xref: /PHP-5.4/ext/standard/http_fopen_wrapper.c (revision 45facd15)
1 /*
2    +----------------------------------------------------------------------+
3    | PHP Version 5                                                        |
4    +----------------------------------------------------------------------+
5    | Copyright (c) 1997-2014 The PHP Group                                |
6    +----------------------------------------------------------------------+
7    | This source file is subject to version 3.01 of the PHP license,      |
8    | that is bundled with this package in the file LICENSE, and is        |
9    | available through the world-wide-web at the following url:           |
10    | http://www.php.net/license/3_01.txt                                  |
11    | If you did not receive a copy of the PHP license and are unable to   |
12    | obtain it through the world-wide-web, please send a note to          |
13    | license@php.net so we can mail you a copy immediately.               |
14    +----------------------------------------------------------------------+
15    | Authors: Rasmus Lerdorf <rasmus@php.net>                             |
16    |          Jim Winstead <jimw@php.net>                                 |
17    |          Hartmut Holzgraefe <hholzgra@php.net>                       |
18    |          Wez Furlong <wez@thebrainroom.com>                          |
19    |          Sara Golemon <pollita@php.net>                              |
20    +----------------------------------------------------------------------+
21  */
22 /* $Id$ */
23 
24 #include "php.h"
25 #include "php_globals.h"
26 #include "php_streams.h"
27 #include "php_network.h"
28 #include "php_ini.h"
29 #include "ext/standard/basic_functions.h"
30 #include "ext/standard/php_smart_str.h"
31 
32 #include <stdio.h>
33 #include <stdlib.h>
34 #include <errno.h>
35 #include <sys/types.h>
36 #include <sys/stat.h>
37 #include <fcntl.h>
38 
39 #ifdef PHP_WIN32
40 #define O_RDONLY _O_RDONLY
41 #include "win32/param.h"
42 #else
43 #include <sys/param.h>
44 #endif
45 
46 #include "php_standard.h"
47 
48 #include <sys/types.h>
49 #if HAVE_SYS_SOCKET_H
50 #include <sys/socket.h>
51 #endif
52 
53 #ifdef PHP_WIN32
54 #include <winsock2.h>
55 #elif defined(NETWARE) && defined(USE_WINSOCK)
56 #include <novsock2.h>
57 #else
58 #include <netinet/in.h>
59 #include <netdb.h>
60 #if HAVE_ARPA_INET_H
61 #include <arpa/inet.h>
62 #endif
63 #endif
64 
65 #if defined(PHP_WIN32) || defined(__riscos__) || defined(NETWARE)
66 #undef AF_UNIX
67 #endif
68 
69 #if defined(AF_UNIX)
70 #include <sys/un.h>
71 #endif
72 
73 #include "php_fopen_wrappers.h"
74 
75 #define HTTP_HEADER_BLOCK_SIZE		1024
76 #define PHP_URL_REDIRECT_MAX		20
77 #define HTTP_HEADER_USER_AGENT		1
78 #define HTTP_HEADER_HOST			2
79 #define HTTP_HEADER_AUTH			4
80 #define HTTP_HEADER_FROM			8
81 #define HTTP_HEADER_CONTENT_LENGTH	16
82 #define HTTP_HEADER_TYPE			32
83 
84 #define HTTP_WRAPPER_HEADER_INIT    1
85 #define HTTP_WRAPPER_REDIRECTED     2
86 
strip_header(char * header_bag,char * lc_header_bag,const char * lc_header_name)87 static inline void strip_header(char *header_bag, char *lc_header_bag,
88 		const char *lc_header_name)
89 {
90 	char *lc_header_start = strstr(lc_header_bag, lc_header_name);
91 	char *header_start = header_bag + (lc_header_start - lc_header_bag);
92 
93 	if (lc_header_start
94 	&& (lc_header_start == lc_header_bag || *(lc_header_start-1) == '\n')
95 	) {
96 		char *lc_eol = strchr(lc_header_start, '\n');
97 		char *eol = header_start + (lc_eol - lc_header_start);
98 
99 		if (lc_eol) {
100 			size_t eollen = strlen(lc_eol);
101 
102 			memmove(lc_header_start, lc_eol+1, eollen);
103 			memmove(header_start, eol+1, eollen);
104 		} else {
105 			*lc_header_start = '\0';
106 			*header_start = '\0';
107 		}
108 	}
109 }
110 
php_stream_url_wrap_http_ex(php_stream_wrapper * wrapper,char * path,char * mode,int options,char ** opened_path,php_stream_context * context,int redirect_max,int flags STREAMS_DC TSRMLS_DC)111 php_stream *php_stream_url_wrap_http_ex(php_stream_wrapper *wrapper, char *path, char *mode, int options, char **opened_path, php_stream_context *context, int redirect_max, int flags STREAMS_DC TSRMLS_DC) /* {{{ */
112 {
113 	php_stream *stream = NULL;
114 	php_url *resource = NULL;
115 	int use_ssl;
116 	int use_proxy = 0;
117 	char *scratch = NULL;
118 	char *tmp = NULL;
119 	char *ua_str = NULL;
120 	zval **ua_zval = NULL, **tmpzval = NULL;
121 	int scratch_len = 0;
122 	int body = 0;
123 	char location[HTTP_HEADER_BLOCK_SIZE];
124 	zval *response_header = NULL;
125 	int reqok = 0;
126 	char *http_header_line = NULL;
127 	char tmp_line[128];
128 	size_t chunk_size = 0, file_size = 0;
129 	int eol_detect = 0;
130 	char *transport_string, *errstr = NULL;
131 	int transport_len, have_header = 0, request_fulluri = 0, ignore_errors = 0;
132 	char *protocol_version = NULL;
133 	int protocol_version_len = 3; /* Default: "1.0" */
134 	struct timeval timeout;
135 	char *user_headers = NULL;
136 	int header_init = ((flags & HTTP_WRAPPER_HEADER_INIT) != 0);
137 	int redirected = ((flags & HTTP_WRAPPER_REDIRECTED) != 0);
138 	int follow_location = 1;
139 	php_stream_filter *transfer_encoding = NULL;
140 	int response_code;
141 
142 	tmp_line[0] = '\0';
143 
144 	if (redirect_max < 1) {
145 		php_stream_wrapper_log_error(wrapper, options TSRMLS_CC, "Redirection limit reached, aborting");
146 		return NULL;
147 	}
148 
149 	resource = php_url_parse(path);
150 	if (resource == NULL) {
151 		return NULL;
152 	}
153 
154 	if (strncasecmp(resource->scheme, "http", sizeof("http")) && strncasecmp(resource->scheme, "https", sizeof("https"))) {
155 		if (!context ||
156 			php_stream_context_get_option(context, wrapper->wops->label, "proxy", &tmpzval) == FAILURE ||
157 			Z_TYPE_PP(tmpzval) != IS_STRING ||
158 			Z_STRLEN_PP(tmpzval) <= 0) {
159 			php_url_free(resource);
160 			return php_stream_open_wrapper_ex(path, mode, REPORT_ERRORS, NULL, context);
161 		}
162 		/* Called from a non-http wrapper with http proxying requested (i.e. ftp) */
163 		request_fulluri = 1;
164 		use_ssl = 0;
165 		use_proxy = 1;
166 
167 		transport_len = Z_STRLEN_PP(tmpzval);
168 		transport_string = estrndup(Z_STRVAL_PP(tmpzval), Z_STRLEN_PP(tmpzval));
169 	} else {
170 		/* Normal http request (possibly with proxy) */
171 
172 		if (strpbrk(mode, "awx+")) {
173 			php_stream_wrapper_log_error(wrapper, options TSRMLS_CC, "HTTP wrapper does not support writeable connections");
174 			php_url_free(resource);
175 			return NULL;
176 		}
177 
178 		use_ssl = resource->scheme && (strlen(resource->scheme) > 4) && resource->scheme[4] == 's';
179 		/* choose default ports */
180 		if (use_ssl && resource->port == 0)
181 			resource->port = 443;
182 		else if (resource->port == 0)
183 			resource->port = 80;
184 
185 		if (context &&
186 			php_stream_context_get_option(context, wrapper->wops->label, "proxy", &tmpzval) == SUCCESS &&
187 			Z_TYPE_PP(tmpzval) == IS_STRING &&
188 			Z_STRLEN_PP(tmpzval) > 0) {
189 			use_proxy = 1;
190 			transport_len = Z_STRLEN_PP(tmpzval);
191 			transport_string = estrndup(Z_STRVAL_PP(tmpzval), Z_STRLEN_PP(tmpzval));
192 		} else {
193 			transport_len = spprintf(&transport_string, 0, "%s://%s:%d", use_ssl ? "ssl" : "tcp", resource->host, resource->port);
194 		}
195 	}
196 
197 	if (context && php_stream_context_get_option(context, wrapper->wops->label, "timeout", &tmpzval) == SUCCESS) {
198 		SEPARATE_ZVAL(tmpzval);
199 		convert_to_double_ex(tmpzval);
200 		timeout.tv_sec = (time_t) Z_DVAL_PP(tmpzval);
201 		timeout.tv_usec = (size_t) ((Z_DVAL_PP(tmpzval) - timeout.tv_sec) * 1000000);
202 	} else {
203 		timeout.tv_sec = FG(default_socket_timeout);
204 		timeout.tv_usec = 0;
205 	}
206 
207 	stream = php_stream_xport_create(transport_string, transport_len, options,
208 			STREAM_XPORT_CLIENT | STREAM_XPORT_CONNECT,
209 			NULL, &timeout, context, &errstr, NULL);
210 
211 	if (stream) {
212 		php_stream_set_option(stream, PHP_STREAM_OPTION_READ_TIMEOUT, 0, &timeout);
213 	}
214 
215 	if (errstr) {
216 		php_stream_wrapper_log_error(wrapper, options TSRMLS_CC, "%s", errstr);
217 		efree(errstr);
218 		errstr = NULL;
219 	}
220 
221 	efree(transport_string);
222 
223 	if (stream && use_proxy && use_ssl) {
224 		smart_str header = {0};
225 
226 		smart_str_appendl(&header, "CONNECT ", sizeof("CONNECT ")-1);
227 		smart_str_appends(&header, resource->host);
228 		smart_str_appendc(&header, ':');
229 		smart_str_append_unsigned(&header, resource->port);
230 		smart_str_appendl(&header, " HTTP/1.0\r\n", sizeof(" HTTP/1.0\r\n")-1);
231 
232 	    /* check if we have Proxy-Authorization header */
233 		if (context && php_stream_context_get_option(context, "http", "header", &tmpzval) == SUCCESS) {
234 			char *s, *p;
235 
236 			if (Z_TYPE_PP(tmpzval) == IS_ARRAY) {
237 				HashPosition pos;
238 				zval **tmpheader = NULL;
239 
240 				for (zend_hash_internal_pointer_reset_ex(Z_ARRVAL_PP(tmpzval), &pos);
241 					SUCCESS == zend_hash_get_current_data_ex(Z_ARRVAL_PP(tmpzval), (void *)&tmpheader, &pos);
242 					zend_hash_move_forward_ex(Z_ARRVAL_PP(tmpzval), &pos)) {
243 					if (Z_TYPE_PP(tmpheader) == IS_STRING) {
244 						s = Z_STRVAL_PP(tmpheader);
245 						do {
246 							while (*s == ' ' || *s == '\t') s++;
247 							p = s;
248 							while (*p != 0 && *p != ':' && *p != '\r' && *p !='\n') p++;
249 							if (*p == ':') {
250 								p++;
251 								if (p - s == sizeof("Proxy-Authorization:") - 1 &&
252 								    zend_binary_strcasecmp(s, sizeof("Proxy-Authorization:") - 1,
253 								        "Proxy-Authorization:", sizeof("Proxy-Authorization:") - 1) == 0) {
254 									while (*p != 0 && *p != '\r' && *p !='\n') p++;
255 									smart_str_appendl(&header, s, p - s);
256 									smart_str_appendl(&header, "\r\n", sizeof("\r\n")-1);
257 									goto finish;
258 								} else {
259 									while (*p != 0 && *p != '\r' && *p !='\n') p++;
260 								}
261 							}
262 							s = p;
263 							while (*s == '\r' || *s == '\n') s++;
264 						} while (*s != 0);
265 					}
266 				}
267 			} else if (Z_TYPE_PP(tmpzval) == IS_STRING && Z_STRLEN_PP(tmpzval)) {
268 				s = Z_STRVAL_PP(tmpzval);
269 				do {
270 					while (*s == ' ' || *s == '\t') s++;
271 					p = s;
272 					while (*p != 0 && *p != ':' && *p != '\r' && *p !='\n') p++;
273 					if (*p == ':') {
274 						p++;
275 						if (p - s == sizeof("Proxy-Authorization:") - 1 &&
276 						    zend_binary_strcasecmp(s, sizeof("Proxy-Authorization:") - 1,
277 						        "Proxy-Authorization:", sizeof("Proxy-Authorization:") - 1) == 0) {
278 							while (*p != 0 && *p != '\r' && *p !='\n') p++;
279 							smart_str_appendl(&header, s, p - s);
280 							smart_str_appendl(&header, "\r\n", sizeof("\r\n")-1);
281 							goto finish;
282 						} else {
283 							while (*p != 0 && *p != '\r' && *p !='\n') p++;
284 						}
285 					}
286 					s = p;
287 					while (*s == '\r' || *s == '\n') s++;
288 				} while (*s != 0);
289 			}
290 		}
291 finish:
292 		smart_str_appendl(&header, "\r\n", sizeof("\r\n")-1);
293 
294 		if (php_stream_write(stream, header.c, header.len) != header.len) {
295 			php_stream_wrapper_log_error(wrapper, options TSRMLS_CC, "Cannot connect to HTTPS server through proxy");
296 			php_stream_close(stream);
297 			stream = NULL;
298 		}
299  	 	smart_str_free(&header);
300 
301  	 	if (stream) {
302  	 		char header_line[HTTP_HEADER_BLOCK_SIZE];
303 
304 			/* get response header */
305 			while (php_stream_gets(stream, header_line, HTTP_HEADER_BLOCK_SIZE-1) != NULL) {
306 				if (header_line[0] == '\n' ||
307 				    header_line[0] == '\r' ||
308 				    header_line[0] == '\0') {
309 				  break;
310 				}
311 			}
312 		}
313 
314 		/* enable SSL transport layer */
315 		if (stream) {
316 			if (php_stream_xport_crypto_setup(stream, STREAM_CRYPTO_METHOD_SSLv23_CLIENT, NULL TSRMLS_CC) < 0 ||
317 			    php_stream_xport_crypto_enable(stream, 1 TSRMLS_CC) < 0) {
318 				php_stream_wrapper_log_error(wrapper, options TSRMLS_CC, "Cannot connect to HTTPS server through proxy");
319 				php_stream_close(stream);
320 				stream = NULL;
321 			}
322 		}
323 	}
324 
325 	if (stream == NULL)
326 		goto out;
327 
328 	/* avoid buffering issues while reading header */
329 	if (options & STREAM_WILL_CAST)
330 		chunk_size = php_stream_set_chunk_size(stream, 1);
331 
332 	/* avoid problems with auto-detecting when reading the headers -> the headers
333 	 * are always in canonical \r\n format */
334 	eol_detect = stream->flags & (PHP_STREAM_FLAG_DETECT_EOL | PHP_STREAM_FLAG_EOL_MAC);
335 	stream->flags &= ~(PHP_STREAM_FLAG_DETECT_EOL | PHP_STREAM_FLAG_EOL_MAC);
336 
337 	php_stream_context_set(stream, context);
338 
339 	php_stream_notify_info(context, PHP_STREAM_NOTIFY_CONNECT, NULL, 0);
340 
341 	if (header_init && context && php_stream_context_get_option(context, "http", "max_redirects", &tmpzval) == SUCCESS) {
342 		SEPARATE_ZVAL(tmpzval);
343 		convert_to_long_ex(tmpzval);
344 		redirect_max = Z_LVAL_PP(tmpzval);
345 	}
346 
347 	if (context && php_stream_context_get_option(context, "http", "method", &tmpzval) == SUCCESS) {
348 		if (Z_TYPE_PP(tmpzval) == IS_STRING && Z_STRLEN_PP(tmpzval) > 0) {
349 			/* As per the RFC, automatically redirected requests MUST NOT use other methods than
350 			 * GET and HEAD unless it can be confirmed by the user */
351 			if (!redirected
352 				|| (Z_STRLEN_PP(tmpzval) == 3 && memcmp("GET", Z_STRVAL_PP(tmpzval), 3) == 0)
353 				|| (Z_STRLEN_PP(tmpzval) == 4 && memcmp("HEAD",Z_STRVAL_PP(tmpzval), 4) == 0)
354 			) {
355 				scratch_len = strlen(path) + 29 + Z_STRLEN_PP(tmpzval);
356 				scratch = emalloc(scratch_len);
357 				strlcpy(scratch, Z_STRVAL_PP(tmpzval), Z_STRLEN_PP(tmpzval) + 1);
358 				strncat(scratch, " ", 1);
359 			}
360 		}
361 	}
362 
363 	if (context && php_stream_context_get_option(context, "http", "protocol_version", &tmpzval) == SUCCESS) {
364 		SEPARATE_ZVAL(tmpzval);
365 		convert_to_double_ex(tmpzval);
366 		protocol_version_len = spprintf(&protocol_version, 0, "%.1F", Z_DVAL_PP(tmpzval));
367 	}
368 
369 	if (!scratch) {
370 		scratch_len = strlen(path) + 29 + protocol_version_len;
371 		scratch = emalloc(scratch_len);
372 		strncpy(scratch, "GET ", scratch_len);
373 	}
374 
375 	/* Should we send the entire path in the request line, default to no. */
376 	if (!request_fulluri &&
377 		context &&
378 		php_stream_context_get_option(context, "http", "request_fulluri", &tmpzval) == SUCCESS) {
379 		zval ztmp = **tmpzval;
380 
381 		zval_copy_ctor(&ztmp);
382 		convert_to_boolean(&ztmp);
383 		request_fulluri = Z_BVAL(ztmp) ? 1 : 0;
384 		zval_dtor(&ztmp);
385 	}
386 
387 	if (request_fulluri) {
388 		/* Ask for everything */
389 		strcat(scratch, path);
390 	} else {
391 		/* Send the traditional /path/to/file?query_string */
392 
393 		/* file */
394 		if (resource->path && *resource->path) {
395 			strlcat(scratch, resource->path, scratch_len);
396 		} else {
397 			strlcat(scratch, "/", scratch_len);
398 		}
399 
400 		/* query string */
401 		if (resource->query) {
402 			strlcat(scratch, "?", scratch_len);
403 			strlcat(scratch, resource->query, scratch_len);
404 		}
405 	}
406 
407 	/* protocol version we are speaking */
408 	if (protocol_version) {
409 		strlcat(scratch, " HTTP/", scratch_len);
410 		strlcat(scratch, protocol_version, scratch_len);
411 		strlcat(scratch, "\r\n", scratch_len);
412 		efree(protocol_version);
413 		protocol_version = NULL;
414 	} else {
415 		strlcat(scratch, " HTTP/1.0\r\n", scratch_len);
416 	}
417 
418 	/* send it */
419 	php_stream_write(stream, scratch, strlen(scratch));
420 
421 	if (context && php_stream_context_get_option(context, "http", "header", &tmpzval) == SUCCESS) {
422 		tmp = NULL;
423 
424 		if (Z_TYPE_PP(tmpzval) == IS_ARRAY) {
425 			HashPosition pos;
426 			zval **tmpheader = NULL;
427 			smart_str tmpstr = {0};
428 
429 			for (zend_hash_internal_pointer_reset_ex(Z_ARRVAL_PP(tmpzval), &pos);
430 				SUCCESS == zend_hash_get_current_data_ex(Z_ARRVAL_PP(tmpzval), (void *)&tmpheader, &pos);
431 				zend_hash_move_forward_ex(Z_ARRVAL_PP(tmpzval), &pos)
432 			) {
433 				if (Z_TYPE_PP(tmpheader) == IS_STRING) {
434 					smart_str_appendl(&tmpstr, Z_STRVAL_PP(tmpheader), Z_STRLEN_PP(tmpheader));
435 					smart_str_appendl(&tmpstr, "\r\n", sizeof("\r\n") - 1);
436 				}
437 			}
438 			smart_str_0(&tmpstr);
439 			/* Remove newlines and spaces from start and end. there's at least one extra \r\n at the end that needs to go. */
440 			if (tmpstr.c) {
441 				tmp = php_trim(tmpstr.c, strlen(tmpstr.c), NULL, 0, NULL, 3 TSRMLS_CC);
442 				smart_str_free(&tmpstr);
443 			}
444 		}
445 		if (Z_TYPE_PP(tmpzval) == IS_STRING && Z_STRLEN_PP(tmpzval)) {
446 			/* Remove newlines and spaces from start and end php_trim will estrndup() */
447 			tmp = php_trim(Z_STRVAL_PP(tmpzval), Z_STRLEN_PP(tmpzval), NULL, 0, NULL, 3 TSRMLS_CC);
448 		}
449 		if (tmp && strlen(tmp) > 0) {
450 			char *s;
451 
452 			user_headers = estrdup(tmp);
453 
454 			/* Make lowercase for easy comparison against 'standard' headers */
455 			php_strtolower(tmp, strlen(tmp));
456 
457 			if (!header_init) {
458 				/* strip POST headers on redirect */
459 				strip_header(user_headers, tmp, "content-length:");
460 				strip_header(user_headers, tmp, "content-type:");
461 			}
462 
463 			if ((s = strstr(tmp, "user-agent:")) &&
464 			    (s == tmp || *(s-1) == '\r' || *(s-1) == '\n' ||
465 			                 *(s-1) == '\t' || *(s-1) == ' ')) {
466 				 have_header |= HTTP_HEADER_USER_AGENT;
467 			}
468 			if ((s = strstr(tmp, "host:")) &&
469 			    (s == tmp || *(s-1) == '\r' || *(s-1) == '\n' ||
470 			                 *(s-1) == '\t' || *(s-1) == ' ')) {
471 				 have_header |= HTTP_HEADER_HOST;
472 			}
473 			if ((s = strstr(tmp, "from:")) &&
474 			    (s == tmp || *(s-1) == '\r' || *(s-1) == '\n' ||
475 			                 *(s-1) == '\t' || *(s-1) == ' ')) {
476 				 have_header |= HTTP_HEADER_FROM;
477 				}
478 			if ((s = strstr(tmp, "authorization:")) &&
479 			    (s == tmp || *(s-1) == '\r' || *(s-1) == '\n' ||
480 			                 *(s-1) == '\t' || *(s-1) == ' ')) {
481 				 have_header |= HTTP_HEADER_AUTH;
482 			}
483 			if ((s = strstr(tmp, "content-length:")) &&
484 			    (s == tmp || *(s-1) == '\r' || *(s-1) == '\n' ||
485 			                 *(s-1) == '\t' || *(s-1) == ' ')) {
486 				 have_header |= HTTP_HEADER_CONTENT_LENGTH;
487 			}
488 			if ((s = strstr(tmp, "content-type:")) &&
489 			    (s == tmp || *(s-1) == '\r' || *(s-1) == '\n' ||
490 			                 *(s-1) == '\t' || *(s-1) == ' ')) {
491 				 have_header |= HTTP_HEADER_TYPE;
492 			}
493 			/* remove Proxy-Authorization header */
494 			if (use_proxy && use_ssl && (s = strstr(tmp, "proxy-authorization:")) &&
495 			    (s == tmp || *(s-1) == '\r' || *(s-1) == '\n' ||
496 			                 *(s-1) == '\t' || *(s-1) == ' ')) {
497 				char *p = s + sizeof("proxy-authorization:") - 1;
498 
499 				while (s > tmp && (*(s-1) == ' ' || *(s-1) == '\t')) s--;
500 				while (*p != 0 && *p != '\r' && *p != '\n') p++;
501 				while (*p == '\r' || *p == '\n') p++;
502 				if (*p == 0) {
503 					if (s == tmp) {
504 						efree(user_headers);
505 						user_headers = NULL;
506 					} else {
507 						while (s > tmp && (*(s-1) == '\r' || *(s-1) == '\n')) s--;
508 						user_headers[s - tmp] = 0;
509 					}
510 				} else {
511 					memmove(user_headers + (s - tmp), user_headers + (p - tmp), strlen(p) + 1);
512 				}
513 			}
514 
515 		}
516 		if (tmp) {
517 			efree(tmp);
518 		}
519 	}
520 
521 	/* auth header if it was specified */
522 	if (((have_header & HTTP_HEADER_AUTH) == 0) && resource->user) {
523 		/* decode the strings first */
524 		php_url_decode(resource->user, strlen(resource->user));
525 
526 		/* scratch is large enough, since it was made large enough for the whole URL */
527 		strcpy(scratch, resource->user);
528 		strcat(scratch, ":");
529 
530 		/* Note: password is optional! */
531 		if (resource->pass) {
532 			php_url_decode(resource->pass, strlen(resource->pass));
533 			strcat(scratch, resource->pass);
534 		}
535 
536 		tmp = (char*)php_base64_encode((unsigned char*)scratch, strlen(scratch), NULL);
537 
538 		if (snprintf(scratch, scratch_len, "Authorization: Basic %s\r\n", tmp) > 0) {
539 			php_stream_write(stream, scratch, strlen(scratch));
540 			php_stream_notify_info(context, PHP_STREAM_NOTIFY_AUTH_REQUIRED, NULL, 0);
541 		}
542 
543 		efree(tmp);
544 		tmp = NULL;
545 	}
546 
547 	/* if the user has configured who they are, send a From: line */
548 	if (((have_header & HTTP_HEADER_FROM) == 0) && FG(from_address)) {
549 		if (snprintf(scratch, scratch_len, "From: %s\r\n", FG(from_address)) > 0)
550 			php_stream_write(stream, scratch, strlen(scratch));
551 	}
552 
553 	/* Send Host: header so name-based virtual hosts work */
554 	if ((have_header & HTTP_HEADER_HOST) == 0) {
555 		if ((use_ssl && resource->port != 443 && resource->port != 0) ||
556 			(!use_ssl && resource->port != 80 && resource->port != 0)) {
557 			if (snprintf(scratch, scratch_len, "Host: %s:%i\r\n", resource->host, resource->port) > 0)
558 				php_stream_write(stream, scratch, strlen(scratch));
559 		} else {
560 			if (snprintf(scratch, scratch_len, "Host: %s\r\n", resource->host) > 0) {
561 				php_stream_write(stream, scratch, strlen(scratch));
562 			}
563 		}
564 	}
565 
566 	if (context &&
567 	    php_stream_context_get_option(context, "http", "user_agent", &ua_zval) == SUCCESS &&
568 		Z_TYPE_PP(ua_zval) == IS_STRING) {
569 		ua_str = Z_STRVAL_PP(ua_zval);
570 	} else if (FG(user_agent)) {
571 		ua_str = FG(user_agent);
572 	}
573 
574 	if (((have_header & HTTP_HEADER_USER_AGENT) == 0) && ua_str) {
575 #define _UA_HEADER "User-Agent: %s\r\n"
576 		char *ua;
577 		size_t ua_len;
578 
579 		ua_len = sizeof(_UA_HEADER) + strlen(ua_str);
580 
581 		/* ensure the header is only sent if user_agent is not blank */
582 		if (ua_len > sizeof(_UA_HEADER)) {
583 			ua = emalloc(ua_len + 1);
584 			if ((ua_len = slprintf(ua, ua_len, _UA_HEADER, ua_str)) > 0) {
585 				ua[ua_len] = 0;
586 				php_stream_write(stream, ua, ua_len);
587 			} else {
588 				php_error_docref(NULL TSRMLS_CC, E_WARNING, "Cannot construct User-agent header");
589 			}
590 
591 			if (ua) {
592 				efree(ua);
593 			}
594 		}
595 	}
596 
597 	if (user_headers) {
598 		/* A bit weird, but some servers require that Content-Length be sent prior to Content-Type for POST
599 		 * see bug #44603 for details. Since Content-Type maybe part of user's headers we need to do this check first.
600 		 */
601 		if (
602 				header_init &&
603 				context &&
604 				!(have_header & HTTP_HEADER_CONTENT_LENGTH) &&
605 				php_stream_context_get_option(context, "http", "content", &tmpzval) == SUCCESS &&
606 				Z_TYPE_PP(tmpzval) == IS_STRING && Z_STRLEN_PP(tmpzval) > 0
607 		) {
608 			scratch_len = slprintf(scratch, scratch_len, "Content-Length: %d\r\n", Z_STRLEN_PP(tmpzval));
609 			php_stream_write(stream, scratch, scratch_len);
610 			have_header |= HTTP_HEADER_CONTENT_LENGTH;
611 		}
612 
613 		php_stream_write(stream, user_headers, strlen(user_headers));
614 		php_stream_write(stream, "\r\n", sizeof("\r\n")-1);
615 		efree(user_headers);
616 	}
617 
618 	/* Request content, such as for POST requests */
619 	if (header_init && context &&
620 		php_stream_context_get_option(context, "http", "content", &tmpzval) == SUCCESS &&
621 		Z_TYPE_PP(tmpzval) == IS_STRING && Z_STRLEN_PP(tmpzval) > 0) {
622 		if (!(have_header & HTTP_HEADER_CONTENT_LENGTH)) {
623 			scratch_len = slprintf(scratch, scratch_len, "Content-Length: %d\r\n", Z_STRLEN_PP(tmpzval));
624 			php_stream_write(stream, scratch, scratch_len);
625 		}
626 		if (!(have_header & HTTP_HEADER_TYPE)) {
627 			php_stream_write(stream, "Content-Type: application/x-www-form-urlencoded\r\n",
628 				sizeof("Content-Type: application/x-www-form-urlencoded\r\n") - 1);
629 			php_error_docref(NULL TSRMLS_CC, E_NOTICE, "Content-type not specified assuming application/x-www-form-urlencoded");
630 		}
631 		php_stream_write(stream, "\r\n", sizeof("\r\n")-1);
632 		php_stream_write(stream, Z_STRVAL_PP(tmpzval), Z_STRLEN_PP(tmpzval));
633 	} else {
634 		php_stream_write(stream, "\r\n", sizeof("\r\n")-1);
635 	}
636 
637 	location[0] = '\0';
638 
639 	if (!EG(active_symbol_table)) {
640 		zend_rebuild_symbol_table(TSRMLS_C);
641 	}
642 
643 	if (header_init) {
644 		zval *ztmp;
645 		MAKE_STD_ZVAL(ztmp);
646 		array_init(ztmp);
647 		ZEND_SET_SYMBOL(EG(active_symbol_table), "http_response_header", ztmp);
648 	}
649 
650 	{
651 		zval **rh;
652 		if(zend_hash_find(EG(active_symbol_table), "http_response_header", sizeof("http_response_header"), (void **) &rh) != SUCCESS || Z_TYPE_PP(rh) != IS_ARRAY) {
653 			php_stream_wrapper_log_error(wrapper, options TSRMLS_CC, "HTTP request failed, http_response_header overwritten");
654 			goto out;
655 		}
656 		response_header = *rh;
657 		Z_ADDREF_P(response_header);
658 	}
659 
660 	if (!php_stream_eof(stream)) {
661 		size_t tmp_line_len;
662 		/* get response header */
663 
664 		if (php_stream_get_line(stream, tmp_line, sizeof(tmp_line) - 1, &tmp_line_len) != NULL) {
665 			zval *http_response;
666 
667 			if (tmp_line_len > 9) {
668 				response_code = atoi(tmp_line + 9);
669 			} else {
670 				response_code = 0;
671 			}
672 			if (context && SUCCESS==php_stream_context_get_option(context, "http", "ignore_errors", &tmpzval)) {
673 				ignore_errors = zend_is_true(*tmpzval);
674 			}
675 			/* when we request only the header, don't fail even on error codes */
676 			if ((options & STREAM_ONLY_GET_HEADERS) || ignore_errors) {
677 				reqok = 1;
678 			}
679 			/* all status codes in the 2xx range are defined by the specification as successful;
680 			 * all status codes in the 3xx range are for redirection, and so also should never
681 			 * fail */
682 			if (response_code >= 200 && response_code < 400) {
683 				reqok = 1;
684 			} else {
685 				switch(response_code) {
686 					case 403:
687 						php_stream_notify_error(context, PHP_STREAM_NOTIFY_AUTH_RESULT,
688 								tmp_line, response_code);
689 						break;
690 					default:
691 						/* safety net in the event tmp_line == NULL */
692 						if (!tmp_line_len) {
693 							tmp_line[0] = '\0';
694 						}
695 						php_stream_notify_error(context, PHP_STREAM_NOTIFY_FAILURE,
696 								tmp_line, response_code);
697 				}
698 			}
699 			if (tmp_line[tmp_line_len - 1] == '\n') {
700 				--tmp_line_len;
701 				if (tmp_line[tmp_line_len - 1] == '\r') {
702 					--tmp_line_len;
703 				}
704 			}
705 			MAKE_STD_ZVAL(http_response);
706 			ZVAL_STRINGL(http_response, tmp_line, tmp_line_len, 1);
707 			zend_hash_next_index_insert(Z_ARRVAL_P(response_header), &http_response, sizeof(zval *), NULL);
708 		}
709 	} else {
710 		php_stream_wrapper_log_error(wrapper, options TSRMLS_CC, "HTTP request failed, unexpected end of socket!");
711 		goto out;
712 	}
713 
714 	/* read past HTTP headers */
715 
716 	http_header_line = emalloc(HTTP_HEADER_BLOCK_SIZE);
717 
718 	while (!body && !php_stream_eof(stream)) {
719 		size_t http_header_line_length;
720 		if (php_stream_get_line(stream, http_header_line, HTTP_HEADER_BLOCK_SIZE, &http_header_line_length) && *http_header_line != '\n' && *http_header_line != '\r') {
721 			char *e = http_header_line + http_header_line_length - 1;
722 			if (*e != '\n') {
723 				do { /* partial header */
724 					if (php_stream_get_line(stream, http_header_line, HTTP_HEADER_BLOCK_SIZE, &http_header_line_length) == NULL) {
725 						php_stream_wrapper_log_error(wrapper, options TSRMLS_CC, "Failed to read HTTP headers");
726 						goto out;
727 					}
728 					e = http_header_line + http_header_line_length - 1;
729 				} while (*e != '\n');
730 				continue;
731 			}
732 			while (*e == '\n' || *e == '\r') {
733 				e--;
734 			}
735 			http_header_line_length = e - http_header_line + 1;
736 			http_header_line[http_header_line_length] = '\0';
737 
738 			if (!strncasecmp(http_header_line, "Location: ", 10)) {
739 				if (context && php_stream_context_get_option(context, "http", "follow_location", &tmpzval) == SUCCESS) {
740 					SEPARATE_ZVAL(tmpzval);
741 					convert_to_long_ex(tmpzval);
742 					follow_location = Z_LVAL_PP(tmpzval);
743 				} else if (!(response_code >= 300 && response_code < 304 || 307 == response_code || 308 == response_code)) {
744 					/* we shouldn't redirect automatically
745 					if follow_location isn't set and response_code not in (300, 301, 302, 303 and 307)
746 					see http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html#sec10.3.1
747 					RFC 7238 defines 308: http://tools.ietf.org/html/rfc7238 */
748 					follow_location = 0;
749 				}
750 				strlcpy(location, http_header_line + 10, sizeof(location));
751 			} else if (!strncasecmp(http_header_line, "Content-Type: ", 14)) {
752 				php_stream_notify_info(context, PHP_STREAM_NOTIFY_MIME_TYPE_IS, http_header_line + 14, 0);
753 			} else if (!strncasecmp(http_header_line, "Content-Length: ", 16)) {
754 				file_size = atoi(http_header_line + 16);
755 				php_stream_notify_file_size(context, file_size, http_header_line, 0);
756 			} else if (!strncasecmp(http_header_line, "Transfer-Encoding: chunked", sizeof("Transfer-Encoding: chunked"))) {
757 
758 				/* create filter to decode response body */
759 				if (!(options & STREAM_ONLY_GET_HEADERS)) {
760 					long decode = 1;
761 
762 					if (context && php_stream_context_get_option(context, "http", "auto_decode", &tmpzval) == SUCCESS) {
763 						SEPARATE_ZVAL(tmpzval);
764 						convert_to_boolean(*tmpzval);
765 						decode = Z_LVAL_PP(tmpzval);
766 					}
767 					if (decode) {
768 						transfer_encoding = php_stream_filter_create("dechunk", NULL, php_stream_is_persistent(stream) TSRMLS_CC);
769 						if (transfer_encoding) {
770 							/* don't store transfer-encodeing header */
771 							continue;
772 						}
773 					}
774 				}
775 			}
776 
777 			if (http_header_line[0] == '\0') {
778 				body = 1;
779 			} else {
780 				zval *http_header;
781 
782 				MAKE_STD_ZVAL(http_header);
783 
784 				ZVAL_STRINGL(http_header, http_header_line, http_header_line_length, 1);
785 
786 				zend_hash_next_index_insert(Z_ARRVAL_P(response_header), &http_header, sizeof(zval *), NULL);
787 			}
788 		} else {
789 			break;
790 		}
791 	}
792 
793 	if (!reqok || (location[0] != '\0' && follow_location)) {
794 		if (!follow_location || (((options & STREAM_ONLY_GET_HEADERS) || ignore_errors) && redirect_max <= 1)) {
795 			goto out;
796 		}
797 
798 		if (location[0] != '\0')
799 			php_stream_notify_info(context, PHP_STREAM_NOTIFY_REDIRECTED, location, 0);
800 
801 		php_stream_close(stream);
802 		stream = NULL;
803 
804 		if (location[0] != '\0') {
805 
806 			char new_path[HTTP_HEADER_BLOCK_SIZE];
807 			char loc_path[HTTP_HEADER_BLOCK_SIZE];
808 
809 			*new_path='\0';
810 			if (strlen(location)<8 || (strncasecmp(location, "http://", sizeof("http://")-1) &&
811 							strncasecmp(location, "https://", sizeof("https://")-1) &&
812 							strncasecmp(location, "ftp://", sizeof("ftp://")-1) &&
813 							strncasecmp(location, "ftps://", sizeof("ftps://")-1)))
814 			{
815 				if (*location != '/') {
816 					if (*(location+1) != '\0' && resource->path) {
817 						char *s = strrchr(resource->path, '/');
818 						if (!s) {
819 							s = resource->path;
820 							if (!s[0]) {
821 								efree(s);
822 								s = resource->path = estrdup("/");
823 							} else {
824 								*s = '/';
825 							}
826 						}
827 						s[1] = '\0';
828 						if (resource->path && *(resource->path) == '/' && *(resource->path + 1) == '\0') {
829 							snprintf(loc_path, sizeof(loc_path) - 1, "%s%s", resource->path, location);
830 						} else {
831 							snprintf(loc_path, sizeof(loc_path) - 1, "%s/%s", resource->path, location);
832 						}
833 					} else {
834 						snprintf(loc_path, sizeof(loc_path) - 1, "/%s", location);
835 					}
836 				} else {
837 					strlcpy(loc_path, location, sizeof(loc_path));
838 				}
839 				if ((use_ssl && resource->port != 443) || (!use_ssl && resource->port != 80)) {
840 					snprintf(new_path, sizeof(new_path) - 1, "%s://%s:%d%s", resource->scheme, resource->host, resource->port, loc_path);
841 				} else {
842 					snprintf(new_path, sizeof(new_path) - 1, "%s://%s%s", resource->scheme, resource->host, loc_path);
843 				}
844 			} else {
845 				strlcpy(new_path, location, sizeof(new_path));
846 			}
847 
848 			php_url_free(resource);
849 			/* check for invalid redirection URLs */
850 			if ((resource = php_url_parse(new_path)) == NULL) {
851 				php_stream_wrapper_log_error(wrapper, options TSRMLS_CC, "Invalid redirect URL! %s", new_path);
852 				goto out;
853 			}
854 
855 #define CHECK_FOR_CNTRL_CHARS(val) { \
856 	if (val) { \
857 		unsigned char *s, *e; \
858 		int l; \
859 		l = php_url_decode(val, strlen(val)); \
860 		s = (unsigned char*)val; e = s + l; \
861 		while (s < e) { \
862 			if (iscntrl(*s)) { \
863 				php_stream_wrapper_log_error(wrapper, options TSRMLS_CC, "Invalid redirect URL! %s", new_path); \
864 				goto out; \
865 			} \
866 			s++; \
867 		} \
868 	} \
869 }
870 			/* check for control characters in login, password & path */
871 			if (strncasecmp(new_path, "http://", sizeof("http://") - 1) || strncasecmp(new_path, "https://", sizeof("https://") - 1)) {
872 				CHECK_FOR_CNTRL_CHARS(resource->user)
873 				CHECK_FOR_CNTRL_CHARS(resource->pass)
874 				CHECK_FOR_CNTRL_CHARS(resource->path)
875 			}
876 			stream = php_stream_url_wrap_http_ex(wrapper, new_path, mode, options, opened_path, context, --redirect_max, HTTP_WRAPPER_REDIRECTED STREAMS_CC TSRMLS_CC);
877 		} else {
878 			php_stream_wrapper_log_error(wrapper, options TSRMLS_CC, "HTTP request failed! %s", tmp_line);
879 		}
880 	}
881 out:
882 	if (protocol_version) {
883 		efree(protocol_version);
884 	}
885 
886 	if (http_header_line) {
887 		efree(http_header_line);
888 	}
889 
890 	if (scratch) {
891 		efree(scratch);
892 	}
893 
894 	if (resource) {
895 		php_url_free(resource);
896 	}
897 
898 	if (stream) {
899 		if (header_init) {
900 			stream->wrapperdata = response_header;
901 		} else {
902 			if(response_header) {
903 				Z_DELREF_P(response_header);
904 			}
905 		}
906 		php_stream_notify_progress_init(context, 0, file_size);
907 
908 		/* Restore original chunk size now that we're done with headers */
909 		if (options & STREAM_WILL_CAST)
910 			php_stream_set_chunk_size(stream, chunk_size);
911 
912 		/* restore the users auto-detect-line-endings setting */
913 		stream->flags |= eol_detect;
914 
915 		/* as far as streams are concerned, we are now at the start of
916 		 * the stream */
917 		stream->position = 0;
918 
919 		/* restore mode */
920 		strlcpy(stream->mode, mode, sizeof(stream->mode));
921 
922 		if (transfer_encoding) {
923 			php_stream_filter_append(&stream->readfilters, transfer_encoding);
924 		}
925 	} else {
926 		if(response_header) {
927 			Z_DELREF_P(response_header);
928 		}
929 		if (transfer_encoding) {
930 			php_stream_filter_free(transfer_encoding TSRMLS_CC);
931 		}
932 	}
933 
934 	return stream;
935 }
936 /* }}} */
937 
php_stream_url_wrap_http(php_stream_wrapper * wrapper,char * path,char * mode,int options,char ** opened_path,php_stream_context * context STREAMS_DC TSRMLS_DC)938 php_stream *php_stream_url_wrap_http(php_stream_wrapper *wrapper, char *path, char *mode, int options, char **opened_path, php_stream_context *context STREAMS_DC TSRMLS_DC) /* {{{ */
939 {
940 	return php_stream_url_wrap_http_ex(wrapper, path, mode, options, opened_path, context, PHP_URL_REDIRECT_MAX, HTTP_WRAPPER_HEADER_INIT STREAMS_CC TSRMLS_CC);
941 }
942 /* }}} */
943 
php_stream_http_stream_stat(php_stream_wrapper * wrapper,php_stream * stream,php_stream_statbuf * ssb TSRMLS_DC)944 static int php_stream_http_stream_stat(php_stream_wrapper *wrapper, php_stream *stream, php_stream_statbuf *ssb TSRMLS_DC) /* {{{ */
945 {
946 	/* one day, we could fill in the details based on Date: and Content-Length:
947 	 * headers.  For now, we return with a failure code to prevent the underlying
948 	 * file's details from being used instead. */
949 	return -1;
950 }
951 /* }}} */
952 
953 static php_stream_wrapper_ops http_stream_wops = {
954 	php_stream_url_wrap_http,
955 	NULL, /* stream_close */
956 	php_stream_http_stream_stat,
957 	NULL, /* stat_url */
958 	NULL, /* opendir */
959 	"http",
960 	NULL, /* unlink */
961 	NULL, /* rename */
962 	NULL, /* mkdir */
963 	NULL  /* rmdir */
964 };
965 
966 PHPAPI php_stream_wrapper php_stream_http_wrapper = {
967 	&http_stream_wops,
968 	NULL,
969 	1 /* is_url */
970 };
971 
972 /*
973  * Local variables:
974  * tab-width: 4
975  * c-basic-offset: 4
976  * End:
977  * vim600: sw=4 ts=4 fdm=marker
978  * vim<600: sw=4 ts=4
979  */
980