2 +--------------------------------------------------------------------+
4 +--------------------------------------------------------------------+
5 | Redistribution and use in source and binary forms, with or without |
6 | modification, are permitted provided that the conditions mentioned |
7 | in the accompanying LICENSE file are met. |
8 +--------------------------------------------------------------------+
9 | Copyright (c) 2004-2011, Michael Wallner <mike@php.net> |
10 +--------------------------------------------------------------------+
13 #include "php_http_api.h"
15 typedef struct php_http_message_parser_state_spec
{
16 php_http_message_parser_state_t state
;
18 } php_http_message_parser_state_spec_t
;
20 static const php_http_message_parser_state_spec_t php_http_message_parser_states
[] = {
21 {PHP_HTTP_MESSAGE_PARSER_STATE_START
, 1},
22 {PHP_HTTP_MESSAGE_PARSER_STATE_HEADER
, 1},
23 {PHP_HTTP_MESSAGE_PARSER_STATE_HEADER_DONE
, 0},
24 {PHP_HTTP_MESSAGE_PARSER_STATE_BODY
, 0},
25 {PHP_HTTP_MESSAGE_PARSER_STATE_BODY_DUMB
, 1},
26 {PHP_HTTP_MESSAGE_PARSER_STATE_BODY_LENGTH
, 1},
27 {PHP_HTTP_MESSAGE_PARSER_STATE_BODY_CHUNKED
, 1},
28 {PHP_HTTP_MESSAGE_PARSER_STATE_BODY_DONE
, 0},
29 {PHP_HTTP_MESSAGE_PARSER_STATE_DONE
, 0}
32 PHP_HTTP_API php_http_message_parser_t
*php_http_message_parser_init(php_http_message_parser_t
*parser TSRMLS_DC
)
35 parser
= emalloc(sizeof(*parser
));
37 memset(parser
, 0, sizeof(*parser
));
39 TSRMLS_SET_CTX(parser
->ts
);
41 php_http_header_parser_init(&parser
->header TSRMLS_CC
);
42 zend_stack_init(&parser
->stack
);
47 PHP_HTTP_API php_http_message_parser_state_t
php_http_message_parser_state_push(php_http_message_parser_t
*parser
, unsigned argc
, ...)
51 va_start(va_args
, argc
);
52 php_http_message_parser_state_t state
;
54 for (i
= 0; i
< argc
; ++i
) {
55 state
= va_arg(va_args
, php_http_message_parser_state_t
);
56 zend_stack_push(&parser
->stack
, &state
, sizeof(state
));
63 PHP_HTTP_API php_http_message_parser_state_t
php_http_message_parser_state_is(php_http_message_parser_t
*parser
)
65 php_http_message_parser_state_t
*state
;
67 if (SUCCESS
== zend_stack_top(&parser
->stack
, (void *) &state
)) {
70 return PHP_HTTP_MESSAGE_PARSER_STATE_START
;
73 PHP_HTTP_API php_http_message_parser_state_t
php_http_message_parser_state_pop(php_http_message_parser_t
*parser
)
75 php_http_message_parser_state_t state
, *state_ptr
;
76 if (SUCCESS
== zend_stack_top(&parser
->stack
, (void *) &state_ptr
)) {
78 zend_stack_del_top(&parser
->stack
);
81 return PHP_HTTP_MESSAGE_PARSER_STATE_START
;
84 PHP_HTTP_API
void php_http_message_parser_dtor(php_http_message_parser_t
*parser
)
86 php_http_header_parser_dtor(&parser
->header
);
87 zend_stack_destroy(&parser
->stack
);
88 if (parser
->dechunk
) {
89 php_http_encoding_stream_free(&parser
->dechunk
);
91 if (parser
->inflate
) {
92 php_http_encoding_stream_free(&parser
->inflate
);
96 PHP_HTTP_API
void php_http_message_parser_free(php_http_message_parser_t
**parser
)
99 php_http_message_parser_dtor(*parser
);
105 PHP_HTTP_API php_http_message_parser_state_t
php_http_message_parser_parse_stream(php_http_message_parser_t
*parser
, php_stream
*s
, php_http_message_t
**message
)
107 php_http_buffer_t buf
;
108 TSRMLS_FETCH_FROM_CTX(parser
->ts
);
110 php_http_buffer_init_ex(&buf
, 0x1000, PHP_HTTP_BUFFER_INIT_PREALLOC
);
112 while (!php_stream_eof(s
)) {
115 switch (php_http_message_parser_state_is(parser
)) {
116 case PHP_HTTP_MESSAGE_PARSER_STATE_START
:
117 case PHP_HTTP_MESSAGE_PARSER_STATE_HEADER
:
118 case PHP_HTTP_MESSAGE_PARSER_STATE_HEADER_DONE
:
120 php_stream_get_line(s
, buf
.data
, buf
.free
, &len
);
125 case PHP_HTTP_MESSAGE_PARSER_STATE_BODY_DUMB
:
127 len
= php_stream_read(s
, buf
.data
, buf
.free
);
132 case PHP_HTTP_MESSAGE_PARSER_STATE_BODY_LENGTH
:
133 /* read body_length */
134 len
= php_stream_read(s
, buf
.data
, MIN(buf
.free
, parser
->body_length
));
139 case PHP_HTTP_MESSAGE_PARSER_STATE_BODY_CHUNKED
:
140 /* duh, this is very naive */
142 size_t read
= php_stream_read(s
, buf
.data
, MIN(len
, buf
.free
));
149 php_stream_get_line(s
, buf
.data
, buf
.free
, &len
);
153 len
= strtoul(buf
.data
- len
, NULL
, 16);
157 case PHP_HTTP_MESSAGE_PARSER_STATE_BODY
:
158 case PHP_HTTP_MESSAGE_PARSER_STATE_BODY_DONE
:
159 /* should not occur */
163 case PHP_HTTP_MESSAGE_PARSER_STATE_DONE
:
164 case PHP_HTTP_MESSAGE_PARSER_STATE_FAILURE
:
165 php_http_buffer_dtor(&buf
);
166 return php_http_message_parser_state_is(parser
);
169 php_http_message_parser_parse(parser
, &buf
, 0, message
);
172 php_http_buffer_dtor(&buf
);
173 return PHP_HTTP_MESSAGE_PARSER_STATE_DONE
;
177 PHP_HTTP_API php_http_message_parser_state_t
php_http_message_parser_parse(php_http_message_parser_t
*parser
, php_http_buffer_t
*buffer
, unsigned flags
, php_http_message_t
**message
)
179 TSRMLS_FETCH_FROM_CTX(parser
->ts
);
184 while (buffer
->used
|| !php_http_message_parser_states
[php_http_message_parser_state_is(parser
)].need_data
) {
186 const char *state
[] = {"START", "HEADER", "HEADER_DONE", "BODY", "BODY_DUMB", "BODY_LENGTH", "BODY_CHUNK", "BODY_DONE", "DONE"};
187 fprintf(stderr
, "#MP: %s (%d)\n", php_http_message_parser_state_is(parser
) < 0 ? "FAILURE" : state
[php_http_message_parser_state_is(parser
)], message
&& *message
? (*message
)->type
: -1);
188 _dpf(0, buffer
->data
, buffer
->used
);
191 switch (php_http_message_parser_state_pop(parser
))
193 case PHP_HTTP_MESSAGE_PARSER_STATE_FAILURE
:
194 return php_http_message_parser_state_push(parser
, 1, PHP_HTTP_MESSAGE_PARSER_STATE_FAILURE
);
196 case PHP_HTTP_MESSAGE_PARSER_STATE_START
:
198 char *ptr
= buffer
->data
;
200 while (ptr
- buffer
->data
< buffer
->used
&& PHP_HTTP_IS_CTYPE(space
, *ptr
)) {
204 php_http_buffer_cut(buffer
, 0, ptr
- buffer
->data
);
207 php_http_message_parser_state_push(parser
, 1, PHP_HTTP_MESSAGE_PARSER_STATE_HEADER
);
212 case PHP_HTTP_MESSAGE_PARSER_STATE_HEADER
:
214 unsigned header_parser_flags
= (flags
& PHP_HTTP_MESSAGE_PARSER_CLEANUP
) ? PHP_HTTP_HEADER_PARSER_CLEANUP
: 0;
216 switch (php_http_header_parser_parse(&parser
->header
, buffer
, header_parser_flags
, *message
? &(*message
)->hdrs
: NULL
, (php_http_info_callback_t
) php_http_message_info_callback
, message
)) {
217 case PHP_HTTP_HEADER_PARSER_STATE_FAILURE
:
218 return PHP_HTTP_MESSAGE_PARSER_STATE_FAILURE
;
220 case PHP_HTTP_HEADER_PARSER_STATE_DONE
:
221 php_http_message_parser_state_push(parser
, 1, PHP_HTTP_MESSAGE_PARSER_STATE_HEADER_DONE
);
225 php_http_message_parser_state_push(parser
, 1, PHP_HTTP_MESSAGE_PARSER_STATE_HEADER
);
227 return PHP_HTTP_MESSAGE_PARSER_STATE_HEADER
;
233 case PHP_HTTP_MESSAGE_PARSER_STATE_HEADER_DONE
:
235 zval
*h
, *h_loc
= NULL
, *h_con
= NULL
, **h_cl
= NULL
, **h_cr
= NULL
, **h_te
= NULL
;
237 if ((h
= php_http_message_header(*message
, ZEND_STRL("Transfer-Encoding"), 1))) {
238 zend_hash_update(&(*message
)->hdrs
, "X-Original-Transfer-Encoding", sizeof("X-Original-Transfer-Encoding"), &h
, sizeof(zval
*), (void *) &h_te
);
239 zend_hash_del(&(*message
)->hdrs
, "Transfer-Encoding", sizeof("Transfer-Encoding"));
241 if ((h
= php_http_message_header(*message
, ZEND_STRL("Content-Length"), 1))) {
242 zend_hash_update(&(*message
)->hdrs
, "X-Original-Content-Length", sizeof("X-Original-Content-Length"), &h
, sizeof(zval
*), (void *) &h_cl
);
244 if ((h
= php_http_message_header(*message
, ZEND_STRL("Content-Range"), 1))) {
245 zend_hash_update(&(*message
)->hdrs
, "X-Original-Content-Range", sizeof("X-Original-Content-Range"), &h
, sizeof(zval
*), (void *) &h_cr
);
246 zend_hash_del(&(*message
)->hdrs
, "Content-Range", sizeof("Content-Range"));
252 zend_hash_update(&(*message
)->hdrs
, "Content-Length", sizeof("Content-Length"), &h
, sizeof(zval
*), NULL
);
254 /* so, if curl sees a 3xx code, a Location header and a Connection:close header
255 * it decides not to read the response body.
257 if ((flags
& PHP_HTTP_MESSAGE_PARSER_EMPTY_REDIRECTS
)
258 && (*message
)->type
== PHP_HTTP_RESPONSE
259 && (*message
)->http
.info
.response
.code
/100 == 3
260 && (h_loc
= php_http_message_header(*message
, ZEND_STRL("Location"), 1))
261 && (h_con
= php_http_message_header(*message
, ZEND_STRL("Connection"), 1))
263 if (php_http_match(Z_STRVAL_P(h_con
), "close", PHP_HTTP_MATCH_WORD
)) {
264 php_http_message_parser_state_push(parser
, 1, PHP_HTTP_MESSAGE_PARSER_STATE_DONE
);
265 zval_ptr_dtor(&h_loc
);
266 zval_ptr_dtor(&h_con
);
271 zval_ptr_dtor(&h_loc
);
274 zval_ptr_dtor(&h_con
);
277 if ((h
= php_http_message_header(*message
, ZEND_STRL("Content-Encoding"), 1))) {
278 if (php_http_match(Z_STRVAL_P(h
), "gzip", PHP_HTTP_MATCH_WORD
)
279 || php_http_match(Z_STRVAL_P(h
), "x-gzip", PHP_HTTP_MATCH_WORD
)
280 || php_http_match(Z_STRVAL_P(h
), "deflate", PHP_HTTP_MATCH_WORD
)
282 if (parser
->inflate
) {
283 php_http_encoding_stream_reset(&parser
->inflate
);
285 parser
->inflate
= php_http_encoding_stream_init(NULL
, php_http_encoding_stream_get_inflate_ops(), 0 TSRMLS_CC
);
287 zend_hash_update(&(*message
)->hdrs
, "X-Original-Content-Encoding", sizeof("X-Original-Content-Encoding"), &h
, sizeof(zval
*), NULL
);
288 zend_hash_del(&(*message
)->hdrs
, "Content-Encoding", sizeof("Content-Encoding"));
294 if ((flags
& PHP_HTTP_MESSAGE_PARSER_DUMB_BODIES
)) {
295 php_http_message_parser_state_push(parser
, 1, PHP_HTTP_MESSAGE_PARSER_STATE_BODY_DUMB
);
298 if (strstr(Z_STRVAL_PP(h_te
), "chunked")) {
299 parser
->dechunk
= php_http_encoding_stream_init(parser
->dechunk
, php_http_encoding_stream_get_dechunk_ops(), 0 TSRMLS_CC
);
300 php_http_message_parser_state_push(parser
, 1, PHP_HTTP_MESSAGE_PARSER_STATE_BODY_CHUNKED
);
308 parser
->body_length
= strtoul(Z_STRVAL_PP(h_cl
), &stop
, 10);
310 if (stop
!= Z_STRVAL_PP(h_cl
)) {
311 php_http_message_parser_state_push(parser
, 1, PHP_HTTP_MESSAGE_PARSER_STATE_BODY_LENGTH
);
317 ulong total
= 0, start
= 0, end
= 0;
319 if (!strncasecmp(Z_STRVAL_PP(h_cr
), "bytes", lenof("bytes"))
320 && ( Z_STRVAL_P(h
)[lenof("bytes")] == ':'
321 || Z_STRVAL_P(h
)[lenof("bytes")] == ' '
322 || Z_STRVAL_P(h
)[lenof("bytes")] == '='
325 char *total_at
= NULL
, *end_at
= NULL
;
326 char *start_at
= Z_STRVAL_PP(h_cr
) + sizeof("bytes");
328 start
= strtoul(start_at
, &end_at
, 10);
330 end
= strtoul(end_at
+ 1, &total_at
, 10);
331 if (total_at
&& strncmp(total_at
+ 1, "*", 1)) {
332 total
= strtoul(total_at
+ 1, NULL
, 10);
335 if (end
>= start
&& (!total
|| end
< total
)) {
336 parser
->body_length
= end
+ 1 - start
;
337 php_http_message_parser_state_push(parser
, 1, PHP_HTTP_MESSAGE_PARSER_STATE_BODY_LENGTH
);
345 if ((*message
)->type
== PHP_HTTP_REQUEST
) {
346 php_http_message_parser_state_push(parser
, 1, PHP_HTTP_MESSAGE_PARSER_STATE_DONE
);
348 php_http_message_parser_state_push(parser
, 1, PHP_HTTP_MESSAGE_PARSER_STATE_BODY_DUMB
);
354 case PHP_HTTP_MESSAGE_PARSER_STATE_BODY
:
359 if (parser
->inflate
) {
360 char *dec_str
= NULL
;
363 if (SUCCESS
!= php_http_encoding_stream_update(parser
->inflate
, str
, len
, &dec_str
, &dec_len
)) {
364 return php_http_message_parser_state_push(parser
, 1, PHP_HTTP_MESSAGE_PARSER_STATE_FAILURE
);
367 if (str
!= buffer
->data
) {
374 php_stream_write(php_http_message_body_stream(&(*message
)->body
), str
, len
);
378 ZVAL_LONG(zcl
, php_http_message_body_size(&(*message
)->body
));
379 zend_hash_update(&(*message
)->hdrs
, "Content-Length", sizeof("Content-Length"), &zcl
, sizeof(zval
*), NULL
);
383 php_http_buffer_cut(buffer
, 0, cut
);
386 if (str
!= buffer
->data
) {
396 case PHP_HTTP_MESSAGE_PARSER_STATE_BODY_DUMB
:
402 php_http_message_parser_state_push(parser
, 2, PHP_HTTP_MESSAGE_PARSER_STATE_BODY_DONE
, PHP_HTTP_MESSAGE_PARSER_STATE_BODY
);
406 case PHP_HTTP_MESSAGE_PARSER_STATE_BODY_LENGTH
:
408 len
= MIN(parser
->body_length
, buffer
->used
);
412 parser
->body_length
-= len
;
414 php_http_message_parser_state_push(parser
, 2, !parser
->body_length
?PHP_HTTP_MESSAGE_PARSER_STATE_BODY_DONE
:PHP_HTTP_MESSAGE_PARSER_STATE_BODY_LENGTH
, PHP_HTTP_MESSAGE_PARSER_STATE_BODY
);
418 case PHP_HTTP_MESSAGE_PARSER_STATE_BODY_CHUNKED
:
421 * - pass available data through the dechunk stream
422 * - pass decoded data along
423 * - if stream zeroed:
424 * Y: - cut processed string out of buffer, but leave length of unprocessed dechunk stream data untouched
428 char *dec_str
= NULL
;
431 if (SUCCESS
!= php_http_encoding_stream_update(parser
->dechunk
, buffer
->data
, buffer
->used
, &dec_str
, &dec_len
)) {
438 if (php_http_encoding_stream_done(parser
->dechunk
)) {
439 cut
= buffer
->used
- PHP_HTTP_BUFFER_LEN(parser
->dechunk
->ctx
);
440 php_http_message_parser_state_push(parser
, 2, PHP_HTTP_MESSAGE_PARSER_STATE_BODY_DONE
, PHP_HTTP_MESSAGE_PARSER_STATE_BODY
);
443 php_http_message_parser_state_push(parser
, 2, PHP_HTTP_MESSAGE_PARSER_STATE_BODY_CHUNKED
, PHP_HTTP_MESSAGE_PARSER_STATE_BODY
);
448 case PHP_HTTP_MESSAGE_PARSER_STATE_BODY_DONE
:
450 php_http_message_parser_state_push(parser
, 1, PHP_HTTP_MESSAGE_PARSER_STATE_DONE
);
452 if (parser
->dechunk
) {
453 char *dec_str
= NULL
;
456 if (SUCCESS
!= php_http_encoding_stream_finish(parser
->dechunk
, &dec_str
, &dec_len
)) {
457 return php_http_message_parser_state_push(parser
, 1, PHP_HTTP_MESSAGE_PARSER_STATE_FAILURE
);
459 php_http_encoding_stream_dtor(parser
->dechunk
);
461 if (dec_str
&& dec_len
) {
465 php_http_message_parser_state_push(parser
, 1, PHP_HTTP_MESSAGE_PARSER_STATE_BODY
);
472 case PHP_HTTP_MESSAGE_PARSER_STATE_DONE
: {
473 char *ptr
= buffer
->data
;
475 while (ptr
- buffer
->data
< buffer
->used
&& PHP_HTTP_IS_CTYPE(space
, *ptr
)) {
479 php_http_buffer_cut(buffer
, 0, ptr
- buffer
->data
);
485 return php_http_message_parser_state_is(parser
);
493 * vim600: noet sw=4 ts=4 fdm=marker
494 * vim<600: noet sw=4 ts=4