Bugfixes to parser.

4 files changed, 46 insertions, 39 deletions

diff --git a/src/upb_msg.c b/src/upb_msg.c
index deefb27..a923bb8 100644
--- a/src/upb_msg.c
+++ b/src/upb_msg.c
@@ -1,6 +1,7 @@
 /*
  * upb - a minimalist implementation of protocol buffers.
  *
+ * Copyright (c) 2009 Joshua Haberman.  See LICENSE for details.
  */
 
 #include <inttypes.h>
@@ -281,10 +282,13 @@ static void *value_cb(void *udata, void *buf, void *end,
   struct upb_msg_parse_state *s = udata;
   struct upb_msg_field *f = user_field_desc;
   union upb_value_ptr p = get_value_ptr(s->top->data, f);
-  return upb_parse_value(buf, end, f->type, p, errjmp);
+  void *ret = upb_parse_value(buf, end, f->type, p, errjmp);
+  google_protobuf_FieldDescriptorProto *fd = upb_msg_field_descriptor(f, s->top->m);
+  //upb_text_printfield(&s->p, *fd->name, f->type, p, stdout);
+  return ret;
 }
 
-static upb_status_t str_cb(void *udata, struct upb_string *str,
+static void str_cb(void *udata, struct upb_string *str,
                            void *user_field_desc)
 {
   struct upb_msg_parse_state *s = udata;
@@ -297,7 +301,8 @@ static upb_status_t str_cb(void *udata, struct upb_string *str,
     upb_msg_reuse_str(p.str, str->byte_len);
     upb_strcpy(*p.str, str);
   }
-  return UPB_STATUS_OK;
+  google_protobuf_FieldDescriptorProto *fd = upb_msg_field_descriptor(f, s->top->m);
+  //upb_text_printfield(&s->p, *fd->name, f->type, p, stdout);
 }
 
 static void submsg_start_cb(void *udata, void *user_field_desc)
@@ -311,12 +316,21 @@ static void submsg_start_cb(void *udata, void *user_field_desc)
   s->top->m = f->ref.msg;
   s->top->data = *p.msg;
   if(!s->merge) upb_msg_clear(s->top->data, s->top->m);
+  //upb_text_push(&s->p, *s->top->m->descriptor->name, stdout);
+}
+
+static void submsg_end_cb(void *udata)
+{
+  struct upb_msg_parse_state *s = udata;
+  s->top--;
+  //upb_text_pop(&s->p, stdout);
 }
 
 void upb_msg_parse_reset(struct upb_msg_parse_state *s, void *msg,
                          struct upb_msg *m, bool merge, bool byref)
 {
   upb_parse_reset(&s->s, s);
+  //upb_text_printer_init(&s->p, false);
   s->merge = merge;
   s->byref = byref;
   if(!merge && msg == NULL) msg = upb_msgdata_new(m);
@@ -328,6 +342,7 @@ void upb_msg_parse_reset(struct upb_msg_parse_state *s, void *msg,
   s->s.value_cb = value_cb;
   s->s.str_cb = str_cb;
   s->s.submsg_start_cb = submsg_start_cb;
+  s->s.submsg_end_cb = submsg_end_cb;
 }
 
 void upb_msg_parse_init(struct upb_msg_parse_state *s, void *msg,
diff --git a/src/upb_msg.h b/src/upb_msg.h
index 0cb3cd5..9eaf6c2 100644
--- a/src/upb_msg.h
+++ b/src/upb_msg.h
@@ -101,7 +101,6 @@
 #include "upb.h"
 #include "upb_table.h"
 #include "upb_parse.h"
-#include "descriptor.h"
 
 #ifdef __cplusplus
 extern "C" {
@@ -195,16 +194,6 @@ INLINE struct upb_msg_field *upb_msg_fieldbyname(struct upb_msg *m,
   return e ? &e->f : NULL;
 }
 
-INLINE bool upb_issubmsgtype(upb_field_type_t type) {
-  return type == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_GROUP  ||
-         type == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_MESSAGE;
-}
-
-INLINE bool upb_isstringtype(upb_field_type_t type) {
-  return type == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_STRING  ||
-         type == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_BYTES;
-}
-
 INLINE bool upb_issubmsg(struct upb_msg_field *f) {
   return upb_issubmsgtype(f->type);
 }
@@ -311,12 +300,14 @@ struct upb_msg_parse_frame {
   void *data;
 };
 
+//#include "upb_text.h"
 struct upb_msg_parse_state {
   struct upb_parse_state s;
   bool merge;
   bool byref;
   struct upb_msg *m;
   struct upb_msg_parse_frame stack[UPB_MAX_NESTING], *top;
+  //struct upb_text_printer p;
 };
 
 /* Initializes/frees a message parser.  The parser will write the data to the
diff --git a/src/upb_parse.c b/src/upb_parse.c
index fa37a56..3bbbf4a 100644
--- a/src/upb_parse.c
+++ b/src/upb_parse.c
@@ -10,7 +10,6 @@
 #include <stdlib.h>
 
 /* May want to move this to upb.c if enough other things warrant it. */
-#include "descriptor.h"
 #define alignof(t) offsetof(struct { char c; t x; }, x)
 struct upb_type_info upb_type_info[] = {
   [GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_DOUBLE]  = {alignof(double),   sizeof(double),   UPB_WIRE_TYPE_64BIT},
@@ -258,6 +257,7 @@ void upb_parse_reset(struct upb_parse_state *state, void *udata)
   /* The top-level message is not delimited (we can keep receiving data for
    * it indefinitely), so we treat it like a group. */
   *state->top = 0;
+  state->completed_offset = 0;
   state->udata = udata;
 }
 
@@ -272,34 +272,23 @@ void upb_parse_free(struct upb_parse_state *state)
   free(state->stack);
 }
 
-static void *pop_stack_frame(struct upb_parse_state *s,
-                             uint8_t *buf, uint8_t *submsg_end)
+static void *pop_stack_frame(struct upb_parse_state *s, uint8_t *buf)
 {
   if(s->submsg_end_cb) s->submsg_end_cb(s->udata);
-  uint32_t final_submsg_len = *s->top - (buf - submsg_end);
   s->top--;
-  *s->top -= final_submsg_len;
-  return (char*)buf + (*s->top > 0 ? *s->top : 0);
+  return (char*)buf + (*s->top > 0 ? (*s->top - s->completed_offset) : 0);
 }
 
 /* Returns the next end offset. */
 static void *push_stack_frame(struct upb_parse_state *s,
-                              uint8_t *buf, uint8_t *submsg_end, uint32_t len,
+                              uint8_t *buf, uint32_t len,
                               void *user_field_desc, jmp_buf errjmp)
 {
-  *s->top -= len;
-  if(*s->top < 0) *s->top -= (buf - submsg_end);
   s->top++;
   if(s->top > s->limit) longjmp(errjmp, UPB_ERROR_STACK_OVERFLOW);
-  *s->top = len;
+  *s->top = s->completed_offset + len;
   if(s->submsg_start_cb) s->submsg_start_cb(s->udata, user_field_desc);
-  return (char*)buf + *s->top;
-}
-
-upb_status_t upb_isstringtype(upb_field_type_t type)
-{
-  return type == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_STRING ||
-         type == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_BYTES;
+  return (char*)buf + (*s->top > 0 ? (*s->top - s->completed_offset) : 0);
 }
 
 upb_status_t upb_parse(struct upb_parse_state *s, void *_buf, size_t len,
@@ -320,7 +309,7 @@ upb_status_t upb_parse(struct upb_parse_state *s, void *_buf, size_t len,
     struct upb_tag tag;
     buf = parse_tag(buf, end, &tag, errjmp);
     if(tag.wire_type == UPB_WIRE_TYPE_END_GROUP) {
-      submsg_end = pop_stack_frame(s, buf, submsg_end);
+      submsg_end = pop_stack_frame(s, start);
       completed = buf;
       continue;
     }
@@ -344,7 +333,7 @@ upb_status_t upb_parse(struct upb_parse_state *s, void *_buf, size_t len,
       }
 
       if(ft == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_MESSAGE) {
-        submsg_end = push_stack_frame(s, buf, submsg_end, delim_len, user_field_desc, errjmp);
+        submsg_end = push_stack_frame(s, start, delim_end - start, user_field_desc, errjmp);
       } else {  /* Delimited data for which we require (and have) all data. */
         if(ft == 0) {
           /* Do nothing -- client has elected to skip. */
@@ -361,16 +350,17 @@ upb_status_t upb_parse(struct upb_parse_state *s, void *_buf, size_t len,
       if(ft == 0)  /* Client elected to skip. */
         buf = skip_wire_value(buf, end, tag.wire_type, errjmp);
       else if(ft == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_GROUP)
-        submsg_end = push_stack_frame(s, buf, submsg_end, 0, user_field_desc, errjmp);
+        submsg_end = push_stack_frame(s, start, 0, user_field_desc, errjmp);
       else
         buf = s->value_cb(s->udata, buf, end, user_field_desc, errjmp);
     }
 
-    while(buf == submsg_end) submsg_end = pop_stack_frame(s, buf, submsg_end);
+    while(buf == submsg_end) submsg_end = pop_stack_frame(s, start);
     completed = buf;
   }
 
 done:
   *read = (char*)completed - (char*)start;
+  s->completed_offset += *read;
   return status;
 }
diff --git a/src/upb_parse.h b/src/upb_parse.h
index 4a432d8..8975052 100644
--- a/src/upb_parse.h
+++ b/src/upb_parse.h
@@ -16,6 +16,7 @@
 #include <stdbool.h>
 #include <stdint.h>
 #include "upb.h"
+#include "descriptor.h"
 
 #ifdef __cplusplus
 extern "C" {
@@ -48,6 +49,16 @@ struct upb_tag {
   upb_wire_type_t wire_type;
 };
 
+INLINE bool upb_issubmsgtype(upb_field_type_t type) {
+  return type == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_GROUP  ||
+         type == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_MESSAGE;
+}
+
+INLINE bool upb_isstringtype(upb_field_type_t type) {
+  return type == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_STRING  ||
+         type == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_BYTES;
+}
+
 /* High-level parsing interface. **********************************************/
 
 /* The general scheme is that the client registers callbacks that will be
@@ -90,9 +101,8 @@ typedef void *(*upb_value_cb)(void *udata, void *buf, void *end,
                               void *user_field_desc, jmp_buf errjmp);
 
 /* The callback that is called when a string is parsed. */
-typedef upb_status_t (*upb_str_cb)(void *udata,
-                                   struct upb_string *str,
-                                   void *user_field_desc);
+typedef void (*upb_str_cb)(void *udata, struct upb_string *str,
+                           void *user_field_desc);
 
 /* Callbacks that are called when a submessage begins and ends, respectively.
  * Both are called with the submessage's stack frame at the top of the stack. */
@@ -103,7 +113,8 @@ typedef void (*upb_submsg_end_cb)(void *udata);
 struct upb_parse_state {
   /* For delimited submsgs, counts from the submsg len down to zero.
    * For group submsgs, counts from zero down to the negative len. */
-  int32_t stack[UPB_MAX_NESTING], *top, *limit;
+  uint32_t stack[UPB_MAX_NESTING], *top, *limit;
+  size_t completed_offset;
   void *udata;
   upb_tag_cb          tag_cb;
   upb_value_cb        value_cb;