1 const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.13 2006/12/06 19:50:54 fabiankeil Exp $";
2 /*********************************************************************
4 * File : $Source: /cvsroot/ijbswa/current/urlmatch.c,v $
6 * Purpose : Declares functions to match URLs against URL
9 * Copyright : Written by and Copyright (C) 2001-2003, 2006 the SourceForge
10 * Privoxy team. http://www.privoxy.org/
12 * Based on the Internet Junkbuster originally written
13 * by and Copyright (C) 1997 Anonymous Coders and
14 * Junkbusters Corporation. http://www.junkbusters.com
16 * This program is free software; you can redistribute it
17 * and/or modify it under the terms of the GNU General
18 * Public License as published by the Free Software
19 * Foundation; either version 2 of the License, or (at
20 * your option) any later version.
22 * This program is distributed in the hope that it will
23 * be useful, but WITHOUT ANY WARRANTY; without even the
24 * implied warranty of MERCHANTABILITY or FITNESS FOR A
25 * PARTICULAR PURPOSE. See the GNU General Public
26 * License for more details.
28 * The GNU General Public License should be included with
29 * this file. If not, you can view it at
30 * http://www.gnu.org/copyleft/gpl.html
31 * or write to the Free Software Foundation, Inc., 59
32 * Temple Place - Suite 330, Boston, MA 02111-1307, USA.
35 * $Log: urlmatch.c,v $
36 * Revision 1.13 2006/12/06 19:50:54 fabiankeil
37 * parse_http_url() now handles intercepted
38 * HTTP request lines as well. Moved parts
39 * of parse_http_url()'s code into
40 * init_domain_components() so that it can
41 * be reused in chat().
43 * Revision 1.12 2006/07/18 14:48:47 david__schmidt
44 * Reorganizing the repository: swapping out what was HEAD (the old 3.1 branch)
45 * with what was really the latest development (the v_3_0_branch branch)
47 * Revision 1.10.2.7 2003/05/17 15:57:24 oes
48 * - parse_http_url now checks memory allocation failure for
49 * duplication of "*" URL and rejects "*something" URLs
51 * - Added a comment to what might look like a bug in
52 * create_url_spec (see !bug #736931)
55 * Revision 1.10.2.6 2003/05/07 12:39:48 oes
56 * Fix typo: Default port for https URLs is 443, not 143.
57 * Thanks to Scott Tregear for spotting this one.
59 * Revision 1.10.2.5 2003/02/28 13:09:29 oes
60 * Fixed a rare double free condition as per Bug #694713
62 * Revision 1.10.2.4 2003/02/28 12:57:44 oes
63 * Moved freeing of http request structure to its owner
64 * as per Dan Price's observations in Bug #694713
66 * Revision 1.10.2.3 2002/11/12 16:50:40 oes
67 * Fixed memory leak in parse_http_request() reported by Oliver Stoeneberg. Fixes bug #637073
69 * Revision 1.10.2.2 2002/09/25 14:53:15 oes
70 * Added basic support for OPTIONS and TRACE HTTP methods:
71 * parse_http_url now recognizes the "*" URI as well as
72 * the OPTIONS and TRACE method keywords.
74 * Revision 1.10.2.1 2002/06/06 19:06:44 jongfoster
75 * Adding support for proprietary Microsoft WebDAV extensions
77 * Revision 1.10 2002/05/12 21:40:37 jongfoster
78 * - Removing some unused code
80 * Revision 1.9 2002/04/04 00:36:36 gliptak
81 * always use pcre for matching
83 * Revision 1.8 2002/04/03 23:32:47 jongfoster
84 * Fixing memory leak on error
86 * Revision 1.7 2002/03/26 22:29:55 swa
87 * we have a new homepage!
89 * Revision 1.6 2002/03/24 13:25:43 swa
90 * name change related issues
92 * Revision 1.5 2002/03/13 00:27:05 jongfoster
95 * Revision 1.4 2002/03/07 03:46:17 oes
96 * Fixed compiler warnings
98 * Revision 1.3 2002/03/03 14:51:11 oes
99 * Fixed CLF logging: Added ocmd member for client's request to struct http_request
101 * Revision 1.2 2002/01/21 00:14:09 jongfoster
102 * Correcting comment style
103 * Fixing an uninitialized memory bug in create_url_spec()
105 * Revision 1.1 2002/01/17 20:53:46 jongfoster
106 * Moving all our URL and URL pattern parsing code to the same file - it
107 * was scattered around in filters.c, loaders.c and parsers.c.
109 * Providing a single, simple url_match(pattern,url) function - rather than
110 * the 3-line match routine which was repeated all over the place.
112 * Renaming free_url to free_url_spec, since it frees a struct url_spec.
114 * Providing parse_http_url() so that URLs can be parsed without faking a
115 * HTTP request line for parse_http_request() or repeating the parsing
116 * code (both of which were techniques that were actually in use).
118 * Standardizing that struct http_request is used to represent a URL, and
119 * struct url_spec is used to represent a URL pattern. (Before, URLs were
120 * represented as seperate variables and a partially-filled-in url_spec).
123 *********************************************************************/
130 #include <sys/types.h>
138 #if !defined(_WIN32) && !defined(__OS2__)
143 #include "urlmatch.h"
145 #include "miscutil.h"
148 const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
151 /*********************************************************************
153 * Function : free_http_request
155 * Description : Freez a http_request structure
158 * 1 : http = points to a http_request structure to free
162 *********************************************************************/
163 void free_http_request(struct http_request *http)
172 freez(http->hostport);
175 freez(http->host_ip_addr_str);
176 freez(http->dbuffer);
181 /*********************************************************************
183 * Function : init_domain_components
185 * Description : Splits the domain name so we can compare it
186 * against wildcards. It used to be part of
187 * parse_http_url, but was separated because the
188 * same code is required in chat in case of
189 * intercepted requests.
192 * 1 : http = pointer to the http structure to hold elements.
194 * Returns : JB_ERR_OK on success
195 * JB_ERR_MEMORY on out of memory
196 * JB_ERR_PARSE on malformed command/URL
197 * or >100 domains deep.
199 *********************************************************************/
200 jb_err init_domain_components(struct http_request *http)
202 char *vec[BUFFER_SIZE];
206 http->dbuffer = strdup(http->host);
207 if (NULL == http->dbuffer)
209 return JB_ERR_MEMORY;
212 /* map to lower case */
213 for (p = http->dbuffer; *p ; p++)
215 *p = (char)tolower((int)(unsigned char)*p);
218 /* split the domain name into components */
219 http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec), 1, 1);
221 if (http->dcount <= 0)
224 * Error: More than SZ(vec) components in domain
225 * or: no components in domain
227 log_error(LOG_LEVEL_ERROR, "More than SZ(vec) components in domain or none at all.");
231 /* save a copy of the pointers in dvec */
232 size = (size_t)http->dcount * sizeof(*http->dvec);
234 http->dvec = (char **)malloc(size);
235 if (NULL == http->dvec)
237 return JB_ERR_MEMORY;
240 memcpy(http->dvec, vec, size);
245 /*********************************************************************
247 * Function : parse_http_url
249 * Description : Parse out the host and port from the URL. Find the
250 * hostname & path, port (if ':'), and/or password (if '@')
253 * 1 : url = URL (or is it URI?) to break down
254 * 2 : http = pointer to the http structure to hold elements.
255 * Will be zeroed before use. Note that this
256 * function sets the http->gpc and http->ver
258 * 3 : csp = Current client state (buffers, headers, etc...)
260 * Returns : JB_ERR_OK on success
261 * JB_ERR_MEMORY on out of memory
262 * JB_ERR_PARSE on malformed command/URL
263 * or >100 domains deep.
265 *********************************************************************/
266 jb_err parse_http_url(const char * url,
267 struct http_request *http,
268 const struct client_state *csp)
270 int host_available = 1; /* A proxy can dream. */
273 * Zero out the results structure
275 memset(http, '\0', sizeof(*http));
279 * Save our initial URL
281 http->url = strdup(url);
282 if (http->url == NULL)
284 return JB_ERR_MEMORY;
289 * Check for * URI. If found, we're done.
291 if (*http->url == '*')
293 if ( NULL == (http->path = strdup("*"))
294 || NULL == (http->hostport = strdup("")) )
296 return JB_ERR_MEMORY;
298 if (http->url[1] != '\0')
307 * Split URL into protocol,hostport,path.
317 return JB_ERR_MEMORY;
320 /* Find the start of the URL in our scratch space */
322 if (strncmpic(url_noproto, "http://", 7) == 0)
327 else if (strncmpic(url_noproto, "https://", 8) == 0)
332 else if (*url_noproto == '/')
335 * Short request line without protocol and host.
336 * Most likely because the client's request
337 * was intercepted and redirected into Privoxy.
348 url_path = strchr(url_noproto, '/');
349 if (url_path != NULL)
354 * NOTE: The following line ignores the path for HTTPS URLS.
355 * This means that you get consistent behaviour if you type a
356 * https URL in and it's parsed by the function. (When the
357 * URL is actually retrieved, SSL hides the path part).
359 http->path = strdup(http->ssl ? "/" : url_path);
361 http->hostport = strdup(url_noproto);
366 * Repair broken HTTP requests that don't contain a path,
367 * or CONNECT requests
369 http->path = strdup("/");
370 http->hostport = strdup(url_noproto);
375 if ( (http->path == NULL)
376 || (http->hostport == NULL))
378 return JB_ERR_MEMORY;
384 /* Without host, there is nothing left to do here */
389 * Split hostport into user/password (ignored), host, port.
396 buf = strdup(http->hostport);
399 return JB_ERR_MEMORY;
402 /* check if url contains username and/or password */
403 host = strchr(buf, '@');
406 /* Contains username/password, skip it and the @ sign. */
411 /* No username or password. */
415 /* check if url contains port */
416 port = strchr(host, ':');
420 /* Terminate hostname and point to start of port string */
422 http->port = atoi(port);
426 /* No port specified. */
427 http->port = (http->ssl ? 443 : 80);
430 http->host = strdup(host);
434 if (http->host == NULL)
436 return JB_ERR_MEMORY;
441 * Split domain name so we can compare it against wildcards
443 return init_domain_components(http);
448 /*********************************************************************
450 * Function : parse_http_request
452 * Description : Parse out the host and port from the URL. Find the
453 * hostname & path, port (if ':'), and/or password (if '@')
456 * 1 : req = HTTP request line to break down
457 * 2 : http = pointer to the http structure to hold elements
458 * 3 : csp = Current client state (buffers, headers, etc...)
460 * Returns : JB_ERR_OK on success
461 * JB_ERR_MEMORY on out of memory
462 * JB_ERR_CGI_PARAMS on malformed command/URL
463 * or >100 domains deep.
465 *********************************************************************/
466 jb_err parse_http_request(const char *req,
467 struct http_request *http,
468 const struct client_state *csp)
476 memset(http, '\0', sizeof(*http));
481 return JB_ERR_MEMORY;
484 n = ssplit(buf, " \r\n", v, SZ(v), 1, 1);
487 log_error(LOG_LEVEL_ERROR, "Trouble ssplitting: %s", buf);
492 /* this could be a CONNECT request */
493 if (strcmpic(v[0], "connect") == 0)
498 /* or it could be any other basic HTTP request type */
499 else if ((0 == strcmpic(v[0], "get"))
500 || (0 == strcmpic(v[0], "head"))
501 || (0 == strcmpic(v[0], "post"))
502 || (0 == strcmpic(v[0], "put"))
503 || (0 == strcmpic(v[0], "delete"))
504 || (0 == strcmpic(v[0], "options"))
505 || (0 == strcmpic(v[0], "trace"))
507 /* or a webDAV extension (RFC2518) */
508 || (0 == strcmpic(v[0], "propfind"))
509 || (0 == strcmpic(v[0], "proppatch"))
510 || (0 == strcmpic(v[0], "move"))
511 || (0 == strcmpic(v[0], "copy"))
512 || (0 == strcmpic(v[0], "mkcol"))
513 || (0 == strcmpic(v[0], "lock"))
514 || (0 == strcmpic(v[0], "unlock"))
516 /* Or a Microsoft webDAV extension for Exchange 2000. See: */
517 /* http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html */
518 /* http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp */
519 || (0 == strcmpic(v[0], "bcopy"))
520 || (0 == strcmpic(v[0], "bmove"))
521 || (0 == strcmpic(v[0], "bdelete"))
522 || (0 == strcmpic(v[0], "bpropfind"))
523 || (0 == strcmpic(v[0], "bproppatch"))
525 /* Or another Microsoft webDAV extension for Exchange 2000. See: */
526 /* http://systems.cs.colorado.edu/grunwald/MobileComputing/Papers/draft-cohen-gena-p-base-00.txt */
527 /* http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html */
528 /* http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp */
529 || (0 == strcmpic(v[0], "subscribe"))
530 || (0 == strcmpic(v[0], "unsubscribe"))
531 || (0 == strcmpic(v[0], "notify"))
532 || (0 == strcmpic(v[0], "poll"))
540 /* Unknown HTTP method */
541 log_error(LOG_LEVEL_ERROR, "Unknown HTTP method detected: %s", v[0]);
546 err = parse_http_url(v[1], http, csp);
554 * Copy the details into the structure
556 http->ssl = is_connect;
557 http->cmd = strdup(req);
558 http->gpc = strdup(v[0]);
559 http->ver = strdup(v[2]);
561 if ( (http->cmd == NULL)
562 || (http->gpc == NULL)
563 || (http->ver == NULL) )
566 return JB_ERR_MEMORY;
575 /*********************************************************************
577 * Function : simple_domaincmp
579 * Description : Domain-wise Compare fqdn's. The comparison is
580 * both left- and right-anchored. The individual
581 * domain names are compared with simplematch().
582 * This is only used by domain_match.
585 * 1 : pv = array of patterns to compare
586 * 2 : fv = array of domain components to compare
587 * 3 : len = length of the arrays (both arrays are the
588 * same length - if they weren't, it couldn't
589 * possibly be a match).
591 * Returns : 0 => domains are equivalent, else no match.
593 *********************************************************************/
594 static int simple_domaincmp(char **pv, char **fv, int len)
598 for (n = 0; n < len; n++)
600 if (simplematch(pv[n], fv[n]))
611 /*********************************************************************
613 * Function : domain_match
615 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
616 * pattern->unachored, the comparison is un-, left-,
617 * right-anchored, or both.
618 * The individual domain names are compared with
622 * 1 : pattern = a domain that may contain a '*' as a wildcard.
623 * 2 : fqdn = domain name against which the patterns are compared.
625 * Returns : 0 => domains are equivalent, else no match.
627 *********************************************************************/
628 static int domain_match(const struct url_spec *pattern, const struct http_request *fqdn)
630 char **pv, **fv; /* vectors */
632 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
634 plen = pattern->dcount;
639 /* fqdn is too short to match this pattern */
646 if (unanchored == ANCHOR_LEFT)
651 * Convert this into a fully anchored pattern with
652 * the fqdn and pattern the same length
654 fv += (flen - plen); /* flen - plen >= 0 due to check above */
655 return simple_domaincmp(pv, fv, plen);
657 else if (unanchored == 0)
659 /* Fully anchored, check length */
664 return simple_domaincmp(pv, fv, plen);
666 else if (unanchored == ANCHOR_RIGHT)
668 /* Left anchored, ignore all extra in fqdn */
669 return simple_domaincmp(pv, fv, plen);
675 int maxn = flen - plen;
676 for (n = 0; n <= maxn; n++)
678 if (!simple_domaincmp(pv, fv, plen))
683 * Doesn't match from start of fqdn
684 * Try skipping first part of fqdn
694 /*********************************************************************
696 * Function : create_url_spec
698 * Description : Creates a "url_spec" structure from a string.
699 * When finished, free with free_url_spec().
702 * 1 : url = Target url_spec to be filled in. Will be
704 * 2 : buf = Source pattern, null terminated. NOTE: The
705 * contents of this buffer are destroyed by this
706 * function. If this function succeeds, the
707 * buffer is copied to url->spec. If this
708 * function fails, the contents of the buffer
711 * Returns : JB_ERR_OK - Success
712 * JB_ERR_MEMORY - Out of memory
713 * JB_ERR_PARSE - Cannot parse regex (Detailed message
714 * written to system log)
716 *********************************************************************/
717 jb_err create_url_spec(struct url_spec * url, const char * buf)
727 memset(url, '\0', sizeof(*url));
730 * Save a copy of the orignal specification
732 if ((url->spec = strdup(buf)) == NULL)
734 return JB_ERR_MEMORY;
737 if ((p = strchr(buf, '/')) != NULL)
739 if (NULL == (url->path = strdup(p)))
742 return JB_ERR_MEMORY;
744 url->pathlen = strlen(url->path);
755 char rebuf[BUFFER_SIZE];
757 if (NULL == (url->preg = zalloc(sizeof(*url->preg))))
761 return JB_ERR_MEMORY;
764 snprintf(rebuf, sizeof(rebuf), "^(%s)", url->path);
766 errcode = regcomp(url->preg, rebuf,
767 (REG_EXTENDED|REG_NOSUB|REG_ICASE));
770 size_t errlen = regerror(errcode,
771 url->preg, rebuf, sizeof(rebuf));
773 if (errlen > (sizeof(rebuf) - (size_t)1))
775 errlen = sizeof(rebuf) - (size_t)1;
777 rebuf[errlen] = '\0';
779 log_error(LOG_LEVEL_ERROR, "error compiling %s: %s",
790 if ((p = strchr(buf, ':')) == NULL)
808 if (buf[strlen(buf) - 1] == '.')
810 url->unanchored |= ANCHOR_RIGHT;
814 url->unanchored |= ANCHOR_LEFT;
818 * Split domain into components
820 url->dbuffer = strdup(buf);
821 if (NULL == url->dbuffer)
827 return JB_ERR_MEMORY;
833 for (p = url->dbuffer; *p ; p++)
835 *p = (char)tolower((int)(unsigned char)*p);
839 * Split the domain name into components
841 url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1);
851 return JB_ERR_MEMORY;
853 else if (url->dcount != 0)
857 * Save a copy of the pointers in dvec
859 size = (size_t)url->dcount * sizeof(*url->dvec);
861 url->dvec = (char **)malloc(size);
862 if (NULL == url->dvec)
870 return JB_ERR_MEMORY;
873 memcpy(url->dvec, v, size);
876 * else dcount == 0 in which case we needn't do anything,
877 * since dvec will never be accessed and the pattern will
887 /*********************************************************************
889 * Function : free_url_spec
891 * Description : Called from the "unloaders". Freez the url
892 * structure elements.
895 * 1 : url = pointer to a url_spec structure.
899 *********************************************************************/
900 void free_url_spec(struct url_spec *url)
902 if (url == NULL) return;
916 /*********************************************************************
918 * Function : url_match
920 * Description : Compare a URL against a URL pattern.
923 * 1 : pattern = a URL pattern
924 * 2 : url = URL to match
926 * Returns : 0 iff the URL matches the pattern, else nonzero.
928 *********************************************************************/
929 int url_match(const struct url_spec *pattern,
930 const struct http_request *url)
932 return ((pattern->port == 0) || (pattern->port == url->port))
933 && ((pattern->dbuffer == NULL) || (domain_match(pattern, url) == 0))
934 && ((pattern->path == NULL) ||
935 (regexec(pattern->preg, url->path, 0, NULL, 0) == 0)