1 const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.20 2007/09/02 15:31:20 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, 2006-2007 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 1.20 2007/09/02 15:31:20 fabiankeil
37 * Move match_portlist() from filter.c to urlmatch.c.
38 * It's used for url matching, not for filtering.
40 * Revision 1.19 2007/09/02 13:42:11 fabiankeil
41 * - Allow port lists in url patterns.
42 * - Ditch unused url_spec member pathlen.
44 * Revision 1.18 2007/07/30 16:42:21 fabiankeil
45 * Move the method check into unknown_method()
46 * and loop through the known methods instead
47 * of using a screen-long OR chain.
49 * Revision 1.17 2007/04/15 16:39:21 fabiankeil
50 * Introduce tags as alternative way to specify which
51 * actions apply to a request. At the moment tags can be
52 * created based on client and server headers.
54 * Revision 1.16 2007/02/13 13:59:24 fabiankeil
55 * Remove redundant log message.
57 * Revision 1.15 2007/01/28 16:11:23 fabiankeil
58 * Accept WebDAV methods for subversion
59 * in parse_http_request(). Closes FR 1581425.
61 * Revision 1.14 2007/01/06 14:23:56 fabiankeil
62 * Fix gcc43 warnings. Mark *csp as immutable
63 * for parse_http_url() and url_match().
64 * Replace a sprintf call with snprintf.
66 * Revision 1.13 2006/12/06 19:50:54 fabiankeil
67 * parse_http_url() now handles intercepted
68 * HTTP request lines as well. Moved parts
69 * of parse_http_url()'s code into
70 * init_domain_components() so that it can
71 * be reused in chat().
73 * Revision 1.12 2006/07/18 14:48:47 david__schmidt
74 * Reorganizing the repository: swapping out what was HEAD (the old 3.1 branch)
75 * with what was really the latest development (the v_3_0_branch branch)
77 * Revision 1.10.2.7 2003/05/17 15:57:24 oes
78 * - parse_http_url now checks memory allocation failure for
79 * duplication of "*" URL and rejects "*something" URLs
81 * - Added a comment to what might look like a bug in
82 * create_url_spec (see !bug #736931)
85 * Revision 1.10.2.6 2003/05/07 12:39:48 oes
86 * Fix typo: Default port for https URLs is 443, not 143.
87 * Thanks to Scott Tregear for spotting this one.
89 * Revision 1.10.2.5 2003/02/28 13:09:29 oes
90 * Fixed a rare double free condition as per Bug #694713
92 * Revision 1.10.2.4 2003/02/28 12:57:44 oes
93 * Moved freeing of http request structure to its owner
94 * as per Dan Price's observations in Bug #694713
96 * Revision 1.10.2.3 2002/11/12 16:50:40 oes
97 * Fixed memory leak in parse_http_request() reported by Oliver Stoeneberg. Fixes bug #637073
99 * Revision 1.10.2.2 2002/09/25 14:53:15 oes
100 * Added basic support for OPTIONS and TRACE HTTP methods:
101 * parse_http_url now recognizes the "*" URI as well as
102 * the OPTIONS and TRACE method keywords.
104 * Revision 1.10.2.1 2002/06/06 19:06:44 jongfoster
105 * Adding support for proprietary Microsoft WebDAV extensions
107 * Revision 1.10 2002/05/12 21:40:37 jongfoster
108 * - Removing some unused code
110 * Revision 1.9 2002/04/04 00:36:36 gliptak
111 * always use pcre for matching
113 * Revision 1.8 2002/04/03 23:32:47 jongfoster
114 * Fixing memory leak on error
116 * Revision 1.7 2002/03/26 22:29:55 swa
117 * we have a new homepage!
119 * Revision 1.6 2002/03/24 13:25:43 swa
120 * name change related issues
122 * Revision 1.5 2002/03/13 00:27:05 jongfoster
125 * Revision 1.4 2002/03/07 03:46:17 oes
126 * Fixed compiler warnings
128 * Revision 1.3 2002/03/03 14:51:11 oes
129 * Fixed CLF logging: Added ocmd member for client's request to struct http_request
131 * Revision 1.2 2002/01/21 00:14:09 jongfoster
132 * Correcting comment style
133 * Fixing an uninitialized memory bug in create_url_spec()
135 * Revision 1.1 2002/01/17 20:53:46 jongfoster
136 * Moving all our URL and URL pattern parsing code to the same file - it
137 * was scattered around in filters.c, loaders.c and parsers.c.
139 * Providing a single, simple url_match(pattern,url) function - rather than
140 * the 3-line match routine which was repeated all over the place.
142 * Renaming free_url to free_url_spec, since it frees a struct url_spec.
144 * Providing parse_http_url() so that URLs can be parsed without faking a
145 * HTTP request line for parse_http_request() or repeating the parsing
146 * code (both of which were techniques that were actually in use).
148 * Standardizing that struct http_request is used to represent a URL, and
149 * struct url_spec is used to represent a URL pattern. (Before, URLs were
150 * represented as seperate variables and a partially-filled-in url_spec).
153 *********************************************************************/
160 #include <sys/types.h>
168 #if !defined(_WIN32) && !defined(__OS2__)
173 #include "urlmatch.h"
175 #include "miscutil.h"
178 const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
181 /*********************************************************************
183 * Function : free_http_request
185 * Description : Freez a http_request structure
188 * 1 : http = points to a http_request structure to free
192 *********************************************************************/
193 void free_http_request(struct http_request *http)
202 freez(http->hostport);
205 freez(http->host_ip_addr_str);
206 freez(http->dbuffer);
211 /*********************************************************************
213 * Function : init_domain_components
215 * Description : Splits the domain name so we can compare it
216 * against wildcards. It used to be part of
217 * parse_http_url, but was separated because the
218 * same code is required in chat in case of
219 * intercepted requests.
222 * 1 : http = pointer to the http structure to hold elements.
224 * Returns : JB_ERR_OK on success
225 * JB_ERR_MEMORY on out of memory
226 * JB_ERR_PARSE on malformed command/URL
227 * or >100 domains deep.
229 *********************************************************************/
230 jb_err init_domain_components(struct http_request *http)
232 char *vec[BUFFER_SIZE];
236 http->dbuffer = strdup(http->host);
237 if (NULL == http->dbuffer)
239 return JB_ERR_MEMORY;
242 /* map to lower case */
243 for (p = http->dbuffer; *p ; p++)
245 *p = (char)tolower((int)(unsigned char)*p);
248 /* split the domain name into components */
249 http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec), 1, 1);
251 if (http->dcount <= 0)
254 * Error: More than SZ(vec) components in domain
255 * or: no components in domain
257 log_error(LOG_LEVEL_ERROR, "More than SZ(vec) components in domain or none at all.");
261 /* save a copy of the pointers in dvec */
262 size = (size_t)http->dcount * sizeof(*http->dvec);
264 http->dvec = (char **)malloc(size);
265 if (NULL == http->dvec)
267 return JB_ERR_MEMORY;
270 memcpy(http->dvec, vec, size);
275 /*********************************************************************
277 * Function : parse_http_url
279 * Description : Parse out the host and port from the URL. Find the
280 * hostname & path, port (if ':'), and/or password (if '@')
283 * 1 : url = URL (or is it URI?) to break down
284 * 2 : http = pointer to the http structure to hold elements.
285 * Will be zeroed before use. Note that this
286 * function sets the http->gpc and http->ver
288 * 3 : csp = Current client state (buffers, headers, etc...)
290 * Returns : JB_ERR_OK on success
291 * JB_ERR_MEMORY on out of memory
292 * JB_ERR_PARSE on malformed command/URL
293 * or >100 domains deep.
295 *********************************************************************/
296 jb_err parse_http_url(const char * url,
297 struct http_request *http,
298 const struct client_state *csp)
300 int host_available = 1; /* A proxy can dream. */
303 * Zero out the results structure
305 memset(http, '\0', sizeof(*http));
309 * Save our initial URL
311 http->url = strdup(url);
312 if (http->url == NULL)
314 return JB_ERR_MEMORY;
319 * Check for * URI. If found, we're done.
321 if (*http->url == '*')
323 if ( NULL == (http->path = strdup("*"))
324 || NULL == (http->hostport = strdup("")) )
326 return JB_ERR_MEMORY;
328 if (http->url[1] != '\0')
337 * Split URL into protocol,hostport,path.
347 return JB_ERR_MEMORY;
350 /* Find the start of the URL in our scratch space */
352 if (strncmpic(url_noproto, "http://", 7) == 0)
357 else if (strncmpic(url_noproto, "https://", 8) == 0)
362 else if (*url_noproto == '/')
365 * Short request line without protocol and host.
366 * Most likely because the client's request
367 * was intercepted and redirected into Privoxy.
378 url_path = strchr(url_noproto, '/');
379 if (url_path != NULL)
384 * NOTE: The following line ignores the path for HTTPS URLS.
385 * This means that you get consistent behaviour if you type a
386 * https URL in and it's parsed by the function. (When the
387 * URL is actually retrieved, SSL hides the path part).
389 http->path = strdup(http->ssl ? "/" : url_path);
391 http->hostport = strdup(url_noproto);
396 * Repair broken HTTP requests that don't contain a path,
397 * or CONNECT requests
399 http->path = strdup("/");
400 http->hostport = strdup(url_noproto);
405 if ( (http->path == NULL)
406 || (http->hostport == NULL))
408 return JB_ERR_MEMORY;
414 /* Without host, there is nothing left to do here */
419 * Split hostport into user/password (ignored), host, port.
426 buf = strdup(http->hostport);
429 return JB_ERR_MEMORY;
432 /* check if url contains username and/or password */
433 host = strchr(buf, '@');
436 /* Contains username/password, skip it and the @ sign. */
441 /* No username or password. */
445 /* check if url contains port */
446 port = strchr(host, ':');
450 /* Terminate hostname and point to start of port string */
452 http->port = atoi(port);
456 /* No port specified. */
457 http->port = (http->ssl ? 443 : 80);
460 http->host = strdup(host);
464 if (http->host == NULL)
466 return JB_ERR_MEMORY;
471 * Split domain name so we can compare it against wildcards
473 return init_domain_components(http);
478 /*********************************************************************
480 * Function : unknown_method
482 * Description : Checks whether a method is unknown.
485 * 1 : method = points to a http method
487 * Returns : TRUE if it's unknown, FALSE otherwise.
489 *********************************************************************/
490 static int unknown_method(const char *method)
492 static const char *known_http_methods[] = {
493 /* Basic HTTP request type */
494 "GET", "HEAD", "POST", "PUT", "DELETE", "OPTIONS", "TRACE", "CONNECT",
495 /* webDAV extensions (RFC2518) */
496 "PROPFIND", "PROPPATCH", "MOVE", "COPY", "MKCOL", "LOCK", "UNLOCK",
498 * Microsoft webDAV extension for Exchange 2000. See:
499 * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
500 * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
502 "BCOPY", "BMOVE", "BDELETE", "BPROPFIND", "BPROPPATCH",
504 * Another Microsoft webDAV extension for Exchange 2000. See:
505 * http://systems.cs.colorado.edu/grunwald/MobileComputing/Papers/draft-cohen-gena-p-base-00.txt
506 * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
507 * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
509 "SUBSCRIBE", "UNSUBSCRIBE", "NOTIFY", "POLL",
511 * Yet another WebDAV extension, this time for
512 * Web Distributed Authoring and Versioning (RFC3253)
514 "VERSION-CONTROL", "REPORT", "CHECKOUT", "CHECKIN", "UNCHECKOUT",
515 "MKWORKSPACE", "UPDATE", "LABEL", "MERGE", "BASELINE-CONTROL", "MKACTIVITY",
520 for (i = 0; NULL != known_http_methods[i]; i++)
522 if (0 == strcmpic(method, known_http_methods[i]))
533 /*********************************************************************
535 * Function : parse_http_request
537 * Description : Parse out the host and port from the URL. Find the
538 * hostname & path, port (if ':'), and/or password (if '@')
541 * 1 : req = HTTP request line to break down
542 * 2 : http = pointer to the http structure to hold elements
543 * 3 : csp = Current client state (buffers, headers, etc...)
545 * Returns : JB_ERR_OK on success
546 * JB_ERR_MEMORY on out of memory
547 * JB_ERR_CGI_PARAMS on malformed command/URL
548 * or >100 domains deep.
550 *********************************************************************/
551 jb_err parse_http_request(const char *req,
552 struct http_request *http,
553 const struct client_state *csp)
556 char *v[10]; /* XXX: Why 10? We should only need three. */
561 memset(http, '\0', sizeof(*http));
566 return JB_ERR_MEMORY;
569 n = ssplit(buf, " \r\n", v, SZ(v), 1, 1);
577 * Fail in case of unknown methods
578 * which we might not handle correctly.
580 * XXX: There should be a config option
581 * to forward requests with unknown methods
582 * anyway. Most of them don't need special
585 if (unknown_method(v[0]))
587 log_error(LOG_LEVEL_ERROR, "Unknown HTTP method detected: %s", v[0]);
592 if (strcmpic(v[0], "CONNECT") == 0)
597 err = parse_http_url(v[1], http, csp);
605 * Copy the details into the structure
607 http->ssl = is_connect;
608 http->cmd = strdup(req);
609 http->gpc = strdup(v[0]);
610 http->ver = strdup(v[2]);
612 if ( (http->cmd == NULL)
613 || (http->gpc == NULL)
614 || (http->ver == NULL) )
617 return JB_ERR_MEMORY;
626 /*********************************************************************
628 * Function : simple_domaincmp
630 * Description : Domain-wise Compare fqdn's. The comparison is
631 * both left- and right-anchored. The individual
632 * domain names are compared with simplematch().
633 * This is only used by domain_match.
636 * 1 : pv = array of patterns to compare
637 * 2 : fv = array of domain components to compare
638 * 3 : len = length of the arrays (both arrays are the
639 * same length - if they weren't, it couldn't
640 * possibly be a match).
642 * Returns : 0 => domains are equivalent, else no match.
644 *********************************************************************/
645 static int simple_domaincmp(char **pv, char **fv, int len)
649 for (n = 0; n < len; n++)
651 if (simplematch(pv[n], fv[n]))
662 /*********************************************************************
664 * Function : domain_match
666 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
667 * pattern->unachored, the comparison is un-, left-,
668 * right-anchored, or both.
669 * The individual domain names are compared with
673 * 1 : pattern = a domain that may contain a '*' as a wildcard.
674 * 2 : fqdn = domain name against which the patterns are compared.
676 * Returns : 0 => domains are equivalent, else no match.
678 *********************************************************************/
679 static int domain_match(const struct url_spec *pattern, const struct http_request *fqdn)
681 char **pv, **fv; /* vectors */
683 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
685 plen = pattern->dcount;
690 /* fqdn is too short to match this pattern */
697 if (unanchored == ANCHOR_LEFT)
702 * Convert this into a fully anchored pattern with
703 * the fqdn and pattern the same length
705 fv += (flen - plen); /* flen - plen >= 0 due to check above */
706 return simple_domaincmp(pv, fv, plen);
708 else if (unanchored == 0)
710 /* Fully anchored, check length */
715 return simple_domaincmp(pv, fv, plen);
717 else if (unanchored == ANCHOR_RIGHT)
719 /* Left anchored, ignore all extra in fqdn */
720 return simple_domaincmp(pv, fv, plen);
726 int maxn = flen - plen;
727 for (n = 0; n <= maxn; n++)
729 if (!simple_domaincmp(pv, fv, plen))
734 * Doesn't match from start of fqdn
735 * Try skipping first part of fqdn
745 /*********************************************************************
747 * Function : create_url_spec
749 * Description : Creates a "url_spec" structure from a string.
750 * When finished, free with free_url_spec().
753 * 1 : url = Target url_spec to be filled in. Will be
755 * 2 : buf = Source pattern, null terminated. NOTE: The
756 * contents of this buffer are destroyed by this
757 * function. If this function succeeds, the
758 * buffer is copied to url->spec. If this
759 * function fails, the contents of the buffer
762 * Returns : JB_ERR_OK - Success
763 * JB_ERR_MEMORY - Out of memory
764 * JB_ERR_PARSE - Cannot parse regex (Detailed message
765 * written to system log)
767 *********************************************************************/
768 jb_err create_url_spec(struct url_spec * url, const char * buf)
773 char rebuf[BUFFER_SIZE];
781 memset(url, '\0', sizeof(*url));
784 * Save a copy of the orignal specification
786 if ((url->spec = strdup(buf)) == NULL)
788 return JB_ERR_MEMORY;
791 /* Is it tag pattern? */
792 if (0 == strncmpic("TAG:", url->spec, 4))
794 if (NULL == (url->tag_regex = zalloc(sizeof(*url->tag_regex))))
797 return JB_ERR_MEMORY;
800 /* buf + 4 to skip "TAG:" */
801 errcode = regcomp(url->tag_regex, buf + 4, (REG_EXTENDED|REG_NOSUB|REG_ICASE));
804 errlen = regerror(errcode, url->preg, rebuf, sizeof(rebuf));
805 if (errlen > (sizeof(rebuf) - 1))
807 errlen = sizeof(rebuf) - 1;
809 rebuf[errlen] = '\0';
811 log_error(LOG_LEVEL_ERROR, "error compiling %s: %s", url->spec, rebuf);
814 regfree(url->tag_regex);
815 freez(url->tag_regex);
822 /* Only reached for URL patterns */
823 p = strchr(buf, '/');
826 url->path = strdup(p);
827 if (NULL == url->path)
830 return JB_ERR_MEMORY;
840 if (NULL == (url->preg = zalloc(sizeof(*url->preg))))
844 return JB_ERR_MEMORY;
847 snprintf(rebuf, sizeof(rebuf), "^(%s)", url->path);
849 errcode = regcomp(url->preg, rebuf,
850 (REG_EXTENDED|REG_NOSUB|REG_ICASE));
853 errlen = regerror(errcode, url->preg, rebuf, sizeof(rebuf));
855 if (errlen > (sizeof(rebuf) - (size_t)1))
857 errlen = sizeof(rebuf) - (size_t)1;
859 rebuf[errlen] = '\0';
861 log_error(LOG_LEVEL_ERROR, "error compiling %s: %s",
873 p = strchr(buf, ':');
877 url->port_list = strdup(p);
878 if (NULL == url->port_list)
880 return JB_ERR_MEMORY;
885 url->port_list = NULL;
896 if (buf[strlen(buf) - 1] == '.')
898 url->unanchored |= ANCHOR_RIGHT;
902 url->unanchored |= ANCHOR_LEFT;
906 * Split domain into components
908 url->dbuffer = strdup(buf);
909 if (NULL == url->dbuffer)
915 return JB_ERR_MEMORY;
921 for (p = url->dbuffer; *p ; p++)
923 *p = (char)tolower((int)(unsigned char)*p);
927 * Split the domain name into components
929 url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1);
939 return JB_ERR_MEMORY;
941 else if (url->dcount != 0)
945 * Save a copy of the pointers in dvec
947 size = (size_t)url->dcount * sizeof(*url->dvec);
949 url->dvec = (char **)malloc(size);
950 if (NULL == url->dvec)
958 return JB_ERR_MEMORY;
961 memcpy(url->dvec, v, size);
964 * else dcount == 0 in which case we needn't do anything,
965 * since dvec will never be accessed and the pattern will
975 /*********************************************************************
977 * Function : free_url_spec
979 * Description : Called from the "unloaders". Freez the url
980 * structure elements.
983 * 1 : url = pointer to a url_spec structure.
987 *********************************************************************/
988 void free_url_spec(struct url_spec *url)
990 if (url == NULL) return;
996 freez(url->port_list);
1004 regfree(url->tag_regex);
1005 freez(url->tag_regex);
1010 /*********************************************************************
1012 * Function : url_match
1014 * Description : Compare a URL against a URL pattern.
1017 * 1 : pattern = a URL pattern
1018 * 2 : url = URL to match
1020 * Returns : Nonzero if the URL matches the pattern, else 0.
1022 *********************************************************************/
1023 int url_match(const struct url_spec *pattern,
1024 const struct http_request *url)
1026 /* XXX: these should probably be functions. */
1027 #define PORT_MATCHES ((NULL == pattern->port_list) || match_portlist(pattern->port_list, url->port))
1028 #define DOMAIN_MATCHES ((NULL == pattern->dbuffer) || (0 == domain_match(pattern, url)))
1029 #define PATH_MATCHES ((NULL == pattern->path) || (0 == regexec(pattern->preg, url->path, 0, NULL, 0)))
1031 if (pattern->tag_regex != NULL)
1033 /* It's a tag pattern and shouldn't be matched against URLs */
1037 return (PORT_MATCHES && DOMAIN_MATCHES && PATH_MATCHES);
1042 /*********************************************************************
1044 * Function : match_portlist
1046 * Description : Check if a given number is covered by a comma
1047 * separated list of numbers and ranges (a,b-c,d,..)
1050 * 1 : portlist = String with list
1051 * 2 : port = port to check
1053 * Returns : 0 => no match
1056 *********************************************************************/
1057 int match_portlist(const char *portlist, int port)
1059 char *min, *max, *next, *portlist_copy;
1061 min = next = portlist_copy = strdup(portlist);
1064 * Zero-terminate first item and remember offset for next
1066 if (NULL != (next = strchr(portlist_copy, (int) ',')))
1072 * Loop through all items, checking for match
1076 if (NULL == (max = strchr(min, (int) '-')))
1079 * No dash, check for equality
1081 if (port == atoi(min))
1083 free(portlist_copy);
1090 * This is a range, so check if between min and max,
1091 * or, if max was omitted, between min and 65K
1094 if(port >= atoi(min) && port <= (atoi(max) ? atoi(max) : 65535))
1096 free(portlist_copy);
1108 * Zero-terminate next item and remember offset for n+1
1110 if ((NULL != next) && (NULL != (next = strchr(next, (int) ','))))
1116 free(portlist_copy);