google::protobuf::Message& msg2) {
+  std::string text_str1;
+  std::string text_str2;
+  google::protobuf::TextFormat::PrintToString(msg1, &text_str1);
+  google::protobuf::TextFormat::PrintToString(msg2, &text_str2);
+  fprintf(stderr, "str1: %s, str2: %s\n", text_str1.c_str(), text_str2.c_str());
+}
+
 void parse_and_compare(google::protobuf::Message *msg1,
                        google::protobuf::Message *msg2,
                        const upb::Handlers *protomsg_handlers,
@@ -55,7 +64,7 @@ void parse_and_compare(google::protobuf::Message *msg1,
   upb::pb::CodeCache cache;
   ASSERT(cache.set_allow_jit(allow_jit));
   upb::reffed_ptr<const upb::pb::DecoderMethod> decoder_method(
-      cache.GetDecoderMethodForDestHandlers(protomsg_handlers));
+      cache.GetDecoderMethod(upb::pb::DecoderMethodOptions(protomsg_handlers)));
 
   upb::Status status;
   upb::pb::Decoder decoder(decoder_method.get(), &status);
@@ -67,6 +76,7 @@ void parse_and_compare(google::protobuf::Message *msg1,
   bool ok = upb::BufferSource::PutBuffer(str, len, decoder.input());
   if (!ok) {
     fprintf(stderr, "error parsing: %s\n", status.error_message());
+    print_diff(*msg1, *msg2);
   }
   ASSERT(ok);
   ASSERT(status.ok());
@@ -79,14 +89,8 @@ void parse_and_compare(google::protobuf::Message *msg1,
   std::string str2;
   msg1->SerializeToString(&str1);
   msg2->SerializeToString(&str2);
-
-  std::string text_str1;
-  std::string text_str2;
-  google::protobuf::TextFormat::PrintToString(*msg1, &text_str1);
-  google::protobuf::TextFormat::PrintToString(*msg2, &text_str2);
   if (str1 != str2) {
-    fprintf(stderr, "str1: %s, str2: %s\n",
-            text_str1.c_str(), text_str2.c_str());
+    print_diff(*msg1, *msg2);
   }
   ASSERT(str1 == str2);
   ASSERT(std::string(str, len) == str2);

tools/dump_cinit.lua

@@ -359,15 +359,16 @@ local function dump_defs_c(symtab, basename, append)
     else
       intfmt = "0"
     end
-    -- UPB_FIELDDEF_INIT(label, type, intfmt, tagdelim, is_extension, name,
-    --                   num, msgdef, subdef, selector_base, index,
+    -- UPB_FIELDDEF_INIT(label, type, intfmt, tagdelim, is_extension, lazy,
+    --                   name, num, msgdef, subdef, selector_base, index,
     --                   default_value)
-    append('  UPB_FIELDDEF_INIT(%s, %s, %s, %s, %s, "%s", %d, %s, %s, %d, ' ..
-           '%d, {0},' .. -- TODO: support default value
+    append('  UPB_FIELDDEF_INIT(%s, %s, %s, %s, %s, %s, "%s", %d, %s, %s, ' ..
+           '%d, %d, {0},' .. -- TODO: support default value
            '&reftables[%d], &reftables[%d]),\n',
            const(f, "label"), const(f, "type"), intfmt,
-           boolstr(f:istagdelim()), boolstr(f:is_extension()), f:name(),
-           f:number(), linktab:addr(f:containing_type()), subdef,
+           boolstr(f:istagdelim()), boolstr(f:is_extension()),
+           boolstr(f:lazy()), f:name(), f:number(),
+           linktab:addr(f:containing_type()), subdef,
            f:_selector_base(), f:index(),
            reftable, reftable + 1
            )

upb/bindings/googlepb/bridge.cc

@@ -109,10 +109,6 @@ const MessageDef* DefBuilder::GetMaybeUnfrozenMessageDef(
   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
     md->AddField(NewFieldDef(proto2_f, m), &status);
   }
   ASSERT_STATUS(&status);
@@ -135,6 +131,9 @@ reffed_ptr<FieldDef> DefBuilder::NewFieldDef(const goog::FieldDescriptor* f,
   Status status;
   upb_f->set_number(f->number(), &status);
   upb_f->set_label(FieldDef::ConvertLabel(f->label()));
+#ifdef UPB_GOOGLE3
+  upb_f->set_lazy(f->options().lazy());
+#endif
 
   if (f->is_extension()) {
     upb_f->set_name(f->full_name(), &status);

upb/bindings/googlepb/proto2.cc

@@ -42,9 +42,14 @@ namespace google_opensource { class GMR_Handlers; }
 
 #ifdef UPB_GOOGLE3
 
+// TODO(haberman): Add public functionality to ExtensionSet for populating
+// LazyFields.
+#define private public
+#include "net/proto2/public/extension_set.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"
@@ -66,9 +71,12 @@ namespace me = ::upb::google_google3;
 #include "google/protobuf/generated_message_reflection.h"
 #undef private
 
+#define private public
+#include "google/protobuf/extension_set.h"
+#undef private
+
 #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;
@@ -176,9 +184,14 @@ case goog::FieldDescriptor::cpptype:                                           \
       }
       case goog::FieldDescriptor::CPPTYPE_MESSAGE:
 #ifdef UPB_GOOGLE3
-        if (proto2_f->options().lazy()) {
-          assert(false);
-          return false;  // Not yet implemented.
+        if (proto2_f->options().lazy() &&
+            // proto2 lets you set lazy=true on a repeated field, but doesn't
+            // actually support lazy repeated messages, so just ignore
+            // lazy=true for repeated messages.
+            !proto2_f->is_repeated()) {
+          // Supports lazy fields and lazy extensions.
+          SetLazyFieldHandlers(proto2_f, m, r, upb_f, h);
+          return true;
         }
 #endif
         if (proto2_f->is_extension()) {
@@ -267,12 +280,14 @@ case goog::FieldDescriptor::cpptype:                                           \
         const goog::FieldDescriptor* proto2_f,
         const goog::internal::GeneratedMessageReflection* r)
         : offset_(r->extensions_offset_),
-          number_(proto2_f->number()),
-          type_(proto2_f->type()) {
+          field_descriptor_(proto2_f) {
     }
 
-    int number() const { return number_; }
-    goog::internal::FieldType type() const { return type_; }
+    int number() const { return field_descriptor_->number(); }
+    goog::internal::FieldType type() const { return field_descriptor_->type(); }
+    const goog::FieldDescriptor* field_descriptor() const {
+      return field_descriptor_;
+    }
 
     goog::internal::ExtensionSet* GetExtensionSet(goog::Message* m) const {
       return GetPointer<goog::internal::ExtensionSet>(m, offset_);
@@ -280,8 +295,10 @@ case goog::FieldDescriptor::cpptype:                                           \
 
    private:
     const size_t offset_;
-    int number_;
-    goog::internal::FieldType type_;
+    // We know it will outlive because we require that the input message used to
+    // build these handlers outlives us, and the descriptor will outlive the
+    // message.
+    const goog::FieldDescriptor* field_descriptor_;
   };
 
   // StartSequence /////////////////////////////////////////////////////////////
@@ -693,10 +710,27 @@ case goog::FieldDescriptor::cpptype:                                           \
 #ifdef UPB_GOOGLE3
   // Handlers for types/features only included in internal proto2 release:
   // Cord, StringPiece, LazyField, and MessageSet.
-  // TODO(haberman): LazyField, MessageSet.
+  // TODO(haberman): MessageSet.
 
   // Cord //////////////////////////////////////////////////////////////////////
 
+  static void AppendBufToCord(const char* buf, size_t n,
+                              const upb::BufferHandle* handle, Cord* c) {
+    const Cord* source_cord = handle->GetAttachedObject<Cord>();
+    if (source_cord) {
+      // This TODO is copied from CordReader::CopyToCord():
+      // "We could speed this up by using CordReader internals."
+      Cord piece(*source_cord);
+      piece.RemovePrefix(handle->object_offset() + (buf - handle->buffer()));
+      assert(piece.size() >= n);
+      piece.RemoveSuffix(piece.size() - n);
+
+      c->Append(piece);
+    } else {
+      c->Append(StringPiece(buf, n));
+    }
+  }
+
   static void SetCordHandlers(
       const proto2::FieldDescriptor* proto2_f,
       const proto2::internal::GeneratedMessageReflection* r,
@@ -723,19 +757,7 @@ case goog::FieldDescriptor::cpptype:                                           \
 
   static void OnCordBuf(Cord* c, const char* buf, size_t n,
                         const upb::BufferHandle* handle) {
-    const Cord* source_cord = handle->GetAttachedObject<Cord>();
-    if (source_cord) {
-      // This TODO is copied from CordReader::CopyToCord():
-      // "We could speed this up by using CordReader internals."
-      Cord piece(*source_cord);
-      piece.RemovePrefix(handle->object_offset() + (buf - handle->buffer()));
-      assert(piece.size() >= n);
-      piece.RemoveSuffix(piece.size() - n);
-
-      c->Append(piece);
-    } else {
-      c->Append(StringPiece(buf, n));
-    }
+    AppendBufToCord(buf, n, handle, c);
   }
 
   static Cord* StartRepeatedCord(proto2::RepeatedField<Cord>* r,
@@ -795,6 +817,164 @@ case goog::FieldDescriptor::cpptype:                                           \
     return field;
   }
 
+  // LazyField /////////////////////////////////////////////////////////////////
+
+  // For lazy fields we set both lazy and eager handlers.  The user can
+  // configure the data source to call either, though lazy handlers may only be
+  // used when the source data is binary protobuf.
+  static void SetLazyFieldHandlers(
+      const proto2::FieldDescriptor* proto2_f,
+      const proto2::Message& m,
+      const proto2::internal::GeneratedMessageReflection* r,
+      const upb::FieldDef* f, upb::Handlers* h) {
+    assert(!proto2_f->is_repeated());
+    const goog::Message* field_prototype = GetFieldPrototype(m, proto2_f);
+    CHKRET(h->SetStringHandler(f, UpbMakeHandler(OnLazyFieldBuf)));
+    if (proto2_f->is_extension()) {
+      CHKRET(h->SetStartStringHandler(
+          f, UpbBind(StartLazyExtension, new ExtensionFieldData(proto2_f, r))));
+      CHKRET(h->SetStartSubMessageHandler(
+          f, UpbBind(StartSubMessageExtension,
+                     new SubMessageExtensionHandlerData(proto2_f, r,
+                                                        field_prototype))));
+    } else {
+      CHKRET(h->SetStartStringHandler(
+          f, UpbBind(StartLazyField, new FieldOffset(proto2_f, r))));
+      CHKRET(h->SetStartSubMessageHandler(
+          f, UpbBind(StartLazyFieldEager,
+                     new SubMessageHandlerData(proto2_f, r, field_prototype))));
+    }
+  }
+
+  static proto2::internal::LazyField* StartLazyField(proto2::Message* m,
+                                                     const FieldOffset* offset,
+                                                     size_t size_hint) {
+    UPB_UNUSED(size_hint);
+    offset->SetHasbit(m);
+    proto2::internal::LazyField* field =
+        offset->GetFieldPointer<proto2::internal::LazyField>(m);
+    field->Clear();
+    return field;
+  }
+
+  // For when the field has a lazy representation but we parse it eagerly anyway
+  // (either because we want to or because we're parsing from a format other
+  // than binary protobuf).
+  static proto2::Message* StartLazyFieldEager(
+      proto2::Message* m, const SubMessageHandlerData* data) {
+    data->SetHasbit(m);
+    proto2::internal::LazyField* field =
+        data->GetFieldPointer<proto2::internal::LazyField>(m);
+    return field->MutableByPrototype(*data->prototype());
+  }
+
+  class LazyMessageExtensionImpl
+      : public proto2::internal::ExtensionSet::LazyMessageExtension {
+   public:
+    LazyMessageExtensionImpl() {}
+    virtual ~LazyMessageExtensionImpl() {}
+
+    virtual LazyMessageExtension* New() const {
+      return new LazyMessageExtensionImpl();
+    }
+
+    virtual const proto2::MessageLite& GetMessage(
+        const proto2::MessageLite& prototype) const {
+      return lazy_field_.GetByPrototype(
+          static_cast<const proto2::Message&>(prototype));
+    }
+
+    virtual proto2::MessageLite* MutableMessage(
+        const proto2::MessageLite& prototype) {
+      return lazy_field_.MutableByPrototype(
+          static_cast<const proto2::Message&>(prototype));
+    }
+
+    virtual void SetAllocatedMessage(proto2::MessageLite* message) {
+      return lazy_field_.SetAllocated(static_cast<proto2::Message*>(message));
+    }
+
+    virtual proto2::MessageLite* ReleaseMessage(
+        const proto2::MessageLite& prototype) {
+      return lazy_field_.ReleaseByPrototype(
+          static_cast<const proto2::Message&>(prototype));
+    }
+
+    virtual bool IsInitialized() const { return true; }
+
+    virtual int ByteSize() const { return lazy_field_.MessageByteSize(); }
+
+    int SpaceUsed() const {
+      return sizeof(*this) + lazy_field_.SpaceUsedExcludingSelf();
+    }
+
+    virtual void MergeFrom(const LazyMessageExtension& other) {
+      MergeFrom(*static_cast<const LazyMessageExtensionImpl*>(&other));
+    }
+
+    virtual void MergeFrom(const LazyMessageExtensionImpl& other) {
+      lazy_field_.MergeFrom(other.lazy_field_);
+    }
+
+    virtual void Clear() { lazy_field_.Clear(); }
+
+    virtual bool ReadMessage(const proto2::MessageLite& prototype,
+                             proto2::io::CodedInputStream* input) {
+      return lazy_field_.Read(input);
+    }
+
+    virtual void WriteMessage(int number,
+                              proto2::io::CodedOutputStream* output) const {
+      lazy_field_.Write(number, output);
+    }
+
+    virtual uint8* WriteMessageToArray(int number, uint8* target) const {
+      return lazy_field_.WriteToArray(number, target);
+    }
+
+    proto2::internal::LazyField& lazy_field() { return lazy_field_; }
+
+   private:
+    proto2::internal::LazyField lazy_field_;
+    DISALLOW_COPY_AND_ASSIGN(LazyMessageExtensionImpl);
+  };
+
+  static proto2::internal::LazyField* StartLazyExtension(
+      proto2::Message* m, const ExtensionFieldData* data, size_t size_hint) {
+    proto2::internal::ExtensionSet* set = data->GetExtensionSet(m);
+
+    // We have to break encapsulation here since no public accessors expose the
+    // LazyField.
+    //
+    // TODO(haberman): add a function to ExtensionSet that allows us to set the
+    // lazy field directly.
+    proto2::internal::ExtensionSet::Extension* item;
+    LazyMessageExtensionImpl* lazy_extension;
+    if (set->MaybeNewExtension(data->number(), data->field_descriptor(),
+                               &item)) {
+      lazy_extension = new LazyMessageExtensionImpl();
+      item->type = UPB_DESCRIPTOR_TYPE_MESSAGE;
+      item->is_repeated = false;
+      item->is_lazy = true;
+      item->lazymessage_value = lazy_extension;
+    } else {
+      lazy_extension =
+          CheckDownCast<LazyMessageExtensionImpl*>(item->lazymessage_value);
+    }
+
+    item->is_cleared = false;
+
+    return &lazy_extension->lazy_field();
+  }
+
+  static void OnLazyFieldBuf(proto2::internal::LazyField* field,
+                             const char* buf, size_t len,
+                             const upb::BufferHandle* handle) {
+    Cord encoded(field->GetEncoded());
+    AppendBufToCord(buf, len, handle, &encoded);
+    field->SetEncoded(encoded);
+  }
+
 #endif  // UPB_GOOGLE3
 };
 

upb/bindings/lua/upb.c

@@ -326,6 +326,10 @@ static void lupb_fielddef_dosetlabel(lua_State *L, upb_fielddef *f, int narg) {
   upb_fielddef_setlabel(f, label);
 }
 
+static void lupb_fielddef_dosetlazy(lua_State *L, upb_fielddef *f, int narg) {
+  upb_fielddef_setlazy(f, chkbool(L, narg, "lazy"));
+}
+
 static void lupb_fielddef_dosetname(lua_State *L, upb_fielddef *f, int narg) {
   CHK(upb_fielddef_setname(f, chkname(L, narg), &status));
 }
@@ -402,6 +406,12 @@ static int lupb_fielddef_setlabel(lua_State *L) {
   return 0;
 }
 
+static int lupb_fielddef_setlazy(lua_State *L) {
+  upb_fielddef *f = lupb_fielddef_checkmutable(L, 1);
+  lupb_fielddef_dosetlazy(L, f, 2);
+  return 0;
+}
+
 static int lupb_fielddef_setname(lua_State *L) {
   upb_fielddef *f = lupb_fielddef_checkmutable(L, 1);
   lupb_fielddef_dosetname(L, f, 2);
@@ -466,6 +476,7 @@ static int lupb_fielddef_new(lua_State *L) {
     else if (streql(key, "number")) lupb_fielddef_dosetnumber(L, f, v);
     else if (streql(key, "type")) lupb_fielddef_dosettype(L, f, v);
     else if (streql(key, "label")) lupb_fielddef_dosetlabel(L, f, v);
+    else if (streql(key, "lazy")) lupb_fielddef_dosetlazy(L, f, v);
     else if (streql(key, "is_extension"))
       lupb_fielddef_dosetisextension(L, f, v);
     else if (streql(key, "containing_type_name"))
@@ -541,6 +552,12 @@ static int lupb_fielddef_label(lua_State *L) {
   return 1;
 }
 
+static int lupb_fielddef_lazy(lua_State *L) {
+  const upb_fielddef *f = lupb_fielddef_check(L, 1);
+  lua_pushboolean(L, upb_fielddef_lazy(f));
+  return 1;
+}
+
 static int lupb_fielddef_name(lua_State *L) {
   const upb_fielddef *f = lupb_fielddef_check(L, 1);
   lua_pushstring(L, upb_fielddef_name(f));
@@ -653,6 +670,7 @@ static const struct luaL_Reg lupb_fielddef_m[] = {
   {"is_extension", lupb_fielddef_isextension},
   {"istagdelim", lupb_fielddef_istagdelim},
   {"label", lupb_fielddef_label},
+  {"lazy", lupb_fielddef_lazy},
   {"name", lupb_fielddef_name},
   {"number", lupb_fielddef_number},
   {"subdef", lupb_fielddef_subdef},
@@ -663,6 +681,7 @@ static const struct luaL_Reg lupb_fielddef_m[] = {
   {"set_default", lupb_fielddef_setdefault},
   {"set_is_extension", lupb_fielddef_setisextension},
   {"set_label", lupb_fielddef_setlabel},
+  {"set_lazy", lupb_fielddef_setlazy},
   {"set_name", lupb_fielddef_setname},
   {"set_number", lupb_fielddef_setnumber},
   {"set_subdef", lupb_fielddef_setsubdef},

upb/def.c

@@ -145,6 +145,12 @@ static bool upb_validate_field(upb_fielddef *f, upb_status *s) {
     upb_status_seterrmsg(s, "fielddef type was not initialized");
     return false;
   }
+  if (upb_fielddef_lazy(f) &&
+      upb_fielddef_descriptortype(f) != UPB_DESCRIPTOR_TYPE_MESSAGE) {
+    upb_status_seterrmsg(s,
+                         "only length-delimited submessage fields may be lazy");
+    return false;
+  }
   if (upb_fielddef_hassubdef(f)) {
     if (f->subdef_is_symbolic) {
       upb_status_seterrf(s,
@@ -463,6 +469,7 @@ upb_fielddef *upb_fielddef_new(const void *owner) {
   f->type_is_set_ = false;
   f->tagdelim = false;
   f->is_extension_ = false;
+  f->lazy_ = 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
@@ -565,6 +572,10 @@ bool upb_fielddef_isextension(const upb_fielddef *f) {
   return f->is_extension_;
 }
 
+bool upb_fielddef_lazy(const upb_fielddef *f) {
+  return f->lazy_;
+}
+
 const char *upb_fielddef_name(const upb_fielddef *f) {
   return upb_def_fullname(UPB_UPCAST(f));
 }
@@ -824,10 +835,14 @@ upb_descriptortype_t upb_fielddef_descriptortype(const upb_fielddef *f) {
   return 0;
 }
 
-bool upb_fielddef_setisextension(upb_fielddef *f, bool is_extension) {
+void upb_fielddef_setisextension(upb_fielddef *f, bool is_extension) {
   assert(!upb_fielddef_isfrozen(f));
   f->is_extension_ = is_extension;
-  return true;
+}
+
+void upb_fielddef_setlazy(upb_fielddef *f, bool lazy) {
+  assert(!upb_fielddef_isfrozen(f));
+  f->lazy_ = lazy;
 }
 
 void upb_fielddef_setlabel(upb_fielddef *f, upb_label_t label) {
@@ -836,17 +851,16 @@ void upb_fielddef_setlabel(upb_fielddef *f, upb_label_t label) {
   f->label_ = label;
 }
 
-bool upb_fielddef_setintfmt(upb_fielddef *f, upb_intfmt_t fmt) {
+void upb_fielddef_setintfmt(upb_fielddef *f, upb_intfmt_t fmt) {
   assert(!upb_fielddef_isfrozen(f));
   assert(upb_fielddef_checkintfmt(fmt));
   f->intfmt = fmt;
-  return true;
 }
 
-bool upb_fielddef_settagdelim(upb_fielddef *f, bool tag_delim) {
+void upb_fielddef_settagdelim(upb_fielddef *f, bool tag_delim) {
   assert(!upb_fielddef_isfrozen(f));
   f->tagdelim = tag_delim;
-  return true;
+  f->tagdelim = tag_delim;
 }
 
 static bool checksetdefault(upb_fielddef *f, upb_fieldtype_t type) {

upb/def.h

@@ -274,6 +274,20 @@ class upb::FieldDef /* : public upb::Def */ {
   uint32_t number() const;   // Returns 0 if uninitialized.
   bool is_extension() const;
 
+  // For UPB_TYPE_MESSAGE fields only where is_tag_delimited() == false,
+  // indicates whether this field should have lazy parsing handlers that yield
+  // the unparsed string for the submessage.
+  //
+  // When we freeze, we ensure that this can only be true for length-delimited
+  // message fields.  Prior to freezing this can be true or false with no
+  // restrictions.
+  //
+  // TODO(haberman): I think we want to move this into a FieldOptions container
+  // when we add support for custom options (the FieldOptions struct will
+  // contain both regular FieldOptions like "lazy" *and* custom options).
+  bool lazy() const;
+  void set_lazy(bool lazy);
+
   // An integer that can be used as an index into an array of fields for
   // whatever message this field belongs to.  Guaranteed to be less than
   // f->containing_type()->field_count().  May only be accessed once the def has
@@ -450,6 +464,7 @@ struct upb_fielddef {
   bool default_is_string;
   bool type_is_set_;     // False until type is explicitly set.
   bool is_extension_;
+  bool lazy_;
   upb_intfmt_t intfmt;
   bool tagdelim;
   upb_fieldtype_t type_;
@@ -459,14 +474,14 @@ struct upb_fielddef {
   uint32_t index_;
 };
 
-#define UPB_FIELDDEF_INIT(label, type, intfmt, tagdelim, is_extension, name,   \
-                          num, msgdef, subdef, selector_base, index,           \
+#define UPB_FIELDDEF_INIT(label, type, intfmt, tagdelim, is_extension, lazy,   \
+                          name, num, msgdef, subdef, selector_base, index,     \
                           defaultval, refs, ref2s)                             \
   {                                                                            \
     UPB_DEF_INIT(name, UPB_DEF_FIELD, refs, ref2s), defaultval, {msgdef},      \
         {subdef}, false, false,                                                \
         type == UPB_TYPE_STRING || type == UPB_TYPE_BYTES, true, is_extension, \
-        intfmt, tagdelim, type, label, num, selector_base, index               \
+        lazy, intfmt, tagdelim, type, label, num, selector_base, index         \
   }
 
 // Native C API.
@@ -496,6 +511,7 @@ 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);
 bool upb_fielddef_isextension(const upb_fielddef *f);
+bool upb_fielddef_lazy(const upb_fielddef *f);
 const upb_msgdef *upb_fielddef_containingtype(const upb_fielddef *f);
 upb_msgdef *upb_fielddef_containingtype_mutable(upb_fielddef *f);
 const char *upb_fielddef_containingtypename(upb_fielddef *f);
@@ -528,9 +544,10 @@ bool upb_fielddef_setnumber(upb_fielddef *f, uint32_t number, upb_status *s);
 bool upb_fielddef_setname(upb_fielddef *f, const char *name, upb_status *s);
 bool upb_fielddef_setcontainingtypename(upb_fielddef *f, const char *name,
                                         upb_status *s);
-bool upb_fielddef_setisextension(upb_fielddef *f, bool is_extension);
-bool upb_fielddef_setintfmt(upb_fielddef *f, upb_intfmt_t fmt);
-bool upb_fielddef_settagdelim(upb_fielddef *f, bool tag_delim);
+void upb_fielddef_setisextension(upb_fielddef *f, bool is_extension);
+void upb_fielddef_setlazy(upb_fielddef *f, bool lazy);
+void upb_fielddef_setintfmt(upb_fielddef *f, upb_intfmt_t fmt);
+void upb_fielddef_settagdelim(upb_fielddef *f, bool tag_delim);
 void upb_fielddef_setdefaultint64(upb_fielddef *f, int64_t val);
 void upb_fielddef_setdefaultint32(upb_fielddef *f, int32_t val);
 void upb_fielddef_setdefaultuint64(upb_fielddef *f, uint64_t val);
@@ -1071,6 +1088,12 @@ inline const char* FieldDef::name() const { return upb_fielddef_name(this); }
 inline bool FieldDef::is_extension() const {
   return upb_fielddef_isextension(this);
 }
+inline bool FieldDef::lazy() const {
+  return upb_fielddef_lazy(this);
+}
+inline void FieldDef::set_lazy(bool lazy) {
+  upb_fielddef_setlazy(this, lazy);
+}
 inline const MessageDef* FieldDef::containing_type() const {
   return upb_fielddef_containingtype(this);
 }

upb/descriptor/descriptor.proto

@@ -59,6 +59,11 @@ message FileDescriptorProto {
 
   // Names of files imported by this file.
   repeated string dependency = 3;
+  // Indexes of the public imported files in the dependency list above.
+  repeated int32 public_dependency = 10;
+  // Indexes of the weak imported files in the dependency list.
+  // For Google-internal migration only. Do not use.
+  repeated int32 weak_dependency = 11;
 
   // All top-level definitions in this file.
   repeated DescriptorProto message_type = 4;
@@ -101,13 +106,13 @@ message FieldDescriptorProto {
     // Order is weird for historical reasons.
     TYPE_DOUBLE         = 1;
     TYPE_FLOAT          = 2;
-    TYPE_INT64          = 3;   // Not ZigZag encoded.  Negative numbers
-                               // take 10 bytes.  Use TYPE_SINT64 if negative
-                               // values are likely.
+    // Not ZigZag encoded.  Negative numbers take 10 bytes.  Use TYPE_SINT64 if
+    // negative values are likely.
+    TYPE_INT64          = 3;
     TYPE_UINT64         = 4;
-    TYPE_INT32          = 5;   // Not ZigZag encoded.  Negative numbers
-                               // take 10 bytes.  Use TYPE_SINT32 if negative
-                               // values are likely.
+    // Not ZigZag encoded.  Negative numbers take 10 bytes.  Use TYPE_SINT32 if
+    // negative values are likely.
+    TYPE_INT32          = 5;
     TYPE_FIXED64        = 6;
     TYPE_FIXED32        = 7;
     TYPE_BOOL           = 8;
@@ -199,6 +204,7 @@ message MethodDescriptorProto {
   optional MethodOptions options = 4;
 }
 
+
 // ===================================================================
 // Options
 
@@ -220,10 +226,15 @@ message MethodDescriptorProto {
 //   through 99999.  It is up to you to ensure that you do not use the
 //   same number for multiple options.
 // * For options which will be published and used publicly by multiple
-//   independent entities, e-mail kenton@google.com to reserve extension
-//   numbers.  Simply tell me how many you need and I'll send you back a
-//   set of numbers to use -- there's no need to explain how you intend to
-//   use them.  If this turns out to be popular, a web service will be set up
+//   independent entities, e-mail protobuf-global-extension-registry@google.com
+//   to reserve extension numbers. Simply provide your project name (e.g.
+//   Object-C plugin) and your porject website (if available) -- there's no need
+//   to explain how you intend to use them. Usually you only need one extension
+//   number. You can declare multiple options with only one extension number by
+//   putting them in a sub-message. See the Custom Options section of the docs
+//   for examples:
+//   http://code.google.com/apis/protocolbuffers/docs/proto.html#options
+//   If this turns out to be popular, a web service will be set up
 //   to automatically assign option numbers.
 
 
@@ -266,6 +277,9 @@ message FileOptions {
   }
   optional OptimizeMode optimize_for = 9 [default=SPEED];
 
+  // Sets the Go package where structs generated from this .proto will be
+  // placed.  There is no default.
+  optional string go_package = 11;
 
 
 
@@ -344,6 +358,37 @@ message FieldOptions {
   optional bool packed = 2;
 
 
+
+  // Should this field be parsed lazily?  Lazy applies only to message-type
+  // fields.  It means that when the outer message is initially parsed, the
+  // inner message's contents will not be parsed but instead stored in encoded
+  // form.  The inner message will actually be parsed when it is first accessed.
+  //
+  // This is only a hint.  Implementations are free to choose whether to use
+  // eager or lazy parsing regardless of the value of this option.  However,
+  // setting this option true suggests that the protocol author believes that
+  // using lazy parsing on this field is worth the additional bookkeeping
+  // overhead typically needed to implement it.
+  //
+  // This option does not affect the public interface of any generated code;
+  // all method signatures remain the same.  Furthermore, thread-safety of the
+  // interface is not affected by this option; const methods remain safe to
+  // call from multiple threads concurrently, while non-const methods continue
+  // to require exclusive access.
+  //
+  //
+  // Note that implementations may choose not to check required fields within
+  // a lazy sub-message.  That is, calling IsInitialized() on the outher message
+  // may return true even if the inner message has missing required fields.
+  // This is necessary because otherwise the inner message would have to be
+  // parsed in order to perform the check, defeating the purpose of lazy
+  // parsing.  An implementation which chooses not to check required fields
+  // must be consistent about it.  That is, for any particular sub-message, the
+  // implementation must either *always* check its required fields, or *never*
+  // check its required fields, regardless of whether or not the message has
+  // been parsed.
+  optional bool lazy = 5 [default=false];
+
   // Is this field deprecated?
   // Depending on the target platform, this can emit Deprecated annotations
   // for accessors, or it will be completely ignored; in the very least, this
@@ -364,6 +409,9 @@ message FieldOptions {
   // TODO: Fully-implement this, then remove the "experimental_" prefix.
   optional string experimental_map_key = 9;
 
+  // For Google-internal migration only. Do not use.
+  optional bool weak = 10 [default=false];
+
   // The parser stores options it doesn't recognize here. See above.
   repeated UninterpretedOption uninterpreted_option = 999;
 
@@ -373,6 +421,10 @@ message FieldOptions {
 
 message EnumOptions {
 
+  // Set this option to false to disallow mapping different tag names to a same
+  // value.
+  optional bool allow_alias = 2 [default=true];
+
   // The parser stores options it doesn't recognize here. See above.
   repeated UninterpretedOption uninterpreted_option = 999;
 
@@ -416,6 +468,7 @@ message MethodOptions {
   extensions 1000 to max;
 }
 
+
 // A message representing a option the parser does not recognize. This only
 // appears in options protos created by the compiler::Parser class.
 // DescriptorPool resolves these when building Descriptor objects. Therefore,
@@ -527,7 +580,41 @@ message SourceCodeInfo {
     // 1 to each before displaying to a user.
     repeated int32 span = 2 [packed=true];
 
-    // TODO(kenton):  Record comments appearing before and after the
-    // declaration.
+    // If this SourceCodeInfo represents a complete declaration, these are any
+    // comments appearing before and after the declaration which appear to be
+    // attached to the declaration.
+    //
+    // A series of line comments appearing on consecutive lines, with no other
+    // tokens appearing on those lines, will be treated as a single comment.
+    //
+    // Only the comment content is provided; comment markers (e.g. //) are
+    // stripped out.  For block comments, leading whitespace and an asterisk
+    // will be stripped from the beginning of each line other than the first.
+    // Newlines are included in the output.
+    //
+    // Examples:
+    //
+    //   optional int32 foo = 1;  // Comment attached to foo.
+    //   // Comment attached to bar.
+    //   optional int32 bar = 2;
+    //
+    //   optional string baz = 3;
+    //   // Comment attached to baz.
+    //   // Another line attached to baz.
+    //
+    //   // Comment attached to qux.
+    //   //
+    //   // Another line attached to qux.
+    //   optional double qux = 4;
+    //
+    //   optional string corge = 5;
+    //   /* Block comment attached
+    //    * to corge.  Leading asterisks
+    //    * will be removed. */
+    //   /* Block comment attached to
+    //    * grault. */
+    //   optional int32 grault = 6;
+    optional string leading_comments = 3;
+    optional string trailing_comments = 4;
   }
 }

upb/descriptor/descriptor.upb.c

@@ -5,272 +5,292 @@
 #include "upb/def.h"
 
 const upb_msgdef google_protobuf_msgs[20];
-const upb_fielddef google_protobuf_fields[73];
+const upb_fielddef google_protobuf_fields[81];
 const upb_enumdef google_protobuf_enums[4];
-const upb_tabent google_protobuf_strentries[192];
+const upb_tabent google_protobuf_strentries[204];
 const upb_tabent google_protobuf_intentries[14];
-const _upb_value google_protobuf_arrays[224];
+const _upb_value google_protobuf_arrays[232];
 
 #ifdef UPB_DEBUG_REFS
-static upb_inttable reftables[194];
+static upb_inttable reftables[210];
 #endif
 
 const upb_msgdef google_protobuf_msgs[20] = {
   UPB_MSGDEF_INIT("google.protobuf.DescriptorProto", 27, 6, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[0], 8, 7), UPB_STRTABLE_INIT(7, 15, UPB_CTYPE_PTR, 4, &google_protobuf_strentries[0]),&reftables[0], &reftables[1]),
   UPB_MSGDEF_INIT("google.protobuf.DescriptorProto.ExtensionRange", 4, 0, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[8], 3, 2), UPB_STRTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[16]),&reftables[2], &reftables[3]),
   UPB_MSGDEF_INIT("google.protobuf.EnumDescriptorProto", 11, 2, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[11], 4, 3), UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[20]),&reftables[4], &reftables[5]),
-  UPB_MSGDEF_INIT("google.protobuf.EnumOptions", 6, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &google_protobuf_intentries[0], &google_protobuf_arrays[15], 4, 0), UPB_STRTABLE_INIT(1, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[24]),&reftables[6], &reftables[7]),
-  UPB_MSGDEF_INIT("google.protobuf.EnumValueDescriptorProto", 8, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[19], 4, 3), UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[28]),&reftables[8], &reftables[9]),
-  UPB_MSGDEF_INIT("google.protobuf.EnumValueOptions", 6, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &google_protobuf_intentries[2], &google_protobuf_arrays[23], 4, 0), UPB_STRTABLE_INIT(1, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[32]),&reftables[10], &reftables[11]),
-  UPB_MSGDEF_INIT("google.protobuf.FieldDescriptorProto", 19, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[27], 9, 8), UPB_STRTABLE_INIT(8, 15, UPB_CTYPE_PTR, 4, &google_protobuf_strentries[36]),&reftables[12], &reftables[13]),
-  UPB_MSGDEF_INIT("google.protobuf.FieldOptions", 12, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &google_protobuf_intentries[4], &google_protobuf_arrays[36], 32, 4), UPB_STRTABLE_INIT(5, 7, UPB_CTYPE_PTR, 3, &google_protobuf_strentries[52]),&reftables[14], &reftables[15]),
-  UPB_MSGDEF_INIT("google.protobuf.FileDescriptorProto", 33, 6, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[68], 10, 9), UPB_STRTABLE_INIT(9, 15, UPB_CTYPE_PTR, 4, &google_protobuf_strentries[60]),&reftables[16], &reftables[17]),
-  UPB_MSGDEF_INIT("google.protobuf.FileDescriptorSet", 6, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[78], 2, 1), UPB_STRTABLE_INIT(1, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[76]),&reftables[18], &reftables[19]),
-  UPB_MSGDEF_INIT("google.protobuf.FileOptions", 18, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &google_protobuf_intentries[6], &google_protobuf_arrays[80], 64, 8), UPB_STRTABLE_INIT(9, 15, UPB_CTYPE_PTR, 4, &google_protobuf_strentries[80]),&reftables[20], &reftables[21]),
-  UPB_MSGDEF_INIT("google.protobuf.MessageOptions", 8, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &google_protobuf_intentries[8], &google_protobuf_arrays[144], 16, 2), UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[96]),&reftables[22], &reftables[23]),
-  UPB_MSGDEF_INIT("google.protobuf.MethodDescriptorProto", 13, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[160], 5, 4), UPB_STRTABLE_INIT(4, 7, UPB_CTYPE_PTR, 3, &google_protobuf_strentries[100]),&reftables[24], &reftables[25]),
-  UPB_MSGDEF_INIT("google.protobuf.MethodOptions", 6, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &google_protobuf_intentries[10], &google_protobuf_arrays[165], 4, 0), UPB_STRTABLE_INIT(1, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[108]),&reftables[26], &reftables[27]),
-  UPB_MSGDEF_INIT("google.protobuf.ServiceDescriptorProto", 11, 2, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[169], 4, 3), UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[112]),&reftables[28], &reftables[29]),
-  UPB_MSGDEF_INIT("google.protobuf.ServiceOptions", 6, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &google_protobuf_intentries[12], &google_protobuf_arrays[173], 4, 0), UPB_STRTABLE_INIT(1, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[116]),&reftables[30], &reftables[31]),
-  UPB_MSGDEF_INIT("google.protobuf.SourceCodeInfo", 6, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[177], 2, 1), UPB_STRTABLE_INIT(1, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[120]),&reftables[32], &reftables[33]),
-  UPB_MSGDEF_INIT("google.protobuf.SourceCodeInfo.Location", 8, 0, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[179], 3, 2), UPB_STRTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[124]),&reftables[34], &reftables[35]),
-  UPB_MSGDEF_INIT("google.protobuf.UninterpretedOption", 18, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[182], 9, 7), UPB_STRTABLE_INIT(7, 15, UPB_CTYPE_PTR, 4, &google_protobuf_strentries[128]),&reftables[36], &reftables[37]),
-  UPB_MSGDEF_INIT("google.protobuf.UninterpretedOption.NamePart", 6, 0, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[191], 3, 2), UPB_STRTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[144]),&reftables[38], &reftables[39]),
+  UPB_MSGDEF_INIT("google.protobuf.EnumOptions", 7, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &google_protobuf_intentries[0], &google_protobuf_arrays[15], 8, 1), UPB_STRTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[24]),&reftables[6], &reftables[7]),
+  UPB_MSGDEF_INIT("google.protobuf.EnumValueDescriptorProto", 8, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[23], 4, 3), UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[28]),&reftables[8], &reftables[9]),
+  UPB_MSGDEF_INIT("google.protobuf.EnumValueOptions", 6, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &google_protobuf_intentries[2], &google_protobuf_arrays[27], 4, 0), UPB_STRTABLE_INIT(1, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[32]),&reftables[10], &reftables[11]),
+  UPB_MSGDEF_INIT("google.protobuf.FieldDescriptorProto", 19, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[31], 9, 8), UPB_STRTABLE_INIT(8, 15, UPB_CTYPE_PTR, 4, &google_protobuf_strentries[36]),&reftables[12], &reftables[13]),
+  UPB_MSGDEF_INIT("google.protobuf.FieldOptions", 14, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &google_protobuf_intentries[4], &google_protobuf_arrays[40], 32, 6), UPB_STRTABLE_INIT(7, 15, UPB_CTYPE_PTR, 4, &google_protobuf_strentries[52]),&reftables[14], &reftables[15]),
+  UPB_MSGDEF_INIT("google.protobuf.FileDescriptorProto", 39, 6, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[72], 12, 11), UPB_STRTABLE_INIT(11, 15, UPB_CTYPE_PTR, 4, &google_protobuf_strentries[68]),&reftables[16], &reftables[17]),
+  UPB_MSGDEF_INIT("google.protobuf.FileDescriptorSet", 6, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[84], 2, 1), UPB_STRTABLE_INIT(1, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[84]),&reftables[18], &reftables[19]),
+  UPB_MSGDEF_INIT("google.protobuf.FileOptions", 21, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &google_protobuf_intentries[6], &google_protobuf_arrays[86], 64, 9), UPB_STRTABLE_INIT(10, 15, UPB_CTYPE_PTR, 4, &google_protobuf_strentries[88]),&reftables[20], &reftables[21]),
+  UPB_MSGDEF_INIT("google.protobuf.MessageOptions", 8, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &google_protobuf_intentries[8], &google_protobuf_arrays[150], 16, 2), UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[104]),&reftables[22], &reftables[23]),
+  UPB_MSGDEF_INIT("google.protobuf.MethodDescriptorProto", 13, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[166], 5, 4), UPB_STRTABLE_INIT(4, 7, UPB_CTYPE_PTR, 3, &google_protobuf_strentries[108]),&reftables[24], &reftables[25]),
+  UPB_MSGDEF_INIT("google.protobuf.MethodOptions", 6, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &google_protobuf_intentries[10], &google_protobuf_arrays[171], 4, 0), UPB_STRTABLE_INIT(1, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[116]),&reftables[26], &reftables[27]),
+  UPB_MSGDEF_INIT("google.protobuf.ServiceDescriptorProto", 11, 2, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[175], 4, 3), UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[120]),&reftables[28], &reftables[29]),
+  UPB_MSGDEF_INIT("google.protobuf.ServiceOptions", 6, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &google_protobuf_intentries[12], &google_protobuf_arrays[179], 4, 0), UPB_STRTABLE_INIT(1, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[124]),&reftables[30], &reftables[31]),
+  UPB_MSGDEF_INIT("google.protobuf.SourceCodeInfo", 6, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[183], 2, 1), UPB_STRTABLE_INIT(1, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[128]),&reftables[32], &reftables[33]),
+  UPB_MSGDEF_INIT("google.protobuf.SourceCodeInfo.Location", 14, 0, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[185], 5, 4), UPB_STRTABLE_INIT(4, 7, UPB_CTYPE_PTR, 3, &google_protobuf_strentries[132]),&reftables[34], &reftables[35]),
+  UPB_MSGDEF_INIT("google.protobuf.UninterpretedOption", 18, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[190], 9, 7), UPB_STRTABLE_INIT(7, 15, UPB_CTYPE_PTR, 4, &google_protobuf_strentries[140]),&reftables[36], &reftables[37]),
+  UPB_MSGDEF_INIT("google.protobuf.UninterpretedOption.NamePart", 6, 0, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &google_protobuf_arrays[199], 3, 2), UPB_STRTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &google_protobuf_strentries[156]),&reftables[38], &reftables[39]),
 };
 
-const upb_fielddef google_protobuf_fields[73] = {
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, "aggregate_value", 8, &google_protobuf_msgs[18], NULL, 15, 6, {0},&reftables[40], &reftables[41]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, "cc_generic_services", 16, &google_protobuf_msgs[10], NULL, 14, 5, {0},&reftables[42], &reftables[43]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_ENUM, 0, false, false, "ctype", 1, &google_protobuf_msgs[7], UPB_UPCAST(&google_protobuf_enums[2]), 6, 1, {0},&reftables[44], &reftables[45]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, "default_value", 7, &google_protobuf_msgs[6], NULL, 16, 7, {0},&reftables[46], &reftables[47]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_STRING, 0, false, false, "dependency", 3, &google_protobuf_msgs[8], NULL, 30, 8, {0},&reftables[48], &reftables[49]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, "deprecated", 3, &google_protobuf_msgs[7], NULL, 8, 3, {0},&reftables[50], &reftables[51]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_DOUBLE, 0, false, false, "double_value", 6, &google_protobuf_msgs[18], NULL, 11, 4, {0},&reftables[52], &reftables[53]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_INT32, UPB_INTFMT_VARIABLE, false, false, "end", 2, &google_protobuf_msgs[1], NULL, 3, 1, {0},&reftables[54], &reftables[55]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "enum_type", 4, &google_protobuf_msgs[0], UPB_UPCAST(&google_protobuf_msgs[2]), 16, 2, {0},&reftables[56], &reftables[57]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "enum_type", 5, &google_protobuf_msgs[8], UPB_UPCAST(&google_protobuf_msgs[2]), 13, 1, {0},&reftables[58], &reftables[59]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, "experimental_map_key", 9, &google_protobuf_msgs[7], NULL, 9, 4, {0},&reftables[60], &reftables[61]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, "extendee", 2, &google_protobuf_msgs[6], NULL, 7, 2, {0},&reftables[62], &reftables[63]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "extension", 7, &google_protobuf_msgs[8], UPB_UPCAST(&google_protobuf_msgs[6]), 19, 3, {0},&reftables[64], &reftables[65]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "extension", 6, &google_protobuf_msgs[0], UPB_UPCAST(&google_protobuf_msgs[6]), 22, 4, {0},&reftables[66], &reftables[67]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "extension_range", 5, &google_protobuf_msgs[0], UPB_UPCAST(&google_protobuf_msgs[1]), 19, 3, {0},&reftables[68], &reftables[69]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "field", 2, &google_protobuf_msgs[0], UPB_UPCAST(&google_protobuf_msgs[6]), 10, 0, {0},&reftables[70], &reftables[71]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "file", 1, &google_protobuf_msgs[9], UPB_UPCAST(&google_protobuf_msgs[8]), 5, 0, {0},&reftables[72], &reftables[73]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, "identifier_value", 3, &google_protobuf_msgs[18], NULL, 6, 1, {0},&reftables[74], &reftables[75]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, "input_type", 2, &google_protobuf_msgs[12], NULL, 7, 2, {0},&reftables[76], &reftables[77]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REQUIRED, UPB_TYPE_BOOL, 0, false, false, "is_extension", 2, &google_protobuf_msgs[19], NULL, 5, 1, {0},&reftables[78], &reftables[79]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, "java_generate_equals_and_hash", 20, &google_protobuf_msgs[10], NULL, 17, 8, {0},&reftables[80], &reftables[81]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, "java_generic_services", 17, &google_protobuf_msgs[10], NULL, 15, 6, {0},&reftables[82], &reftables[83]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, "java_multiple_files", 10, &google_protobuf_msgs[10], NULL, 13, 4, {0},&reftables[84], &reftables[85]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, "java_outer_classname", 8, &google_protobuf_msgs[10], NULL, 9, 2, {0},&reftables[86], &reftables[87]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, "java_package", 1, &google_protobuf_msgs[10], NULL, 6, 1, {0},&reftables[88], &reftables[89]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_ENUM, 0, false, false, "label", 4, &google_protobuf_msgs[6], UPB_UPCAST(&google_protobuf_enums[0]), 11, 4, {0},&reftables[90], &reftables[91]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "location", 1, &google_protobuf_msgs[16], UPB_UPCAST(&google_protobuf_msgs[17]), 5, 0, {0},&reftables[92], &reftables[93]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, "message_set_wire_format", 1, &google_protobuf_msgs[11], NULL, 6, 1, {0},&reftables[94], &reftables[95]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "message_type", 4, &google_protobuf_msgs[8], UPB_UPCAST(&google_protobuf_msgs[0]), 10, 0, {0},&reftables[96], &reftables[97]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "method", 2, &google_protobuf_msgs[14], UPB_UPCAST(&google_protobuf_msgs[12]), 6, 0, {0},&reftables[98], &reftables[99]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "name", 2, &google_protobuf_msgs[18], UPB_UPCAST(&google_protobuf_msgs[19]), 5, 0, {0},&reftables[100], &reftables[101]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, "name", 1, &google_protobuf_msgs[8], NULL, 22, 6, {0},&reftables[102], &reftables[103]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, "name", 1, &google_protobuf_msgs[14], NULL, 8, 2, {0},&reftables[104], &reftables[105]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, "name", 1, &google_protobuf_msgs[12], NULL, 4, 1, {0},&reftables[106], &reftables[107]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, "name", 1, &google_protobuf_msgs[6], NULL, 4, 1, {0},&reftables[108], &reftables[109]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, "name", 1, &google_protobuf_msgs[4], NULL, 4, 1, {0},&reftables[110], &reftables[111]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, "name", 1, &google_protobuf_msgs[2], NULL, 8, 2, {0},&reftables[112], &reftables[113]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, "name", 1, &google_protobuf_msgs[0], NULL, 24, 6, {0},&reftables[114], &reftables[115]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REQUIRED, UPB_TYPE_STRING, 0, false, false, "name_part", 1, &google_protobuf_msgs[19], NULL, 2, 0, {0},&reftables[116], &reftables[117]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_INT64, UPB_INTFMT_VARIABLE, false, false, "negative_int_value", 5, &google_protobuf_msgs[18], NULL, 10, 3, {0},&reftables[118], &reftables[119]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "nested_type", 3, &google_protobuf_msgs[0], UPB_UPCAST(&google_protobuf_msgs[0]), 13, 1, {0},&reftables[120], &reftables[121]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, "no_standard_descriptor_accessor", 2, &google_protobuf_msgs[11], NULL, 7, 2, {0},&reftables[122], &reftables[123]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_INT32, UPB_INTFMT_VARIABLE, false, false, "number", 2, &google_protobuf_msgs[4], NULL, 7, 2, {0},&reftables[124], &reftables[125]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_INT32, UPB_INTFMT_VARIABLE, false, false, "number", 3, &google_protobuf_msgs[6], NULL, 10, 3, {0},&reftables[126], &reftables[127]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_ENUM, 0, false, false, "optimize_for", 9, &google_protobuf_msgs[10], UPB_UPCAST(&google_protobuf_enums[3]), 12, 3, {0},&reftables[128], &reftables[129]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, false, "options", 3, &google_protobuf_msgs[2], UPB_UPCAST(&google_protobuf_msgs[3]), 7, 1, {0},&reftables[130], &reftables[131]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, false, "options", 8, &google_protobuf_msgs[8], UPB_UPCAST(&google_protobuf_msgs[10]), 20, 4, {0},&reftables[132], &reftables[133]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, false, "options", 3, &google_protobuf_msgs[14], UPB_UPCAST(&google_protobuf_msgs[15]), 7, 1, {0},&reftables[134], &reftables[135]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, false, "options", 7, &google_protobuf_msgs[0], UPB_UPCAST(&google_protobuf_msgs[11]), 23, 5, {0},&reftables[136], &reftables[137]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, false, "options", 4, &google_protobuf_msgs[12], UPB_UPCAST(&google_protobuf_msgs[13]), 3, 0, {0},&reftables[138], &reftables[139]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, false, "options", 3, &google_protobuf_msgs[4], UPB_UPCAST(&google_protobuf_msgs[5]), 3, 0, {0},&reftables[140], &reftables[141]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, false, "options", 8, &google_protobuf_msgs[6], UPB_UPCAST(&google_protobuf_msgs[7]), 3, 0, {0},&reftables[142], &reftables[143]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, "output_type", 3, &google_protobuf_msgs[12], NULL, 10, 3, {0},&reftables[144], &reftables[145]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, "package", 2, &google_protobuf_msgs[8], NULL, 25, 7, {0},&reftables[146], &reftables[147]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, "packed", 2, &google_protobuf_msgs[7], NULL, 7, 2, {0},&reftables[148], &reftables[149]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_INT32, UPB_INTFMT_VARIABLE, false, false, "path", 1, &google_protobuf_msgs[17], NULL, 4, 0, {0},&reftables[150], &reftables[151]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_UINT64, UPB_INTFMT_VARIABLE, false, false, "positive_int_value", 4, &google_protobuf_msgs[18], NULL, 9, 2, {0},&reftables[152], &reftables[153]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, "py_generic_services", 18, &google_protobuf_msgs[10], NULL, 16, 7, {0},&reftables[154], &reftables[155]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "service", 6, &google_protobuf_msgs[8], UPB_UPCAST(&google_protobuf_msgs[14]), 16, 2, {0},&reftables[156], &reftables[157]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, false, "source_code_info", 9, &google_protobuf_msgs[8], UPB_UPCAST(&google_protobuf_msgs[16]), 21, 5, {0},&reftables[158], &reftables[159]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_INT32, UPB_INTFMT_VARIABLE, false, false, "span", 2, &google_protobuf_msgs[17], NULL, 7, 1, {0},&reftables[160], &reftables[161]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_INT32, UPB_INTFMT_VARIABLE, false, false, "start", 1, &google_protobuf_msgs[1], NULL, 2, 0, {0},&reftables[162], &reftables[163]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BYTES, 0, false, false, "string_value", 7, &google_protobuf_msgs[18], NULL, 12, 5, {0},&reftables[164], &reftables[165]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_ENUM, 0, false, false, "type", 5, &google_protobuf_msgs[6], UPB_UPCAST(&google_protobuf_enums[1]), 12, 5, {0},&reftables[166], &reftables[167]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, "type_name", 6, &google_protobuf_msgs[6], NULL, 13, 6, {0},&reftables[168], &reftables[169]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "uninterpreted_option", 999, &google_protobuf_msgs[3], UPB_UPCAST(&google_protobuf_msgs[18]), 5, 0, {0},&reftables[170], &reftables[171]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "uninterpreted_option", 999, &google_protobuf_msgs[15], UPB_UPCAST(&google_protobuf_msgs[18]), 5, 0, {0},&reftables[172], &reftables[173]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "uninterpreted_option", 999, &google_protobuf_msgs[11], UPB_UPCAST(&google_protobuf_msgs[18]), 5, 0, {0},&reftables[174], &reftables[175]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "uninterpreted_option", 999, &google_protobuf_msgs[10], UPB_UPCAST(&google_protobuf_msgs[18]), 5, 0, {0},&reftables[176], &reftables[177]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "uninterpreted_option", 999, &google_protobuf_msgs[7], UPB_UPCAST(&google_protobuf_msgs[18]), 5, 0, {0},&reftables[178], &reftables[179]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "uninterpreted_option", 999, &google_protobuf_msgs[13], UPB_UPCAST(&google_protobuf_msgs[18]), 5, 0, {0},&reftables[180], &reftables[181]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "uninterpreted_option", 999, &google_protobuf_msgs[5], UPB_UPCAST(&google_protobuf_msgs[18]), 5, 0, {0},&reftables[182], &reftables[183]),
-  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, "value", 2, &google_protobuf_msgs[2], UPB_UPCAST(&google_protobuf_msgs[4]), 6, 0, {0},&reftables[184], &reftables[185]),
+const upb_fielddef google_protobuf_fields[81] = {
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "aggregate_value", 8, &google_protobuf_msgs[18], NULL, 15, 6, {0},&reftables[40], &reftables[41]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, false, "allow_alias", 2, &google_protobuf_msgs[3], NULL, 6, 1, {0},&reftables[42], &reftables[43]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, false, "cc_generic_services", 16, &google_protobuf_msgs[10], NULL, 17, 6, {0},&reftables[44], &reftables[45]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_ENUM, 0, false, false, false, "ctype", 1, &google_protobuf_msgs[7], UPB_UPCAST(&google_protobuf_enums[2]), 6, 1, {0},&reftables[46], &reftables[47]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "default_value", 7, &google_protobuf_msgs[6], NULL, 16, 7, {0},&reftables[48], &reftables[49]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_STRING, 0, false, false, false, "dependency", 3, &google_protobuf_msgs[8], NULL, 30, 8, {0},&reftables[50], &reftables[51]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, false, "deprecated", 3, &google_protobuf_msgs[7], NULL, 8, 3, {0},&reftables[52], &reftables[53]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_DOUBLE, 0, false, false, false, "double_value", 6, &google_protobuf_msgs[18], NULL, 11, 4, {0},&reftables[54], &reftables[55]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_INT32, UPB_INTFMT_VARIABLE, false, false, false, "end", 2, &google_protobuf_msgs[1], NULL, 3, 1, {0},&reftables[56], &reftables[57]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "enum_type", 5, &google_protobuf_msgs[8], UPB_UPCAST(&google_protobuf_msgs[2]), 13, 1, {0},&reftables[58], &reftables[59]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "enum_type", 4, &google_protobuf_msgs[0], UPB_UPCAST(&google_protobuf_msgs[2]), 16, 2, {0},&reftables[60], &reftables[61]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "experimental_map_key", 9, &google_protobuf_msgs[7], NULL, 10, 5, {0},&reftables[62], &reftables[63]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "extendee", 2, &google_protobuf_msgs[6], NULL, 7, 2, {0},&reftables[64], &reftables[65]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "extension", 7, &google_protobuf_msgs[8], UPB_UPCAST(&google_protobuf_msgs[6]), 19, 3, {0},&reftables[66], &reftables[67]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "extension", 6, &google_protobuf_msgs[0], UPB_UPCAST(&google_protobuf_msgs[6]), 22, 4, {0},&reftables[68], &reftables[69]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "extension_range", 5, &google_protobuf_msgs[0], UPB_UPCAST(&google_protobuf_msgs[1]), 19, 3, {0},&reftables[70], &reftables[71]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "field", 2, &google_protobuf_msgs[0], UPB_UPCAST(&google_protobuf_msgs[6]), 10, 0, {0},&reftables[72], &reftables[73]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "file", 1, &google_protobuf_msgs[9], UPB_UPCAST(&google_protobuf_msgs[8]), 5, 0, {0},&reftables[74], &reftables[75]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "go_package", 11, &google_protobuf_msgs[10], NULL, 14, 5, {0},&reftables[76], &reftables[77]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "identifier_value", 3, &google_protobuf_msgs[18], NULL, 6, 1, {0},&reftables[78], &reftables[79]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "input_type", 2, &google_protobuf_msgs[12], NULL, 7, 2, {0},&reftables[80], &reftables[81]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REQUIRED, UPB_TYPE_BOOL, 0, false, false, false, "is_extension", 2, &google_protobuf_msgs[19], NULL, 5, 1, {0},&reftables[82], &reftables[83]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, false, "java_generate_equals_and_hash", 20, &google_protobuf_msgs[10], NULL, 20, 9, {0},&reftables[84], &reftables[85]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, false, "java_generic_services", 17, &google_protobuf_msgs[10], NULL, 18, 7, {0},&reftables[86], &reftables[87]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, false, "java_multiple_files", 10, &google_protobuf_msgs[10], NULL, 13, 4, {0},&reftables[88], &reftables[89]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "java_outer_classname", 8, &google_protobuf_msgs[10], NULL, 9, 2, {0},&reftables[90], &reftables[91]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "java_package", 1, &google_protobuf_msgs[10], NULL, 6, 1, {0},&reftables[92], &reftables[93]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_ENUM, 0, false, false, false, "label", 4, &google_protobuf_msgs[6], UPB_UPCAST(&google_protobuf_enums[0]), 11, 4, {0},&reftables[94], &reftables[95]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, false, "lazy", 5, &google_protobuf_msgs[7], NULL, 9, 4, {0},&reftables[96], &reftables[97]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "leading_comments", 3, &google_protobuf_msgs[17], NULL, 8, 2, {0},&reftables[98], &reftables[99]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "location", 1, &google_protobuf_msgs[16], UPB_UPCAST(&google_protobuf_msgs[17]), 5, 0, {0},&reftables[100], &reftables[101]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, false, "message_set_wire_format", 1, &google_protobuf_msgs[11], NULL, 6, 1, {0},&reftables[102], &reftables[103]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "message_type", 4, &google_protobuf_msgs[8], UPB_UPCAST(&google_protobuf_msgs[0]), 10, 0, {0},&reftables[104], &reftables[105]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "method", 2, &google_protobuf_msgs[14], UPB_UPCAST(&google_protobuf_msgs[12]), 6, 0, {0},&reftables[106], &reftables[107]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "name", 1, &google_protobuf_msgs[6], NULL, 4, 1, {0},&reftables[108], &reftables[109]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "name", 1, &google_protobuf_msgs[14], NULL, 8, 2, {0},&reftables[110], &reftables[111]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "name", 1, &google_protobuf_msgs[2], NULL, 8, 2, {0},&reftables[112], &reftables[113]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "name", 1, &google_protobuf_msgs[4], NULL, 4, 1, {0},&reftables[114], &reftables[115]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "name", 1, &google_protobuf_msgs[8], NULL, 22, 6, {0},&reftables[116], &reftables[117]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "name", 1, &google_protobuf_msgs[0], NULL, 24, 6, {0},&reftables[118], &reftables[119]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "name", 1, &google_protobuf_msgs[12], NULL, 4, 1, {0},&reftables[120], &reftables[121]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "name", 2, &google_protobuf_msgs[18], UPB_UPCAST(&google_protobuf_msgs[19]), 5, 0, {0},&reftables[122], &reftables[123]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REQUIRED, UPB_TYPE_STRING, 0, false, false, false, "name_part", 1, &google_protobuf_msgs[19], NULL, 2, 0, {0},&reftables[124], &reftables[125]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_INT64, UPB_INTFMT_VARIABLE, false, false, false, "negative_int_value", 5, &google_protobuf_msgs[18], NULL, 10, 3, {0},&reftables[126], &reftables[127]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "nested_type", 3, &google_protobuf_msgs[0], UPB_UPCAST(&google_protobuf_msgs[0]), 13, 1, {0},&reftables[128], &reftables[129]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, false, "no_standard_descriptor_accessor", 2, &google_protobuf_msgs[11], NULL, 7, 2, {0},&reftables[130], &reftables[131]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_INT32, UPB_INTFMT_VARIABLE, false, false, false, "number", 3, &google_protobuf_msgs[6], NULL, 10, 3, {0},&reftables[132], &reftables[133]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_INT32, UPB_INTFMT_VARIABLE, false, false, false, "number", 2, &google_protobuf_msgs[4], NULL, 7, 2, {0},&reftables[134], &reftables[135]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_ENUM, 0, false, false, false, "optimize_for", 9, &google_protobuf_msgs[10], UPB_UPCAST(&google_protobuf_enums[3]), 12, 3, {0},&reftables[136], &reftables[137]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, false, false, "options", 4, &google_protobuf_msgs[12], UPB_UPCAST(&google_protobuf_msgs[13]), 3, 0, {0},&reftables[138], &reftables[139]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, false, false, "options", 3, &google_protobuf_msgs[14], UPB_UPCAST(&google_protobuf_msgs[15]), 7, 1, {0},&reftables[140], &reftables[141]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, false, false, "options", 3, &google_protobuf_msgs[2], UPB_UPCAST(&google_protobuf_msgs[3]), 7, 1, {0},&reftables[142], &reftables[143]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, false, false, "options", 8, &google_protobuf_msgs[8], UPB_UPCAST(&google_protobuf_msgs[10]), 20, 4, {0},&reftables[144], &reftables[145]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, false, false, "options", 8, &google_protobuf_msgs[6], UPB_UPCAST(&google_protobuf_msgs[7]), 3, 0, {0},&reftables[146], &reftables[147]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, false, false, "options", 3, &google_protobuf_msgs[4], UPB_UPCAST(&google_protobuf_msgs[5]), 3, 0, {0},&reftables[148], &reftables[149]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, false, false, "options", 7, &google_protobuf_msgs[0], UPB_UPCAST(&google_protobuf_msgs[11]), 23, 5, {0},&reftables[150], &reftables[151]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "output_type", 3, &google_protobuf_msgs[12], NULL, 10, 3, {0},&reftables[152], &reftables[153]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "package", 2, &google_protobuf_msgs[8], NULL, 25, 7, {0},&reftables[154], &reftables[155]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, false, "packed", 2, &google_protobuf_msgs[7], NULL, 7, 2, {0},&reftables[156], &reftables[157]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_INT32, UPB_INTFMT_VARIABLE, false, false, false, "path", 1, &google_protobuf_msgs[17], NULL, 4, 0, {0},&reftables[158], &reftables[159]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_UINT64, UPB_INTFMT_VARIABLE, false, false, false, "positive_int_value", 4, &google_protobuf_msgs[18], NULL, 9, 2, {0},&reftables[160], &reftables[161]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_INT32, UPB_INTFMT_VARIABLE, false, false, false, "public_dependency", 10, &google_protobuf_msgs[8], NULL, 35, 9, {0},&reftables[162], &reftables[163]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, false, "py_generic_services", 18, &google_protobuf_msgs[10], NULL, 19, 8, {0},&reftables[164], &reftables[165]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "service", 6, &google_protobuf_msgs[8], UPB_UPCAST(&google_protobuf_msgs[14]), 16, 2, {0},&reftables[166], &reftables[167]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_MESSAGE, 0, false, false, false, "source_code_info", 9, &google_protobuf_msgs[8], UPB_UPCAST(&google_protobuf_msgs[16]), 21, 5, {0},&reftables[168], &reftables[169]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_INT32, UPB_INTFMT_VARIABLE, false, false, false, "span", 2, &google_protobuf_msgs[17], NULL, 7, 1, {0},&reftables[170], &reftables[171]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_INT32, UPB_INTFMT_VARIABLE, false, false, false, "start", 1, &google_protobuf_msgs[1], NULL, 2, 0, {0},&reftables[172], &reftables[173]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BYTES, 0, false, false, false, "string_value", 7, &google_protobuf_msgs[18], NULL, 12, 5, {0},&reftables[174], &reftables[175]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "trailing_comments", 4, &google_protobuf_msgs[17], NULL, 11, 3, {0},&reftables[176], &reftables[177]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_ENUM, 0, false, false, false, "type", 5, &google_protobuf_msgs[6], UPB_UPCAST(&google_protobuf_enums[1]), 12, 5, {0},&reftables[178], &reftables[179]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_STRING, 0, false, false, false, "type_name", 6, &google_protobuf_msgs[6], NULL, 13, 6, {0},&reftables[180], &reftables[181]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "uninterpreted_option", 999, &google_protobuf_msgs[15], UPB_UPCAST(&google_protobuf_msgs[18]), 5, 0, {0},&reftables[182], &reftables[183]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "uninterpreted_option", 999, &google_protobuf_msgs[11], UPB_UPCAST(&google_protobuf_msgs[18]), 5, 0, {0},&reftables[184], &reftables[185]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "uninterpreted_option", 999, &google_protobuf_msgs[13], UPB_UPCAST(&google_protobuf_msgs[18]), 5, 0, {0},&reftables[186], &reftables[187]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "uninterpreted_option", 999, &google_protobuf_msgs[7], UPB_UPCAST(&google_protobuf_msgs[18]), 5, 0, {0},&reftables[188], &reftables[189]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "uninterpreted_option", 999, &google_protobuf_msgs[3], UPB_UPCAST(&google_protobuf_msgs[18]), 5, 0, {0},&reftables[190], &reftables[191]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "uninterpreted_option", 999, &google_protobuf_msgs[5], UPB_UPCAST(&google_protobuf_msgs[18]), 5, 0, {0},&reftables[192], &reftables[193]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "uninterpreted_option", 999, &google_protobuf_msgs[10], UPB_UPCAST(&google_protobuf_msgs[18]), 5, 0, {0},&reftables[194], &reftables[195]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_MESSAGE, 0, false, false, false, "value", 2, &google_protobuf_msgs[2], UPB_UPCAST(&google_protobuf_msgs[4]), 6, 0, {0},&reftables[196], &reftables[197]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_OPTIONAL, UPB_TYPE_BOOL, 0, false, false, false, "weak", 10, &google_protobuf_msgs[7], NULL, 13, 6, {0},&reftables[198], &reftables[199]),
+  UPB_FIELDDEF_INIT(UPB_LABEL_REPEATED, UPB_TYPE_INT32, UPB_INTFMT_VARIABLE, false, false, false, "weak_dependency", 11, &google_protobuf_msgs[8], NULL, 38, 10, {0},&reftables[200], &reftables[201]),
 };
 
 const upb_enumdef google_protobuf_enums[4] = {
-  UPB_ENUMDEF_INIT("google.protobuf.FieldDescriptorProto.Label", UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_INT32, 2, &google_protobuf_strentries[148]), UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_CSTR, 0, NULL, &google_protobuf_arrays[194], 4, 3), 0, &reftables[186], &reftables[187]),
-  UPB_ENUMDEF_INIT("google.protobuf.FieldDescriptorProto.Type", UPB_STRTABLE_INIT(18, 31, UPB_CTYPE_INT32, 5, &google_protobuf_strentries[152]), UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_CSTR, 0, NULL, &google_protobuf_arrays[198], 19, 18), 0, &reftables[188], &reftables[189]),
-  UPB_ENUMDEF_INIT("google.protobuf.FieldOptions.CType", UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_INT32, 2, &google_protobuf_strentries[184]), UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_CSTR, 0, NULL, &google_protobuf_arrays[217], 3, 3), 0, &reftables[190], &reftables[191]),
-  UPB_ENUMDEF_INIT("google.protobuf.FileOptions.OptimizeMode", UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_INT32, 2, &google_protobuf_strentries[188]), UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_CSTR, 0, NULL, &google_protobuf_arrays[220], 4, 3), 0, &reftables[192], &reftables[193]),
+  UPB_ENUMDEF_INIT("google.protobuf.FieldDescriptorProto.Label", UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_INT32, 2, &google_protobuf_strentries[160]), UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_CSTR, 0, NULL, &google_protobuf_arrays[202], 4, 3), 0, &reftables[202], &reftables[203]),
+  UPB_ENUMDEF_INIT("google.protobuf.FieldDescriptorProto.Type", UPB_STRTABLE_INIT(18, 31, UPB_CTYPE_INT32, 5, &google_protobuf_strentries[164]), UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_CSTR, 0, NULL, &google_protobuf_arrays[206], 19, 18), 0, &reftables[204], &reftables[205]),
+  UPB_ENUMDEF_INIT("google.protobuf.FieldOptions.CType", UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_INT32, 2, &google_protobuf_strentries[196]), UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_CSTR, 0, NULL, &google_protobuf_arrays[225], 3, 3), 0, &reftables[206], &reftables[207]),
+  UPB_ENUMDEF_INIT("google.protobuf.FileOptions.OptimizeMode", UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_INT32, 2, &google_protobuf_strentries[200]), UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_CSTR, 0, NULL, &google_protobuf_arrays[228], 4, 3), 0, &reftables[208], &reftables[209]),
 };
 
-const upb_tabent google_protobuf_strentries[192] = {
-  {UPB_TABKEY_STR("extension"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[13]), NULL},
+const upb_tabent google_protobuf_strentries[204] = {
+  {UPB_TABKEY_STR("extension"), 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("name"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[37]), NULL},
+  {UPB_TABKEY_STR("name"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[39]), 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_STR("field"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[16]), NULL},
+  {UPB_TABKEY_STR("extension_range"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[15]), 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_STR("nested_type"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[44]), 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[48]), 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_STR("options"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[55]), NULL},
+  {UPB_TABKEY_STR("enum_type"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[10]), &google_protobuf_strentries[14]},
+  {UPB_TABKEY_STR("start"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[66]), NULL},
+  {UPB_TABKEY_STR("end"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[8]), 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[45]), NULL},
+  {UPB_TABKEY_STR("value"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[78]), 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[36]), &google_protobuf_strentries[22]},
-  {UPB_TABKEY_STR("uninterpreted_option"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[65]), NULL},
+  {UPB_TABKEY_STR("uninterpreted_option"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[75]), NULL},
   {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_STR("allow_alias"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[1]), NULL},
   {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_STR("number"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[47]), NULL},
   {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_STR("number"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[42]), NULL},
+  {UPB_TABKEY_STR("options"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[54]), NULL},
+  {UPB_TABKEY_STR("name"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[37]), &google_protobuf_strentries[30]},
+  {UPB_TABKEY_STR("uninterpreted_option"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[76]), NULL},
   {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_STR("options"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[50]), NULL},
-  {UPB_TABKEY_STR("name"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[35]), &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_STR("label"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[25]), NULL},
+  {UPB_TABKEY_STR("label"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[27]), 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_STR("number"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[43]), &google_protobuf_strentries[49]},
+  {UPB_TABKEY_STR("number"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[46]), &google_protobuf_strentries[49]},
   {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[51]), NULL},
-  {UPB_TABKEY_STR("experimental_map_key"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[10]), &google_protobuf_strentries[58]},
+  {UPB_TABKEY_STR("type_name"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[70]), NULL},
+  {UPB_TABKEY_STR("extendee"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[12]), NULL},
+  {UPB_TABKEY_STR("type"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[69]), &google_protobuf_strentries[48]},
+  {UPB_TABKEY_STR("default_value"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[4]), NULL},
+  {UPB_TABKEY_STR("options"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[53]), NULL},
+  {UPB_TABKEY_STR("experimental_map_key"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[11]), &google_protobuf_strentries[67]},
   {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_STR("ctype"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[2]), NULL},
+  {UPB_TABKEY_STR("weak"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[79]), 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_STR("name"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[31]), NULL},
-  {UPB_TABKEY_STR("service"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[58]), NULL},
+  {UPB_TABKEY_STR("packed"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[58]), NULL},
+  {UPB_TABKEY_STR("lazy"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[28]), 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_STR("ctype"), 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_STR("deprecated"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[6]), 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_STR("uninterpreted_option"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[74]), NULL},
+  {UPB_TABKEY_STR("extension"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[13]), NULL},
+  {UPB_TABKEY_STR("weak_dependency"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[80]), NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_STR("name"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[38]), NULL},
+  {UPB_TABKEY_STR("service"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[63]), NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_STR("source_code_info"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[64]), NULL},
   {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_STR("options"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[46]), 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_STR("file"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[16]), NULL},
   {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_STR("dependency"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[5]), NULL},
+  {UPB_TABKEY_STR("message_type"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[32]), NULL},
+  {UPB_TABKEY_STR("package"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[57]), NULL},
+  {UPB_TABKEY_STR("options"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[52]), &google_protobuf_strentries[82]},
+  {UPB_TABKEY_STR("enum_type"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[9]), NULL},
+  {UPB_TABKEY_STR("public_dependency"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[61]), &google_protobuf_strentries[81]},
   {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_STR("uninterpreted_option"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[68]), NULL},
+  {UPB_TABKEY_STR("file"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[17]), 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_STR("java_multiple_files"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[22]), NULL},
+  {UPB_TABKEY_STR("uninterpreted_option"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[77]), 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_STR("cc_generic_services"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[2]), NULL},
   {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_STR("java_multiple_files"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[24]), NULL},
   {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_STR("java_generic_services"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[23]), &google_protobuf_strentries[102]},
+  {UPB_TABKEY_STR("java_generate_equals_and_hash"), 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_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_STR("uninterpreted_option"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[67]), NULL},
-  {UPB_TABKEY_STR("no_standard_descriptor_accessor"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[41]), NULL},
+  {UPB_TABKEY_STR("go_package"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[18]), NULL},
+  {UPB_TABKEY_STR("java_package"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[26]), NULL},
+  {UPB_TABKEY_STR("optimize_for"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[48]), NULL},
+  {UPB_TABKEY_STR("py_generic_services"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[62]), NULL},
+  {UPB_TABKEY_STR("java_outer_classname"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[25]), NULL},
+  {UPB_TABKEY_STR("message_set_wire_format"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[31]), &google_protobuf_strentries[106]},
   {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_STR("uninterpreted_option"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[72]), NULL},
+  {UPB_TABKEY_STR("no_standard_descriptor_accessor"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[45]), 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[33]), 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_STR("output_type"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[52]), NULL},
+  {UPB_TABKEY_STR("name"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[40]), NULL},
+  {UPB_TABKEY_STR("input_type"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[20]), NULL},
+  {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
+  {UPB_TABKEY_STR("output_type"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[56]), NULL},
   {UPB_TABKEY_STR("options"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[49]), NULL},
-  {UPB_TABKEY_STR("uninterpreted_option"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[70]), NULL},
+  {UPB_TABKEY_STR("uninterpreted_option"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[73]), 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[47]), &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[66]), NULL},
+  {UPB_TABKEY_STR("options"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[50]), &google_protobuf_strentries[122]},
+  {UPB_TABKEY_STR("method"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[33]), NULL},
+  {UPB_TABKEY_STR("name"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[35]), &google_protobuf_strentries[121]},
+  {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_STR("location"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[26]), NULL},
+  {UPB_TABKEY_STR("location"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[30]), 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_STR("span"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[65]), &google_protobuf_strentries[139]},
   {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_STR("name"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[30]), NULL},
+  {UPB_TABKEY_STR("trailing_comments"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[68]), NULL},
+  {UPB_TABKEY_STR("leading_comments"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[29]), &google_protobuf_strentries[137]},
+  {UPB_TABKEY_STR("path"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[59]), NULL},
+  {UPB_TABKEY_STR("double_value"), 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_STR("name"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[41]), 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_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[43]), 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_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_STR("positive_int_value"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[60]), NULL},
+  {UPB_TABKEY_STR("identifier_value"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[19]), NULL},
+  {UPB_TABKEY_STR("string_value"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[67]), &google_protobuf_strentries[154]},
   {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_STR("is_extension"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[21]), NULL},
+  {UPB_TABKEY_STR("name_part"), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[42]), NULL},
+  {UPB_TABKEY_STR("LABEL_REQUIRED"), UPB_VALUE_INIT_INT32(2), &google_protobuf_strentries[162]},
   {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},
@@ -280,17 +300,17 @@ const upb_tabent google_protobuf_strentries[192] = {
   {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_FLOAT"), UPB_VALUE_INIT_INT32(2), &google_protobuf_strentries[193]},
   {UPB_TABKEY_STR("TYPE_DOUBLE"), UPB_VALUE_INIT_INT32(1), 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_STR("TYPE_MESSAGE"), UPB_VALUE_INIT_INT32(11), &google_protobuf_strentries[182]},
+  {UPB_TABKEY_STR("TYPE_MESSAGE"), UPB_VALUE_INIT_INT32(11), &google_protobuf_strentries[194]},
   {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_STR("TYPE_INT64"), UPB_VALUE_INIT_INT32(3), &google_protobuf_strentries[191]},
   {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
   {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
@@ -298,7 +318,7 @@ const upb_tabent google_protobuf_strentries[192] = {
   {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_STR("TYPE_UINT64"), UPB_VALUE_INIT_INT32(4), &google_protobuf_strentries[178]},
+  {UPB_TABKEY_STR("TYPE_UINT64"), UPB_VALUE_INIT_INT32(4), &google_protobuf_strentries[190]},
   {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},
@@ -308,80 +328,83 @@ const upb_tabent google_protobuf_strentries[192] = {
   {UPB_TABKEY_STR("TYPE_SINT32"), UPB_VALUE_INIT_INT32(17), 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"), UPB_VALUE_INIT_INT32(0), &google_protobuf_strentries[197]},
   {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_STR("SPEED"), UPB_VALUE_INIT_INT32(1), &google_protobuf_strentries[203]},
   {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[14] = {
   {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NUM(999), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[65]), NULL},
+  {UPB_TABKEY_NUM(999), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[75]), 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(999), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[76]), 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(999), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[74]), 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(999), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[77]), NULL},
   {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NUM(999), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[67]), NULL},
+  {UPB_TABKEY_NUM(999), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[72]), NULL},
   {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NUM(999), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[70]), NULL},
+  {UPB_TABKEY_NUM(999), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[73]), NULL},
   {UPB_TABKEY_NONE, UPB__VALUE_INIT_NONE, NULL},
-  {UPB_TABKEY_NUM(999), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[66]), NULL},
+  {UPB_TABKEY_NUM(999), UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[71]), NULL},
 };
 
-const _upb_value google_protobuf_arrays[224] = {
+const _upb_value google_protobuf_arrays[232] = {
   UPB_ARRAY_EMPTYENT,
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[37]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[39]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[16]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[44]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[10]),
   UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[15]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[40]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[8]),
   UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[14]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[13]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[48]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[55]),
   UPB_ARRAY_EMPTYENT,
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[61]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[7]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[66]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[8]),
   UPB_ARRAY_EMPTYENT,
   UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[36]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[72]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[45]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[78]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[51]),
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[1]),
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[35]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[42]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[50]),
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[37]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[47]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[54]),
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[34]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[11]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[43]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[25]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[63]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[64]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[3]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[51]),
   UPB_ARRAY_EMPTYENT,
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[2]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[54]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[5]),
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[34]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[12]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[46]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[27]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[69]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[70]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[4]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[53]),
   UPB_ARRAY_EMPTYENT,
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[3]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[58]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[6]),
   UPB_ARRAY_EMPTYENT,
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[28]),
   UPB_ARRAY_EMPTYENT,
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[10]),
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[11]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[79]),
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
@@ -403,38 +426,41 @@ const _upb_value google_protobuf_arrays[224] = {
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[31]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[53]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[4]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[28]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[9]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[58]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[12]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[46]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[59]),
   UPB_ARRAY_EMPTYENT,
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[16]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[38]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[57]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[5]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[32]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[9]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[63]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[13]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[52]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[64]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[61]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[80]),
   UPB_ARRAY_EMPTYENT,
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[24]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[17]),
   UPB_ARRAY_EMPTYENT,
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[26]),
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[23]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[44]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[22]),
   UPB_ARRAY_EMPTYENT,
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[25]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[48]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[24]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[18]),
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[1]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[21]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[57]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[2]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[23]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[62]),
   UPB_ARRAY_EMPTYENT,
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[20]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[22]),
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
@@ -479,8 +505,8 @@ const _upb_value google_protobuf_arrays[224] = {
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[27]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[41]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[31]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[45]),
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
@@ -495,39 +521,41 @@ const _upb_value google_protobuf_arrays[224] = {
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[33]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[18]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[52]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[40]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[20]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[56]),
   UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[49]),
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[32]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[29]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[47]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[35]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[33]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[50]),
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[26]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[30]),
   UPB_ARRAY_EMPTYENT,
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[55]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[60]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[59]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[65]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[29]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[68]),
   UPB_ARRAY_EMPTYENT,
   UPB_ARRAY_EMPTYENT,
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[30]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[17]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[56]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[39]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[6]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[62]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[41]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[19]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[60]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[43]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[7]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[67]),
   UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[0]),
   UPB_ARRAY_EMPTYENT,
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[38]),
-  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[19]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[42]),
+  UPB_VALUE_INIT_CONSTPTR(&google_protobuf_fields[21]),
   UPB_ARRAY_EMPTYENT,
   UPB_VALUE_INIT_CONSTPTR("LABEL_OPTIONAL"),
   UPB_VALUE_INIT_CONSTPTR("LABEL_REQUIRED"),
@@ -561,7 +589,23 @@ const _upb_value google_protobuf_arrays[224] = {
 };
 
 #ifdef UPB_DEBUG_REFS
-static upb_inttable reftables[194] = {
+static upb_inttable reftables[210] = {
+  UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR),
+  UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR),
+  UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR),
+  UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR),
+  UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR),
+  UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR),
+  UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR),
+  UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR),
+  UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR),
+  UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR),
+  UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR),
+  UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR),
+  UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR),
+  UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR),
+  UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR),
+  UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR),
   UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR),
   UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR),
   UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR),

upb/descriptor/descriptor.upb.h

@@ -55,7 +55,7 @@ typedef enum {
 // Do not refer to these forward declarations; use the constants
 // below.
 extern const upb_msgdef google_protobuf_msgs[20];
-extern const upb_fielddef google_protobuf_fields[73];
+extern const upb_fielddef google_protobuf_fields[81];
 extern const upb_enumdef google_protobuf_enums[4];
 
 // Constants for references to defs.
@@ -85,55 +85,89 @@ extern const upb_enumdef google_protobuf_enums[4];
 
 // Selector definitions.
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_ENUM_TYPE_ENDSEQ 15
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_ENUM_TYPE_ENDSTR 18
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_ENUM_TYPE_ENDSUBMSG 16
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_ENUM_TYPE_STARTSEQ 14
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_ENUM_TYPE_STARTSTR 17
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_ENUM_TYPE_STARTSUBMSG 4
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_ENUM_TYPE_STRING 19
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSIONRANGE_END_INT32 3
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSIONRANGE_START_INT32 2
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_ENDSEQ 21
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_ENDSTR 24
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_ENDSUBMSG 22
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_RANGE_ENDSEQ 18
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_RANGE_ENDSTR 21
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_RANGE_ENDSUBMSG 19
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_RANGE_STARTSEQ 17
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_RANGE_STARTSTR 20
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_RANGE_STARTSUBMSG 5
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_RANGE_STRING 22
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_STARTSEQ 20
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_STARTSTR 23
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_STARTSUBMSG 6
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_EXTENSION_STRING 25
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_FIELD_ENDSEQ 9
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_FIELD_ENDSTR 12
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_FIELD_ENDSUBMSG 10
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_FIELD_STARTSEQ 8
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_FIELD_STARTSTR 11
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_FIELD_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_FIELD_STRING 13
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_NAME_ENDSTR 26
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_NAME_STARTSTR 25
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_NAME_STRING 24
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_NESTED_TYPE_ENDSEQ 12
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_NESTED_TYPE_ENDSTR 15
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_NESTED_TYPE_ENDSUBMSG 13
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_NESTED_TYPE_STARTSEQ 11
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_NESTED_TYPE_STARTSTR 14
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_NESTED_TYPE_STARTSUBMSG 3
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_NESTED_TYPE_STRING 16
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_OPTIONS_ENDSTR 25
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_OPTIONS_ENDSUBMSG 23
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_OPTIONS_STARTSTR 24
 #define GOOGLE_PROTOBUF_DESCRIPTORPROTO_OPTIONS_STARTSUBMSG 7
+#define GOOGLE_PROTOBUF_DESCRIPTORPROTO_OPTIONS_STRING 26
 #define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_NAME_ENDSTR 10
 #define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_NAME_STARTSTR 9
 #define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_NAME_STRING 8
+#define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_OPTIONS_ENDSTR 9
 #define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_OPTIONS_ENDSUBMSG 7
+#define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_OPTIONS_STARTSTR 8
 #define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_OPTIONS_STARTSUBMSG 3
+#define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_OPTIONS_STRING 10
 #define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_VALUE_ENDSEQ 5
+#define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_VALUE_ENDSTR 8
 #define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_VALUE_ENDSUBMSG 6
 #define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_VALUE_STARTSEQ 4
+#define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_VALUE_STARTSTR 7
 #define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_VALUE_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_ENUMDESCRIPTORPROTO_VALUE_STRING 9
+#define GOOGLE_PROTOBUF_ENUMOPTIONS_ALLOW_ALIAS_BOOL 6
 #define GOOGLE_PROTOBUF_ENUMOPTIONS_UNINTERPRETED_OPTION_ENDSEQ 4
+#define GOOGLE_PROTOBUF_ENUMOPTIONS_UNINTERPRETED_OPTION_ENDSTR 7
 #define GOOGLE_PROTOBUF_ENUMOPTIONS_UNINTERPRETED_OPTION_ENDSUBMSG 5
 #define GOOGLE_PROTOBUF_ENUMOPTIONS_UNINTERPRETED_OPTION_STARTSEQ 3
+#define GOOGLE_PROTOBUF_ENUMOPTIONS_UNINTERPRETED_OPTION_STARTSTR 6
 #define GOOGLE_PROTOBUF_ENUMOPTIONS_UNINTERPRETED_OPTION_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_ENUMOPTIONS_UNINTERPRETED_OPTION_STRING 8
 #define GOOGLE_PROTOBUF_ENUMVALUEDESCRIPTORPROTO_NAME_ENDSTR 6
 #define GOOGLE_PROTOBUF_ENUMVALUEDESCRIPTORPROTO_NAME_STARTSTR 5
 #define GOOGLE_PROTOBUF_ENUMVALUEDESCRIPTORPROTO_NAME_STRING 4
 #define GOOGLE_PROTOBUF_ENUMVALUEDESCRIPTORPROTO_NUMBER_INT32 7
+#define GOOGLE_PROTOBUF_ENUMVALUEDESCRIPTORPROTO_OPTIONS_ENDSTR 5
 #define GOOGLE_PROTOBUF_ENUMVALUEDESCRIPTORPROTO_OPTIONS_ENDSUBMSG 3
+#define GOOGLE_PROTOBUF_ENUMVALUEDESCRIPTORPROTO_OPTIONS_STARTSTR 4
 #define GOOGLE_PROTOBUF_ENUMVALUEDESCRIPTORPROTO_OPTIONS_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_ENUMVALUEDESCRIPTORPROTO_OPTIONS_STRING 6
 #define GOOGLE_PROTOBUF_ENUMVALUEOPTIONS_UNINTERPRETED_OPTION_ENDSEQ 4
+#define GOOGLE_PROTOBUF_ENUMVALUEOPTIONS_UNINTERPRETED_OPTION_ENDSTR 7
 #define GOOGLE_PROTOBUF_ENUMVALUEOPTIONS_UNINTERPRETED_OPTION_ENDSUBMSG 5
 #define GOOGLE_PROTOBUF_ENUMVALUEOPTIONS_UNINTERPRETED_OPTION_STARTSEQ 3
+#define GOOGLE_PROTOBUF_ENUMVALUEOPTIONS_UNINTERPRETED_OPTION_STARTSTR 6
 #define GOOGLE_PROTOBUF_ENUMVALUEOPTIONS_UNINTERPRETED_OPTION_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_ENUMVALUEOPTIONS_UNINTERPRETED_OPTION_STRING 8
 #define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_DEFAULT_VALUE_ENDSTR 18
 #define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_DEFAULT_VALUE_STARTSTR 17
 #define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_DEFAULT_VALUE_STRING 16
@@ -145,60 +179,98 @@ extern const upb_enumdef google_protobuf_enums[4];
 #define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_NAME_STARTSTR 5
 #define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_NAME_STRING 4
 #define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_NUMBER_INT32 10
+#define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_OPTIONS_ENDSTR 5
 #define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_OPTIONS_ENDSUBMSG 3
+#define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_OPTIONS_STARTSTR 4
 #define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_OPTIONS_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_OPTIONS_STRING 6
 #define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_INT32 12
 #define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_NAME_ENDSTR 15
 #define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_NAME_STARTSTR 14
 #define GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_NAME_STRING 13
 #define GOOGLE_PROTOBUF_FIELDOPTIONS_CTYPE_INT32 6
 #define GOOGLE_PROTOBUF_FIELDOPTIONS_DEPRECATED_BOOL 8
-#define GOOGLE_PROTOBUF_FIELDOPTIONS_EXPERIMENTAL_MAP_KEY_ENDSTR 11
-#define GOOGLE_PROTOBUF_FIELDOPTIONS_EXPERIMENTAL_MAP_KEY_STARTSTR 10
-#define GOOGLE_PROTOBUF_FIELDOPTIONS_EXPERIMENTAL_MAP_KEY_STRING 9
+#define GOOGLE_PROTOBUF_FIELDOPTIONS_EXPERIMENTAL_MAP_KEY_ENDSTR 12
+#define GOOGLE_PROTOBUF_FIELDOPTIONS_EXPERIMENTAL_MAP_KEY_STARTSTR 11
+#define GOOGLE_PROTOBUF_FIELDOPTIONS_EXPERIMENTAL_MAP_KEY_STRING 10
+#define GOOGLE_PROTOBUF_FIELDOPTIONS_LAZY_BOOL 9
 #define GOOGLE_PROTOBUF_FIELDOPTIONS_PACKED_BOOL 7
 #define GOOGLE_PROTOBUF_FIELDOPTIONS_UNINTERPRETED_OPTION_ENDSEQ 4
+#define GOOGLE_PROTOBUF_FIELDOPTIONS_UNINTERPRETED_OPTION_ENDSTR 7
 #define GOOGLE_PROTOBUF_FIELDOPTIONS_UNINTERPRETED_OPTION_ENDSUBMSG 5
 #define GOOGLE_PROTOBUF_FIELDOPTIONS_UNINTERPRETED_OPTION_STARTSEQ 3
+#define GOOGLE_PROTOBUF_FIELDOPTIONS_UNINTERPRETED_OPTION_STARTSTR 6
 #define GOOGLE_PROTOBUF_FIELDOPTIONS_UNINTERPRETED_OPTION_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_FIELDOPTIONS_UNINTERPRETED_OPTION_STRING 8
+#define GOOGLE_PROTOBUF_FIELDOPTIONS_WEAK_BOOL 13
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_DEPENDENCY_ENDSEQ 29
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_DEPENDENCY_ENDSTR 32
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_DEPENDENCY_STARTSEQ 28
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_DEPENDENCY_STARTSTR 31
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_DEPENDENCY_STRING 30
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_ENUM_TYPE_ENDSEQ 12
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_ENUM_TYPE_ENDSTR 15
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_ENUM_TYPE_ENDSUBMSG 13
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_ENUM_TYPE_STARTSEQ 11
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_ENUM_TYPE_STARTSTR 14
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_ENUM_TYPE_STARTSUBMSG 3
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_ENUM_TYPE_STRING 16
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_EXTENSION_ENDSEQ 18
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_EXTENSION_ENDSTR 21
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_EXTENSION_ENDSUBMSG 19
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_EXTENSION_STARTSEQ 17
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_EXTENSION_STARTSTR 20
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_EXTENSION_STARTSUBMSG 5
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_EXTENSION_STRING 22
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_MESSAGE_TYPE_ENDSEQ 9
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_MESSAGE_TYPE_ENDSTR 12
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_MESSAGE_TYPE_ENDSUBMSG 10
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_MESSAGE_TYPE_STARTSEQ 8
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_MESSAGE_TYPE_STARTSTR 11
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_MESSAGE_TYPE_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_MESSAGE_TYPE_STRING 13
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_NAME_ENDSTR 24
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_NAME_STARTSTR 23
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_NAME_STRING 22
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_OPTIONS_ENDSTR 22
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_OPTIONS_ENDSUBMSG 20
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_OPTIONS_STARTSTR 21
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_OPTIONS_STARTSUBMSG 6
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_OPTIONS_STRING 23
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_PACKAGE_ENDSTR 27
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_PACKAGE_STARTSTR 26
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_PACKAGE_STRING 25
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_PUBLIC_DEPENDENCY_ENDSEQ 34
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_PUBLIC_DEPENDENCY_INT32 35
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_PUBLIC_DEPENDENCY_STARTSEQ 33
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_SERVICE_ENDSEQ 15
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_SERVICE_ENDSTR 18
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_SERVICE_ENDSUBMSG 16
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_SERVICE_STARTSEQ 14
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_SERVICE_STARTSTR 17
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_SERVICE_STARTSUBMSG 4
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_SERVICE_STRING 19
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_SOURCE_CODE_INFO_ENDSTR 23
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_SOURCE_CODE_INFO_ENDSUBMSG 21
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_SOURCE_CODE_INFO_STARTSTR 22
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_SOURCE_CODE_INFO_STARTSUBMSG 7
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_SOURCE_CODE_INFO_STRING 24
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_WEAK_DEPENDENCY_ENDSEQ 37
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_WEAK_DEPENDENCY_INT32 38
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORPROTO_WEAK_DEPENDENCY_STARTSEQ 36
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORSET_FILE_ENDSEQ 4
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORSET_FILE_ENDSTR 7
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORSET_FILE_ENDSUBMSG 5
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORSET_FILE_STARTSEQ 3
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORSET_FILE_STARTSTR 6
 #define GOOGLE_PROTOBUF_FILEDESCRIPTORSET_FILE_STARTSUBMSG 2
-#define GOOGLE_PROTOBUF_FILEOPTIONS_CC_GENERIC_SERVICES_BOOL 14
-#define GOOGLE_PROTOBUF_FILEOPTIONS_JAVA_GENERATE_EQUALS_AND_HASH_BOOL 17
-#define GOOGLE_PROTOBUF_FILEOPTIONS_JAVA_GENERIC_SERVICES_BOOL 15
+#define GOOGLE_PROTOBUF_FILEDESCRIPTORSET_FILE_STRING 8
+#define GOOGLE_PROTOBUF_FILEOPTIONS_CC_GENERIC_SERVICES_BOOL 17
+#define GOOGLE_PROTOBUF_FILEOPTIONS_GO_PACKAGE_ENDSTR 16
+#define GOOGLE_PROTOBUF_FILEOPTIONS_GO_PACKAGE_STARTSTR 15
+#define GOOGLE_PROTOBUF_FILEOPTIONS_GO_PACKAGE_STRING 14
+#define GOOGLE_PROTOBUF_FILEOPTIONS_JAVA_GENERATE_EQUALS_AND_HASH_BOOL 20
+#define GOOGLE_PROTOBUF_FILEOPTIONS_JAVA_GENERIC_SERVICES_BOOL 18
 #define GOOGLE_PROTOBUF_FILEOPTIONS_JAVA_MULTIPLE_FILES_BOOL 13
 #define GOOGLE_PROTOBUF_FILEOPTIONS_JAVA_OUTER_CLASSNAME_ENDSTR 11
 #define GOOGLE_PROTOBUF_FILEOPTIONS_JAVA_OUTER_CLASSNAME_STARTSTR 10
@@ -207,47 +279,72 @@ extern const upb_enumdef google_protobuf_enums[4];
 #define GOOGLE_PROTOBUF_FILEOPTIONS_JAVA_PACKAGE_STARTSTR 7
 #define GOOGLE_PROTOBUF_FILEOPTIONS_JAVA_PACKAGE_STRING 6
 #define GOOGLE_PROTOBUF_FILEOPTIONS_OPTIMIZE_FOR_INT32 12
-#define GOOGLE_PROTOBUF_FILEOPTIONS_PY_GENERIC_SERVICES_BOOL 16
+#define GOOGLE_PROTOBUF_FILEOPTIONS_PY_GENERIC_SERVICES_BOOL 19
 #define GOOGLE_PROTOBUF_FILEOPTIONS_UNINTERPRETED_OPTION_ENDSEQ 4
+#define GOOGLE_PROTOBUF_FILEOPTIONS_UNINTERPRETED_OPTION_ENDSTR 7
 #define GOOGLE_PROTOBUF_FILEOPTIONS_UNINTERPRETED_OPTION_ENDSUBMSG 5
 #define GOOGLE_PROTOBUF_FILEOPTIONS_UNINTERPRETED_OPTION_STARTSEQ 3
+#define GOOGLE_PROTOBUF_FILEOPTIONS_UNINTERPRETED_OPTION_STARTSTR 6
 #define GOOGLE_PROTOBUF_FILEOPTIONS_UNINTERPRETED_OPTION_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_FILEOPTIONS_UNINTERPRETED_OPTION_STRING 8
 #define GOOGLE_PROTOBUF_MESSAGEOPTIONS_MESSAGE_SET_WIRE_FORMAT_BOOL 6
 #define GOOGLE_PROTOBUF_MESSAGEOPTIONS_NO_STANDARD_DESCRIPTOR_ACCESSOR_BOOL 7
 #define GOOGLE_PROTOBUF_MESSAGEOPTIONS_UNINTERPRETED_OPTION_ENDSEQ 4
+#define GOOGLE_PROTOBUF_MESSAGEOPTIONS_UNINTERPRETED_OPTION_ENDSTR 7
 #define GOOGLE_PROTOBUF_MESSAGEOPTIONS_UNINTERPRETED_OPTION_ENDSUBMSG 5
 #define GOOGLE_PROTOBUF_MESSAGEOPTIONS_UNINTERPRETED_OPTION_STARTSEQ 3
+#define GOOGLE_PROTOBUF_MESSAGEOPTIONS_UNINTERPRETED_OPTION_STARTSTR 6
 #define GOOGLE_PROTOBUF_MESSAGEOPTIONS_UNINTERPRETED_OPTION_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_MESSAGEOPTIONS_UNINTERPRETED_OPTION_STRING 8
 #define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_INPUT_TYPE_ENDSTR 9
 #define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_INPUT_TYPE_STARTSTR 8
 #define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_INPUT_TYPE_STRING 7
 #define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_NAME_ENDSTR 6
 #define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_NAME_STARTSTR 5
 #define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_NAME_STRING 4
+#define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_OPTIONS_ENDSTR 5
 #define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_OPTIONS_ENDSUBMSG 3
+#define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_OPTIONS_STARTSTR 4
 #define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_OPTIONS_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_OPTIONS_STRING 6
 #define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_OUTPUT_TYPE_ENDSTR 12
 #define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_OUTPUT_TYPE_STARTSTR 11
 #define GOOGLE_PROTOBUF_METHODDESCRIPTORPROTO_OUTPUT_TYPE_STRING 10
 #define GOOGLE_PROTOBUF_METHODOPTIONS_UNINTERPRETED_OPTION_ENDSEQ 4
+#define GOOGLE_PROTOBUF_METHODOPTIONS_UNINTERPRETED_OPTION_ENDSTR 7
 #define GOOGLE_PROTOBUF_METHODOPTIONS_UNINTERPRETED_OPTION_ENDSUBMSG 5
 #define GOOGLE_PROTOBUF_METHODOPTIONS_UNINTERPRETED_OPTION_STARTSEQ 3
+#define GOOGLE_PROTOBUF_METHODOPTIONS_UNINTERPRETED_OPTION_STARTSTR 6
 #define GOOGLE_PROTOBUF_METHODOPTIONS_UNINTERPRETED_OPTION_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_METHODOPTIONS_UNINTERPRETED_OPTION_STRING 8
 #define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_METHOD_ENDSEQ 5
+#define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_METHOD_ENDSTR 8
 #define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_METHOD_ENDSUBMSG 6
 #define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_METHOD_STARTSEQ 4
+#define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_METHOD_STARTSTR 7
 #define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_METHOD_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_METHOD_STRING 9
 #define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_NAME_ENDSTR 10
 #define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_NAME_STARTSTR 9
 #define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_NAME_STRING 8
+#define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_OPTIONS_ENDSTR 9
 #define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_OPTIONS_ENDSUBMSG 7
+#define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_OPTIONS_STARTSTR 8
 #define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_OPTIONS_STARTSUBMSG 3
+#define GOOGLE_PROTOBUF_SERVICEDESCRIPTORPROTO_OPTIONS_STRING 10
 #define GOOGLE_PROTOBUF_SERVICEOPTIONS_UNINTERPRETED_OPTION_ENDSEQ 4
+#define GOOGLE_PROTOBUF_SERVICEOPTIONS_UNINTERPRETED_OPTION_ENDSTR 7
 #define GOOGLE_PROTOBUF_SERVICEOPTIONS_UNINTERPRETED_OPTION_ENDSUBMSG 5
 #define GOOGLE_PROTOBUF_SERVICEOPTIONS_UNINTERPRETED_OPTION_STARTSEQ 3
+#define GOOGLE_PROTOBUF_SERVICEOPTIONS_UNINTERPRETED_OPTION_STARTSTR 6
 #define GOOGLE_PROTOBUF_SERVICEOPTIONS_UNINTERPRETED_OPTION_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_SERVICEOPTIONS_UNINTERPRETED_OPTION_STRING 8
 #define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_ENDSEQ 4
+#define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_ENDSTR 7
 #define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_ENDSUBMSG 5
+#define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_LEADING_COMMENTS_ENDSTR 10
+#define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_LEADING_COMMENTS_STARTSTR 9
+#define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_LEADING_COMMENTS_STRING 8
 #define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_PATH_ENDSEQ 3
 #define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_PATH_INT32 4
 #define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_PATH_STARTSEQ 2
@@ -255,7 +352,12 @@ extern const upb_enumdef google_protobuf_enums[4];
 #define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_SPAN_INT32 7
 #define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_SPAN_STARTSEQ 5
 #define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_STARTSEQ 3
+#define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_STARTSTR 6
 #define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_STRING 8
+#define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_TRAILING_COMMENTS_ENDSTR 13
+#define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_TRAILING_COMMENTS_STARTSTR 12
+#define GOOGLE_PROTOBUF_SOURCECODEINFO_LOCATION_TRAILING_COMMENTS_STRING 11
 #define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_AGGREGATE_VALUE_ENDSTR 17
 #define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_AGGREGATE_VALUE_STARTSTR 16
 #define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_AGGREGATE_VALUE_STRING 15
@@ -268,9 +370,12 @@ extern const upb_enumdef google_protobuf_enums[4];
 #define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_NAMEPART_NAME_PART_STARTSTR 3
 #define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_NAMEPART_NAME_PART_STRING 2
 #define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_NAME_ENDSEQ 4
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_NAME_ENDSTR 7
 #define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_NAME_ENDSUBMSG 5
 #define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_NAME_STARTSEQ 3
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_NAME_STARTSTR 6
 #define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_NAME_STARTSUBMSG 2
+#define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_NAME_STRING 8
 #define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_NEGATIVE_INT_VALUE_INT64 10
 #define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_POSITIVE_INT_VALUE_UINT64 9
 #define GOOGLE_PROTOBUF_UNINTERPRETEDOPTION_STRING_VALUE_ENDSTR 14

upb/descriptor/reader.c

@@ -376,6 +376,13 @@ static bool field_endmsg(void *closure, const void *hd, upb_status *status) {
   return true;
 }
 
+static bool field_onlazy(void *closure, const void *hd, bool val) {
+  UPB_UNUSED(hd);
+  upb_descreader *r = closure;
+  upb_fielddef_setlazy(r->f, val);
+  return true;
+}
+
 static bool field_ontype(void *closure, const void *hd, int32_t val) {
   UPB_UNUSED(hd);
   upb_descreader *r = closure;
@@ -540,6 +547,8 @@ static void reghandlers(void *closure, upb_handlers *h) {
     upb_handlers_setstring(h, f(h, "type_name"), &field_ontypename, NULL);
     upb_handlers_setstring(h, f(h, "extendee"), &field_onextendee, NULL);
     upb_handlers_setstring(h, f(h, "default_value"), &field_ondefaultval, NULL);
+  } else if (m == GOOGLE_PROTOBUF_FIELDOPTIONS) {
+    upb_handlers_setbool(h, f(h, "lazy"), &field_onlazy, NULL);
   }
 }
 

upb/handlers.c

@@ -52,6 +52,9 @@ typedef struct {
   void *closure;
 } dfs_state;
 
+// TODO(haberman): discard upb_handlers* objects that do not actually have any
+// handlers set and cannot reach any upb_handlers* object that does.  This is
+// slightly tricky to do correctly.
 static upb_handlers *newformsg(const upb_msgdef *m, const void *owner,
                                dfs_state *s) {
   upb_handlers *h = upb_handlers_new(m, owner);
@@ -425,6 +428,39 @@ bool upb_handlers_freeze(upb_handlers *const*handlers, int n, upb_status *s) {
       }
 
       if (upb_fielddef_issubmsg(f)) {
+        bool hashandler = false;
+        if (upb_handlers_gethandler(h, getsel(h, f, UPB_HANDLER_STARTSUBMSG)) ||
+            upb_handlers_gethandler(h, getsel(h, f, UPB_HANDLER_ENDSUBMSG))) {
+          hashandler = true;
+        }
+
+        if (upb_fielddef_isseq(f) &&
+            (upb_handlers_gethandler(h, getsel(h, f, UPB_HANDLER_STARTSEQ)) ||
+             upb_handlers_gethandler(h, getsel(h, f, UPB_HANDLER_ENDSEQ)))) {
+          hashandler = true;
+        }
+
+        if (hashandler && !upb_handlers_getsubhandlers(h, f)) {
+          // For now we add an empty subhandlers in this case.  It makes the
+          // decoder code generator simpler, because it only has to handle two
+          // cases (submessage has handlers or not) as opposed to three
+          // (submessage has handlers in enclosing message but no subhandlers).
+          //
+          // This makes parsing less efficient in the case that we want to
+          // notice a submessage but skip its contents (like if we're testing
+          // for submessage presence or counting the number of repeated
+          // submessages).  In this case we will end up parsing the submessage
+          // field by field and throwing away the results for each, instead of
+          // skipping the whole delimited thing at once.  If this is an issue we
+          // can revisit it, but do remember that this only arises when you have
+          // handlers (startseq/startsubmsg/endsubmsg/endseq) set for the
+          // submessage but no subhandlers.  The uses cases for this are
+          // limited.
+          upb_handlers *sub = upb_handlers_new(upb_fielddef_msgsubdef(f), &sub);
+          upb_handlers_setsubhandlers(h, f, sub);
+          upb_handlers_unref(sub, &sub);
+        }
+
         // TODO(haberman): check type of submessage.
         // This is slightly tricky; also consider whether we should check that
         // they match at setsubhandlers time.
@@ -470,16 +506,27 @@ bool upb_handlers_getselector(const upb_fielddef *f, upb_handlertype_t type,
       *s = f->selector_base;
       break;
     case UPB_HANDLER_STRING:
-      if (!upb_fielddef_isstring(f)) return false;
-      *s = f->selector_base;
+      if (upb_fielddef_isstring(f)) {
+        *s = f->selector_base;
+      } else if (upb_fielddef_issubmsg(f)) {
+        *s = f->selector_base + 3;
+      } else {
+        return false;
+      }
       break;
     case UPB_HANDLER_STARTSTR:
-      if (!upb_fielddef_isstring(f)) return false;
-      *s = f->selector_base + 1;
+      if (upb_fielddef_isstring(f) || upb_fielddef_issubmsg(f)) {
+        *s = f->selector_base + 1;
+      } else {
+        return false;
+      }
       break;
     case UPB_HANDLER_ENDSTR:
-      if (!upb_fielddef_isstring(f)) return false;
-      *s = f->selector_base + 2;
+      if (upb_fielddef_isstring(f) || upb_fielddef_issubmsg(f)) {
+        *s = f->selector_base + 2;
+      } else {
+        return false;
+      }
       break;
     case UPB_HANDLER_STARTSEQ:
       if (!upb_fielddef_isseq(f)) return false;
@@ -501,7 +548,6 @@ bool upb_handlers_getselector(const upb_fielddef *f, upb_handlertype_t type,
       if (!upb_fielddef_issubmsg(f)) return false;
       *s = f->selector_base;
       break;
-    // Subhandler slot is selector_base + 2.
   }
   assert(*s < upb_fielddef_containingtype(f)->selector_count);
   return true;
@@ -514,9 +560,15 @@ 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_isseq(f)) ret += 2;    // STARTSEQ/ENDSEQ
-  if (upb_fielddef_isstring(f)) ret += 2; // [STARTSTR]/STRING/ENDSTR
-   // ENDSUBMSG (STARTSUBMSG is at table beginning)
-  if (upb_fielddef_issubmsg(f)) ret += 0;
+  if (upb_fielddef_isstring(f)) ret += 2; // [STRING]/STARTSTR/ENDSTR
+  if (upb_fielddef_issubmsg(f)) {
+    // ENDSUBMSG (STARTSUBMSG is at table beginning)
+    ret += 0;
+    if (upb_fielddef_lazy(f)) {
+      // STARTSTR/ENDSTR/STRING (for lazy)
+      ret += 3;
+    }
+  }
   return ret;
 }
 

upb/pb/compile_decoder.c

@@ -19,13 +19,6 @@
 #define MAXLABEL 5
 #define EMPTYLABEL -1
 
-static const void *methodkey(const upb_msgdef *md, const upb_handlers *h) {
-  const void *ret = h ? (const void*)h : (const void*)md;
-  assert(ret);
-  return ret;
-}
-
-
 /* mgroup *********************************************************************/
 
 static void freegroup(upb_refcounted *r) {
@@ -80,10 +73,8 @@ static void visitmethod(const upb_refcounted *r, upb_refcounted_visit *visit,
   visit(r, m->group, closure);
 }
 
-static upb_pbdecodermethod *newmethod(const upb_msgdef *msg,
-                                      const upb_handlers *dest_handlers,
-                                      mgroup *group,
-                                      const void *key) {
+static upb_pbdecodermethod *newmethod(const upb_handlers *dest_handlers,
+                                      mgroup *group) {
   static const struct upb_refcounted_vtbl vtbl = {visitmethod, freemethod};
   upb_pbdecodermethod *ret = malloc(sizeof(*ret));
   upb_refcounted_init(UPB_UPCAST(ret), &vtbl, &ret);
@@ -92,11 +83,10 @@ static upb_pbdecodermethod *newmethod(const upb_msgdef *msg,
   // The method references the group and vice-versa, in a circular reference.
   upb_ref2(ret, group);
   upb_ref2(group, ret);
-  upb_inttable_insertptr(&group->methods, key, upb_value_ptr(ret));  // Owns ref
+  upb_inttable_insertptr(&group->methods, dest_handlers, upb_value_ptr(ret));
   upb_refcounted_unref(UPB_UPCAST(ret), &ret);
 
   ret->group = UPB_UPCAST(group);
-  ret->schema_ = msg;
   ret->dest_handlers_ = dest_handlers;
   ret->is_native_ = false;  // If we JIT, it will update this later.
   upb_inttable_init(&ret->dispatch, UPB_CTYPE_UINT64);
@@ -126,10 +116,6 @@ void upb_pbdecodermethod_checkref(const upb_pbdecodermethod *m,
   upb_refcounted_checkref(UPB_UPCAST(m), owner);
 }
 
-const upb_msgdef *upb_pbdecodermethod_schema(const upb_pbdecodermethod *m) {
-  return m->schema_;
-}
-
 const upb_handlers *upb_pbdecodermethod_desthandlers(
     const upb_pbdecodermethod *m) {
   return m->dest_handlers_;
@@ -144,12 +130,12 @@ bool upb_pbdecodermethod_isnative(const upb_pbdecodermethod *m) {
   return m->is_native_;
 }
 
-const upb_pbdecodermethod *upb_pbdecodermethod_newfordesthandlers(
-    const upb_handlers *dest, const void *owner) {
+const upb_pbdecodermethod *upb_pbdecodermethod_new(
+    const upb_pbdecodermethodopts *opts, const void *owner) {
   upb_pbcodecache cache;
   upb_pbcodecache_init(&cache);
   const upb_pbdecodermethod *ret =
-      upb_pbcodecache_getdecodermethodfordesthandlers(&cache, dest);
+      upb_pbcodecache_getdecodermethod(&cache, opts);
   upb_pbdecodermethod_ref(ret, owner);
   upb_pbcodecache_uninit(&cache);
   return ret;
@@ -165,11 +151,15 @@ typedef struct {
   uint32_t *pc;
   int fwd_labels[MAXLABEL];
   int back_labels[MAXLABEL];
+
+  // For fields marked "lazy", parse them lazily or eagerly?
+  bool lazy;
 } compiler;
 
-static compiler *newcompiler(mgroup *group) {
+static compiler *newcompiler(mgroup *group, bool lazy) {
   compiler *ret = malloc(sizeof(*ret));
   ret->group = group;
+  ret->lazy = lazy;
   for (int i = 0; i < MAXLABEL; i++) {
     ret->fwd_labels[i] = EMPTYLABEL;
     ret->back_labels[i] = EMPTYLABEL;
@@ -300,11 +290,11 @@ static void putop(compiler *c, opcode op, ...) {
     }
     case OP_STARTMSG:
     case OP_ENDMSG:
-    case OP_PUSHTAGDELIM:
     case OP_PUSHLENDELIM:
     case OP_POP:
     case OP_SETDELIM:
     case OP_HALT:
+    case OP_RET:
       put32(c, op);
       break;
     case OP_PARSE_DOUBLE:
@@ -321,13 +311,13 @@ static void putop(compiler *c, opcode op, ...) {
     case OP_PARSE_SINT32:
     case OP_PARSE_SINT64:
     case OP_STARTSEQ:
-    case OP_SETGROUPNUM:
     case OP_ENDSEQ:
     case OP_STARTSUBMSG:
     case OP_ENDSUBMSG:
     case OP_STARTSTR:
     case OP_STRING:
     case OP_ENDSTR:
+    case OP_PUSHTAGDELIM:
       put32(c, op | va_arg(ap, upb_selector_t) << 8);
       break;
     case OP_SETBIGGROUPNUM:
@@ -382,10 +372,10 @@ const char *upb_pbdecoder_getopname(unsigned int op) {
     T(DOUBLE), T(FLOAT), T(INT64), T(UINT64), T(INT32), T(FIXED64), T(FIXED32),
     T(BOOL), T(UINT32), T(SFIXED32), T(SFIXED64), T(SINT32), T(SINT64),
     OP(STARTMSG), OP(ENDMSG), OP(STARTSEQ), OP(ENDSEQ), OP(STARTSUBMSG),
-    OP(ENDSUBMSG), OP(STARTSTR), OP(STRING), OP(ENDSTR), OP(CALL),
+    OP(ENDSUBMSG), OP(STARTSTR), OP(STRING), OP(ENDSTR), OP(CALL), OP(RET),
     OP(PUSHLENDELIM), OP(PUSHTAGDELIM), OP(SETDELIM), OP(CHECKDELIM),
     OP(BRANCH), OP(TAG1), OP(TAG2), OP(TAGN), OP(SETDISPATCH), OP(POP),
-    OP(SETGROUPNUM), OP(SETBIGGROUPNUM), OP(HALT),
+    OP(SETBIGGROUPNUM), OP(HALT),
   };
   return op > OP_HALT ? names[0] : names[op];
 #undef OP
@@ -413,16 +403,17 @@ static void dumpbc(uint32_t *p, uint32_t *end, FILE *f) {
         const upb_pbdecodermethod *method =
             (void *)((char *)dispatch -
                      offsetof(upb_pbdecodermethod, dispatch));
-        fprintf(f, " %s", upb_msgdef_fullname(method->schema_));
+        fprintf(f, " %s", upb_msgdef_fullname(
+                              upb_handlers_msgdef(method->dest_handlers_)));
         break;
       }
       case OP_STARTMSG:
       case OP_ENDMSG:
       case OP_PUSHLENDELIM:
-      case OP_PUSHTAGDELIM:
       case OP_POP:
       case OP_SETDELIM:
       case OP_HALT:
+      case OP_RET:
         break;
       case OP_PARSE_DOUBLE:
       case OP_PARSE_FLOAT:
@@ -444,7 +435,7 @@ static void dumpbc(uint32_t *p, uint32_t *end, FILE *f) {
       case OP_STARTSTR:
       case OP_STRING:
       case OP_ENDSTR:
-      case OP_SETGROUPNUM:
+      case OP_PUSHTAGDELIM:
         fprintf(f, " %d", instr >> 8);
         break;
       case OP_SETBIGGROUPNUM:
@@ -537,11 +528,11 @@ static void putpush(compiler *c, const upb_fielddef *f) {
     putop(c, OP_PUSHLENDELIM);
   } else {
     uint32_t fn = upb_fielddef_number(f);
-    putop(c, OP_PUSHTAGDELIM);
     if (fn >= 1 << 24) {
+      putop(c, OP_PUSHTAGDELIM, 0);
       putop(c, OP_SETBIGGROUPNUM, fn);
     } else {
-      putop(c, OP_SETGROUPNUM, fn);
+      putop(c, OP_PUSHTAGDELIM, fn);
     }
   }
 }
@@ -549,13 +540,35 @@ static void putpush(compiler *c, const upb_fielddef *f) {
 static upb_pbdecodermethod *find_submethod(const compiler *c,
                                            const upb_pbdecodermethod *method,
                                            const upb_fielddef *f) {
-  const upb_handlers *sub = method->dest_handlers_ ?
-      upb_handlers_getsubhandlers(method->dest_handlers_, f) : NULL;
-  const void *key = methodkey(upb_downcast_msgdef(upb_fielddef_subdef(f)), sub);
+  const upb_handlers *sub =
+      upb_handlers_getsubhandlers(method->dest_handlers_, f);
   upb_value v;
-  bool ok = upb_inttable_lookupptr(&c->group->methods, key, &v);
-  UPB_ASSERT_VAR(ok, ok);
-  return upb_value_getptr(v);
+  return upb_inttable_lookupptr(&c->group->methods, sub, &v)
+             ? upb_value_getptr(v)
+             : NULL;
+}
+
+static void putsel(compiler *c, opcode op, upb_selector_t sel,
+                   const upb_handlers *h) {
+  if (upb_handlers_gethandler(h, sel)) {
+    putop(c, op, sel);
+  }
+}
+
+// Puts an opcode to call a callback, but only if a callback actually exists for
+// this field and handler type.
+static void putcb(compiler *c, opcode op, const upb_handlers *h,
+                  const upb_fielddef *f, upb_handlertype_t type) {
+  putsel(c, op, getsel(f, type), h);
+}
+
+static bool haslazyhandlers(const upb_handlers *h, const upb_fielddef *f) {
+  if (!upb_fielddef_lazy(f))
+    return false;
+
+  return upb_handlers_gethandler(h, getsel(f, UPB_HANDLER_STARTSTR)) ||
+         upb_handlers_gethandler(h, getsel(f, UPB_HANDLER_STRING)) ||
+         upb_handlers_gethandler(h, getsel(f, UPB_HANDLER_ENDSTR));
 }
 
 // Adds bytecode for parsing the given message to the given decoderplan,
@@ -596,177 +609,178 @@ static void compile_method(compiler *c, upb_pbdecodermethod *method) {
   upb_inttable_uninit(&method->dispatch);
   upb_inttable_init(&method->dispatch, UPB_CTYPE_UINT64);
 
+  const upb_handlers *h = upb_pbdecodermethod_desthandlers(method);
+  const upb_msgdef *md = upb_handlers_msgdef(h);
+
  method->code_base.ofs = pcofs(c);
   putop(c, OP_SETDISPATCH, &method->dispatch);
-  putop(c, OP_STARTMSG);
+  putsel(c, OP_STARTMSG, UPB_STARTMSG_SELECTOR, h);
  label(c, LABEL_FIELD);
   upb_msg_iter i;
-  for(upb_msg_begin(&i, method->schema_); !upb_msg_done(&i); upb_msg_next(&i)) {
+  for(upb_msg_begin(&i, md); !upb_msg_done(&i); upb_msg_next(&i)) {
     const upb_fielddef *f = upb_msg_iter_field(&i);
-    upb_descriptortype_t type = upb_fielddef_descriptortype(f);
+    upb_descriptortype_t descriptor_type = upb_fielddef_descriptortype(f);
+    upb_fieldtype_t type = upb_fielddef_type(f);
 
     // From a decoding perspective, ENUM is the same as INT32.
-    if (type == UPB_DESCRIPTOR_TYPE_ENUM)
-      type = UPB_DESCRIPTOR_TYPE_INT32;
-
-    label(c, LABEL_FIELD);
-
-    switch (upb_fielddef_type(f)) {
-      case UPB_TYPE_MESSAGE: {
-        const upb_pbdecodermethod *sub_m = find_submethod(c, method, f);
-        int wire_type = (type == UPB_DESCRIPTOR_TYPE_MESSAGE) ?
-            UPB_WIRE_TYPE_DELIMITED : UPB_WIRE_TYPE_START_GROUP;
-        if (upb_fielddef_isseq(f)) {
-          putop(c, OP_CHECKDELIM, LABEL_ENDMSG);
-          putchecktag(c, f, wire_type, LABEL_DISPATCH);
-         dispatchtarget(c, method, f, wire_type);
-          putop(c, OP_PUSHTAGDELIM);
-          putop(c, OP_STARTSEQ, getsel(f, UPB_HANDLER_STARTSEQ));
-         label(c, LABEL_LOOPSTART);
-          putpush(c, f);
-          putop(c, OP_STARTSUBMSG, getsel(f, UPB_HANDLER_STARTSUBMSG));
-          putop(c, OP_CALL, sub_m);
-          putop(c, OP_POP);
-          putop(c, OP_ENDSUBMSG, getsel(f, UPB_HANDLER_ENDSUBMSG));
-          if (wire_type == UPB_WIRE_TYPE_DELIMITED) {
-            putop(c, OP_SETDELIM);
-          }
-          putop(c, OP_CHECKDELIM, LABEL_LOOPBREAK);
-          putchecktag(c, f, wire_type, LABEL_LOOPBREAK);
-          putop(c, OP_BRANCH, -LABEL_LOOPSTART);
-         label(c, LABEL_LOOPBREAK);
-          putop(c, OP_POP);
-          putop(c, OP_ENDSEQ, getsel(f, UPB_HANDLER_ENDSEQ));
-        } else {
-          putop(c, OP_CHECKDELIM, LABEL_ENDMSG);
-          putchecktag(c, f, wire_type, LABEL_DISPATCH);
-         dispatchtarget(c, method, f, wire_type);
-          putpush(c, f);
-          putop(c, OP_STARTSUBMSG, getsel(f, UPB_HANDLER_STARTSUBMSG));
-          putop(c, OP_CALL, sub_m);
-          putop(c, OP_POP);
-          putop(c, OP_ENDSUBMSG, getsel(f, UPB_HANDLER_ENDSUBMSG));
-          if (wire_type == UPB_WIRE_TYPE_DELIMITED) {
-            putop(c, OP_SETDELIM);
-          }
-        }
-        break;
+    if (descriptor_type == UPB_DESCRIPTOR_TYPE_ENUM)
+      descriptor_type = UPB_DESCRIPTOR_TYPE_INT32;
+
+    if (type == UPB_TYPE_MESSAGE && !(haslazyhandlers(h, f) && c->lazy)) {
+      const upb_pbdecodermethod *sub_m = find_submethod(c, method, f);
+      if (!sub_m) {
+        // Don't emit any code for this field at all; it will be parsed as an
+        // unknown field.
+        continue;
       }
-      case UPB_TYPE_STRING:
-      case UPB_TYPE_BYTES:
-        if (upb_fielddef_isseq(f)) {
-          putop(c, OP_CHECKDELIM, LABEL_ENDMSG);
-          putchecktag(c, f, UPB_WIRE_TYPE_DELIMITED, LABEL_DISPATCH);
-         dispatchtarget(c, method, f, UPB_WIRE_TYPE_DELIMITED);
-          putop(c, OP_PUSHTAGDELIM);
-          putop(c, OP_STARTSEQ, getsel(f, UPB_HANDLER_STARTSEQ));
-         label(c, LABEL_LOOPSTART);
-          putop(c, OP_PUSHLENDELIM);
-          putop(c, OP_STARTSTR, getsel(f, UPB_HANDLER_STARTSTR));
-          putop(c, OP_STRING, getsel(f, UPB_HANDLER_STRING));
-          putop(c, OP_POP);
-          putop(c, OP_ENDSTR, getsel(f, UPB_HANDLER_ENDSTR));
-          putop(c, OP_SETDELIM);
-          putop(c, OP_CHECKDELIM, LABEL_LOOPBREAK);
-          putchecktag(c, f, UPB_WIRE_TYPE_DELIMITED, LABEL_LOOPBREAK);
-          putop(c, OP_BRANCH, -LABEL_LOOPSTART);
-         label(c, LABEL_LOOPBREAK);
-          putop(c, OP_POP);
-          putop(c, OP_ENDSEQ, getsel(f, UPB_HANDLER_ENDSEQ));
-        } else {
-          putop(c, OP_CHECKDELIM, LABEL_ENDMSG);
-          putchecktag(c, f, UPB_WIRE_TYPE_DELIMITED, LABEL_DISPATCH);
-         dispatchtarget(c, method, f, UPB_WIRE_TYPE_DELIMITED);
-          putop(c, OP_PUSHLENDELIM);
-          putop(c, OP_STARTSTR, getsel(f, UPB_HANDLER_STARTSTR));
-          putop(c, OP_STRING, getsel(f, UPB_HANDLER_STRING));
-          putop(c, OP_POP);
-          putop(c, OP_ENDSTR, getsel(f, UPB_HANDLER_ENDSTR));
+
+      label(c, LABEL_FIELD);
+
+      int wire_type = (descriptor_type == UPB_DESCRIPTOR_TYPE_MESSAGE)
+                          ? UPB_WIRE_TYPE_DELIMITED
+                          : UPB_WIRE_TYPE_START_GROUP;
+      if (upb_fielddef_isseq(f)) {
+        putop(c, OP_CHECKDELIM, LABEL_ENDMSG);
+        putchecktag(c, f, wire_type, LABEL_DISPATCH);
+       dispatchtarget(c, method, f, wire_type);
+        putop(c, OP_PUSHTAGDELIM, 0);
+        putop(c, OP_STARTSEQ, getsel(f, UPB_HANDLER_STARTSEQ));
+       label(c, LABEL_LOOPSTART);
+        putpush(c, f);
+        putop(c, OP_STARTSUBMSG, getsel(f, UPB_HANDLER_STARTSUBMSG));
+        putop(c, OP_CALL, sub_m);
+        putop(c, OP_POP);
+        putcb(c, OP_ENDSUBMSG, h, f, UPB_HANDLER_ENDSUBMSG);
+        if (wire_type == UPB_WIRE_TYPE_DELIMITED) {
           putop(c, OP_SETDELIM);
         }
-        break;
-      default: {
-        opcode parse_type = (opcode)type;
-        assert((int)parse_type >= 0 && parse_type <= OP_MAX);
-        upb_selector_t sel = getsel(f, upb_handlers_getprimitivehandlertype(f));
-        int wire_type = native_wire_types[upb_fielddef_descriptortype(f)];
-        if (upb_fielddef_isseq(f)) {
-          putop(c, OP_CHECKDELIM, LABEL_ENDMSG);
-          putchecktag(c, f, UPB_WIRE_TYPE_DELIMITED, LABEL_DISPATCH);
-         dispatchtarget(c, method, f, UPB_WIRE_TYPE_DELIMITED);
-          putop(c, OP_PUSHLENDELIM);
-          putop(c, OP_STARTSEQ, getsel(f, UPB_HANDLER_STARTSEQ));  // Packed
-         label(c, LABEL_LOOPSTART);
-          putop(c, parse_type, sel);
-          putop(c, OP_CHECKDELIM, LABEL_LOOPBREAK);
-          putop(c, OP_BRANCH, -LABEL_LOOPSTART);
-         dispatchtarget(c, method, f, wire_type);
-          putop(c, OP_PUSHTAGDELIM);
-          putop(c, OP_STARTSEQ, getsel(f, UPB_HANDLER_STARTSEQ));  // Non-packed
-         label(c, LABEL_LOOPSTART);
-          putop(c, parse_type, sel);
-          putop(c, OP_CHECKDELIM, LABEL_LOOPBREAK);
-          putchecktag(c, f, wire_type, LABEL_LOOPBREAK);
-          putop(c, OP_BRANCH, -LABEL_LOOPSTART);
-         label(c, LABEL_LOOPBREAK);
-          putop(c, OP_POP);  // Packed and non-packed join.
-          putop(c, OP_ENDSEQ, getsel(f, UPB_HANDLER_ENDSEQ));
-          putop(c, OP_SETDELIM);  // Could remove for non-packed by dup ENDSEQ.
-        } else {
-          putop(c, OP_CHECKDELIM, LABEL_ENDMSG);
-          putchecktag(c, f, wire_type, LABEL_DISPATCH);
-         dispatchtarget(c, method, f, wire_type);
-          putop(c, parse_type, sel);
+        putop(c, OP_CHECKDELIM, LABEL_LOOPBREAK);
+        putchecktag(c, f, wire_type, LABEL_LOOPBREAK);
+        putop(c, OP_BRANCH, -LABEL_LOOPSTART);
+       label(c, LABEL_LOOPBREAK);
+        putop(c, OP_POP);
+        putcb(c, OP_ENDSEQ, h, f, UPB_HANDLER_ENDSEQ);
+      } else {
+        putop(c, OP_CHECKDELIM, LABEL_ENDMSG);
+        putchecktag(c, f, wire_type, LABEL_DISPATCH);
+       dispatchtarget(c, method, f, wire_type);
+        putpush(c, f);
+        putop(c, OP_STARTSUBMSG, getsel(f, UPB_HANDLER_STARTSUBMSG));
+        putop(c, OP_CALL, sub_m);
+        putop(c, OP_POP);
+        putcb(c, OP_ENDSUBMSG, h, f, UPB_HANDLER_ENDSUBMSG);
+        if (wire_type == UPB_WIRE_TYPE_DELIMITED) {
+          putop(c, OP_SETDELIM);
         }
       }
+    } else if (type == UPB_TYPE_STRING || type == UPB_TYPE_BYTES ||
+               type == UPB_TYPE_MESSAGE) {
+      label(c, LABEL_FIELD);
+      if (upb_fielddef_isseq(f)) {
+        putop(c, OP_CHECKDELIM, LABEL_ENDMSG);
+        putchecktag(c, f, UPB_WIRE_TYPE_DELIMITED, LABEL_DISPATCH);
+       dispatchtarget(c, method, f, UPB_WIRE_TYPE_DELIMITED);
+        putop(c, OP_PUSHTAGDELIM, 0);
+        putop(c, OP_STARTSEQ, getsel(f, UPB_HANDLER_STARTSEQ));
+       label(c, LABEL_LOOPSTART);
+        putop(c, OP_PUSHLENDELIM);
+        putop(c, OP_STARTSTR, getsel(f, UPB_HANDLER_STARTSTR));
+        // Need to emit even if no handler to skip past the string.
+        putop(c, OP_STRING, getsel(f, UPB_HANDLER_STRING));
+        putop(c, OP_POP);
+        putcb(c, OP_ENDSTR, h, f, UPB_HANDLER_ENDSTR);
+        putop(c, OP_SETDELIM);
+        putop(c, OP_CHECKDELIM, LABEL_LOOPBREAK);
+        putchecktag(c, f, UPB_WIRE_TYPE_DELIMITED, LABEL_LOOPBREAK);
+        putop(c, OP_BRANCH, -LABEL_LOOPSTART);
+       label(c, LABEL_LOOPBREAK);
+        putop(c, OP_POP);
+        putcb(c, OP_ENDSEQ, h, f, UPB_HANDLER_ENDSEQ);
+      } else {
+        putop(c, OP_CHECKDELIM, LABEL_ENDMSG);
+        putchecktag(c, f, UPB_WIRE_TYPE_DELIMITED, LABEL_DISPATCH);
+       dispatchtarget(c, method, f, UPB_WIRE_TYPE_DELIMITED);
+        putop(c, OP_PUSHLENDELIM);
+        putop(c, OP_STARTSTR, getsel(f, UPB_HANDLER_STARTSTR));
+        putop(c, OP_STRING, getsel(f, UPB_HANDLER_STRING));
+        putop(c, OP_POP);
+        putcb(c, OP_ENDSTR, h, f, UPB_HANDLER_ENDSTR);
+        putop(c, OP_SETDELIM);
+      }
+    } else {
+      label(c, LABEL_FIELD);
+      opcode parse_type = (opcode)descriptor_type;
+      assert((int)parse_type >= 0 && parse_type <= OP_MAX);
+      upb_selector_t sel = getsel(f, upb_handlers_getprimitivehandlertype(f));
+      int wire_type = native_wire_types[upb_fielddef_descriptortype(f)];
+      if (upb_fielddef_isseq(f)) {
+        putop(c, OP_CHECKDELIM, LABEL_ENDMSG);
+        putchecktag(c, f, UPB_WIRE_TYPE_DELIMITED, LABEL_DISPATCH);
+       dispatchtarget(c, method, f, UPB_WIRE_TYPE_DELIMITED);
+        putop(c, OP_PUSHLENDELIM);
+        putop(c, OP_STARTSEQ, getsel(f, UPB_HANDLER_STARTSEQ));  // Packed
+       label(c, LABEL_LOOPSTART);
+        putop(c, parse_type, sel);
+        putop(c, OP_CHECKDELIM, LABEL_LOOPBREAK);
+        putop(c, OP_BRANCH, -LABEL_LOOPSTART);
+       dispatchtarget(c, method, f, wire_type);
+        putop(c, OP_PUSHTAGDELIM, 0);
+        putop(c, OP_STARTSEQ, getsel(f, UPB_HANDLER_STARTSEQ));  // Non-packed
+       label(c, LABEL_LOOPSTART);
+        putop(c, parse_type, sel);
+        putop(c, OP_CHECKDELIM, LABEL_LOOPBREAK);
+        putchecktag(c, f, wire_type, LABEL_LOOPBREAK);
+        putop(c, OP_BRANCH, -LABEL_LOOPSTART);
+       label(c, LABEL_LOOPBREAK);
+        putop(c, OP_POP);  // Packed and non-packed join.
+        putcb(c, OP_ENDSEQ, h, f, UPB_HANDLER_ENDSEQ);
+        putop(c, OP_SETDELIM);  // Could remove for non-packed by dup ENDSEQ.
+      } else {
+        putop(c, OP_CHECKDELIM, LABEL_ENDMSG);
+        putchecktag(c, f, wire_type, LABEL_DISPATCH);
+       dispatchtarget(c, method, f, wire_type);
+        putop(c, parse_type, sel);
+      }
     }
   }
+
   // For now we just loop back to the last field of the message (or if none,
   // the DISPATCH opcode for the message.
   putop(c, OP_BRANCH, -LABEL_FIELD);
+
+  // Insert both a label and a dispatch table entry for this end-of-msg.
  label(c, LABEL_ENDMSG);
-  putop(c, OP_ENDMSG);
+  upb_value val = upb_value_uint64(pcofs(c) - method->code_base.ofs);
+  upb_inttable_insert(&method->dispatch, DISPATCH_ENDMSG, val);
+
+  putsel(c, OP_ENDMSG, UPB_ENDMSG_SELECTOR, h);
+  putop(c, OP_RET);
 
   upb_inttable_compact(&method->dispatch);
 }
 
-// Populate "methods" with new upb_pbdecodermethod objects reachable from "md".
-// "h" can be NULL, in which case the methods will not be statically bound to
-// destination handlers.
-//
-// Returns the method for this msgdef/handlers.
+// Populate "methods" with new upb_pbdecodermethod objects reachable from "h".
+// Returns the method for these handlers.
 //
-// Note that there is a deep difference between keying the method table on
-// upb_msgdef and keying it on upb_handlers.  Since upb_msgdef : upb_handlers
-// can be 1:many, binding a handlers statically can result in *more* methods
-// being generated than if the methods are dynamically-bound.
-//
-// On the other hand, if/when the optimization mentioned below is implemented,
-// binding to a upb_handlers can result in *fewer* methods being generated if
-// many of the submessages have no handlers bound to them.
-static void find_methods(compiler *c, const upb_msgdef *md,
-                         const upb_handlers *h) {
-  const void *key = methodkey(md, h);
+// Generates a new method for every destination handlers reachable from "h".
+static void find_methods(compiler *c, const upb_handlers *h) {
   upb_value v;
-  if (upb_inttable_lookupptr(&c->group->methods, key, &v))
+  if (upb_inttable_lookupptr(&c->group->methods, h, &v))
     return;
-  newmethod(md, h, c->group, key);
+  newmethod(h, c->group);
 
   // Find submethods.
   upb_msg_iter i;
+  const upb_msgdef *md = upb_handlers_msgdef(h);
   for(upb_msg_begin(&i, md); !upb_msg_done(&i); upb_msg_next(&i)) {
     const upb_fielddef *f = upb_msg_iter_field(&i);
-    if (upb_fielddef_type(f) != UPB_TYPE_MESSAGE)
-      continue;
-    const upb_handlers *sub_h = h ? upb_handlers_getsubhandlers(h, f) : NULL;
-
-    if (h && !sub_h &&
-        upb_fielddef_descriptortype(f) == UPB_DESCRIPTOR_TYPE_MESSAGE) {
-      // OPT: We could optimize away the sub-method, but would have to make sure
-      // this field is compiled as a string instead of a submessage.
+    const upb_handlers *sub_h;
+    if (upb_fielddef_type(f) == UPB_TYPE_MESSAGE &&
+        (sub_h = upb_handlers_getsubhandlers(h, f)) != NULL) {
+      // We only generate a decoder method for submessages with handlers.
+      // Others will be parsed as unknown fields.
+      find_methods(c, sub_h);
     }
-
-    find_methods(c, upb_downcast_msgdef(upb_fielddef_subdef(f)), sub_h);
   }
 }
 
@@ -814,12 +828,6 @@ static void sethandlers(mgroup *g, bool allowjit) {
   }
 }
 
-static bool bind_dynamic(bool allowjit) {
-  // For the moment, JIT handlers always bind statically, but bytecode handlers
-  // never do.
-  return !allowjit;
-}
-
 #else  // UPB_USE_JIT_X64
 
 static void sethandlers(mgroup *g, bool allowjit) {
@@ -828,33 +836,19 @@ static void sethandlers(mgroup *g, bool allowjit) {
   set_bytecode_handlers(g);
 }
 
-static bool bind_dynamic(bool allowjit) {
-  // Bytecode handlers never bind statically.
-  UPB_UNUSED(allowjit);
-  return true;
-}
-
 #endif  // UPB_USE_JIT_X64
 
 
 // TODO(haberman): allow this to be constructed for an arbitrary set of dest
 // handlers and other mgroups (but verify we have a transitive closure).
-const mgroup *mgroup_new(const upb_handlers *dest, bool allowjit,
+const mgroup *mgroup_new(const upb_handlers *dest, bool allowjit, bool lazy,
                          const void *owner) {
   UPB_UNUSED(allowjit);
   assert(upb_handlers_isfrozen(dest));
-  const upb_msgdef *md = upb_handlers_msgdef(dest);
 
   mgroup *g = newgroup(owner);
-  compiler *c = newcompiler(g);
-
-  if (bind_dynamic(allowjit)) {
-    // If binding dynamically, remove the reference against destination
-    // handlers.
-    dest = NULL;
-  }
-
-  find_methods(c, md, dest);
+  compiler *c = newcompiler(g, lazy);
+  find_methods(c, dest);
 
   // We compile in two passes:
   // 1. all messages are assigned relative offsets from the beginning of the
@@ -909,20 +903,28 @@ bool upb_pbcodecache_setallowjit(upb_pbcodecache *c, bool allow) {
   return true;
 }
 
-const upb_pbdecodermethod *upb_pbcodecache_getdecodermethodfordesthandlers(
-    upb_pbcodecache *c, const upb_handlers *handlers) {
+const upb_pbdecodermethod *upb_pbcodecache_getdecodermethod(
+    upb_pbcodecache *c, const upb_pbdecodermethodopts *opts) {
   // Right now we build a new DecoderMethod every time.
   // TODO(haberman): properly cache methods by their true key.
-  const mgroup *g = mgroup_new(handlers, c->allow_jit_, c);
+  const mgroup *g = mgroup_new(opts->handlers, c->allow_jit_, opts->lazy, c);
   upb_inttable_push(&c->groups, upb_value_constptr(g));
 
-  const upb_msgdef *md = upb_handlers_msgdef(handlers);
-  if (bind_dynamic(c->allow_jit_)) {
-    handlers = NULL;
-  }
-
   upb_value v;
-  bool ok = upb_inttable_lookupptr(&g->methods, methodkey(md, handlers), &v);
+  bool ok = upb_inttable_lookupptr(&g->methods, opts->handlers, &v);
   UPB_ASSERT_VAR(ok, ok);
   return upb_value_getptr(v);
 }
+
+
+/* upb_pbdecodermethodopts ****************************************************/
+
+void upb_pbdecodermethodopts_init(upb_pbdecodermethodopts *opts,
+                                  const upb_handlers *h) {
+  opts->handlers = h;
+  opts->lazy = false;
+}
+
+void upb_pbdecodermethodopts_setlazy(upb_pbdecodermethodopts *opts, bool lazy) {
+  opts->lazy = lazy;
+}

upb/pb/compile_decoder_x64.c

@@ -194,6 +194,8 @@ static void patchdispatch(jitcompiler *jc) {
 }
 
 // Define for JIT debugging.
+//#define UPB_JIT_LOAD_SO
+
 #ifdef UPB_JIT_LOAD_SO
 static void load_so(jitcompiler *jc) {
   // Dump to a .so file in /tmp and load that, so all the tooling works right

upb/pb/compile_decoder_x64.dasc

@@ -42,6 +42,9 @@
 |  mov  DECODER->top, FRAME
 |  mov  DECODER->ptr, PTR
 |  mov  DECODER->data_end, DATAEND
+|  // We don't guarantee that delim_end is NULL when out of range like the
+|  // interpreter does.
+|  mov  DECODER->delim_end, DELIMEND
 |  sub  DELIMEND, DECODER->buf
 |  add  DELIMEND, DECODER->bufstart_ofs
 |  mov  FRAME->end_ofs, DELIMEND
@@ -205,6 +208,8 @@ static void emit_static_asm(jitcompiler *jc) {
   |
   |  mov   DECODER, rdi
   |  callp upb_pbdecoder_resume  // Same args as us; reuse regs.
+  |  test  eax, eax
+  |  jns   >1
   |  mov   DECODER->saved_rsp, rsp
   |  mov   rax, rbx
   |  load_regs
@@ -212,12 +217,13 @@ static void emit_static_asm(jitcompiler *jc) {
   |  // Test whether we have a saved stack to resume.
   |  mov   ARG3_64, DECODER->call_len
   |  test  ARG3_64, ARG3_64
-  |  jnz   >1
+  |  jnz   >2
   |
   |  call  rax
   |
   |  mov   rax, DECODER->size_param
   |  mov   qword DECODER->call_len, 0
+  |1:
   |  add   rsp, 8  // Counter previous alignment.
   |  pop   rbx
   |  pop   r12
@@ -227,7 +233,7 @@ static void emit_static_asm(jitcompiler *jc) {
   |  pop   rbp
   |  ret
   |
-  |1:
+  |2:
   |  // Resume decoder.
   |  lea   ARG2_64, DECODER->callstack
   |  sub   rsp, ARG3_64
@@ -293,6 +299,7 @@ static void emit_static_asm(jitcompiler *jc) {
   |  add   DELIMEND, rdx
   |  cmp   FRAME, DECODER->limit
   |  je    >3   // Stack overflow
+  |  mov   dword FRAME->groupnum, 0
   |  test  rcx, rcx
   |  jz    >2
   |  mov   DATAEND, DECODER->end
@@ -850,24 +857,7 @@ static void jittag(jitcompiler *jc, uint64_t tag, int n, int ofs,
   |5:
 }
 
-// Emit message-specific assembly.  Overall code layout is:
-// +---------------------------------------------------------------------------+
-// | Message A                                                                 |
-// | 1. function prologue (startmsg), jmps to OP_CHECKDELIM_RET before first   |
-// |    OP_TAG* in 4.                                                          |
-// | 2. function epilogue (endmsg), returns from function.                     |
-// | 3. dispatch function (returns fptr to 4)                                  |
-// |  - loops internally to skip unknown fields                                |
-// |  - after each unknown field does OP_CHECKDELIM_RET (returns 2)            |
-// |  - also returns 2 for END_GROUP.
-// | 4. code for each op:                                                      |
-// |  - OP_TAG* on mismatch calls 3 to get addr, then jumps to 4 (or 2 on EOM).|
-// |  - OP_CHECKDELIM_RET jumps to 2                                           |
-// +---------------------------------------------------------------------------+
-// | Message B                                                                 |
-// | 1. ...                                                                    |
-// | ...                                                                       |
-// +---------------------------------------------------------------------------+
+// Compile the bytecode to x64.
 static void jitbytecode(jitcompiler *jc) {
   upb_pbdecodermethod *method = NULL;
   const upb_handlers *h = NULL;
@@ -877,20 +867,21 @@ static void jitbytecode(jitcompiler *jc) {
     uint32_t arg = instr >> 8;
     int32_t longofs = arg;
 
-    if (op != OP_STARTMSG && op != OP_SETDISPATCH) {
+    if (op != OP_SETDISPATCH) {
+      // Skipped for SETDISPATCH because it defines its own asmlabel for the
+      // dispatch code it emits.
       asmlabel(jc, "0x%lx.%s", pcofs(jc), upb_pbdecoder_getopname(op));
+
+      // Skipped for SETDISPATCH because it should point at the function
+      // prologue, not the dispatch function that is emitted first.
+      // TODO: optimize this to only define pclabels that are actually used.
+      |=>define_pclabel(jc, jc->pc):
     }
-    // TODO: optimize this to only define pclabels that are actually used.
-    |=>define_pclabel(jc, jc->pc):
+
     jc->pc++;
 
     switch (op) {
     case OP_STARTMSG: {
-      // This opcode serves as a function prolouge also.
-      const char *msgname = upb_msgdef_fullname(method->schema_);
-      asmlabel(jc, "0x%lx.parse.%s", pcofs(jc), msgname);
-      |=>define_pclabel(jc, method):
-      |  sub   rsp, 8
       upb_func *startmsg = gethandler(h, UPB_STARTMSG_SELECTOR);
       if (startmsg) {
         // bool startmsg(void *closure, const void *hd)
@@ -905,11 +896,12 @@ static void jitbytecode(jitcompiler *jc) {
           |  jmp   <1
           |2:
         }
+      } else {
+        | nop
       }
       break;
     }
     case OP_ENDMSG: {
-      // This opcode serves as a function epiloue also.
       upb_func *endmsg = gethandler(h, UPB_ENDMSG_SELECTOR);
       |9:
       if (endmsg) {
@@ -919,11 +911,12 @@ static void jitbytecode(jitcompiler *jc) {
         |  mov   ARG3_64, DECODER->status
         |  callp endmsg
       }
-      |  add  rsp, 8
-      |  ret
       break;
     }
     case OP_SETDISPATCH: {
+      uint32_t *op_pc = jc->pc - 1;
+
+      // Load info for new method.
       upb_inttable *dispatch;
       memcpy(&dispatch, jc->pc, sizeof(void*));
       jc->pc += sizeof(void*) / sizeof(uint32_t);
@@ -936,9 +929,18 @@ static void jitbytecode(jitcompiler *jc) {
       // case instead of parsing it field by field.  We should also do the skip
       // in the containing message's code.
       h = method->dest_handlers_;
-      const char *msgname = upb_msgdef_fullname(method->schema_);
+      const char *msgname = upb_msgdef_fullname(upb_handlers_msgdef(h));
+
+      // Emit dispatch code for new method.
       asmlabel(jc, "0x%lx.dispatch.%s", pcofs(jc), msgname);
       jitdispatch(jc, method);
+
+      // Emit function prologue for new method.
+      asmlabel(jc, "0x%lx.parse.%s", pcofs(jc), msgname);
+      |=>define_pclabel(jc, op_pc):
+      |=>define_pclabel(jc, method):
+      |  sub   rsp, 8
+
       break;
     }
     case OP_PARSE_DOUBLE:
@@ -1056,6 +1058,7 @@ static void jitbytecode(jitcompiler *jc) {
       |  add   FRAME, sizeof(upb_pbdecoder_frame)
       |  cmp   FRAME, DECODER->limit
       |  je    ->err
+      |  mov   dword FRAME->groupnum, arg
       break;
     case OP_PUSHLENDELIM:
       |  call  ->pushlendelim
@@ -1075,9 +1078,6 @@ static void jitbytecode(jitcompiler *jc) {
       |  mov   DATAEND, DELIMEND
       |1:
       break;
-    case OP_SETGROUPNUM:
-      |  mov   dword FRAME->groupnum, arg
-      break;
     case OP_SETBIGGROUPNUM:
       |  mov   dword FRAME->groupnum, *jc->pc++
       break;
@@ -1086,11 +1086,16 @@ static void jitbytecode(jitcompiler *jc) {
       |  je   =>pclabel(jc, jc->pc + longofs)
       break;
     case OP_CALL:
-      |  call =>pclabel(jc, jc->pc + longofs + 3)
+      |  call =>pclabel(jc, jc->pc + longofs)
       break;
     case OP_BRANCH:
       |  jmp  =>pclabel(jc, jc->pc + longofs);
       break;
+    case OP_RET:
+      |9:
+      |  add  rsp, 8
+      |  ret
+      break;
     case OP_TAG1:
       jittag(jc, (arg >> 8) & 0xff, 1, (int8_t)arg, method);
       break;
@@ -1107,6 +1112,7 @@ static void jitbytecode(jitcompiler *jc) {
       assert(false);
     }
   }
+
   asmlabel(jc, "eof");
   |  nop
 }

upb/pb/decoder.c

@@ -18,7 +18,6 @@
 #endif
 
 #define CHECK_SUSPEND(x) if (!(x)) return upb_pbdecoder_suspend(d);
-#define CHECK_RETURN(x) { int32_t ret = x; if (ret >= 0) return ret; }
 
 // Error messages that are shared between the bytecode and JIT decoders.
 const char *kPbDecoderStackOverflow = "Nesting too deep.";
@@ -45,10 +44,10 @@ static bool consumes_input(opcode op) {
     case OP_PUSHTAGDELIM:
     case OP_POP:
     case OP_SETDELIM:
-    case OP_SETGROUPNUM:
     case OP_SETBIGGROUPNUM:
     case OP_CHECKDELIM:
     case OP_CALL:
+    case OP_RET:
     case OP_BRANCH:
       return false;
     default:
@@ -147,13 +146,12 @@ static void checkpoint(upb_pbdecoder *d) {
 }
 
 // Resumes the decoder from an initial state or from a previous suspend.
-void *upb_pbdecoder_resume(upb_pbdecoder *d, void *p, const char *buf,
-                           size_t size, const upb_bufhandle *handle) {
+int32_t upb_pbdecoder_resume(upb_pbdecoder *d, void *p, const char *buf,
+                             size_t size, const upb_bufhandle *handle) {
   UPB_UNUSED(p);  // Useless; just for the benefit of the JIT.
   d->buf_param = buf;
   d->size_param = size;
   d->handle = handle;
-  d->skip = 0;
   if (d->residual_end > d->residual) {
     // We have residual bytes from the last buffer.
     assert(ptr(d) == d->residual);
@@ -161,7 +159,11 @@ void *upb_pbdecoder_resume(upb_pbdecoder *d, void *p, const char *buf,
     switchtobuf(d, buf, buf + size);
   }
   d->checkpoint = ptr(d);
-  return d;  // For the JIT.
+  if (d->top->groupnum < 0) {
+    CHECK_RETURN(upb_pbdecoder_skipunknown(d, -1, 0));
+    d->checkpoint = ptr(d);
+  }
+  return DECODE_OK;
 }
 
 // Suspends the decoder at the last checkpoint, without saving any residual
@@ -176,10 +178,10 @@ size_t upb_pbdecoder_suspend(upb_pbdecoder *d) {
     assert(!in_residual_buf(d, d->checkpoint));
     assert(d->buf == d->buf_param);
     size_t consumed = d->checkpoint - d->buf;
-    d->bufstart_ofs += consumed + d->skip;
+    d->bufstart_ofs += consumed;
     d->residual_end = d->residual;
     switchtobuf(d, d->residual, d->residual_end);
-    return consumed + d->skip;
+    return consumed;
   }
 }
 
@@ -209,11 +211,11 @@ static size_t suspend_save(upb_pbdecoder *d) {
     assert(save <= sizeof(d->residual));
     memcpy(d->residual, ptr(d), save);
     d->residual_end = d->residual + save;
-    d->bufstart_ofs = offset(d) + d->skip;
+    d->bufstart_ofs = offset(d);
   }
 
   switchtobuf(d, d->residual, d->residual_end);
-  return d->size_param + d->skip;
+  return d->size_param;
 }
 
 static int32_t skip(upb_pbdecoder *d, size_t bytes) {
@@ -221,12 +223,16 @@ static int32_t skip(upb_pbdecoder *d, size_t bytes) {
   if (curbufleft(d) >= bytes) {
     // Skipped data is all in current buffer.
     advance(d, bytes);
+    return DECODE_OK;
   } else {
     // Skipped data extends beyond currently available buffers.
-    d->skip = bytes - curbufleft(d);
-    advance(d, curbufleft(d));
+    d->pc = d->last;
+    size_t skip = bytes - curbufleft(d);
+    d->bufstart_ofs += (d->end - d->buf) + skip;
+    d->residual_end = d->residual;
+    switchtobuf(d, d->residual, d->residual_end);
+    return d->size_param + skip;
   }
-  return DECODE_OK;
 }
 
 FORCEINLINE void consumebytes(upb_pbdecoder *d, void *buf, size_t bytes) {
@@ -247,8 +253,8 @@ static NOINLINE int32_t getbytes_slow(upb_pbdecoder *d, void *buf,
   if (curbufleft(d) >= bytes) {
     consumebytes(d, buf + avail, bytes);
     return DECODE_OK;
-  } else if (d->data_end - d->buf == d->top->end_ofs - d->bufstart_ofs) {
-    seterr(d, "Submessage ended in the middle of a value");
+  } else if (d->data_end == d->delim_end) {
+    seterr(d, "Submessage ended in the middle of a value or group");
     return upb_pbdecoder_suspend(d);
   } else {
     return suspend_save(d);
@@ -378,11 +384,24 @@ static bool push(upb_pbdecoder *d, uint64_t end) {
   fr++;
   fr->end_ofs = end;
   fr->dispatch = NULL;
-  fr->groupnum = -1;
+  fr->groupnum = 0;
   d->top = fr;
   return true;
 }
 
+static bool pushtagdelim(upb_pbdecoder *d, uint32_t arg) {
+  // While we expect to see an "end" tag (either ENDGROUP or a non-sequence
+  // field number) prior to hitting any enclosing submessage end, pushing our
+  // existing delim end prevents us from continuing to parse values from a
+  // corrupt proto that doesn't give us an END tag in time.
+  if (!push(d, d->top->end_ofs))
+    return false;
+  d->top->groupnum = arg;
+  return true;
+}
+
+static void pop(upb_pbdecoder *d) { d->top--; }
+
 NOINLINE int32_t upb_pbdecoder_checktag_slow(upb_pbdecoder *d,
                                              uint64_t expected) {
   uint64_t data = 0;
@@ -400,46 +419,103 @@ NOINLINE int32_t upb_pbdecoder_checktag_slow(upb_pbdecoder *d,
   }
 }
 
-int32_t upb_pbdecoder_skipunknown(upb_pbdecoder *d, uint32_t fieldnum,
+int32_t upb_pbdecoder_skipunknown(upb_pbdecoder *d, int32_t fieldnum,
                                   uint8_t wire_type) {
-  if (fieldnum == 0 || fieldnum > UPB_MAX_FIELDNUMBER) {
-    seterr(d, "Invalid field number");
-    return upb_pbdecoder_suspend(d);
-  }
-
-  if (wire_type == UPB_WIRE_TYPE_END_GROUP) {
-    if (fieldnum != d->top->groupnum) {
-      seterr(d, "Unmatched ENDGROUP tag.");
+  if (fieldnum >= 0)
+    goto have_tag;
+
+  while (true) {
+    uint32_t tag;
+    CHECK_RETURN(decode_v32(d, &tag));
+    wire_type = tag & 0x7;
+    fieldnum = tag >> 3;
+
+have_tag:
+    if (fieldnum == 0) {
+      seterr(d, "Saw invalid field number (0)");
       return upb_pbdecoder_suspend(d);
     }
-    return DECODE_ENDGROUP;
-  }
 
-  // TODO: deliver to unknown field callback.
-  switch (wire_type) {
-    case UPB_WIRE_TYPE_VARINT: {
-      uint64_t u64;
-      return decode_varint(d, &u64);
+    // TODO: deliver to unknown field callback.
+    switch (wire_type) {
+      case UPB_WIRE_TYPE_32BIT:
+        CHECK_RETURN(skip(d, 4));
+        break;
+      case UPB_WIRE_TYPE_64BIT:
+        CHECK_RETURN(skip(d, 8));
+        break;
+      case UPB_WIRE_TYPE_VARINT: {
+        uint64_t u64;
+        CHECK_RETURN(decode_varint(d, &u64));
+        break;
+      }
+      case UPB_WIRE_TYPE_DELIMITED: {
+        uint32_t len;
+        CHECK_RETURN(decode_v32(d, &len));
+        CHECK_RETURN(skip(d, len));
+        break;
+      }
+      case UPB_WIRE_TYPE_START_GROUP:
+        CHECK_SUSPEND(pushtagdelim(d, -fieldnum));
+        break;
+      case UPB_WIRE_TYPE_END_GROUP:
+        if (fieldnum == -d->top->groupnum) {
+          pop(d);
+        } else if (fieldnum == d->top->groupnum) {
+          return DECODE_ENDGROUP;
+        } else {
+          seterr(d, "Unmatched ENDGROUP tag.");
+          return upb_pbdecoder_suspend(d);
+        }
+        break;
+      default:
+        seterr(d, "Invalid wire type");
+        return upb_pbdecoder_suspend(d);
     }
-    case UPB_WIRE_TYPE_32BIT:
-      return skip(d, 4);
-    case UPB_WIRE_TYPE_64BIT:
-      return skip(d, 8);
-    case UPB_WIRE_TYPE_DELIMITED: {
-      uint32_t len;
-      CHECK_RETURN(decode_v32(d, &len));
-      return skip(d, len);
+
+    if (d->top->groupnum >= 0) {
+      return DECODE_OK;
     }
-    case UPB_WIRE_TYPE_START_GROUP:
-      seterr(d, "Can't handle unknown groups yet");
-      return upb_pbdecoder_suspend(d);
-    case UPB_WIRE_TYPE_END_GROUP:
-    default:
-      seterr(d, "Invalid wire type");
+
+    if (ptr(d) == d->delim_end) {
+      seterr(d, "Enclosing submessage ended in the middle of value or group");
+      // Unlike most errors we notice during parsing, right now we have consumed
+      // all of the user's input.
+      //
+      // There are three different options for how to handle this case:
+      //
+      //   1. decode() = short count, error = set
+      //   2. decode() = full count, error = set
+      //   3. decode() = full count, error NOT set, short count and error will
+      //      be reported on next call to decode() (or end())
+      //
+      // (1) and (3) have the advantage that they preserve the invariant that an
+      // error occurs iff decode() returns a short count.
+      //
+      // (2) and (3) have the advantage of reflecting the fact that all of the
+      // bytes were in fact parsed (and possibly delivered to the unknown field
+      // handler, in the future when that is supported).
+      //
+      // (3) requires extra state in the decode (a place to store the "permanent
+      // error" that we should return for all subsequent attempts to decode).
+      // But we likely want this anyway.
+      //
+      // Right now we do (1), thanks to the fact that we checkpoint *after* this
+      // check.  (3) may be a better choice long term; unclear at the moment.
       return upb_pbdecoder_suspend(d);
+    }
+
+    checkpoint(d);
   }
 }
 
+static void goto_endmsg(upb_pbdecoder *d) {
+  upb_value v;
+  bool found = upb_inttable_lookup32(d->top->dispatch, DISPATCH_ENDMSG, &v);
+  UPB_ASSERT_VAR(found, found);
+  d->pc = d->top->base + upb_value_getuint64(v);
+}
+
 static int32_t dispatch(upb_pbdecoder *d) {
   upb_inttable *dispatch = d->top->dispatch;
 
@@ -470,7 +546,7 @@ static int32_t dispatch(upb_pbdecoder *d) {
   int32_t ret = upb_pbdecoder_skipunknown(d, fieldnum, wire_type);
 
   if (ret == DECODE_ENDGROUP) {
-    d->pc = d->top->base - 1;  // Back to OP_ENDMSG.
+    goto_endmsg(d);
     return DECODE_OK;
   } else {
     d->pc = d->last - 1;  // Rewind to CHECKDELIM.
@@ -493,7 +569,11 @@ size_t upb_pbdecoder_decode(void *closure, const void *hd, const char *buf,
   upb_pbdecoder *d = closure;
   const mgroup *group = hd;
   assert(buf);
-  upb_pbdecoder_resume(d, NULL, buf, size, handle);
+  int32_t result = upb_pbdecoder_resume(d, NULL, buf, size, handle);
+  if (result == DECODE_ENDGROUP) {
+    goto_endmsg(d);
+  }
+  CHECK_RETURN(result);
   UPB_UNUSED(group);
 
 #define VMCASE(op, code) \
@@ -552,8 +632,6 @@ size_t upb_pbdecoder_decode(void *closure, const void *hd, const char *buf,
       )
       VMCASE(OP_ENDMSG,
         CHECK_SUSPEND(upb_sink_endmsg(&d->top->sink, d->status));
-        assert(d->call_len > 0);
-        d->pc = d->callstack[--d->call_len];
       )
       VMCASE(OP_STARTSEQ,
         upb_pbdecoder_frame *outer = outer_frame(d);
@@ -579,25 +657,39 @@ size_t upb_pbdecoder_decode(void *closure, const void *hd, const char *buf,
       )
       VMCASE(OP_STRING,
         uint32_t len = curbufleft(d);
-        CHECK_SUSPEND(
-            upb_sink_putstring(&d->top->sink, arg, ptr(d), len, handle));
-        advance(d, len);
-        if (d->delim_end == NULL) {  // String extends beyond this buf?
-          d->pc--;
-          d->bufstart_ofs += size;
-          d->residual_end = d->residual;
-          return size;
+        size_t n = upb_sink_putstring(&d->top->sink, arg, ptr(d), len, handle);
+        if (n > len) {
+          if (n > d->top->end_ofs - offset(d)) {
+            seterr(d, "Tried to skip past end of string.");
+            return upb_pbdecoder_suspend(d);
+          } else {
+            return skip(d, n);
+          }
+        } else if (n < len) {
+          advance(d, n);
+          return upb_pbdecoder_suspend(d);
+        } else {
+          advance(d, n);
+          if (d->delim_end == NULL) {  // String extends beyond this buf?
+            d->pc--;  // Do OP_STRING again when we resume.
+            d->bufstart_ofs += size;
+            d->residual_end = d->residual;
+            return size;
+          }
         }
       )
       VMCASE(OP_ENDSTR,
         CHECK_SUSPEND(upb_sink_endstr(&d->top->sink, arg));
       )
       VMCASE(OP_PUSHTAGDELIM,
-        CHECK_SUSPEND(push(d, d->top->end_ofs));
+        CHECK_SUSPEND(pushtagdelim(d, arg));
+      )
+      VMCASE(OP_SETBIGGROUPNUM,
+        d->top->groupnum = *d->pc++;
       )
       VMCASE(OP_POP,
         assert(d->top > d->stack);
-        d->top--;
+        pop(d);
       )
       VMCASE(OP_PUSHLENDELIM,
         uint32_t len;
@@ -608,13 +700,9 @@ size_t upb_pbdecoder_decode(void *closure, const void *hd, const char *buf,
       VMCASE(OP_SETDELIM,
         set_delim_end(d);
       )
-      VMCASE(OP_SETGROUPNUM,
-        d->top->groupnum = arg;
-      )
-      VMCASE(OP_SETBIGGROUPNUM,
-        d->top->groupnum = *d->pc++;
-      )
       VMCASE(OP_CHECKDELIM,
+        // We are guaranteed of this assert because we never allow ourselves to
+        // consume bytes beyond data_end, which covers delim_end when non-NULL.
         assert(!(d->delim_end && ptr(d) > d->delim_end));
         if (ptr(d) == d->delim_end)
           d->pc += longofs;
@@ -623,6 +711,10 @@ size_t upb_pbdecoder_decode(void *closure, const void *hd, const char *buf,
         d->callstack[d->call_len++] = d->pc;
         d->pc += longofs;
       )
+      VMCASE(OP_RET,
+        assert(d->call_len > 0);
+        d->pc = d->callstack[--d->call_len];
+      )
       VMCASE(OP_BRANCH,
         d->pc += longofs;
       )
@@ -755,6 +847,7 @@ void upb_pbdecoder_init(upb_pbdecoder *d, const upb_pbdecodermethod *m,
 void upb_pbdecoder_reset(upb_pbdecoder *d) {
   d->top = d->stack;
   d->top->end_ofs = UINT64_MAX;
+  d->top->groupnum = 0;
   d->bufstart_ofs = 0;
   d->ptr = d->residual;
   d->buf = d->residual;

upb/pb/decoder.h

@@ -20,19 +20,23 @@ namespace pb {
 class CodeCache;
 class Decoder;
 class DecoderMethod;
+class DecoderMethodOptions;
 }  // namespace pb
 }  // namespace upb
 
 typedef upb::pb::CodeCache upb_pbcodecache;
 typedef upb::pb::Decoder upb_pbdecoder;
 typedef upb::pb::DecoderMethod upb_pbdecodermethod;
+typedef upb::pb::DecoderMethodOptions upb_pbdecodermethodopts;
 #else
 struct upb_pbdecoder;
 struct upb_pbdecodermethod;
+struct upb_pbdecodermethodopts;
 struct upb_pbcodecache;
 
 typedef struct upb_pbdecoder upb_pbdecoder;
 typedef struct upb_pbdecodermethod upb_pbdecodermethod;
+typedef struct upb_pbdecodermethodopts upb_pbdecodermethodopts;
 typedef struct upb_pbcodecache upb_pbcodecache;
 #endif
 
@@ -72,14 +76,40 @@ typedef struct {
   // that calls from the JIT into C work correctly.
   uint64_t end_ofs;
   const uint32_t *base;
-  uint32_t groupnum;
+
+  // 0 indicates a length-delimited field.
+  // A positive number indicates a known group.
+  // A negative number indicates an unknown group.
+  int32_t groupnum;
   upb_inttable *dispatch;  // Not used by the JIT.
 } upb_pbdecoder_frame;
 
 #ifdef __cplusplus
 
-// Represents the code to parse a protobuf according to a specific schema,
-// optionally bound to a set of destination handlers.
+// The parameters one uses to construct a DecoderMethod.
+// TODO(haberman): move allowjit here?  Seems more convenient for users.
+class upb::pb::DecoderMethodOptions {
+ public:
+  // Parameter represents the destination handlers that this method will push
+  // to.
+  explicit DecoderMethodOptions(const Handlers* dest_handlers);
+
+  // Should the decoder push submessages to lazy handlers for fields that have
+  // them?  The caller should set this iff the lazy handlers expect data that is
+  // in protobuf binary format and the caller wishes to lazy parse it.
+  void set_lazy(bool lazy);
+
+ private:
+#else
+struct upb_pbdecodermethodopts {
+#endif
+  const upb_handlers *handlers;
+  bool lazy;
+};
+
+#ifdef __cplusplus
+
+// Represents the code to parse a protobuf according to a destination Handlers.
 class upb::pb::DecoderMethod /* : public upb::RefCounted */ {
  public:
   // From upb::ReferenceCounted.
@@ -88,14 +118,9 @@ class upb::pb::DecoderMethod /* : public upb::RefCounted */ {
   void DonateRef(const void* from, const void* to) const;
   void CheckRef(const void* owner) const;
 
-  // The schema that this method parses.  Never NULL.
-  const MessageDef* schema() const;
-
   // The destination handlers that are statically bound to this method.
   // This method is only capable of outputting to a sink that uses these
   // handlers.
-  //
-  // Will be NULL if this method is not statically bound.
   const Handlers* dest_handlers() const;
 
   // The input handlers for this decoder method.
@@ -106,8 +131,7 @@ class upb::pb::DecoderMethod /* : public upb::RefCounted */ {
 
   // Convenience method for generating a DecoderMethod without explicitly
   // creating a CodeCache.
-  static reffed_ptr<const DecoderMethod> NewForDestHandlers(
-      const upb::Handlers *dest);
+  static reffed_ptr<const DecoderMethod> New(const DecoderMethodOptions& opts);
 
  private:
   UPB_DISALLOW_POD_OPS(DecoderMethod, upb::pb::DecoderMethod);
@@ -138,13 +162,7 @@ struct upb_pbdecodermethod {
   // The handler one calls to invoke this method.
   upb_byteshandler input_handler_;
 
-  // The message type that this method is parsing.
-  const upb_msgdef *schema_;
-
-  // The destination handlers this method is bound to, or NULL if this method
-  // can be bound to a destination handlers instance at runtime.
-  //
-  // If non-NULL, we own a ref.
+  // The destination handlers this method is bound to.  We own a ref.
   const upb_handlers *dest_handlers_;
 
   // The dispatch table layout is:
@@ -183,8 +201,7 @@ class upb::pb::Decoder {
   void Reset();
 
   // Resets the output sink of the Decoder.
-  // The given sink must match method()->schema() as well as
-  // method()->dest_handlers() if the latter is non-NULL.
+  // The given sink must match method()->dest_handlers().
   //
   // 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
@@ -221,9 +238,6 @@ struct upb_pbdecoder {
   // Overall stream offset of "buf."
   uint64_t bufstart_ofs;
 
-  // How many bytes past the end of the user buffer we want to skip.
-  size_t skip;
-
   // Buffer for residual bytes not parsed from the previous buffer.
   // The maximum number of residual bytes we require is 12; a five-byte
   // unknown tag plus an eight-byte value, less one because the value
@@ -290,8 +304,7 @@ class upb::pb::CodeCache {
   // more efficient decoding.  However the returned method may or may not
   // actually be statically bound.  But in all cases, the returned method can
   // push data to the given handlers.
-  const DecoderMethod *GetDecoderMethodForDestHandlers(
-      const upb::Handlers *handlers);
+  const DecoderMethod *GetDecoderMethod(const DecoderMethodOptions& opts);
 
   // If/when someone needs to explicitly create a dynamically-bound
   // DecoderMethod*, we can add a method to get it here.
@@ -320,27 +333,30 @@ const upb_pbdecodermethod *upb_pbdecoder_method(const upb_pbdecoder *d);
 bool upb_pbdecoder_resetoutput(upb_pbdecoder *d, upb_sink *sink);
 upb_bytessink *upb_pbdecoder_input(upb_pbdecoder *d);
 
+void upb_pbdecodermethodopts_init(upb_pbdecodermethodopts *opts,
+                                  const upb_handlers *h);
+void upb_pbdecodermethodopts_setlazy(upb_pbdecodermethodopts *opts, bool lazy);
+
 void upb_pbdecodermethod_ref(const upb_pbdecodermethod *m, const void *owner);
 void upb_pbdecodermethod_unref(const upb_pbdecodermethod *m, const void *owner);
 void upb_pbdecodermethod_donateref(const upb_pbdecodermethod *m,
                                    const void *from, const void *to);
 void upb_pbdecodermethod_checkref(const upb_pbdecodermethod *m,
                                   const void *owner);
-const upb_msgdef *upb_pbdecodermethod_schema(const upb_pbdecodermethod *m);
 const upb_handlers *upb_pbdecodermethod_desthandlers(
     const upb_pbdecodermethod *m);
 const upb_byteshandler *upb_pbdecodermethod_inputhandler(
     const upb_pbdecodermethod *m);
 bool upb_pbdecodermethod_isnative(const upb_pbdecodermethod *m);
-const upb_pbdecodermethod *upb_pbdecodermethod_newfordesthandlers(
-    const upb_handlers *dest, const void *owner);
+const upb_pbdecodermethod *upb_pbdecodermethod_new(
+    const upb_pbdecodermethodopts *opts, const void *owner);
 
 void upb_pbcodecache_init(upb_pbcodecache *c);
 void upb_pbcodecache_uninit(upb_pbcodecache *c);
 bool upb_pbcodecache_allowjit(const upb_pbcodecache *c);
 bool upb_pbcodecache_setallowjit(upb_pbcodecache *c, bool allow);
-const upb_pbdecodermethod *upb_pbcodecache_getdecodermethodfordesthandlers(
-    upb_pbcodecache *c, const upb_handlers *handlers);
+const upb_pbdecodermethod *upb_pbcodecache_getdecodermethod(
+    upb_pbcodecache *c, const upb_pbdecodermethodopts *opts);
 
 #ifdef __cplusplus
 }  /* extern "C" */
@@ -391,6 +407,13 @@ inline BytesSink* Decoder::input() {
   return upb_pbdecoder_input(this);
 }
 
+inline DecoderMethodOptions::DecoderMethodOptions(const Handlers* h) {
+  upb_pbdecodermethodopts_init(this, h);
+}
+inline void DecoderMethodOptions::set_lazy(bool lazy) {
+  upb_pbdecodermethodopts_setlazy(this, lazy);
+}
+
 inline void DecoderMethod::Ref(const void *owner) const {
   upb_pbdecodermethod_ref(this, owner);
 }
@@ -403,9 +426,6 @@ inline void DecoderMethod::DonateRef(const void *from, const void *to) const {
 inline void DecoderMethod::CheckRef(const void *owner) const {
   upb_pbdecodermethod_checkref(this, owner);
 }
-inline const MessageDef* DecoderMethod::schema() const {
-  return upb_pbdecodermethod_schema(this);
-}
 inline const Handlers* DecoderMethod::dest_handlers() const {
   return upb_pbdecodermethod_desthandlers(this);
 }
@@ -416,10 +436,9 @@ inline bool DecoderMethod::is_native() const {
   return upb_pbdecodermethod_isnative(this);
 }
 // static
-inline reffed_ptr<const DecoderMethod> DecoderMethod::NewForDestHandlers(
-    const Handlers *dest) {
-  const upb_pbdecodermethod *m =
-      upb_pbdecodermethod_newfordesthandlers(dest, &m);
+inline reffed_ptr<const DecoderMethod> DecoderMethod::New(
+    const DecoderMethodOptions &opts) {
+  const upb_pbdecodermethod *m = upb_pbdecodermethod_new(&opts, &m);
   return reffed_ptr<const DecoderMethod>(m, &m);
 }
 
@@ -435,9 +454,9 @@ inline bool CodeCache::allow_jit() const {
 inline bool CodeCache::set_allow_jit(bool allow) {
   return upb_pbcodecache_setallowjit(this, allow);
 }
-inline const DecoderMethod* CodeCache::GetDecoderMethodForDestHandlers(
-    const upb::Handlers* handlers) {
-  return upb_pbcodecache_getdecodermethodfordesthandlers(this, handlers);
+inline const DecoderMethod *CodeCache::GetDecoderMethod(
+    const DecoderMethodOptions& opts) {
+  return upb_pbcodecache_getdecodermethod(this, &opts);
 }
 
 }  // namespace pb

upb/pb/decoder.int.h

@@ -40,12 +40,10 @@ typedef enum {
   OP_PUSHLENDELIM   = 24,  // No arg.
   OP_POP            = 25,  // No arg.
   OP_SETDELIM       = 26,  // No arg.
-  OP_SETGROUPNUM    = 27,
-  OP_SETBIGGROUPNUM = 28,  // two words: | unused (24) | opc || groupnum (32) |
-
-  // The arg for these opcodes is a local label reference.
-  OP_CHECKDELIM     = 29,
-  OP_CALL           = 30,
+  OP_SETBIGGROUPNUM = 27,  // two words: | unused (24) | opc || groupnum (32) |
+  OP_CHECKDELIM     = 28,
+  OP_CALL           = 29,
+  OP_RET            = 30,
   OP_BRANCH         = 31,
 
   // Different opcodes depending on how many bytes expected.
@@ -112,10 +110,10 @@ size_t upb_pbdecoder_decode(void *closure, const void *hd, const char *buf,
 bool upb_pbdecoder_end(void *closure, const void *handler_data);
 
 // Decoder-internal functions that the JIT calls to handle fallback paths.
-void *upb_pbdecoder_resume(upb_pbdecoder *d, void *p, const char *buf,
-                           size_t size, const upb_bufhandle *handle);
+int32_t upb_pbdecoder_resume(upb_pbdecoder *d, void *p, const char *buf,
+                             size_t size, const upb_bufhandle *handle);
 size_t upb_pbdecoder_suspend(upb_pbdecoder *d);
-int32_t upb_pbdecoder_skipunknown(upb_pbdecoder *d, uint32_t fieldnum,
+int32_t upb_pbdecoder_skipunknown(upb_pbdecoder *d, int32_t fieldnum,
                                   uint8_t wire_type);
 int32_t upb_pbdecoder_checktag_slow(upb_pbdecoder *d, uint64_t expected);
 int32_t upb_pbdecoder_decode_varint_slow(upb_pbdecoder *d, uint64_t *u64);
@@ -137,8 +135,21 @@ void upb_pbdecoder_freejit(mgroup *group);
 // wherever that takes you."
 #define LABEL_DISPATCH 0
 
+// A special slot in the dispatch table that stores the epilogue (ENDMSG and/or
+// RET) for branching to when we find an appropriate ENDGROUP tag.
+#define DISPATCH_ENDMSG 0
+
+// All of the functions in decoder.c that return int32_t return values according
+// to the following scheme:
+//   1. negative values indicate a return code from the following list.
+//   2. positive values indicate that error or end of buffer was hit, and
+//      that the decode function should immediately return the given value
+//      (the decoder state has already been suspended and is ready to be
+//      resumed).
 #define DECODE_OK -1
 #define DECODE_MISMATCH -2  // Used only from checktag_slow().
-#define DECODE_ENDGROUP -2  // Used only from checkunknown().
+#define DECODE_ENDGROUP -3  // Used only from checkunknown().
+
+#define CHECK_RETURN(x) { int32_t ret = x; if (ret >= 0) return ret; }
 
 #endif  // UPB_DECODER_INT_H_

upb/pb/glue.c

@@ -17,8 +17,10 @@ upb_def **upb_load_defs_from_descriptor(const char *str, size_t len, int *n,
                                         void *owner, upb_status *status) {
   // Create handlers.
   const upb_handlers *reader_h = upb_descreader_newhandlers(&reader_h);
+  upb_pbdecodermethodopts opts;
+  upb_pbdecodermethodopts_init(&opts, reader_h);
   const upb_pbdecodermethod *decoder_m =
-      upb_pbdecodermethod_newfordesthandlers(reader_h, &decoder_m);
+      upb_pbdecodermethod_new(&opts, &decoder_m);
 
   upb_pbdecoder decoder;
   upb_descreader reader;