1 /*
2 +----------------------------------------------------------------------+
3 | Copyright (c) The PHP Group |
4 +----------------------------------------------------------------------+
5 | This source file is subject to version 3.01 of the PHP license, |
6 | that is bundled with this package in the file LICENSE, and is |
7 | available through the world-wide-web at the following url: |
8 | https://www.php.net/license/3_01.txt |
9 | If you did not receive a copy of the PHP license and are unable to |
10 | obtain it through the world-wide-web, please send a note to |
11 | license@php.net so we can mail you a copy immediately. |
12 +----------------------------------------------------------------------+
13 | Authors: Rasmus Lerdorf <rasmus@php.net> |
14 | Jim Winstead <jimw@php.net> |
15 | Hartmut Holzgraefe <hholzgra@php.net> |
16 | Wez Furlong <wez@thebrainroom.com> |
17 | Sara Golemon <pollita@php.net> |
18 +----------------------------------------------------------------------+
19 */
20
21 #include "php.h"
22 #include "php_globals.h"
23 #include "php_streams.h"
24 #include "php_network.h"
25 #include "php_ini.h"
26 #include "ext/standard/basic_functions.h"
27 #include "zend_smart_str.h"
28
29 #include <stdio.h>
30 #include <stdlib.h>
31 #include <errno.h>
32 #include <sys/types.h>
33 #include <sys/stat.h>
34 #include <fcntl.h>
35
36 #ifdef PHP_WIN32
37 #define O_RDONLY _O_RDONLY
38 #include "win32/param.h"
39 #else
40 #include <sys/param.h>
41 #endif
42
43 #include "php_standard.h"
44
45 #include <sys/types.h>
46 #if HAVE_SYS_SOCKET_H
47 #include <sys/socket.h>
48 #endif
49
50 #ifdef PHP_WIN32
51 #include <winsock2.h>
52 #else
53 #include <netinet/in.h>
54 #include <netdb.h>
55 #if HAVE_ARPA_INET_H
56 #include <arpa/inet.h>
57 #endif
58 #endif
59
60 #if defined(PHP_WIN32) || defined(__riscos__)
61 #undef AF_UNIX
62 #endif
63
64 #if defined(AF_UNIX)
65 #include <sys/un.h>
66 #endif
67
68 #include "php_fopen_wrappers.h"
69
70 #define HTTP_HEADER_BLOCK_SIZE 1024
71 #define PHP_URL_REDIRECT_MAX 20
72 #define HTTP_HEADER_USER_AGENT 1
73 #define HTTP_HEADER_HOST 2
74 #define HTTP_HEADER_AUTH 4
75 #define HTTP_HEADER_FROM 8
76 #define HTTP_HEADER_CONTENT_LENGTH 16
77 #define HTTP_HEADER_TYPE 32
78 #define HTTP_HEADER_CONNECTION 64
79
80 #define HTTP_WRAPPER_HEADER_INIT 1
81 #define HTTP_WRAPPER_REDIRECTED 2
82
strip_header(char * header_bag,char * lc_header_bag,const char * lc_header_name)83 static inline void strip_header(char *header_bag, char *lc_header_bag,
84 const char *lc_header_name)
85 {
86 char *lc_header_start = strstr(lc_header_bag, lc_header_name);
87 if (lc_header_start
88 && (lc_header_start == lc_header_bag || *(lc_header_start-1) == '\n')
89 ) {
90 char *header_start = header_bag + (lc_header_start - lc_header_bag);
91 char *lc_eol = strchr(lc_header_start, '\n');
92
93 if (lc_eol) {
94 char *eol = header_start + (lc_eol - lc_header_start);
95 size_t eollen = strlen(lc_eol);
96
97 memmove(lc_header_start, lc_eol+1, eollen);
98 memmove(header_start, eol+1, eollen);
99 } else {
100 *lc_header_start = '\0';
101 *header_start = '\0';
102 }
103 }
104 }
105
check_has_header(const char * headers,const char * header)106 static bool check_has_header(const char *headers, const char *header) {
107 const char *s = headers;
108 while ((s = strstr(s, header))) {
109 if (s == headers || *(s-1) == '\n') {
110 return 1;
111 }
112 s++;
113 }
114 return 0;
115 }
116
php_stream_url_wrap_http_ex(php_stream_wrapper * wrapper,const char * path,const char * mode,int options,zend_string ** opened_path,php_stream_context * context,int redirect_max,int flags,zval * response_header STREAMS_DC)117 static php_stream *php_stream_url_wrap_http_ex(php_stream_wrapper *wrapper,
118 const char *path, const char *mode, int options, zend_string **opened_path,
119 php_stream_context *context, int redirect_max, int flags,
120 zval *response_header STREAMS_DC) /* {{{ */
121 {
122 php_stream *stream = NULL;
123 php_url *resource = NULL;
124 int use_ssl;
125 int use_proxy = 0;
126 zend_string *tmp = NULL;
127 char *ua_str = NULL;
128 zval *ua_zval = NULL, *tmpzval = NULL, ssl_proxy_peer_name;
129 char location[HTTP_HEADER_BLOCK_SIZE];
130 int reqok = 0;
131 char *http_header_line = NULL;
132 char tmp_line[128];
133 size_t chunk_size = 0, file_size = 0;
134 int eol_detect = 0;
135 zend_string *transport_string;
136 zend_string *errstr = NULL;
137 int have_header = 0;
138 bool request_fulluri = 0, ignore_errors = 0;
139 struct timeval timeout;
140 char *user_headers = NULL;
141 int header_init = ((flags & HTTP_WRAPPER_HEADER_INIT) != 0);
142 int redirected = ((flags & HTTP_WRAPPER_REDIRECTED) != 0);
143 bool follow_location = 1;
144 php_stream_filter *transfer_encoding = NULL;
145 int response_code;
146 smart_str req_buf = {0};
147 bool custom_request_method;
148
149 tmp_line[0] = '\0';
150
151 if (redirect_max < 1) {
152 php_stream_wrapper_log_error(wrapper, options, "Redirection limit reached, aborting");
153 return NULL;
154 }
155
156 resource = php_url_parse(path);
157 if (resource == NULL) {
158 return NULL;
159 }
160
161 if (!zend_string_equals_literal_ci(resource->scheme, "http") &&
162 !zend_string_equals_literal_ci(resource->scheme, "https")) {
163 if (!context ||
164 (tmpzval = php_stream_context_get_option(context, wrapper->wops->label, "proxy")) == NULL ||
165 Z_TYPE_P(tmpzval) != IS_STRING ||
166 Z_STRLEN_P(tmpzval) == 0) {
167 php_url_free(resource);
168 return php_stream_open_wrapper_ex(path, mode, REPORT_ERRORS, NULL, context);
169 }
170 /* Called from a non-http wrapper with http proxying requested (i.e. ftp) */
171 request_fulluri = 1;
172 use_ssl = 0;
173 use_proxy = 1;
174 transport_string = zend_string_copy(Z_STR_P(tmpzval));
175 } else {
176 /* Normal http request (possibly with proxy) */
177
178 if (strpbrk(mode, "awx+")) {
179 php_stream_wrapper_log_error(wrapper, options, "HTTP wrapper does not support writeable connections");
180 php_url_free(resource);
181 return NULL;
182 }
183
184 use_ssl = resource->scheme && (ZSTR_LEN(resource->scheme) > 4) && ZSTR_VAL(resource->scheme)[4] == 's';
185 /* choose default ports */
186 if (use_ssl && resource->port == 0)
187 resource->port = 443;
188 else if (resource->port == 0)
189 resource->port = 80;
190
191 if (context &&
192 (tmpzval = php_stream_context_get_option(context, wrapper->wops->label, "proxy")) != NULL &&
193 Z_TYPE_P(tmpzval) == IS_STRING &&
194 Z_STRLEN_P(tmpzval) > 0) {
195 use_proxy = 1;
196 transport_string = zend_string_copy(Z_STR_P(tmpzval));
197 } else {
198 transport_string = zend_strpprintf(0, "%s://%s:%d", use_ssl ? "ssl" : "tcp", ZSTR_VAL(resource->host), resource->port);
199 }
200 }
201
202 if (context && (tmpzval = php_stream_context_get_option(context, wrapper->wops->label, "timeout")) != NULL) {
203 double d = zval_get_double(tmpzval);
204 #ifndef PHP_WIN32
205 timeout.tv_sec = (time_t) d;
206 timeout.tv_usec = (size_t) ((d - timeout.tv_sec) * 1000000);
207 #else
208 timeout.tv_sec = (long) d;
209 timeout.tv_usec = (long) ((d - timeout.tv_sec) * 1000000);
210 #endif
211 } else {
212 #ifndef PHP_WIN32
213 timeout.tv_sec = FG(default_socket_timeout);
214 #else
215 timeout.tv_sec = (long)FG(default_socket_timeout);
216 #endif
217 timeout.tv_usec = 0;
218 }
219
220 stream = php_stream_xport_create(ZSTR_VAL(transport_string), ZSTR_LEN(transport_string), options,
221 STREAM_XPORT_CLIENT | STREAM_XPORT_CONNECT,
222 NULL, &timeout, context, &errstr, NULL);
223
224 if (stream) {
225 php_stream_set_option(stream, PHP_STREAM_OPTION_READ_TIMEOUT, 0, &timeout);
226 }
227
228 if (errstr) {
229 php_stream_wrapper_log_error(wrapper, options, "%s", ZSTR_VAL(errstr));
230 zend_string_release_ex(errstr, 0);
231 errstr = NULL;
232 }
233
234 zend_string_release(transport_string);
235
236 if (stream && use_proxy && use_ssl) {
237 smart_str header = {0};
238
239 /* Set peer_name or name verification will try to use the proxy server name */
240 if (!context || (tmpzval = php_stream_context_get_option(context, "ssl", "peer_name")) == NULL) {
241 ZVAL_STR_COPY(&ssl_proxy_peer_name, resource->host);
242 php_stream_context_set_option(PHP_STREAM_CONTEXT(stream), "ssl", "peer_name", &ssl_proxy_peer_name);
243 zval_ptr_dtor(&ssl_proxy_peer_name);
244 }
245
246 smart_str_appendl(&header, "CONNECT ", sizeof("CONNECT ")-1);
247 smart_str_appends(&header, ZSTR_VAL(resource->host));
248 smart_str_appendc(&header, ':');
249 smart_str_append_unsigned(&header, resource->port);
250 smart_str_appendl(&header, " HTTP/1.0\r\n", sizeof(" HTTP/1.0\r\n")-1);
251
252 /* check if we have Proxy-Authorization header */
253 if (context && (tmpzval = php_stream_context_get_option(context, "http", "header")) != NULL) {
254 char *s, *p;
255
256 if (Z_TYPE_P(tmpzval) == IS_ARRAY) {
257 zval *tmpheader = NULL;
258
259 ZEND_HASH_FOREACH_VAL(Z_ARRVAL_P(tmpzval), tmpheader) {
260 if (Z_TYPE_P(tmpheader) == IS_STRING) {
261 s = Z_STRVAL_P(tmpheader);
262 do {
263 while (*s == ' ' || *s == '\t') s++;
264 p = s;
265 while (*p != 0 && *p != ':' && *p != '\r' && *p !='\n') p++;
266 if (*p == ':') {
267 p++;
268 if (p - s == sizeof("Proxy-Authorization:") - 1 &&
269 zend_binary_strcasecmp(s, sizeof("Proxy-Authorization:") - 1,
270 "Proxy-Authorization:", sizeof("Proxy-Authorization:") - 1) == 0) {
271 while (*p != 0 && *p != '\r' && *p !='\n') p++;
272 smart_str_appendl(&header, s, p - s);
273 smart_str_appendl(&header, "\r\n", sizeof("\r\n")-1);
274 goto finish;
275 } else {
276 while (*p != 0 && *p != '\r' && *p !='\n') p++;
277 }
278 }
279 s = p;
280 while (*s == '\r' || *s == '\n') s++;
281 } while (*s != 0);
282 }
283 } ZEND_HASH_FOREACH_END();
284 } else if (Z_TYPE_P(tmpzval) == IS_STRING && Z_STRLEN_P(tmpzval)) {
285 s = Z_STRVAL_P(tmpzval);
286 do {
287 while (*s == ' ' || *s == '\t') s++;
288 p = s;
289 while (*p != 0 && *p != ':' && *p != '\r' && *p !='\n') p++;
290 if (*p == ':') {
291 p++;
292 if (p - s == sizeof("Proxy-Authorization:") - 1 &&
293 zend_binary_strcasecmp(s, sizeof("Proxy-Authorization:") - 1,
294 "Proxy-Authorization:", sizeof("Proxy-Authorization:") - 1) == 0) {
295 while (*p != 0 && *p != '\r' && *p !='\n') p++;
296 smart_str_appendl(&header, s, p - s);
297 smart_str_appendl(&header, "\r\n", sizeof("\r\n")-1);
298 goto finish;
299 } else {
300 while (*p != 0 && *p != '\r' && *p !='\n') p++;
301 }
302 }
303 s = p;
304 while (*s == '\r' || *s == '\n') s++;
305 } while (*s != 0);
306 }
307 }
308 finish:
309 smart_str_appendl(&header, "\r\n", sizeof("\r\n")-1);
310
311 if (php_stream_write(stream, ZSTR_VAL(header.s), ZSTR_LEN(header.s)) != ZSTR_LEN(header.s)) {
312 php_stream_wrapper_log_error(wrapper, options, "Cannot connect to HTTPS server through proxy");
313 php_stream_close(stream);
314 stream = NULL;
315 }
316 smart_str_free(&header);
317
318 if (stream) {
319 char header_line[HTTP_HEADER_BLOCK_SIZE];
320
321 /* get response header */
322 while (php_stream_gets(stream, header_line, HTTP_HEADER_BLOCK_SIZE-1) != NULL) {
323 if (header_line[0] == '\n' ||
324 header_line[0] == '\r' ||
325 header_line[0] == '\0') {
326 break;
327 }
328 }
329 }
330
331 /* enable SSL transport layer */
332 if (stream) {
333 if (php_stream_xport_crypto_setup(stream, STREAM_CRYPTO_METHOD_SSLv23_CLIENT, NULL) < 0 ||
334 php_stream_xport_crypto_enable(stream, 1) < 0) {
335 php_stream_wrapper_log_error(wrapper, options, "Cannot connect to HTTPS server through proxy");
336 php_stream_close(stream);
337 stream = NULL;
338 }
339 }
340 }
341
342 if (stream == NULL)
343 goto out;
344
345 /* avoid buffering issues while reading header */
346 if (options & STREAM_WILL_CAST)
347 chunk_size = php_stream_set_chunk_size(stream, 1);
348
349 /* avoid problems with auto-detecting when reading the headers -> the headers
350 * are always in canonical \r\n format */
351 eol_detect = stream->flags & (PHP_STREAM_FLAG_DETECT_EOL | PHP_STREAM_FLAG_EOL_MAC);
352 stream->flags &= ~(PHP_STREAM_FLAG_DETECT_EOL | PHP_STREAM_FLAG_EOL_MAC);
353
354 php_stream_context_set(stream, context);
355
356 php_stream_notify_info(context, PHP_STREAM_NOTIFY_CONNECT, NULL, 0);
357
358 if (header_init && context && (tmpzval = php_stream_context_get_option(context, "http", "max_redirects")) != NULL) {
359 redirect_max = (int)zval_get_long(tmpzval);
360 }
361
362 custom_request_method = 0;
363 if (context && (tmpzval = php_stream_context_get_option(context, "http", "method")) != NULL) {
364 if (Z_TYPE_P(tmpzval) == IS_STRING && Z_STRLEN_P(tmpzval) > 0) {
365 /* As per the RFC, automatically redirected requests MUST NOT use other methods than
366 * GET and HEAD unless it can be confirmed by the user */
367 if (!redirected
368 || zend_string_equals_literal(Z_STR_P(tmpzval), "GET")
369 || zend_string_equals_literal(Z_STR_P(tmpzval), "HEAD")
370 ) {
371 custom_request_method = 1;
372 smart_str_append(&req_buf, Z_STR_P(tmpzval));
373 smart_str_appendc(&req_buf, ' ');
374 }
375 }
376 }
377
378 if (!custom_request_method) {
379 smart_str_appends(&req_buf, "GET ");
380 }
381
382 /* Should we send the entire path in the request line, default to no. */
383 if (!request_fulluri && context &&
384 (tmpzval = php_stream_context_get_option(context, "http", "request_fulluri")) != NULL) {
385 request_fulluri = zend_is_true(tmpzval);
386 }
387
388 if (request_fulluri) {
389 /* Ask for everything */
390 smart_str_appends(&req_buf, path);
391 } else {
392 /* Send the traditional /path/to/file?query_string */
393
394 /* file */
395 if (resource->path && ZSTR_LEN(resource->path)) {
396 smart_str_appends(&req_buf, ZSTR_VAL(resource->path));
397 } else {
398 smart_str_appendc(&req_buf, '/');
399 }
400
401 /* query string */
402 if (resource->query) {
403 smart_str_appendc(&req_buf, '?');
404 smart_str_appends(&req_buf, ZSTR_VAL(resource->query));
405 }
406 }
407
408 /* protocol version we are speaking */
409 if (context && (tmpzval = php_stream_context_get_option(context, "http", "protocol_version")) != NULL) {
410 char *protocol_version;
411 spprintf(&protocol_version, 0, "%.1F", zval_get_double(tmpzval));
412
413 smart_str_appends(&req_buf, " HTTP/");
414 smart_str_appends(&req_buf, protocol_version);
415 smart_str_appends(&req_buf, "\r\n");
416 efree(protocol_version);
417 } else {
418 smart_str_appends(&req_buf, " HTTP/1.1\r\n");
419 }
420
421 if (context && (tmpzval = php_stream_context_get_option(context, "http", "header")) != NULL) {
422 tmp = NULL;
423
424 if (Z_TYPE_P(tmpzval) == IS_ARRAY) {
425 zval *tmpheader = NULL;
426 smart_str tmpstr = {0};
427
428 ZEND_HASH_FOREACH_VAL(Z_ARRVAL_P(tmpzval), tmpheader) {
429 if (Z_TYPE_P(tmpheader) == IS_STRING) {
430 smart_str_append(&tmpstr, Z_STR_P(tmpheader));
431 smart_str_appendl(&tmpstr, "\r\n", sizeof("\r\n") - 1);
432 }
433 } ZEND_HASH_FOREACH_END();
434 smart_str_0(&tmpstr);
435 /* Remove newlines and spaces from start and end. there's at least one extra \r\n at the end that needs to go. */
436 if (tmpstr.s) {
437 tmp = php_trim(tmpstr.s, NULL, 0, 3);
438 smart_str_free(&tmpstr);
439 }
440 } else if (Z_TYPE_P(tmpzval) == IS_STRING && Z_STRLEN_P(tmpzval)) {
441 /* Remove newlines and spaces from start and end php_trim will estrndup() */
442 tmp = php_trim(Z_STR_P(tmpzval), NULL, 0, 3);
443 }
444 if (tmp && ZSTR_LEN(tmp)) {
445 char *s;
446 char *t;
447
448 user_headers = estrndup(ZSTR_VAL(tmp), ZSTR_LEN(tmp));
449
450 if (ZSTR_IS_INTERNED(tmp)) {
451 tmp = zend_string_init(ZSTR_VAL(tmp), ZSTR_LEN(tmp), 0);
452 } else if (GC_REFCOUNT(tmp) > 1) {
453 GC_DELREF(tmp);
454 tmp = zend_string_init(ZSTR_VAL(tmp), ZSTR_LEN(tmp), 0);
455 }
456
457 /* Make lowercase for easy comparison against 'standard' headers */
458 zend_str_tolower(ZSTR_VAL(tmp), ZSTR_LEN(tmp));
459 t = ZSTR_VAL(tmp);
460
461 if (!header_init) {
462 /* strip POST headers on redirect */
463 strip_header(user_headers, t, "content-length:");
464 strip_header(user_headers, t, "content-type:");
465 }
466
467 if (check_has_header(t, "user-agent:")) {
468 have_header |= HTTP_HEADER_USER_AGENT;
469 }
470 if (check_has_header(t, "host:")) {
471 have_header |= HTTP_HEADER_HOST;
472 }
473 if (check_has_header(t, "from:")) {
474 have_header |= HTTP_HEADER_FROM;
475 }
476 if (check_has_header(t, "authorization:")) {
477 have_header |= HTTP_HEADER_AUTH;
478 }
479 if (check_has_header(t, "content-length:")) {
480 have_header |= HTTP_HEADER_CONTENT_LENGTH;
481 }
482 if (check_has_header(t, "content-type:")) {
483 have_header |= HTTP_HEADER_TYPE;
484 }
485 if (check_has_header(t, "connection:")) {
486 have_header |= HTTP_HEADER_CONNECTION;
487 }
488
489 /* remove Proxy-Authorization header */
490 if (use_proxy && use_ssl && (s = strstr(t, "proxy-authorization:")) &&
491 (s == t || *(s-1) == '\n')) {
492 char *p = s + sizeof("proxy-authorization:") - 1;
493
494 while (s > t && (*(s-1) == ' ' || *(s-1) == '\t')) s--;
495 while (*p != 0 && *p != '\r' && *p != '\n') p++;
496 while (*p == '\r' || *p == '\n') p++;
497 if (*p == 0) {
498 if (s == t) {
499 efree(user_headers);
500 user_headers = NULL;
501 } else {
502 while (s > t && (*(s-1) == '\r' || *(s-1) == '\n')) s--;
503 user_headers[s - t] = 0;
504 }
505 } else {
506 memmove(user_headers + (s - t), user_headers + (p - t), strlen(p) + 1);
507 }
508 }
509
510 }
511 if (tmp) {
512 zend_string_release_ex(tmp, 0);
513 }
514 }
515
516 /* auth header if it was specified */
517 if (((have_header & HTTP_HEADER_AUTH) == 0) && resource->user) {
518 /* make scratch large enough to hold the whole URL (over-estimate) */
519 size_t scratch_len = strlen(path) + 1;
520 char *scratch = emalloc(scratch_len);
521 zend_string *stmp;
522
523 /* decode the strings first */
524 php_url_decode(ZSTR_VAL(resource->user), ZSTR_LEN(resource->user));
525
526 strcpy(scratch, ZSTR_VAL(resource->user));
527 strcat(scratch, ":");
528
529 /* Note: password is optional! */
530 if (resource->pass) {
531 php_url_decode(ZSTR_VAL(resource->pass), ZSTR_LEN(resource->pass));
532 strcat(scratch, ZSTR_VAL(resource->pass));
533 }
534
535 stmp = php_base64_encode((unsigned char*)scratch, strlen(scratch));
536
537 smart_str_appends(&req_buf, "Authorization: Basic ");
538 smart_str_appends(&req_buf, ZSTR_VAL(stmp));
539 smart_str_appends(&req_buf, "\r\n");
540
541 php_stream_notify_info(context, PHP_STREAM_NOTIFY_AUTH_REQUIRED, NULL, 0);
542
543 zend_string_free(stmp);
544 efree(scratch);
545 }
546
547 /* if the user has configured who they are, send a From: line */
548 if (!(have_header & HTTP_HEADER_FROM) && FG(from_address)) {
549 smart_str_appends(&req_buf, "From: ");
550 smart_str_appends(&req_buf, FG(from_address));
551 smart_str_appends(&req_buf, "\r\n");
552 }
553
554 /* Send Host: header so name-based virtual hosts work */
555 if ((have_header & HTTP_HEADER_HOST) == 0) {
556 smart_str_appends(&req_buf, "Host: ");
557 smart_str_appends(&req_buf, ZSTR_VAL(resource->host));
558 if ((use_ssl && resource->port != 443 && resource->port != 0) ||
559 (!use_ssl && resource->port != 80 && resource->port != 0)) {
560 smart_str_appendc(&req_buf, ':');
561 smart_str_append_unsigned(&req_buf, resource->port);
562 }
563 smart_str_appends(&req_buf, "\r\n");
564 }
565
566 /* Send a Connection: close header to avoid hanging when the server
567 * interprets the RFC literally and establishes a keep-alive connection,
568 * unless the user specifically requests something else by specifying a
569 * Connection header in the context options. Send that header even for
570 * HTTP/1.0 to avoid issues when the server respond with a HTTP/1.1
571 * keep-alive response, which is the preferred response type. */
572 if ((have_header & HTTP_HEADER_CONNECTION) == 0) {
573 smart_str_appends(&req_buf, "Connection: close\r\n");
574 }
575
576 if (context &&
577 (ua_zval = php_stream_context_get_option(context, "http", "user_agent")) != NULL &&
578 Z_TYPE_P(ua_zval) == IS_STRING) {
579 ua_str = Z_STRVAL_P(ua_zval);
580 } else if (FG(user_agent)) {
581 ua_str = FG(user_agent);
582 }
583
584 if (((have_header & HTTP_HEADER_USER_AGENT) == 0) && ua_str) {
585 #define _UA_HEADER "User-Agent: %s\r\n"
586 char *ua;
587 size_t ua_len;
588
589 ua_len = sizeof(_UA_HEADER) + strlen(ua_str);
590
591 /* ensure the header is only sent if user_agent is not blank */
592 if (ua_len > sizeof(_UA_HEADER)) {
593 ua = emalloc(ua_len + 1);
594 if ((ua_len = slprintf(ua, ua_len, _UA_HEADER, ua_str)) > 0) {
595 ua[ua_len] = 0;
596 smart_str_appendl(&req_buf, ua, ua_len);
597 } else {
598 php_error_docref(NULL, E_WARNING, "Cannot construct User-agent header");
599 }
600 efree(ua);
601 }
602 }
603
604 if (user_headers) {
605 /* A bit weird, but some servers require that Content-Length be sent prior to Content-Type for POST
606 * see bug #44603 for details. Since Content-Type maybe part of user's headers we need to do this check first.
607 */
608 if (
609 header_init &&
610 context &&
611 !(have_header & HTTP_HEADER_CONTENT_LENGTH) &&
612 (tmpzval = php_stream_context_get_option(context, "http", "content")) != NULL &&
613 Z_TYPE_P(tmpzval) == IS_STRING && Z_STRLEN_P(tmpzval) > 0
614 ) {
615 smart_str_appends(&req_buf, "Content-Length: ");
616 smart_str_append_unsigned(&req_buf, Z_STRLEN_P(tmpzval));
617 smart_str_appends(&req_buf, "\r\n");
618 have_header |= HTTP_HEADER_CONTENT_LENGTH;
619 }
620
621 smart_str_appends(&req_buf, user_headers);
622 smart_str_appends(&req_buf, "\r\n");
623 efree(user_headers);
624 }
625
626 /* Request content, such as for POST requests */
627 if (header_init && context &&
628 (tmpzval = php_stream_context_get_option(context, "http", "content")) != NULL &&
629 Z_TYPE_P(tmpzval) == IS_STRING && Z_STRLEN_P(tmpzval) > 0) {
630 if (!(have_header & HTTP_HEADER_CONTENT_LENGTH)) {
631 smart_str_appends(&req_buf, "Content-Length: ");
632 smart_str_append_unsigned(&req_buf, Z_STRLEN_P(tmpzval));
633 smart_str_appends(&req_buf, "\r\n");
634 }
635 if (!(have_header & HTTP_HEADER_TYPE)) {
636 smart_str_appends(&req_buf, "Content-Type: application/x-www-form-urlencoded\r\n");
637 php_error_docref(NULL, E_NOTICE, "Content-type not specified assuming application/x-www-form-urlencoded");
638 }
639 smart_str_appends(&req_buf, "\r\n");
640 smart_str_appendl(&req_buf, Z_STRVAL_P(tmpzval), Z_STRLEN_P(tmpzval));
641 } else {
642 smart_str_appends(&req_buf, "\r\n");
643 }
644
645 /* send it */
646 php_stream_write(stream, ZSTR_VAL(req_buf.s), ZSTR_LEN(req_buf.s));
647
648 location[0] = '\0';
649
650 if (Z_ISUNDEF_P(response_header)) {
651 array_init(response_header);
652 }
653
654 {
655 /* get response header */
656 size_t tmp_line_len;
657 if (!php_stream_eof(stream) &&
658 php_stream_get_line(stream, tmp_line, sizeof(tmp_line) - 1, &tmp_line_len) != NULL) {
659 zval http_response;
660
661 if (tmp_line_len > 9) {
662 response_code = atoi(tmp_line + 9);
663 } else {
664 response_code = 0;
665 }
666 if (context && NULL != (tmpzval = php_stream_context_get_option(context, "http", "ignore_errors"))) {
667 ignore_errors = zend_is_true(tmpzval);
668 }
669 /* when we request only the header, don't fail even on error codes */
670 if ((options & STREAM_ONLY_GET_HEADERS) || ignore_errors) {
671 reqok = 1;
672 }
673
674 /* status codes of 1xx are "informational", and will be followed by a real response
675 * e.g "100 Continue". RFC 7231 states that unexpected 1xx status MUST be parsed,
676 * and MAY be ignored. As such, we need to skip ahead to the "real" status*/
677 if (response_code >= 100 && response_code < 200 && response_code != 101) {
678 /* consume lines until we find a line starting 'HTTP/1' */
679 while (
680 !php_stream_eof(stream)
681 && php_stream_get_line(stream, tmp_line, sizeof(tmp_line) - 1, &tmp_line_len) != NULL
682 && ( tmp_line_len < sizeof("HTTP/1") - 1 || strncasecmp(tmp_line, "HTTP/1", sizeof("HTTP/1") - 1) )
683 );
684
685 if (tmp_line_len > 9) {
686 response_code = atoi(tmp_line + 9);
687 } else {
688 response_code = 0;
689 }
690 }
691 /* all status codes in the 2xx range are defined by the specification as successful;
692 * all status codes in the 3xx range are for redirection, and so also should never
693 * fail */
694 if (response_code >= 200 && response_code < 400) {
695 reqok = 1;
696 } else {
697 switch(response_code) {
698 case 403:
699 php_stream_notify_error(context, PHP_STREAM_NOTIFY_AUTH_RESULT,
700 tmp_line, response_code);
701 break;
702 default:
703 /* safety net in the event tmp_line == NULL */
704 if (!tmp_line_len) {
705 tmp_line[0] = '\0';
706 }
707 php_stream_notify_error(context, PHP_STREAM_NOTIFY_FAILURE,
708 tmp_line, response_code);
709 }
710 }
711 if (tmp_line_len >= 1 && tmp_line[tmp_line_len - 1] == '\n') {
712 --tmp_line_len;
713 if (tmp_line_len >= 1 &&tmp_line[tmp_line_len - 1] == '\r') {
714 --tmp_line_len;
715 }
716 }
717 ZVAL_STRINGL(&http_response, tmp_line, tmp_line_len);
718 zend_hash_next_index_insert(Z_ARRVAL_P(response_header), &http_response);
719 } else {
720 php_stream_close(stream);
721 stream = NULL;
722 php_stream_wrapper_log_error(wrapper, options, "HTTP request failed!");
723 goto out;
724 }
725 }
726
727 /* read past HTTP headers */
728
729 while (!php_stream_eof(stream)) {
730 size_t http_header_line_length;
731
732 if (http_header_line != NULL) {
733 efree(http_header_line);
734 }
735 if ((http_header_line = php_stream_get_line(stream, NULL, 0, &http_header_line_length)) && *http_header_line != '\n' && *http_header_line != '\r') {
736 char *e = http_header_line + http_header_line_length - 1;
737 char *http_header_value;
738
739 while (e >= http_header_line && (*e == '\n' || *e == '\r')) {
740 e--;
741 }
742
743 /* The primary definition of an HTTP header in RFC 7230 states:
744 * > Each header field consists of a case-insensitive field name followed
745 * > by a colon (":"), optional leading whitespace, the field value, and
746 * > optional trailing whitespace. */
747
748 /* Strip trailing whitespace */
749 while (e >= http_header_line && (*e == ' ' || *e == '\t')) {
750 e--;
751 }
752
753 /* Terminate header line */
754 e++;
755 *e = '\0';
756 http_header_line_length = e - http_header_line;
757
758 http_header_value = memchr(http_header_line, ':', http_header_line_length);
759 if (http_header_value) {
760 http_header_value++; /* Skip ':' */
761
762 /* Strip leading whitespace */
763 while (http_header_value < e
764 && (*http_header_value == ' ' || *http_header_value == '\t')) {
765 http_header_value++;
766 }
767 } else {
768 /* There is no colon. Set the value to the end of the header line, which is
769 * effectively an empty string. */
770 http_header_value = e;
771 }
772
773 if (!strncasecmp(http_header_line, "Location:", sizeof("Location:")-1)) {
774 if (context && (tmpzval = php_stream_context_get_option(context, "http", "follow_location")) != NULL) {
775 follow_location = zval_is_true(tmpzval);
776 } else if (!((response_code >= 300 && response_code < 304)
777 || 307 == response_code || 308 == response_code)) {
778 /* we shouldn't redirect automatically
779 if follow_location isn't set and response_code not in (300, 301, 302, 303 and 307)
780 see http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html#sec10.3.1
781 RFC 7238 defines 308: http://tools.ietf.org/html/rfc7238 */
782 follow_location = 0;
783 }
784 strlcpy(location, http_header_value, sizeof(location));
785 } else if (!strncasecmp(http_header_line, "Content-Type:", sizeof("Content-Type:")-1)) {
786 php_stream_notify_info(context, PHP_STREAM_NOTIFY_MIME_TYPE_IS, http_header_value, 0);
787 } else if (!strncasecmp(http_header_line, "Content-Length:", sizeof("Content-Length:")-1)) {
788 file_size = atoi(http_header_value);
789 php_stream_notify_file_size(context, file_size, http_header_line, 0);
790 } else if (
791 !strncasecmp(http_header_line, "Transfer-Encoding:", sizeof("Transfer-Encoding:")-1)
792 && !strncasecmp(http_header_value, "Chunked", sizeof("Chunked")-1)
793 ) {
794
795 /* create filter to decode response body */
796 if (!(options & STREAM_ONLY_GET_HEADERS)) {
797 zend_long decode = 1;
798
799 if (context && (tmpzval = php_stream_context_get_option(context, "http", "auto_decode")) != NULL) {
800 decode = zend_is_true(tmpzval);
801 }
802 if (decode) {
803 transfer_encoding = php_stream_filter_create("dechunk", NULL, php_stream_is_persistent(stream));
804 if (transfer_encoding) {
805 /* don't store transfer-encodeing header */
806 continue;
807 }
808 }
809 }
810 }
811
812 {
813 zval http_header;
814 ZVAL_STRINGL(&http_header, http_header_line, http_header_line_length);
815 zend_hash_next_index_insert(Z_ARRVAL_P(response_header), &http_header);
816 }
817 } else {
818 break;
819 }
820 }
821
822 if (!reqok || (location[0] != '\0' && follow_location)) {
823 if (!follow_location || (((options & STREAM_ONLY_GET_HEADERS) || ignore_errors) && redirect_max <= 1)) {
824 goto out;
825 }
826
827 if (location[0] != '\0')
828 php_stream_notify_info(context, PHP_STREAM_NOTIFY_REDIRECTED, location, 0);
829
830 php_stream_close(stream);
831 stream = NULL;
832
833 if (transfer_encoding) {
834 php_stream_filter_free(transfer_encoding);
835 transfer_encoding = NULL;
836 }
837
838 if (location[0] != '\0') {
839
840 char new_path[HTTP_HEADER_BLOCK_SIZE];
841 char loc_path[HTTP_HEADER_BLOCK_SIZE];
842
843 *new_path='\0';
844 if (strlen(location)<8 || (strncasecmp(location, "http://", sizeof("http://")-1) &&
845 strncasecmp(location, "https://", sizeof("https://")-1) &&
846 strncasecmp(location, "ftp://", sizeof("ftp://")-1) &&
847 strncasecmp(location, "ftps://", sizeof("ftps://")-1)))
848 {
849 if (*location != '/') {
850 if (*(location+1) != '\0' && resource->path) {
851 char *s = strrchr(ZSTR_VAL(resource->path), '/');
852 if (!s) {
853 s = ZSTR_VAL(resource->path);
854 if (!ZSTR_LEN(resource->path)) {
855 zend_string_release_ex(resource->path, 0);
856 resource->path = zend_string_init("/", 1, 0);
857 s = ZSTR_VAL(resource->path);
858 } else {
859 *s = '/';
860 }
861 }
862 s[1] = '\0';
863 if (resource->path &&
864 ZSTR_VAL(resource->path)[0] == '/' &&
865 ZSTR_VAL(resource->path)[1] == '\0') {
866 snprintf(loc_path, sizeof(loc_path) - 1, "%s%s", ZSTR_VAL(resource->path), location);
867 } else {
868 snprintf(loc_path, sizeof(loc_path) - 1, "%s/%s", ZSTR_VAL(resource->path), location);
869 }
870 } else {
871 snprintf(loc_path, sizeof(loc_path) - 1, "/%s", location);
872 }
873 } else {
874 strlcpy(loc_path, location, sizeof(loc_path));
875 }
876 if ((use_ssl && resource->port != 443) || (!use_ssl && resource->port != 80)) {
877 snprintf(new_path, sizeof(new_path) - 1, "%s://%s:%d%s", ZSTR_VAL(resource->scheme), ZSTR_VAL(resource->host), resource->port, loc_path);
878 } else {
879 snprintf(new_path, sizeof(new_path) - 1, "%s://%s%s", ZSTR_VAL(resource->scheme), ZSTR_VAL(resource->host), loc_path);
880 }
881 } else {
882 strlcpy(new_path, location, sizeof(new_path));
883 }
884
885 php_url_free(resource);
886 /* check for invalid redirection URLs */
887 if ((resource = php_url_parse(new_path)) == NULL) {
888 php_stream_wrapper_log_error(wrapper, options, "Invalid redirect URL! %s", new_path);
889 goto out;
890 }
891
892 #define CHECK_FOR_CNTRL_CHARS(val) { \
893 if (val) { \
894 unsigned char *s, *e; \
895 ZSTR_LEN(val) = php_url_decode(ZSTR_VAL(val), ZSTR_LEN(val)); \
896 s = (unsigned char*)ZSTR_VAL(val); e = s + ZSTR_LEN(val); \
897 while (s < e) { \
898 if (iscntrl(*s)) { \
899 php_stream_wrapper_log_error(wrapper, options, "Invalid redirect URL! %s", new_path); \
900 goto out; \
901 } \
902 s++; \
903 } \
904 } \
905 }
906 /* check for control characters in login, password & path */
907 if (strncasecmp(new_path, "http://", sizeof("http://") - 1) || strncasecmp(new_path, "https://", sizeof("https://") - 1)) {
908 CHECK_FOR_CNTRL_CHARS(resource->user);
909 CHECK_FOR_CNTRL_CHARS(resource->pass);
910 CHECK_FOR_CNTRL_CHARS(resource->path);
911 }
912 stream = php_stream_url_wrap_http_ex(
913 wrapper, new_path, mode, options, opened_path, context,
914 --redirect_max, HTTP_WRAPPER_REDIRECTED, response_header STREAMS_CC);
915 } else {
916 php_stream_wrapper_log_error(wrapper, options, "HTTP request failed! %s", tmp_line);
917 }
918 }
919 out:
920
921 smart_str_free(&req_buf);
922
923 if (http_header_line) {
924 efree(http_header_line);
925 }
926
927 if (resource) {
928 php_url_free(resource);
929 }
930
931 if (stream) {
932 if (header_init) {
933 ZVAL_COPY(&stream->wrapperdata, response_header);
934 }
935 php_stream_notify_progress_init(context, 0, file_size);
936
937 /* Restore original chunk size now that we're done with headers */
938 if (options & STREAM_WILL_CAST)
939 php_stream_set_chunk_size(stream, (int)chunk_size);
940
941 /* restore the users auto-detect-line-endings setting */
942 stream->flags |= eol_detect;
943
944 /* as far as streams are concerned, we are now at the start of
945 * the stream */
946 stream->position = 0;
947
948 /* restore mode */
949 strlcpy(stream->mode, mode, sizeof(stream->mode));
950
951 if (transfer_encoding) {
952 php_stream_filter_append(&stream->readfilters, transfer_encoding);
953 }
954 }
955
956 return stream;
957 }
958 /* }}} */
959
php_stream_url_wrap_http(php_stream_wrapper * wrapper,const char * path,const char * mode,int options,zend_string ** opened_path,php_stream_context * context STREAMS_DC)960 php_stream *php_stream_url_wrap_http(php_stream_wrapper *wrapper, const char *path, const char *mode, int options, zend_string **opened_path, php_stream_context *context STREAMS_DC) /* {{{ */
961 {
962 php_stream *stream;
963 zval headers;
964 ZVAL_UNDEF(&headers);
965
966 stream = php_stream_url_wrap_http_ex(
967 wrapper, path, mode, options, opened_path, context,
968 PHP_URL_REDIRECT_MAX, HTTP_WRAPPER_HEADER_INIT, &headers STREAMS_CC);
969
970 if (!Z_ISUNDEF(headers)) {
971 if (FAILURE == zend_set_local_var_str(
972 "http_response_header", sizeof("http_response_header")-1, &headers, 0)) {
973 zval_ptr_dtor(&headers);
974 }
975 }
976
977 return stream;
978 }
979 /* }}} */
980
php_stream_http_stream_stat(php_stream_wrapper * wrapper,php_stream * stream,php_stream_statbuf * ssb)981 static int php_stream_http_stream_stat(php_stream_wrapper *wrapper, php_stream *stream, php_stream_statbuf *ssb) /* {{{ */
982 {
983 /* one day, we could fill in the details based on Date: and Content-Length:
984 * headers. For now, we return with a failure code to prevent the underlying
985 * file's details from being used instead. */
986 return -1;
987 }
988 /* }}} */
989
990 static const php_stream_wrapper_ops http_stream_wops = {
991 php_stream_url_wrap_http,
992 NULL, /* stream_close */
993 php_stream_http_stream_stat,
994 NULL, /* stat_url */
995 NULL, /* opendir */
996 "http",
997 NULL, /* unlink */
998 NULL, /* rename */
999 NULL, /* mkdir */
1000 NULL, /* rmdir */
1001 NULL
1002 };
1003
1004 PHPAPI const php_stream_wrapper php_stream_http_wrapper = {
1005 &http_stream_wops,
1006 NULL,
1007 1 /* is_url */
1008 };
1009