A bit more work on generalizing parsing.

7 files changed, 91 insertions, 58 deletions

diff --git a/upb.h b/upb.h
index f5f19d3..563314c 100644
--- a/upb.h
+++ b/upb.h
@@ -112,6 +112,7 @@ union upb_value_ptr {
   bool *_bool;
   struct upb_string **string;
   struct upb_array **array;
+  void **message;
   void *_void;
 };
 
diff --git a/upb_context.c b/upb_context.c
index 9fd6cef..e124583 100644
--- a/upb_context.c
+++ b/upb_context.c
@@ -232,7 +232,7 @@ error:
 }
 
 bool upb_context_parsefd(struct upb_context *c, struct upb_string *fd_str) {
-  google_protobuf_FileDescriptorProto *fd = upb_msg_parse(c->fd_msg, fd_str);
+  google_protobuf_FileDescriptorProto *fd = upb_alloc_and_parse(c->fd_msg, fd_str);
   if(!fd) return false;
   if(!upb_context_addfd(c, fd)) return false;
   c->fd[c->fd_len++] = fd;  /* Need to keep a ref since we own it. */
diff --git a/upb_context.h b/upb_context.h
index e15b4b8..cd62e30 100644
--- a/upb_context.h
+++ b/upb_context.h
@@ -70,8 +70,8 @@ INLINE struct upb_symtab_entry *upb_context_symbegin(struct upb_context *c) {
 }
 
 INLINE struct upb_symtab_entry *upb_context_symnext(
-    struct upb_context *c, struct upb_inttable_entry *cur) {
-  return upb_strtable_next(&c->symtab, cur);
+    struct upb_context *c, struct upb_symtab_entry *cur) {
+  return upb_strtable_next(&c->symtab, &cur->e);
 }
 
 /* Adding symbols. ************************************************************/
diff --git a/upb_msg.c b/upb_msg.c
index bf55764..1c79f57 100644
--- a/upb_msg.c
+++ b/upb_msg.c
@@ -113,7 +113,7 @@ struct mm_upb_array {
   char *data;
 };
 
-uint32_t round_up_to_pow2(uint32_t v)
+static uint32_t round_up_to_pow2(uint32_t v)
 {
 #ifdef __GNUC__
   return (1U<<31) >> (__builtin_clz(v-1)+1);
@@ -168,6 +168,14 @@ struct parse_frame_data {
   void *data;
 };
 
+static void set_frame_data(struct upb_parse_state *s, struct upb_msg *m,
+                           void *data)
+{
+  struct parse_frame_data *frame = (void*)s->top->user_data;
+  frame->m = m;
+  frame->data = data;
+}
+
 static upb_field_type_t tag_cb(struct upb_parse_state *s, struct upb_tag *tag,
                                void **user_field_desc)
 {
@@ -179,32 +187,36 @@ static upb_field_type_t tag_cb(struct upb_parse_state *s, struct upb_tag *tag,
   return f->type;
 }
 
+static union upb_value_ptr get_value_ptr(void *data, struct upb_msg_field *f)
+{
+  union upb_value_ptr p = upb_msg_get_ptr(data, f);
+  if(f->label == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_LABEL_REPEATED) {
+    size_t len = upb_msg_is_set(data, f) ? (*p.array)->len : 0;
+    upb_msg_reuse_array(p.array, len, f->type);
+    (*p.array)->len = len + 1;
+    p = upb_array_getelementptr(*p.array, len, f->type);
+  }
+  return p;
+}
+
 static upb_status_t value_cb(struct upb_parse_state *s, void **buf, void *end,
-                             upb_field_type_t type, void *user_field_desc)
+                             void *user_field_desc)
 {
   struct parse_frame_data *frame = (void*)s->top->user_data;
   struct upb_msg_field *f = user_field_desc;
-  union upb_value_ptr p = upb_msg_get_ptr(frame->data, f);
-  if(f->label == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_LABEL_REPEATED) {
-    upb_msg_reuse_array(p.array, (*p.array)->len, type);
-    p = upb_array_getelementptr(*p.array, (*p.array)->len++, type);
-  }
-  UPB_CHECK(upb_parse_value(buf, end, type, p));
+  union upb_value_ptr p = get_value_ptr(frame->data, f);
+  UPB_CHECK(upb_parse_value(buf, end, f->type, p));
   return UPB_STATUS_OK;
 }
 
-static upb_status_t str_cb(struct upb_parse_state *s, struct upb_string *str,
-                           upb_field_type_t type, void *user_field_desc)
+static upb_status_t str_cb(struct upb_parse_state *_s, struct upb_string *str,
+                           void *user_field_desc)
 {
-  struct parse_frame_data *frame = (void*)s->top->user_data;
+  struct upb_msg_parse_state *s = (void*)_s;
+  struct parse_frame_data *frame = (void*)s->s.top->user_data;
   struct upb_msg_field *f = user_field_desc;
-  union upb_value_ptr p = upb_msg_get_ptr(frame->data, f);
-  if(f->label == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_LABEL_REPEATED) {
-    upb_msg_reuse_array(p.array, (*p.array)->len, type);
-    p = upb_array_getelementptr(*p.array, (*p.array)->len++, type);
-  }
-  bool byref = false;
-  if(byref) {
+  union upb_value_ptr p = get_value_ptr(frame->data, f);
+  if(s->byref) {
     upb_msg_reuse_strref(p.string);
     **p.string = *str;
   } else {
@@ -214,33 +226,39 @@ static upb_status_t str_cb(struct upb_parse_state *s, struct upb_string *str,
   return UPB_STATUS_OK;
 }
 
-static void set_frame_data(struct upb_parse_state *s, struct upb_msg *m,
-                           void *data)
+static void submsg_start_cb(struct upb_parse_state *_s, void *user_field_desc)
 {
-  struct parse_frame_data *frame = (void*)s->top->user_data;
-  frame->m = m;
-  frame->data = data;
+  struct upb_msg_parse_state *s = (void*)_s;
+  struct upb_msg_field *f = user_field_desc;
+  struct parse_frame_data *frame = (void*)s->s.top->user_data;
+  union upb_value_ptr p = upb_msg_get_ptr(frame->data, f);
+  upb_msg_reuse_submsg(*p.message, f->ref.msg);
+  if(!s->merge) upb_msg_clear(frame->data, f->ref.msg);
+  set_frame_data(&s->s, f->ref.msg, *p.message);
+}
+
+
+void upb_msg_parse_init(struct upb_msg_parse_state *s, void *msg,
+                        struct upb_msg *m, bool merge, bool byref)
+{
+  upb_parse_init(&s->s, sizeof(struct parse_frame_data));
+  s->merge = merge;
+  s->byref = byref;
+  if(!merge && msg == NULL) msg = upb_msg_new(m);
+  set_frame_data(&s->s, m, msg);
+  s->s.tag_cb = tag_cb;
+  s->s.value_cb = value_cb;
+  s->s.str_cb = str_cb;
+  s->s.submsg_start_cb = submsg_start_cb;
 }
 
-static void submsg_start_cb(struct upb_parse_state *s, void *user_field_desc)
+void upb_msg_parse_free(struct upb_msg_parse_state *s)
 {
-  struct upb_msg_field *f = user_field_desc;
-  struct parse_frame_data *frame = (void*)s->top->user_data;
-  void **submsg = upb_msg_get_submsg_ptr(frame->data, f);
-  upb_msg_reuse_submsg(submsg, f->ref.msg);
-  set_frame_data(s, f->ref.msg, *submsg);
-}
-
-upb_status_t upb_msg_merge(void *data, struct upb_msg *m, struct upb_string *str)
-{
-  struct upb_parse_state s;
-  upb_parse_state_init(&s, sizeof(struct parse_frame_data));
-  set_frame_data(&s, m, data);
-  s.tag_cb = tag_cb;
-  s.value_cb = value_cb;
-  s.str_cb = str_cb;
-  s.submsg_start_cb = submsg_start_cb;
-  size_t read;
-  UPB_CHECK(upb_parse(&s, str->ptr, str->byte_len, &read));
-  return UPB_STATUS_OK;
+  upb_parse_free(&s->s);
+}
+
+upb_status_t upb_msg_parse(struct upb_msg_parse_state *s,
+                           void *data, size_t len, size_t *read)
+{
+  return upb_parse(&s->s, data, len, read);
 }
diff --git a/upb_msg.h b/upb_msg.h
index 605ca94..48b411f 100644
--- a/upb_msg.h
+++ b/upb_msg.h
@@ -46,6 +46,7 @@
 
 #include "upb.h"
 #include "upb_table.h"
+#include "upb_parse.h"
 
 #ifdef __cplusplus
 extern "C" {
@@ -57,13 +58,14 @@ struct google_protobuf_FieldDescriptorProto;
 
 /* Message definition. ********************************************************/
 
-/* Structure that describes a single field in a message. */
+/* Structure that describes a single field in a message.  This structure is very
+ * consciously designed to fit into 12/16 bytes (32/64 bit, respectively). */
 struct upb_msg_field {
+  union upb_symbol_ref ref;
   uint32_t byte_offset;     /* Where to find the data. */
   uint16_t field_index;     /* Indexes upb_msg.fields. Also indicates set bit */
   upb_field_type_t type;    /* Copied from descriptor for cache-friendliness. */
   upb_label_t label;
-  union upb_symbol_ref ref;
 };
 
 /* Structure that describes a single .proto message type. */
@@ -155,8 +157,8 @@ struct upb_array {
     uint32_t len; \
   };
 
-union upb_value_ptr upb_array_getelementptr(struct upb_array *arr, uint32_t n,
-                                            upb_field_type_t type)
+INLINE union upb_value_ptr upb_array_getelementptr(
+    struct upb_array *arr, uint32_t n, upb_field_type_t type)
 {
   union upb_value_ptr ptr = {
     ._void = ((char*)arr->elements._void + n*upb_type_info[type].size)
@@ -226,7 +228,21 @@ INLINE union upb_value_ptr upb_msg_get_ptr(
 /* Memory management  *********************************************************/
 
 void *upb_msg_new(struct upb_msg *m);
-//void upb_msg_free(void *msg, struct upb_msg *m, bool free_submsgs);
+
+struct upb_msg_parse_state {
+  struct upb_parse_state s;
+  bool merge;
+  bool byref;
+  struct upb_msg *m;
+};
+
+void upb_msg_parse_init(struct upb_msg_parse_state *s, void *msg,
+                        struct upb_msg *m, bool merge, bool byref);
+void upb_msg_parse_free(struct upb_msg_parse_state *s);
+upb_status_t upb_msg_parse(struct upb_msg_parse_state *s,
+                           void *data, size_t len, size_t *read);
+
+void *upb_alloc_and_parse(struct upb_msg *m, struct upb_string *s);
 
 /* Note!  These two may not be use on a upb_string* that was initialized by
  * means other than these functions. */
diff --git a/upb_parse.c b/upb_parse.c
index 94eb19a..559e227 100644
--- a/upb_parse.c
+++ b/upb_parse.c
@@ -308,12 +308,12 @@ static upb_status_t parse_delimited(struct upb_parse_state *s,
     if(ft == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_STRING ||
        ft == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_BYTES) {
       struct upb_string str = {.ptr = *buf, .byte_len = delim_len};
-      s->str_cb(s, &str, ft, user_field_desc);
+      s->str_cb(s, &str, user_field_desc);
       *buf = delim_end;
     } else {
       /* Packed Array. */
       while(*buf < delim_end)
-        UPB_CHECK(s->value_cb(s, buf, end, ft, user_field_desc));
+        UPB_CHECK(s->value_cb(s, buf, end, user_field_desc));
     }
   }
   return UPB_STATUS_OK;
@@ -332,7 +332,7 @@ static upb_status_t parse_nondelimited(struct upb_parse_state *s,
     /* No length specified, an "end group" tag will mark the end. */
     UPB_CHECK(push_stack_frame(s, 0, user_field_desc));
   } else {
-    UPB_CHECK(s->value_cb(s, buf, end, ft, user_field_desc));
+    UPB_CHECK(s->value_cb(s, buf, end, user_field_desc));
   }
   return UPB_STATUS_OK;
 }
diff --git a/upb_parse.h b/upb_parse.h
index 14520f3..83104f1 100644
--- a/upb_parse.h
+++ b/upb_parse.h
@@ -25,8 +25,8 @@ struct upb_parse_state;
 /* Initialize and free (respectively) the given parse state, which must have
  * been previously allocated.  udata_size specifies how much space will be
  * available at parse_stack_frame.user_data in each frame for user data. */
-void upb_parse_state_init(struct upb_parse_state *state, size_t udata_size);
-void upb_parse_state_free(struct upb_parse_state *state);
+void upb_parse_init(struct upb_parse_state *state, size_t udata_size);
+void upb_parse_free(struct upb_parse_state *state);
 
 /* The callback that is called immediately after a tag has been parsed.  The
  * client should determine whether it wants to parse or skip the corresponding
@@ -48,13 +48,11 @@ typedef upb_field_type_t (*upb_tag_cb)(struct upb_parse_state *s,
  * call to tag_cb in the case of packed arrays. */
 typedef upb_status_t (*upb_value_cb)(struct upb_parse_state *s,
                                      void **buf, void *end,
-                                     upb_field_type_t type,
                                      void *user_field_desc);
 
 /* The callback that is called when a string is parsed. */
 typedef upb_status_t (*upb_str_cb)(struct upb_parse_state *s,
                                    struct upb_string *str,
-                                   upb_field_type_t type,
                                    void *user_field_desc);
 
 /* Callbacks that are called when a submessage begins and ends, respectively.