Changes from Google-internal development.

 * JSON parser expanded to handle split buffers.
 * bugfix to the protobuf decoder.

5 files changed, 32 insertions, 5 deletions

diff --git a/upb/pb/compile_decoder.c b/upb/pb/compile_decoder.c
index 64689f6..11aa4e9 100644
--- a/upb/pb/compile_decoder.c
+++ b/upb/pb/compile_decoder.c
@@ -302,6 +302,7 @@ static void putop(compiler *c, opcode op, ...) {
     case OP_SETDELIM:
     case OP_HALT:
     case OP_RET:
+    case OP_DISPATCH:
       put32(c, op);
       break;
     case OP_PARSE_DOUBLE:
@@ -382,7 +383,7 @@ const char *upb_pbdecoder_getopname(unsigned int op) {
     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(SETBIGGROUPNUM), OP(HALT),
+    OP(SETBIGGROUPNUM), OP(DISPATCH), OP(HALT),
   };
   return op > OP_HALT ? names[0] : names[op];
 #undef OP
@@ -414,6 +415,7 @@ static void dumpbc(uint32_t *p, uint32_t *end, FILE *f) {
                               upb_handlers_msgdef(method->dest_handlers_)));
         break;
       }
+      case OP_DISPATCH:
       case OP_STARTMSG:
       case OP_ENDMSG:
       case OP_PUSHLENDELIM:
@@ -759,6 +761,7 @@ static void compile_method(compiler *c, upb_pbdecodermethod *method) {
   putop(c, OP_SETDISPATCH, &method->dispatch);
   putsel(c, OP_STARTMSG, UPB_STARTMSG_SELECTOR, h);
  label(c, LABEL_FIELD);
+  uint32_t* start_pc = c->pc;
   upb_msg_iter i;
   for(upb_msg_begin(&i, md); !upb_msg_done(&i); upb_msg_next(&i)) {
     const upb_fielddef *f = upb_msg_iter_field(&i);
@@ -774,8 +777,18 @@ static void compile_method(compiler *c, upb_pbdecodermethod *method) {
     }
   }
 
+  // If there were no fields, or if no handlers were defined, we need to
+  // generate a non-empty loop body so that we can at least dispatch for unknown
+  // fields and check for the end of the message.
+  if (c->pc == start_pc) {
+    // Check for end-of-message.
+    putop(c, OP_CHECKDELIM, LABEL_ENDMSG);
+    // Unconditionally dispatch.
+    putop(c, OP_DISPATCH, 0);
+  }
+
   // For now we just loop back to the last field of the message (or if none,
-  // the DISPATCH opcode for the message.
+  // 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.
diff --git a/upb/pb/compile_decoder_x64.dasc b/upb/pb/compile_decoder_x64.dasc
index bac1ac2..3181cab 100644
--- a/upb/pb/compile_decoder_x64.dasc
+++ b/upb/pb/compile_decoder_x64.dasc
@@ -1124,6 +1124,9 @@ static void jitbytecode(jitcompiler *jc) {
       jittag(jc, tag, arg >> 8, (int8_t)arg, method);
       break;
     }
+    case OP_DISPATCH:
+      |  call   =>jmptarget(jc, &method->dispatch)
+      break;
     case OP_HALT:
       assert(false);
     }
diff --git a/upb/pb/compile_decoder_x64.h b/upb/pb/compile_decoder_x64.h
index ba3636e..ef4459d 100644
--- a/upb/pb/compile_decoder_x64.h
+++ b/upb/pb/compile_decoder_x64.h
@@ -1680,6 +1680,11 @@ static void jitbytecode(jitcompiler *jc) {
       jittag(jc, tag, arg >> 8, (int8_t)arg, method);
       break;
     }
+    case OP_DISPATCH:
+      //|  call   =>jmptarget(jc, &method->dispatch)
+      dasm_put(Dst, 2151, jmptarget(jc, &method->dispatch));
+# 1129 "upb/pb/compile_decoder_x64.dasc"
+      break;
     case OP_HALT:
       assert(false);
     }
@@ -1688,5 +1693,5 @@ static void jitbytecode(jitcompiler *jc) {
   asmlabel(jc, "eof");
   //|  nop
   dasm_put(Dst, 1909);
-# 1134 "upb/pb/compile_decoder_x64.dasc"
+# 1137 "upb/pb/compile_decoder_x64.dasc"
 }
diff --git a/upb/pb/decoder.c b/upb/pb/decoder.c
index ec8c03f..04ca413 100644
--- a/upb/pb/decoder.c
+++ b/upb/pb/decoder.c
@@ -801,6 +801,9 @@ size_t upb_pbdecoder_decode(void *closure, const void *hd, const char *buf,
         if (result == DECODE_MISMATCH) goto badtag;
         if (result >= 0) return result;
       })
+      VMCASE(OP_DISPATCH, {
+        CHECK_RETURN(dispatch(d));
+      })
       VMCASE(OP_HALT, {
         return size;
       })
@@ -859,7 +862,8 @@ bool upb_pbdecoder_end(void *closure, const void *handler_data) {
       // Rewind from OP_TAG* to OP_CHECKDELIM.
       assert(getop(*d->pc) == OP_TAG1 ||
              getop(*d->pc) == OP_TAG2 ||
-             getop(*d->pc) == OP_TAGN);
+             getop(*d->pc) == OP_TAGN ||
+             getop(*d->pc == OP_DISPATCH));
       d->pc = p;
     }
     upb_pbdecoder_decode(closure, handler_data, &dummy, 0, NULL);
diff --git a/upb/pb/decoder.int.h b/upb/pb/decoder.int.h
index 9b35b70..302701e 100644
--- a/upb/pb/decoder.int.h
+++ b/upb/pb/decoder.int.h
@@ -66,7 +66,9 @@ typedef enum {
                            //   | unused (24)         | opc |
                            //   | upb_inttable* (32 or 64)  |
 
-  OP_HALT           = 36,  // No arg.
+  OP_DISPATCH       = 36,  // No arg.
+
+  OP_HALT           = 37,  // No arg.
 } opcode;
 
 #define OP_MAX OP_HALT