More tests and bugfixes -- parses its first proto!

3 files changed, 61 insertions, 14 deletions

diff --git a/pbstream.c b/pbstream.c
index c8711e7..82fc4d2 100644
--- a/pbstream.c
+++ b/pbstream.c
@@ -159,7 +159,9 @@ WVTOV_DELIMITED(MESSAGE);
 static pbstream_status_t get_STRING(struct pbstream_parse_state *s, char *buf,
                                     struct pbstream_value *d) {
   uint32_t tmp;
-  CHECK(get_v_uint32_t(&buf, &tmp));
+  char *b = buf;
+  CHECK(get_v_uint32_t(&b, &tmp));
+  s->offset += (b-buf);  /* advance past length varint. */
   wvtov_STRING(tmp, &d->v.delimited, s->offset);
   s->offset = d->v.delimited.offset + d->v.delimited.len; /* skip string */
   /* we leave UTF-8 validation to the client. */
@@ -169,7 +171,9 @@ static pbstream_status_t get_STRING(struct pbstream_parse_state *s, char *buf,
 static pbstream_status_t get_BYTES(struct pbstream_parse_state *s, char *buf,
                                    struct pbstream_value *d) {
   uint32_t tmp;
-  CHECK(get_v_uint32_t(&buf, &tmp));
+  char *b = buf;
+  CHECK(get_v_uint32_t(&b, &tmp));
+  s->offset += (b-buf);  /* advance past length varint. */
   wvtov_BYTES(tmp, &d->v.delimited, s->offset);
   s->offset = d->v.delimited.offset + d->v.delimited.len; /* skip bytes */
   return PBSTREAM_STATUS_OK;
@@ -179,9 +183,11 @@ static pbstream_status_t get_MESSAGE(struct pbstream_parse_state *s, char *buf,
                                      struct pbstream_value *d) {
   /* We're entering a sub-message. */
   uint32_t tmp;
+  char *b = buf;
   CHECK(get_v_uint32_t(&buf, &tmp));
+  s->offset += (b-buf);  /* advance past length varint. */
   wvtov_MESSAGE(tmp, &d->v.delimited, s->offset);
-  s->offset = d->v.delimited.offset;  /* skip past only the tag. */
+  /* Unlike STRING and BYTES, we *don't* advance past delimited here. */
   if (unlikely(++s->top == s->limit)) {
     /* Stack has grown beyond its limit, must reallocate. */
     int cur_size = s->top - s->base;
@@ -254,7 +260,8 @@ static struct pbstream_field *find_field(struct pbstream_fieldset* fs,
                                          pbstream_field_number_t num)
 {
   /* TODO: a hybrid array/hashtable structure. */
-  if(num < fs->num_fields) return &fs->fields[num];
+  /* TODO: can zero be a tag number? */
+  if(num <= fs->num_fields) return &fs->fields[num-1];
   else return NULL;
 }
 
@@ -277,7 +284,7 @@ pbstream_status_t pbstream_parse_field(struct pbstream_parse_state *s,
 
   struct pbstream_tag tag;
   CHECK(parse_tag(&b, &tag));
-  size_t val_offset = s->offset + (b-buf);
+  s->offset += (b-buf);
   struct pbstream_field *fd = find_field(s->top->fieldset, tag.field_number);
   pbstream_status_t unknown_value_status;
   if(unlikely(!fd)) {
@@ -297,22 +304,25 @@ pbstream_status_t pbstream_parse_field(struct pbstream_parse_state *s,
 
 unknown_value:
   wv->type = tag.wire_type;
-  CHECK(parse_unknown_value(&b, val_offset, wv));
+  CHECK(parse_unknown_value(&b, s->offset, wv));
   s->offset += (b-buf);
   return unknown_value_status;
 }
 
 void pbstream_init_parser(
     struct pbstream_parse_state *state,
-    struct pbstream_fieldset *toplevel_fieldset,
-    void *user_data)
+    struct pbstream_fieldset *toplevel_fieldset)
 {
   state->offset = 0;
-  state->user_data = user_data;
-  /* Initial stack of <300b most protobufs are unlikely to nest >20 deep. */
+  /* Initial stack of <300b, most protobufs are unlikely to nest >20 deep. */
   const int initial_stack = 20;
   state->top = state->base = malloc(sizeof(*state->base) * initial_stack); 
   state->limit = state->base + initial_stack;
   state->top->fieldset = toplevel_fieldset;
   state->top->end_offset = SIZE_MAX;
 }
+
+void pbstream_free_parser(struct pbstream_parse_state *state)
+{
+  free(state->base);
+}
diff --git a/pbstream.h b/pbstream.h
index 2713974..89e7329 100644
--- a/pbstream.h
+++ b/pbstream.h
@@ -9,7 +9,7 @@
 
 /* A list of types as they can appear in a .proto file. */
 typedef enum pbstream_type {
-  PBSTREAM_TYPE_DOUBLE,
+  PBSTREAM_TYPE_DOUBLE = 0,
   PBSTREAM_TYPE_FLOAT,
   PBSTREAM_TYPE_INT32,
   PBSTREAM_TYPE_INT64,
@@ -101,7 +101,6 @@ struct pbstream_parse_stack_frame {
 /* The stream parser's state. */
 struct pbstream_parse_state {
   size_t offset;
-  void *user_data;
   struct pbstream_parse_stack_frame *base, *top, *limit;
 };
 
@@ -110,8 +109,9 @@ struct pbstream_parse_state {
  * unknown. */
 void pbstream_init_parser(
     struct pbstream_parse_state *state,
-    struct pbstream_fieldset *toplevel_fieldset,
-    void *user_data);
+    struct pbstream_fieldset *toplevel_fieldset);
+
+void pbstream_free_parser(struct pbstream_parse_state *state);
 
 /* Status as returned by pbstream_parse().  Status codes <0 are fatal errors
  * that cannot be recovered.  Status codes >0 are unusual but nonfatal events,
diff --git a/tests.c b/tests.c
index 4277a19..ea87d79 100644
--- a/tests.c
+++ b/tests.c
@@ -42,9 +42,46 @@ void test_get_v_uint64_t()
   assert(status == PBSTREAM_ERROR_UNTERMINATED_VARINT);
 }
 
+void test_simple_proto()
+{
+  struct pbstream_fieldset *fieldset1 = malloc(sizeof(*fieldset1) +
+                                               2*sizeof(struct pbstream_field));
+  fieldset1->num_fields = 2;
+  fieldset1->fields[0].field_number = 1;
+  fieldset1->fields[0].type = PBSTREAM_TYPE_INT32;
+  fieldset1->fields[1].field_number = 2;
+  fieldset1->fields[1].type = PBSTREAM_TYPE_STRING;
+
+  char message1[] = {0x08, 0x96, 0x01};
+  struct pbstream_parse_state s;
+  pbstream_init_parser(&s, fieldset1);
+  assert(s.offset == 0);
+  pbstream_field_number_t fieldnum;
+  struct pbstream_value val;
+  struct pbstream_wire_value wv;
+  assert(pbstream_parse_field(&s, message1, &fieldnum, &val, &wv) ==
+         PBSTREAM_STATUS_OK);
+  assert(val.field == &fieldset1->fields[0]);
+  assert(val.v.int32 == 150);
+  assert(s.offset == 3);
+  pbstream_free_parser(&s);
+
+  char message2[] = {0x12, 0x07, 0x74, 0x65, 0x73, 0x74, 0x69, 0x6e, 0x67};
+  pbstream_init_parser(&s, fieldset1);
+  assert(pbstream_parse_field(&s, message2, &fieldnum, &val, &wv) ==
+         PBSTREAM_STATUS_OK);
+  assert(val.field == &fieldset1->fields[1]);
+  assert(val.v.delimited.offset == 2);
+  assert(val.v.delimited.len == 7);
+  assert(s.offset == 9);
+  pbstream_free_parser(&s);
+  free(fieldset1);
+}
+
 int main()
 {
   test_get_v_uint64_t();
+  test_simple_proto();
   printf("All tests passed.\n");
   return 0;
 }