From 23a5af3513f82ccdd134d0fe84db0af8548f866e Mon Sep 17 00:00:00 2001
From: Joshua Haberman <jhaberman@gmail.com>
Date: Tue, 25 Feb 2020 19:13:27 -0800
Subject: [PATCH] [json] fixed all remaining conformance bugs.

---
 upb/json_decode.c | 49 ++++++++++++++++++++++++++++++++++-------------
 upb/json_encode.c |  7 ++++---
 upb/reflection.c  | 12 ++++++++++++
 upb/reflection.h  |  3 +++
 4 files changed, 55 insertions(+), 16 deletions(-)

diff --git a/upb/json_decode.c b/upb/json_decode.c
index e879231f9f..54a55ad135 100644
--- a/upb/json_decode.c
+++ b/upb/json_decode.c
@@ -203,7 +203,7 @@ static bool jsondec_objnext(jsondec *d) {
 
 /* JSON number ****************************************************************/
 
-static void jsondec_skipdigits(jsondec *d) {
+static bool jsondec_tryskipdigits(jsondec *d) {
   const char *start = d->ptr;
 
   while (d->ptr < d->end) {
@@ -213,7 +213,11 @@ static void jsondec_skipdigits(jsondec *d) {
     d->ptr++;
   }
 
-  if (d->ptr == start) {
+  return d->ptr != start;
+}
+
+static void jsondec_skipdigits(jsondec *d) {
+  if (!jsondec_tryskipdigits(d)) {
     jsondec_err(d, "Expected one or more digits");
   }
 }
@@ -225,9 +229,15 @@ static double jsondec_number(jsondec *d) {
 
   /* Skip over the syntax of a number, as specified by JSON. */
   if (*d->ptr == '-') d->ptr++;
-  if (!jsondec_tryparsech(d, '0')) {
+
+  if (jsondec_tryparsech(d, '0')) {
+    if (jsondec_tryskipdigits(d)) {
+      jsondec_err(d, "number cannot have leading zero");
+    }
+  } else {
     jsondec_skipdigits(d);
   }
+
   if (d->ptr == d->end) goto parse;
   if (jsondec_tryparsech(d, '.')) {
     jsondec_skipdigits(d);
@@ -337,6 +347,8 @@ static size_t jsondec_unicode(jsondec *d, char* out) {
     cp = (high & 0x3ff) << 10;
     cp |= (low & 0x3ff);
     cp += 0x10000;
+  } else if (cp >= 0xdc00 && cp <= 0xdfff) {
+    jsondec_err(d, "Unpaired low surrogate");
   }
 
   /* Write to UTF-8 */
@@ -846,6 +858,12 @@ static upb_msgval jsondec_msg(jsondec *d, const upb_fielddef *f) {
   return val;
 }
 
+static bool jsondec_isvalue(const upb_fielddef *f) {
+  return upb_fielddef_type(f) == UPB_TYPE_MESSAGE &&
+         upb_msgdef_wellknowntype(upb_fielddef_msgsubdef(f)) ==
+             UPB_WELLKNOWN_VALUE;
+}
+
 static void jsondec_field(jsondec *d, upb_msg *msg, const upb_msgdef *m) {
   upb_strview name;
   const upb_fielddef *f;
@@ -863,7 +881,12 @@ static void jsondec_field(jsondec *d, upb_msg *msg, const upb_msgdef *m) {
     return;
   }
 
-  if (jsondec_peek(d) == JD_NULL) {
+  if (upb_fielddef_containingoneof(f) &&
+      upb_msg_hasoneof(msg, upb_fielddef_containingoneof(f))) {
+    jsondec_err(d, "More than one field for this oneof.");
+  }
+
+  if (jsondec_peek(d) == JD_NULL && !jsondec_isvalue(f)) {
     /* JSON "null" indicates a default value, so no need to set anything. */
     return jsondec_null(d);
   }
@@ -1106,47 +1129,45 @@ static void jsondec_wellknownvalue(jsondec *d, upb_msg *msg,
   switch (jsondec_peek(d)) {
     case JD_NUMBER:
       /* double number_value = 2; */
-      val.double_val = jsondec_number(d);
       f = upb_msgdef_itof(m, 2);
+      val.double_val = jsondec_number(d);
       break;
     case JD_STRING:
       /* string string_value = 3; */
-      val.str_val = jsondec_string(d);
       f = upb_msgdef_itof(m, 3);
+      val.str_val = jsondec_string(d);
       break;
     case JD_FALSE:
       /* bool bool_value = 4; */
-      val.bool_val = false;
       f = upb_msgdef_itof(m, 4);
+      val.bool_val = false;
       jsondec_false(d);
       break;
     case JD_TRUE:
       /* bool bool_value = 4; */
-      val.bool_val = true;
       f = upb_msgdef_itof(m, 4);
+      val.bool_val = true;
       jsondec_true(d);
       break;
     case JD_NULL:
       /* NullValue null_value = 1; */
-      val.int32_val = 0;
       f = upb_msgdef_itof(m, 1);
+      val.int32_val = 0;
       jsondec_null(d);
       break;
     /* Note: these cases return, because upb_msg_mutable() is enough. */
-    case JD_OBJECT: {
+    case JD_OBJECT:
       /* Struct struct_value = 5; */
       f = upb_msgdef_itof(m, 5);
       submsg = upb_msg_mutable(msg, f, d->arena).msg;
       jsondec_struct(d, submsg, upb_fielddef_msgsubdef(f));
       return;
-    }
-    case JD_ARRAY: {
+    case JD_ARRAY:
       /* ListValue list_value = 6; */
       f = upb_msgdef_itof(m, 6);
       submsg = upb_msg_mutable(msg, f, d->arena).msg;
       jsondec_listvalue(d, submsg, upb_fielddef_msgsubdef(f));
       return;
-    }
     default:
       UPB_UNREACHABLE();
   }
@@ -1176,6 +1197,8 @@ static upb_strview jsondec_mask(jsondec *d, const char *buf, const char *end) {
     if (ch >= 'A' && ch <= 'Z') {
       *out++ = '_';
       *out++ = ch + 32;
+    } else if (ch == '_') {
+      jsondec_err(d, "field mask may not contain '_'");
     } else {
       *out++ = ch;
     }
diff --git a/upb/json_encode.c b/upb/json_encode.c
index caa0c8a234..285bcecea5 100644
--- a/upb/json_encode.c
+++ b/upb/json_encode.c
@@ -196,7 +196,6 @@ static void jsonenc_bytes(jsonenc *e, upb_strview str) {
       jsonenc_putbytes(e, buf, 4);
       break;
     case 1:
-      fprintf(stderr, "Base64 encode: %d\n", (int)ptr[0]);
       buf[0] = base64[ptr[0] >> 2];
       buf[1] = base64[((ptr[0] & 0x3) << 4)];
       buf[2] = '=';
@@ -349,15 +348,17 @@ static void jsonenc_fieldpath(jsonenc *e, upb_strview path) {
 
   while (ptr < end) {
     char ch = *ptr;
+
     if (ch >= 'A' && ch <= 'Z') {
       jsonenc_err(e, "Field mask element may not have upper-case letter.");
     } else if (ch == '_') {
       if (ptr == end - 1 || *(ptr + 1) < 'a' || *(ptr + 1) > 'z') {
         jsonenc_err(e, "Underscore must be followed by a lowercase letter.");
       }
-    } else {
-      jsonenc_putbytes(e, &ch, 1);
+      ch = *++ptr - 32;
     }
+
+    jsonenc_putbytes(e, &ch, 1);
     ptr++;
   }
 }
diff --git a/upb/reflection.c b/upb/reflection.c
index b31a917469..481e3db943 100644
--- a/upb/reflection.c
+++ b/upb/reflection.c
@@ -91,6 +91,18 @@ bool upb_msg_has(const upb_msg *msg, const upb_fielddef *f) {
   }
 }
 
+bool upb_msg_hasoneof(const upb_msg *msg, const upb_oneofdef *o) {
+  upb_oneof_iter i;
+  const upb_fielddef *f;
+  const upb_msglayout_field *field;
+
+  upb_oneof_begin(&i, o);
+  if (upb_oneof_done(&i)) return false;
+  f = upb_oneof_iter_field(&i);
+  field = upb_fielddef_layout(f);
+  return *oneofcase(msg, field) != 0;
+}
+
 upb_msgval upb_msg_get(const upb_msg *msg, const upb_fielddef *f) {
   if (!upb_fielddef_haspresence(f) || upb_msg_has(msg, f)) {
     return _upb_msg_getraw(msg, f);
diff --git a/upb/reflection.h b/upb/reflection.h
index 95156b7d76..c55fe4a2e6 100644
--- a/upb/reflection.h
+++ b/upb/reflection.h
@@ -44,6 +44,9 @@ upb_mutmsgval upb_msg_mutable(upb_msg *msg, const upb_fielddef *f, upb_arena *a)
 /* May only be called for fields where upb_fielddef_haspresence(f) == true. */
 bool upb_msg_has(const upb_msg *msg, const upb_fielddef *f);
 
+/* Returns whether any field is set in the oneof. */
+bool upb_msg_hasoneof(const upb_msg *msg, const upb_oneofdef *o);
+
 /* Sets the given field to the given value.  For a msg/array/map/string, the
  * value must be in the same arena.  */
 void upb_msg_set(upb_msg *msg, const upb_fielddef *f, upb_msgval val,