Chiebot-Mirror
/
protobuf
mirror of https://github.com/protocolbuffers/protobuf.git
8
0
Fork 0
Code Issues Projects Releases Wiki Activity

A bit more work on generalizing parsing.

Browse Source
pull/13171/head
Joshua Haberman 16 years ago
parent 6fdf83f59f
commit b8481e0e55
7 changed files with 91 additions and 58 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Unified View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 1
      upb.h
  2. 2
      upb_context.c
  3. 4
      upb_context.h
  4. 104
      upb_msg.c
  5. 26
      upb_msg.h
  6. 6
      upb_parse.c
  7. 6
      upb_parse.h

1
upb.h
Unescape View File

@ -112,6 +112,7 @@ union upb_value_ptr {
bool *_bool; bool *_bool;
struct upb_string **string; struct upb_string **string;
struct upb_array **array; struct upb_array **array;
void **message;
void *_void; void *_void;
}; };

2
upb_context.c
Unescape View File

@ -232,7 +232,7 @@ error:
} }
bool upb_context_parsefd(struct upb_context *c, struct upb_string *fd_str) { 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(!fd) return false;
if(!upb_context_addfd(c, 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. */ c->fd[c->fd_len++] = fd; /* Need to keep a ref since we own it. */

4
upb_context.h
Unescape View File

@ -70,8 +70,8 @@ INLINE struct upb_symtab_entry *upb_context_symbegin(struct upb_context *c) {
} }
INLINE struct upb_symtab_entry *upb_context_symnext( INLINE struct upb_symtab_entry *upb_context_symnext(
struct upb_context *c, struct upb_inttable_entry *cur) { struct upb_context *c, struct upb_symtab_entry *cur) {
return upb_strtable_next(&c->symtab, cur); return upb_strtable_next(&c->symtab, &cur->e);
} }
/* Adding symbols. ************************************************************/ /* Adding symbols. ************************************************************/

104
upb_msg.c
Unescape View File

@ -113,7 +113,7 @@ struct mm_upb_array {
char *data; char *data;
}; };
uint32_t round_up_to_pow2(uint32_t v) static uint32_t round_up_to_pow2(uint32_t v)
{ {
#ifdef __GNUC__ #ifdef __GNUC__
return (1U<<31) >> (__builtin_clz(v-1)+1); return (1U<<31) >> (__builtin_clz(v-1)+1);
@ -168,6 +168,14 @@ struct parse_frame_data {
void *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, static upb_field_type_t tag_cb(struct upb_parse_state *s, struct upb_tag *tag,
void **user_field_desc) 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; 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, 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 parse_frame_data *frame = (void*)s->top->user_data;
struct upb_msg_field *f = user_field_desc; struct upb_msg_field *f = user_field_desc;
union upb_value_ptr p = upb_msg_get_ptr(frame->data, f); union upb_value_ptr p = get_value_ptr(frame->data, f);
if(f->label == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_LABEL_REPEATED) { UPB_CHECK(upb_parse_value(buf, end, f->type, p));
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));
return UPB_STATUS_OK; return UPB_STATUS_OK;
} }
static upb_status_t str_cb(struct upb_parse_state *s, struct upb_string *str, static upb_status_t str_cb(struct upb_parse_state *_s, struct upb_string *str,
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_parse_state *s = (void*)_s;
struct parse_frame_data *frame = (void*)s->s.top->user_data;
struct upb_msg_field *f = user_field_desc; struct upb_msg_field *f = user_field_desc;
union upb_value_ptr p = upb_msg_get_ptr(frame->data, f); union upb_value_ptr p = get_value_ptr(frame->data, f);
if(f->label == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_LABEL_REPEATED) { if(s->byref) {
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) {
upb_msg_reuse_strref(p.string); upb_msg_reuse_strref(p.string);
**p.string = *str; **p.string = *str;
} else { } else {
@ -214,33 +226,39 @@ static upb_status_t str_cb(struct upb_parse_state *s, struct upb_string *str,
return UPB_STATUS_OK; return UPB_STATUS_OK;
} }
static void set_frame_data(struct upb_parse_state *s, struct upb_msg *m, static void submsg_start_cb(struct upb_parse_state *_s, void *user_field_desc)
void *data)
{ {
struct parse_frame_data *frame = (void*)s->top->user_data; struct upb_msg_parse_state *s = (void*)_s;
frame->m = m; struct upb_msg_field *f = user_field_desc;
frame->data = data; 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; upb_parse_free(&s->s);
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); upb_status_t upb_msg_parse(struct upb_msg_parse_state *s,
set_frame_data(s, f->ref.msg, *submsg); void *data, size_t len, size_t *read)
} {
return upb_parse(&s->s, data, len, read);
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;
} }

26
upb_msg.h
Unescape View File

@ -46,6 +46,7 @@
#include "upb.h" #include "upb.h"
#include "upb_table.h" #include "upb_table.h"
#include "upb_parse.h"
#ifdef __cplusplus #ifdef __cplusplus
extern "C" { extern "C" {
@ -57,13 +58,14 @@ struct google_protobuf_FieldDescriptorProto;
/* Message definition. ********************************************************/ /* 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 { struct upb_msg_field {
union upb_symbol_ref ref;
uint32_t byte_offset; /* Where to find the data. */ uint32_t byte_offset; /* Where to find the data. */
uint16_t field_index; /* Indexes upb_msg.fields. Also indicates set bit */ uint16_t field_index; /* Indexes upb_msg.fields. Also indicates set bit */
upb_field_type_t type; /* Copied from descriptor for cache-friendliness. */ upb_field_type_t type; /* Copied from descriptor for cache-friendliness. */
upb_label_t label; upb_label_t label;
union upb_symbol_ref ref;
}; };
/* Structure that describes a single .proto message type. */ /* Structure that describes a single .proto message type. */
@ -155,8 +157,8 @@ struct upb_array {
uint32_t len; \ uint32_t len; \
}; };
union upb_value_ptr upb_array_getelementptr(struct upb_array *arr, uint32_t n, INLINE union upb_value_ptr upb_array_getelementptr(
upb_field_type_t type) struct upb_array *arr, uint32_t n, upb_field_type_t type)
{ {
union upb_value_ptr ptr = { union upb_value_ptr ptr = {
._void = ((char*)arr->elements._void + n*upb_type_info[type].size) ._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 *********************************************************/ /* Memory management *********************************************************/
void *upb_msg_new(struct upb_msg *m); 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 /* Note! These two may not be use on a upb_string* that was initialized by
* means other than these functions. */ * means other than these functions. */

6
upb_parse.c
Unescape View File

@ -308,12 +308,12 @@ static upb_status_t parse_delimited(struct upb_parse_state *s,
if(ft == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_STRING || if(ft == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_STRING ||
ft == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_BYTES) { ft == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_BYTES) {
struct upb_string str = {.ptr = *buf, .byte_len = delim_len}; 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; *buf = delim_end;
} else { } else {
/* Packed Array. */ /* Packed Array. */
while(*buf < delim_end) 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; 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. */ /* No length specified, an "end group" tag will mark the end. */
UPB_CHECK(push_stack_frame(s, 0, user_field_desc)); UPB_CHECK(push_stack_frame(s, 0, user_field_desc));
} else { } 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; return UPB_STATUS_OK;
} }

6
upb_parse.h
Unescape View File

@ -25,8 +25,8 @@ struct upb_parse_state;
/* Initialize and free (respectively) the given parse state, which must have /* Initialize and free (respectively) the given parse state, which must have
* been previously allocated. udata_size specifies how much space will be * been previously allocated. udata_size specifies how much space will be
* available at parse_stack_frame.user_data in each frame for user data. */ * 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_init(struct upb_parse_state *state, size_t udata_size);
void upb_parse_state_free(struct upb_parse_state *state); void upb_parse_free(struct upb_parse_state *state);
/* The callback that is called immediately after a tag has been parsed. The /* 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 * 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. */ * call to tag_cb in the case of packed arrays. */
typedef upb_status_t (*upb_value_cb)(struct upb_parse_state *s, typedef upb_status_t (*upb_value_cb)(struct upb_parse_state *s,
void **buf, void *end, void **buf, void *end,
upb_field_type_t type,
void *user_field_desc); void *user_field_desc);
/* The callback that is called when a string is parsed. */ /* The callback that is called when a string is parsed. */
typedef upb_status_t (*upb_str_cb)(struct upb_parse_state *s, typedef upb_status_t (*upb_str_cb)(struct upb_parse_state *s,
struct upb_string *str, struct upb_string *str,
upb_field_type_t type,
void *user_field_desc); void *user_field_desc);
/* Callbacks that are called when a submessage begins and ends, respectively. /* Callbacks that are called when a submessage begins and ends, respectively.

Write Preview
Loading…
Cancel
Save