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) 2004-2005, Michael Wallner <mike@php.net> |
10 +--------------------------------------------------------------------+
20 #include "php_http_encoding_api.h"
22 #include "php_http_api.h"
25 # include "php_http_send_api.h"
26 # include "php_http_headers_api.h"
30 ZEND_EXTERN_MODULE_GLOBALS(http
);
32 /* {{{ char *http_encoding_dechunk(char *, size_t, char **, size_t *) */
33 PHP_HTTP_API
const char *_http_encoding_dechunk(const char *encoded
, size_t encoded_len
, char **decoded
, size_t *decoded_len TSRMLS_DC
)
40 *decoded
= ecalloc(1, encoded_len
);
44 while ((rest
= encoded
+ encoded_len
- e_ptr
) > 0) {
46 int EOL_len
= 0, eol_mismatch
= 0;
49 chunk_len
= strtol(e_ptr
, &n_ptr
, 16);
52 * - we could not read in chunk size
53 * - we got a negative chunk size
54 * - chunk size is greater then remaining size
55 * - chunk size is not followed by (CR)LF|NUL
57 if ( (n_ptr
== e_ptr
) || (chunk_len
< 0) || (chunk_len
> rest
) ||
58 (*n_ptr
&& (eol_mismatch
= (n_ptr
!= http_locate_eol(e_ptr
, &EOL_len
))))) {
59 /* don't fail on apperently not encoded data */
60 if (e_ptr
== encoded
) {
61 memcpy(*decoded
, encoded
, encoded_len
);
62 *decoded_len
= encoded_len
;
63 return encoded
+ encoded_len
;
68 http_error_ex(HE_WARNING
, HTTP_E_ENCODING
, "Invalid character (expected 0x0D 0x0A; got: 0x%X 0x%X)", *n_ptr
, *(n_ptr
+ 1));
70 http_error_ex(HE_WARNING
, HTTP_E_ENCODING
, "Invalid character (expected 0x0A; got: 0x%X)", *n_ptr
);
73 char *error
= estrndup(n_ptr
, strcspn(n_ptr
, "\r\n "));
74 http_error_ex(HE_WARNING
, HTTP_E_ENCODING
, "Invalid chunk size: '%s' at pos %d", error
, n_ptr
- encoded
);
88 memcpy(d_ptr
, e_ptr
+= EOL_len
, chunk_len
);
90 e_ptr
+= chunk_len
+ EOL_len
;
91 *decoded_len
+= chunk_len
;
100 static const char http_encoding_gzip_header
[] = {
101 (const char) 0x1f, // fixed value
102 (const char) 0x8b, // fixed value
103 (const char) Z_DEFLATED
, // compression algorithm
104 (const char) 0, // none of the possible flags defined by the GZIP "RFC"
105 (const char) 0, // no MTIME available (4 bytes)
106 (const char) 0, // =*=
107 (const char) 0, // =*=
108 (const char) 0, // =*=
109 (const char) 0, // two possible flag values for 9 compression levels? o_O
110 (const char) 0x03 // assume *nix OS
113 inline void http_init_gzencode_buffer(z_stream
*Z
, const char *data
, size_t data_len
, char **buf_ptr
)
119 Z
->next_in
= (Bytef
*) data
;
120 Z
->avail_in
= data_len
;
121 Z
->avail_out
= HTTP_ENCODING_BUFLEN(data_len
) + HTTP_ENCODING_SAFPAD
- 1;
123 *buf_ptr
= emalloc(HTTP_ENCODING_BUFLEN(data_len
) + sizeof(http_encoding_gzip_header
) + HTTP_ENCODING_SAFPAD
);
124 memcpy(*buf_ptr
, http_encoding_gzip_header
, sizeof(http_encoding_gzip_header
));
126 Z
->next_out
= (Bytef
*) *buf_ptr
+ sizeof(http_encoding_gzip_header
);
129 inline void http_init_deflate_buffer(z_stream
*Z
, const char *data
, size_t data_len
, char **buf_ptr
)
135 Z
->data_type
= Z_UNKNOWN
;
136 Z
->next_in
= (Bytef
*) data
;
137 Z
->avail_in
= data_len
;
138 Z
->avail_out
= HTTP_ENCODING_BUFLEN(data_len
) - 1;
139 Z
->next_out
= emalloc(HTTP_ENCODING_BUFLEN(data_len
));
141 *buf_ptr
= (char *) Z
->next_out
;
144 inline void http_init_uncompress_buffer(size_t data_len
, char **buf_ptr
, size_t *buf_len
, int *iteration
)
147 *buf_len
= data_len
* 2;
148 *buf_ptr
= emalloc(*buf_len
+ 1);
150 size_t new_len
= *buf_len
<< 2;
151 char *new_ptr
= erealloc_recoverable(*buf_ptr
, new_len
+ 1);
157 *iteration
= INT_MAX
-1; /* avoid integer overflow on increment op */
162 inline void http_init_inflate_buffer(z_stream
*Z
, const char *data
, size_t data_len
, char **buf_ptr
, size_t *buf_len
, int *iteration
)
167 http_init_uncompress_buffer(data_len
, buf_ptr
, buf_len
, iteration
);
169 Z
->next_in
= (Bytef
*) data
;
170 Z
->avail_in
= data_len
;
171 Z
->avail_out
= *buf_len
;
172 Z
->next_out
= (Bytef
*) *buf_ptr
;
175 inline size_t http_finish_buffer(size_t buf_len
, char **buf_ptr
)
177 (*buf_ptr
)[buf_len
] = '\0';
181 inline size_t http_finish_gzencode_buffer(z_stream
*Z
, const char *data
, size_t data_len
, char **buf_ptr
)
186 crc
= crc32(0L, Z_NULL
, 0);
187 crc
= crc32(crc
, (const Bytef
*) data
, data_len
);
189 trailer
= *buf_ptr
+ sizeof(http_encoding_gzip_header
) + Z
->total_out
;
192 trailer
[0] = (char) (crc
& 0xFF);
193 trailer
[1] = (char) ((crc
>> 8) & 0xFF);
194 trailer
[2] = (char) ((crc
>> 16) & 0xFF);
195 trailer
[3] = (char) ((crc
>> 24) & 0xFF);
196 trailer
[4] = (char) ((Z
->total_in
) & 0xFF);
197 trailer
[5] = (char) ((Z
->total_in
>> 8) & 0xFF);
198 trailer
[6] = (char) ((Z
->total_in
>> 16) & 0xFF);
199 trailer
[7] = (char) ((Z
->total_in
>> 24) & 0xFF);
201 return http_finish_buffer(Z
->total_out
+ sizeof(http_encoding_gzip_header
) + 8, buf_ptr
);
204 inline STATUS
http_verify_gzencode_buffer(const char *data
, size_t data_len
, const char **encoded
, size_t *encoded_len
, int error_level TSRMLS_DC
)
206 size_t offset
= sizeof(http_encoding_gzip_header
);
208 if (data_len
< offset
) {
209 goto really_bad_gzip_header
;
212 if (data
[0] != (const char) 0x1F || data
[1] != (const char) 0x8B) {
213 http_error_ex(error_level TSRMLS_CC
, HTTP_E_ENCODING
, "Unrecognized GZIP header start: 0x%02X 0x%02X", (int) data
[0], (int) (data
[1] & 0xFF));
217 if (data
[2] != (const char) Z_DEFLATED
) {
218 http_error_ex(error_level TSRMLS_CC
, HTTP_E_ENCODING
, "Unrecognized compression format (%d)", (int) (data
[2] & 0xFF));
219 /* still try to decode */
221 if ((data
[3] & 0x4) == 0x4) {
222 if (data_len
< offset
+ 2) {
223 goto really_bad_gzip_header
;
225 /* there are extra fields, the length follows the common header as 2 bytes LSB */
226 offset
+= (unsigned) ((data
[offset
] & 0xFF));
228 offset
+= (unsigned) ((data
[offset
] & 0xFF) << 8);
231 if ((data
[3] & 0x8) == 0x8) {
232 if (data_len
<= offset
) {
233 goto really_bad_gzip_header
;
235 /* there's a file name */
236 offset
+= strlen(&data
[offset
]) + 1 /*NUL*/;
238 if ((data
[3] & 0x10) == 0x10) {
239 if (data_len
<= offset
) {
240 goto really_bad_gzip_header
;
242 /* there's a comment */
243 offset
+= strlen(&data
[offset
]) + 1 /* NUL */;
245 if ((data
[3] & 0x2) == 0x2) {
246 /* there's a CRC16 of the header */
248 if (data_len
<= offset
) {
249 goto really_bad_gzip_header
;
251 unsigned long crc
, cmp
;
253 cmp
= (unsigned) ((data
[offset
-2] & 0xFF));
254 cmp
+= (unsigned) ((data
[offset
-1] & 0xFF) << 8);
256 crc
= crc32(0L, Z_NULL
, 0);
257 crc
= crc32(crc
, (const Bytef
*) data
, sizeof(http_encoding_gzip_header
));
259 if (cmp
!= (crc
& 0xFFFF)) {
260 http_error_ex(error_level TSRMLS_CC
, HTTP_E_ENCODING
, "GZIP headers CRC checksums so not match (%lu, %lu)", cmp
, crc
& 0xFFFF);
266 if (data_len
< offset
+ 8) {
267 http_error(error_level TSRMLS_CC
, HTTP_E_ENCODING
, "Missing or truncated GZIP footer");
272 *encoded
= data
+ offset
;
275 *encoded_len
= data_len
- offset
- 8 /* size of the assumed GZIP footer */;
280 really_bad_gzip_header
:
281 http_error(error_level TSRMLS_CC
, HTTP_E_ENCODING
, "Missing or truncated GZIP header");
285 inline STATUS
http_verify_gzdecode_buffer(const char *data
, size_t data_len
, const char *decoded
, size_t decoded_len
, int error_level TSRMLS_DC
)
287 STATUS status
= SUCCESS
;
288 unsigned long len
, cmp
, crc
;
290 crc
= crc32(0L, Z_NULL
, 0);
291 crc
= crc32(crc
, (const Bytef
*) decoded
, decoded_len
);
293 cmp
= (unsigned) ((data
[data_len
-8] & 0xFF));
294 cmp
+= (unsigned) ((data
[data_len
-7] & 0xFF) << 8);
295 cmp
+= (unsigned) ((data
[data_len
-6] & 0xFF) << 16);
296 cmp
+= (unsigned) ((data
[data_len
-5] & 0xFF) << 24);
297 len
= (unsigned) ((data
[data_len
-4] & 0xFF));
298 len
+= (unsigned) ((data
[data_len
-3] & 0xFF) << 8);
299 len
+= (unsigned) ((data
[data_len
-2] & 0xFF) << 16);
300 len
+= (unsigned) ((data
[data_len
-1] & 0xFF) << 24);
303 http_error_ex(error_level TSRMLS_CC
, HTTP_E_ENCODING
, "Could not verify data integrity: CRC checksums do not match (%lu, %lu)", cmp
, crc
);
306 if (len
!= decoded_len
) {
307 http_error_ex(error_level TSRMLS_CC
, HTTP_E_ENCODING
, "Could not verify data integrity: data sizes do not match (%lu, %lu)", len
, decoded_len
);
313 PHP_HTTP_API STATUS
_http_encode(http_encoding_type type
, int level
, const char *data
, size_t data_len
, char **encoded
, size_t *encoded_len TSRMLS_DC
)
315 STATUS status
= SUCCESS
;
319 case HTTP_ENCODING_ANY
:
320 case HTTP_ENCODING_GZIP
:
321 status
= http_encoding_gzencode(level
, data
, data_len
, encoded
, encoded_len
);
324 case HTTP_ENCODING_DEFLATE
:
325 status
= http_encoding_deflate(level
, data
, data_len
, encoded
, encoded_len
);
328 case HTTP_ENCODING_COMPRESS
:
329 status
= http_encoding_compress(level
, data
, data_len
, encoded
, encoded_len
);
332 case HTTP_ENCODING_NONE
:
334 *encoded
= estrndup(data
, data_len
);
335 *encoded_len
= data_len
;
342 PHP_HTTP_API STATUS
_http_decode(http_encoding_type type
, const char *data
, size_t data_len
, char **decoded
, size_t *decoded_len TSRMLS_DC
)
344 STATUS status
= SUCCESS
;
348 case HTTP_ENCODING_ANY
:
349 if ( SUCCESS
!= http_encoding_gzdecode(data
, data_len
, decoded
, decoded_len
) &&
350 SUCCESS
!= http_encoding_inflate(data
, data_len
, decoded
, decoded_len
) &&
351 SUCCESS
!= http_encoding_uncompress(data
, data_len
, decoded
, decoded_len
)) {
356 case HTTP_ENCODING_GZIP
:
357 status
= http_encoding_gzdecode(data
, data_len
, decoded
, decoded_len
);
360 case HTTP_ENCODING_DEFLATE
:
361 status
= http_encoding_inflate(data
, data_len
, decoded
, decoded_len
);
364 case HTTP_ENCODING_COMPRESS
:
365 status
= http_encoding_uncompress(data
, data_len
, decoded
, decoded_len
);
368 case HTTP_ENCODING_NONE
:
370 *decoded
= estrndup(data
, data_len
);
371 *decoded_len
= data_len
;
378 PHP_HTTP_API STATUS
_http_encoding_gzencode(int level
, const char *data
, size_t data_len
, char **encoded
, size_t *encoded_len TSRMLS_DC
)
381 STATUS status
= Z_OK
;
383 http_init_gzencode_buffer(&Z
, data
, data_len
, encoded
);
385 if ( (Z_OK
== (status
= deflateInit2(&Z
, level
, Z_DEFLATED
, -MAX_WBITS
, MAX_MEM_LEVEL
, Z_DEFAULT_STRATEGY
))) &&
386 (Z_STREAM_END
== (status
= deflate(&Z
, Z_FINISH
))) &&
387 (Z_OK
== (status
= deflateEnd(&Z
)))) {
388 *encoded_len
= http_finish_gzencode_buffer(&Z
, data
, data_len
, encoded
);
393 http_error_ex(HE_WARNING
, HTTP_E_ENCODING
, "Could not gzencode data: %s", zError(status
));
397 PHP_HTTP_API STATUS
_http_encoding_deflate(int level
, const char *data
, size_t data_len
, char **encoded
, size_t *encoded_len TSRMLS_DC
)
400 STATUS status
= Z_OK
;
402 http_init_deflate_buffer(&Z
, data
, data_len
, encoded
);
404 if ( (Z_OK
== (status
= deflateInit2(&Z
, level
, Z_DEFLATED
, -MAX_WBITS
, MAX_MEM_LEVEL
, Z_DEFAULT_STRATEGY
))) &&
405 (Z_STREAM_END
== (status
= deflate(&Z
, Z_FINISH
))) &&
406 (Z_OK
== (status
= deflateEnd(&Z
)))) {
407 *encoded_len
= http_finish_buffer(Z
.total_out
, encoded
);
412 http_error_ex(HE_WARNING
, HTTP_E_ENCODING
, "Could not deflate data: %s", zError(status
));
416 PHP_HTTP_API STATUS
_http_encoding_compress(int level
, const char *data
, size_t data_len
, char **encoded
, size_t *encoded_len TSRMLS_DC
)
420 *encoded
= emalloc(*encoded_len
= HTTP_ENCODING_BUFLEN(data_len
));
422 if (Z_OK
== (status
= compress2((Bytef
*) *encoded
, (uLongf
*) encoded_len
, (const Bytef
*) data
, data_len
, level
))) {
423 http_finish_buffer(*encoded_len
, encoded
);
428 http_error_ex(HE_WARNING
, HTTP_E_ENCODING
, "Could not compress data: %s", zError(status
));
432 PHP_HTTP_API STATUS
_http_encoding_gzdecode(const char *data
, size_t data_len
, char **decoded
, size_t *decoded_len TSRMLS_DC
)
437 if ( (SUCCESS
== http_verify_gzencode_buffer(data
, data_len
, &encoded
, &encoded_len
, HE_NOTICE
)) &&
438 (SUCCESS
== http_encoding_inflate(encoded
, encoded_len
, decoded
, decoded_len
))) {
439 http_verify_gzdecode_buffer(data
, data_len
, *decoded
, *decoded_len
, HE_NOTICE
);
446 PHP_HTTP_API STATUS
_http_encoding_inflate(const char *data
, size_t data_len
, char **decoded
, size_t *decoded_len TSRMLS_DC
)
453 http_init_inflate_buffer(&Z
, data
, data_len
, decoded
, decoded_len
, &max
);
454 if (Z_OK
== (status
= inflateInit2(&Z
, -MAX_WBITS
))) {
455 if (Z_STREAM_END
== (status
= inflate(&Z
, Z_FINISH
))) {
456 if (Z_OK
== (status
= inflateEnd(&Z
))) {
457 *decoded_len
= http_finish_buffer(Z
.total_out
, decoded
);
462 } while (++max
< HTTP_ENCODING_MAXTRY
&& status
== Z_BUF_ERROR
);
465 http_error_ex(HE_WARNING
, HTTP_E_ENCODING
, "Could not inflate data: %s", zError(status
));
469 PHP_HTTP_API STATUS
_http_encoding_uncompress(const char *data
, size_t data_len
, char **decoded
, size_t *decoded_len TSRMLS_DC
)
475 http_init_uncompress_buffer(data_len
, decoded
, decoded_len
, &max
);
476 if (Z_OK
== (status
= uncompress((Bytef
*) *decoded
, (uLongf
*) decoded_len
, (const Bytef
*) data
, data_len
))) {
477 http_finish_buffer(*decoded_len
, decoded
);
480 } while (++max
< HTTP_ENCODING_MAXTRY
&& status
== Z_BUF_ERROR
);
483 http_error_ex(HE_WARNING
, HTTP_E_ENCODING
, "Could not uncompress data: %s", zError(status
));
487 #define HTTP_ENCODING_STREAM_ERROR(status, tofree) \
489 if (tofree) efree(tofree); \
490 http_error_ex(HE_WARNING, HTTP_E_ENCODING, "GZIP stream error: %s", zError(status)); \
494 PHP_HTTP_API STATUS
_http_encoding_stream_init(http_encoding_stream
*s
, int gzip
, int level
, char **encoded
, size_t *encoded_len TSRMLS_DC
)
498 memset(s
, 0, sizeof(http_encoding_stream
));
499 if (Z_OK
!= (status
= deflateInit2(&s
->Z
, level
, Z_DEFLATED
, -MAX_WBITS
, MAX_MEM_LEVEL
, Z_DEFAULT_STRATEGY
))) {
500 HTTP_ENCODING_STREAM_ERROR(status
, NULL
);
503 if (s
->gzip
= gzip
) {
504 s
->crc
= crc32(0L, Z_NULL
, 0);
505 *encoded_len
= sizeof(http_encoding_gzip_header
);
506 *encoded
= emalloc(*encoded_len
);
507 memcpy(*encoded
, http_encoding_gzip_header
, *encoded_len
);
516 PHP_HTTP_API STATUS
_http_encoding_stream_update(http_encoding_stream
*s
, const char *data
, size_t data_len
, char **encoded
, size_t *encoded_len TSRMLS_DC
)
520 *encoded_len
= HTTP_ENCODING_BUFLEN(data_len
);
521 *encoded
= emalloc(*encoded_len
);
523 s
->Z
.next_in
= (Bytef
*) data
;
524 s
->Z
.avail_in
= data_len
;
525 s
->Z
.next_out
= (Bytef
*) *encoded
;
526 s
->Z
.avail_out
= *encoded_len
;
528 status
= deflate(&s
->Z
, Z_SYNC_FLUSH
);
530 if (Z_OK
!= status
&& Z_STREAM_END
!= status
) {
531 HTTP_ENCODING_STREAM_ERROR(status
, *encoded
);
533 *encoded_len
-= s
->Z
.avail_out
;
536 s
->crc
= crc32(s
->crc
, (const Bytef
*) data
, data_len
);
542 PHP_HTTP_API STATUS
_http_encoding_stream_finish(http_encoding_stream
*s
, char **encoded
, size_t *encoded_len TSRMLS_DC
)
547 *encoded
= emalloc(*encoded_len
);
549 s
->Z
.next_out
= (Bytef
*) *encoded
;
550 s
->Z
.avail_out
= *encoded_len
;
552 if (Z_STREAM_END
!= (status
= deflate(&s
->Z
, Z_FINISH
)) || Z_OK
!= (status
= deflateEnd(&s
->Z
))) {
553 HTTP_ENCODING_STREAM_ERROR(status
, *encoded
);
556 *encoded_len
-= s
->Z
.avail_out
;
558 if (s
->Z
.avail_out
< 8) {
559 *encoded
= erealloc(*encoded
, *encoded_len
+ 8);
561 (*encoded
)[(*encoded_len
)++] = (char) (s
->crc
& 0xFF);
562 (*encoded
)[(*encoded_len
)++] = (char) ((s
->crc
>> 8) & 0xFF);
563 (*encoded
)[(*encoded_len
)++] = (char) ((s
->crc
>> 16) & 0xFF);
564 (*encoded
)[(*encoded_len
)++] = (char) ((s
->crc
>> 24) & 0xFF);
565 (*encoded
)[(*encoded_len
)++] = (char) ((s
->Z
.total_in
) & 0xFF);
566 (*encoded
)[(*encoded_len
)++] = (char) ((s
->Z
.total_in
>> 8) & 0xFF);
567 (*encoded
)[(*encoded_len
)++] = (char) ((s
->Z
.total_in
>> 16) & 0xFF);
568 (*encoded
)[(*encoded_len
)++] = (char) ((s
->Z
.total_in
>> 24) & 0xFF);
574 #endif /* HTTP_HAVE_ZLIB */
576 PHP_HTTP_API zend_bool
_http_encoding_response_start(size_t content_length TSRMLS_DC
)
578 if ( php_ob_handler_used("ob_gzhandler" TSRMLS_CC
) ||
579 php_ob_handler_used("zlib output compression" TSRMLS_CC
)) {
580 HTTP_G(send
).gzip_encoding
= 0;
582 if (!HTTP_G(send
).gzip_encoding
) {
583 /* emit a content-length header */
584 if (content_length
) {
585 char cl_header_str
[128];
586 size_t cl_header_len
;
587 cl_header_len
= snprintf(cl_header_str
, lenof(cl_header_str
), "Content-Length: %lu", (unsigned long) content_length
);
588 http_send_header_string_ex(cl_header_str
, cl_header_len
, 1);
591 #ifndef HTTP_HAVE_ZLIB
592 HTTP_G(send
).gzip_encoding
= 0;
593 php_start_ob_buffer_named("ob_gzhandler", 0, 0 TSRMLS_CC
);
598 INIT_PZVAL(&zsupported
);
599 array_init(&zsupported
);
600 add_next_index_stringl(&zsupported
, "gzip", lenof("gzip"), 1);
601 add_next_index_stringl(&zsupported
, "deflate", lenof("deflate"), 1);
603 HTTP_G(send
).gzip_encoding
= 0;
605 if (selected
= http_negotiate_encoding(&zsupported
)) {
607 char *encoding
= NULL
;
610 if (HASH_KEY_IS_STRING
== zend_hash_get_current_key(selected
, &encoding
, &idx
, 0) && encoding
) {
611 if (!strcmp(encoding
, "gzip")) {
612 if (SUCCESS
== (hs
= http_send_header_string("Content-Encoding: gzip"))) {
613 HTTP_G(send
).gzip_encoding
= HTTP_ENCODING_GZIP
;
615 } else if (!strcmp(encoding
, "deflate")) {
616 if (SUCCESS
== (hs
= http_send_header_string("Content-Encoding: deflate"))) {
617 HTTP_G(send
).gzip_encoding
= HTTP_ENCODING_DEFLATE
;
621 http_send_header_string("Vary: Accept-Encoding");
625 zend_hash_destroy(selected
);
626 FREE_HASHTABLE(selected
);
629 zval_dtor(&zsupported
);
630 return HTTP_G(send
).gzip_encoding
;
642 * vim600: noet sw=4 ts=4 fdm=marker
643 * vim<600: noet sw=4 ts=4