1 const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.32 2008/04/12 12:38:06 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, 2006-2008 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 1.32 2008/04/12 12:38:06 fabiankeil
37 * Factor out duplicated code to compile host, path and tag patterns.
39 * Revision 1.31 2008/04/10 14:41:04 fabiankeil
40 * Ditch url_spec's path member now that it's no longer used.
42 * Revision 1.30 2008/04/10 04:24:24 fabiankeil
43 * Stop duplicating the plain text representation of the path regex
44 * (and keeping the copy around). Once the regex is compiled it's no
47 * Revision 1.29 2008/04/10 04:17:56 fabiankeil
48 * In url_match(), check the right member for NULL when determining
49 * whether there's a path regex to execute. Looking for a plain-text
50 * representation works as well, but it looks "interesting" and that
51 * member will be removed soonish anyway.
53 * Revision 1.28 2008/04/08 16:07:39 fabiankeil
54 * Make it harder to mistake url_match()'s
55 * second parameter for an url_spec.
57 * Revision 1.27 2008/04/08 15:44:33 fabiankeil
58 * Save a bit of memory (and a few cpu cycles) by not bothering to
59 * compile slash-only path regexes that don't affect the result.
61 * Revision 1.26 2008/04/07 16:57:18 fabiankeil
62 * - Use free_url_spec() more consistently.
63 * - Let it reset url->dcount just in case.
65 * Revision 1.25 2008/04/06 15:18:38 fabiankeil
66 * Oh well, rename the --enable-pcre-host-patterns option to
67 * --enable-extended-host-patterns as it's not really PCRE syntax.
69 * Revision 1.24 2008/04/06 14:54:26 fabiankeil
70 * Use PCRE syntax in host patterns when configured
71 * with --enable-pcre-host-patterns.
73 * Revision 1.23 2008/04/05 12:19:20 fabiankeil
74 * Factor compile_host_pattern() out of create_url_spec().
76 * Revision 1.22 2008/03/30 15:02:32 fabiankeil
77 * SZitify unknown_method().
79 * Revision 1.21 2007/12/24 16:34:23 fabiankeil
80 * Band-aid (and micro-optimization) that makes it less likely to run out of
81 * stack space with overly-complex path patterns. Probably masks the problem
82 * reported by Lee in #1856679. Hohoho.
84 * Revision 1.20 2007/09/02 15:31:20 fabiankeil
85 * Move match_portlist() from filter.c to urlmatch.c.
86 * It's used for url matching, not for filtering.
88 * Revision 1.19 2007/09/02 13:42:11 fabiankeil
89 * - Allow port lists in url patterns.
90 * - Ditch unused url_spec member pathlen.
92 * Revision 1.18 2007/07/30 16:42:21 fabiankeil
93 * Move the method check into unknown_method()
94 * and loop through the known methods instead
95 * of using a screen-long OR chain.
97 * Revision 1.17 2007/04/15 16:39:21 fabiankeil
98 * Introduce tags as alternative way to specify which
99 * actions apply to a request. At the moment tags can be
100 * created based on client and server headers.
102 * Revision 1.16 2007/02/13 13:59:24 fabiankeil
103 * Remove redundant log message.
105 * Revision 1.15 2007/01/28 16:11:23 fabiankeil
106 * Accept WebDAV methods for subversion
107 * in parse_http_request(). Closes FR 1581425.
109 * Revision 1.14 2007/01/06 14:23:56 fabiankeil
110 * Fix gcc43 warnings. Mark *csp as immutable
111 * for parse_http_url() and url_match().
112 * Replace a sprintf call with snprintf.
114 * Revision 1.13 2006/12/06 19:50:54 fabiankeil
115 * parse_http_url() now handles intercepted
116 * HTTP request lines as well. Moved parts
117 * of parse_http_url()'s code into
118 * init_domain_components() so that it can
119 * be reused in chat().
121 * Revision 1.12 2006/07/18 14:48:47 david__schmidt
122 * Reorganizing the repository: swapping out what was HEAD (the old 3.1 branch)
123 * with what was really the latest development (the v_3_0_branch branch)
125 * Revision 1.10.2.7 2003/05/17 15:57:24 oes
126 * - parse_http_url now checks memory allocation failure for
127 * duplication of "*" URL and rejects "*something" URLs
129 * - Added a comment to what might look like a bug in
130 * create_url_spec (see !bug #736931)
131 * - Comment cosmetics
133 * Revision 1.10.2.6 2003/05/07 12:39:48 oes
134 * Fix typo: Default port for https URLs is 443, not 143.
135 * Thanks to Scott Tregear for spotting this one.
137 * Revision 1.10.2.5 2003/02/28 13:09:29 oes
138 * Fixed a rare double free condition as per Bug #694713
140 * Revision 1.10.2.4 2003/02/28 12:57:44 oes
141 * Moved freeing of http request structure to its owner
142 * as per Dan Price's observations in Bug #694713
144 * Revision 1.10.2.3 2002/11/12 16:50:40 oes
145 * Fixed memory leak in parse_http_request() reported by Oliver Stoeneberg. Fixes bug #637073
147 * Revision 1.10.2.2 2002/09/25 14:53:15 oes
148 * Added basic support for OPTIONS and TRACE HTTP methods:
149 * parse_http_url now recognizes the "*" URI as well as
150 * the OPTIONS and TRACE method keywords.
152 * Revision 1.10.2.1 2002/06/06 19:06:44 jongfoster
153 * Adding support for proprietary Microsoft WebDAV extensions
155 * Revision 1.10 2002/05/12 21:40:37 jongfoster
156 * - Removing some unused code
158 * Revision 1.9 2002/04/04 00:36:36 gliptak
159 * always use pcre for matching
161 * Revision 1.8 2002/04/03 23:32:47 jongfoster
162 * Fixing memory leak on error
164 * Revision 1.7 2002/03/26 22:29:55 swa
165 * we have a new homepage!
167 * Revision 1.6 2002/03/24 13:25:43 swa
168 * name change related issues
170 * Revision 1.5 2002/03/13 00:27:05 jongfoster
173 * Revision 1.4 2002/03/07 03:46:17 oes
174 * Fixed compiler warnings
176 * Revision 1.3 2002/03/03 14:51:11 oes
177 * Fixed CLF logging: Added ocmd member for client's request to struct http_request
179 * Revision 1.2 2002/01/21 00:14:09 jongfoster
180 * Correcting comment style
181 * Fixing an uninitialized memory bug in create_url_spec()
183 * Revision 1.1 2002/01/17 20:53:46 jongfoster
184 * Moving all our URL and URL pattern parsing code to the same file - it
185 * was scattered around in filters.c, loaders.c and parsers.c.
187 * Providing a single, simple url_match(pattern,url) function - rather than
188 * the 3-line match routine which was repeated all over the place.
190 * Renaming free_url to free_url_spec, since it frees a struct url_spec.
192 * Providing parse_http_url() so that URLs can be parsed without faking a
193 * HTTP request line for parse_http_request() or repeating the parsing
194 * code (both of which were techniques that were actually in use).
196 * Standardizing that struct http_request is used to represent a URL, and
197 * struct url_spec is used to represent a URL pattern. (Before, URLs were
198 * represented as seperate variables and a partially-filled-in url_spec).
201 *********************************************************************/
208 #include <sys/types.h>
216 #if !defined(_WIN32) && !defined(__OS2__)
221 #include "urlmatch.h"
223 #include "miscutil.h"
226 const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
228 enum regex_anchoring {NO_ANCHORING, LEFT_ANCHORED, RIGHT_ANCHORED};
230 /*********************************************************************
232 * Function : free_http_request
234 * Description : Freez a http_request structure
237 * 1 : http = points to a http_request structure to free
241 *********************************************************************/
242 void free_http_request(struct http_request *http)
251 freez(http->hostport);
254 freez(http->host_ip_addr_str);
255 freez(http->dbuffer);
261 /*********************************************************************
263 * Function : init_domain_components
265 * Description : Splits the domain name so we can compare it
266 * against wildcards. It used to be part of
267 * parse_http_url, but was separated because the
268 * same code is required in chat in case of
269 * intercepted requests.
272 * 1 : http = pointer to the http structure to hold elements.
274 * Returns : JB_ERR_OK on success
275 * JB_ERR_MEMORY on out of memory
276 * JB_ERR_PARSE on malformed command/URL
277 * or >100 domains deep.
279 *********************************************************************/
280 jb_err init_domain_components(struct http_request *http)
282 char *vec[BUFFER_SIZE];
286 http->dbuffer = strdup(http->host);
287 if (NULL == http->dbuffer)
289 return JB_ERR_MEMORY;
292 /* map to lower case */
293 for (p = http->dbuffer; *p ; p++)
295 *p = (char)tolower((int)(unsigned char)*p);
298 /* split the domain name into components */
299 http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec), 1, 1);
301 if (http->dcount <= 0)
304 * Error: More than SZ(vec) components in domain
305 * or: no components in domain
307 log_error(LOG_LEVEL_ERROR, "More than SZ(vec) components in domain or none at all.");
311 /* save a copy of the pointers in dvec */
312 size = (size_t)http->dcount * sizeof(*http->dvec);
314 http->dvec = (char **)malloc(size);
315 if (NULL == http->dvec)
317 return JB_ERR_MEMORY;
320 memcpy(http->dvec, vec, size);
326 /*********************************************************************
328 * Function : parse_http_url
330 * Description : Parse out the host and port from the URL. Find the
331 * hostname & path, port (if ':'), and/or password (if '@')
334 * 1 : url = URL (or is it URI?) to break down
335 * 2 : http = pointer to the http structure to hold elements.
336 * Will be zeroed before use. Note that this
337 * function sets the http->gpc and http->ver
339 * 3 : csp = Current client state (buffers, headers, etc...)
341 * Returns : JB_ERR_OK on success
342 * JB_ERR_MEMORY on out of memory
343 * JB_ERR_PARSE on malformed command/URL
344 * or >100 domains deep.
346 *********************************************************************/
347 jb_err parse_http_url(const char * url,
348 struct http_request *http,
349 const struct client_state *csp)
351 int host_available = 1; /* A proxy can dream. */
354 * Zero out the results structure
356 memset(http, '\0', sizeof(*http));
360 * Save our initial URL
362 http->url = strdup(url);
363 if (http->url == NULL)
365 return JB_ERR_MEMORY;
370 * Check for * URI. If found, we're done.
372 if (*http->url == '*')
374 if ( NULL == (http->path = strdup("*"))
375 || NULL == (http->hostport = strdup("")) )
377 return JB_ERR_MEMORY;
379 if (http->url[1] != '\0')
388 * Split URL into protocol,hostport,path.
398 return JB_ERR_MEMORY;
401 /* Find the start of the URL in our scratch space */
403 if (strncmpic(url_noproto, "http://", 7) == 0)
408 else if (strncmpic(url_noproto, "https://", 8) == 0)
413 else if (*url_noproto == '/')
416 * Short request line without protocol and host.
417 * Most likely because the client's request
418 * was intercepted and redirected into Privoxy.
429 url_path = strchr(url_noproto, '/');
430 if (url_path != NULL)
435 * NOTE: The following line ignores the path for HTTPS URLS.
436 * This means that you get consistent behaviour if you type a
437 * https URL in and it's parsed by the function. (When the
438 * URL is actually retrieved, SSL hides the path part).
440 http->path = strdup(http->ssl ? "/" : url_path);
442 http->hostport = strdup(url_noproto);
447 * Repair broken HTTP requests that don't contain a path,
448 * or CONNECT requests
450 http->path = strdup("/");
451 http->hostport = strdup(url_noproto);
456 if ( (http->path == NULL)
457 || (http->hostport == NULL))
459 return JB_ERR_MEMORY;
465 /* Without host, there is nothing left to do here */
470 * Split hostport into user/password (ignored), host, port.
477 buf = strdup(http->hostport);
480 return JB_ERR_MEMORY;
483 /* check if url contains username and/or password */
484 host = strchr(buf, '@');
487 /* Contains username/password, skip it and the @ sign. */
492 /* No username or password. */
496 /* check if url contains port */
497 port = strchr(host, ':');
501 /* Terminate hostname and point to start of port string */
503 http->port = atoi(port);
507 /* No port specified. */
508 http->port = (http->ssl ? 443 : 80);
511 http->host = strdup(host);
515 if (http->host == NULL)
517 return JB_ERR_MEMORY;
522 * Split domain name so we can compare it against wildcards
524 return init_domain_components(http);
529 /*********************************************************************
531 * Function : unknown_method
533 * Description : Checks whether a method is unknown.
536 * 1 : method = points to a http method
538 * Returns : TRUE if it's unknown, FALSE otherwise.
540 *********************************************************************/
541 static int unknown_method(const char *method)
543 static const char *known_http_methods[] = {
544 /* Basic HTTP request type */
545 "GET", "HEAD", "POST", "PUT", "DELETE", "OPTIONS", "TRACE", "CONNECT",
546 /* webDAV extensions (RFC2518) */
547 "PROPFIND", "PROPPATCH", "MOVE", "COPY", "MKCOL", "LOCK", "UNLOCK",
549 * Microsoft webDAV extension for Exchange 2000. See:
550 * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
551 * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
553 "BCOPY", "BMOVE", "BDELETE", "BPROPFIND", "BPROPPATCH",
555 * Another Microsoft webDAV extension for Exchange 2000. See:
556 * http://systems.cs.colorado.edu/grunwald/MobileComputing/Papers/draft-cohen-gena-p-base-00.txt
557 * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
558 * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
560 "SUBSCRIBE", "UNSUBSCRIBE", "NOTIFY", "POLL",
562 * Yet another WebDAV extension, this time for
563 * Web Distributed Authoring and Versioning (RFC3253)
565 "VERSION-CONTROL", "REPORT", "CHECKOUT", "CHECKIN", "UNCHECKOUT",
566 "MKWORKSPACE", "UPDATE", "LABEL", "MERGE", "BASELINE-CONTROL", "MKACTIVITY",
570 for (i = 0; i < SZ(known_http_methods); i++)
572 if (0 == strcmpic(method, known_http_methods[i]))
583 /*********************************************************************
585 * Function : parse_http_request
587 * Description : Parse out the host and port from the URL. Find the
588 * hostname & path, port (if ':'), and/or password (if '@')
591 * 1 : req = HTTP request line to break down
592 * 2 : http = pointer to the http structure to hold elements
593 * 3 : csp = Current client state (buffers, headers, etc...)
595 * Returns : JB_ERR_OK on success
596 * JB_ERR_MEMORY on out of memory
597 * JB_ERR_CGI_PARAMS on malformed command/URL
598 * or >100 domains deep.
600 *********************************************************************/
601 jb_err parse_http_request(const char *req,
602 struct http_request *http,
603 const struct client_state *csp)
606 char *v[10]; /* XXX: Why 10? We should only need three. */
611 memset(http, '\0', sizeof(*http));
616 return JB_ERR_MEMORY;
619 n = ssplit(buf, " \r\n", v, SZ(v), 1, 1);
627 * Fail in case of unknown methods
628 * which we might not handle correctly.
630 * XXX: There should be a config option
631 * to forward requests with unknown methods
632 * anyway. Most of them don't need special
635 if (unknown_method(v[0]))
637 log_error(LOG_LEVEL_ERROR, "Unknown HTTP method detected: %s", v[0]);
642 if (strcmpic(v[0], "CONNECT") == 0)
647 err = parse_http_url(v[1], http, csp);
655 * Copy the details into the structure
657 http->ssl = is_connect;
658 http->cmd = strdup(req);
659 http->gpc = strdup(v[0]);
660 http->ver = strdup(v[2]);
662 if ( (http->cmd == NULL)
663 || (http->gpc == NULL)
664 || (http->ver == NULL) )
667 return JB_ERR_MEMORY;
676 /*********************************************************************
678 * Function : compile_pattern
680 * Description : Compiles a host, domain or TAG pattern.
683 * 1 : pattern = The pattern to compile.
684 * 2 : anchoring = How the regex should be anchored.
685 * Can be either one of NO_ANCHORING,
686 * LEFT_ANCHORED or RIGHT_ANCHORED.
687 * 3 : url = In case of failures, the spec member is
688 * logged and the structure freed.
689 * 4 : regex = Where the compiled regex should be stored.
691 * Returns : JB_ERR_OK - Success
692 * JB_ERR_MEMORY - Out of memory
693 * JB_ERR_PARSE - Cannot parse regex
695 *********************************************************************/
696 static jb_err compile_pattern(const char *pattern, enum regex_anchoring anchoring,
697 struct url_spec *url, regex_t **regex)
700 char rebuf[BUFFER_SIZE];
704 assert(strlen(pattern) < sizeof(rebuf) - 2);
706 if (pattern[0] == '\0')
724 log_error(LOG_LEVEL_FATAL,
725 "Invalid anchoring in compile_pattern %d", anchoring);
728 *regex = zalloc(sizeof(**regex));
732 return JB_ERR_MEMORY;
735 snprintf(rebuf, sizeof(rebuf), fmt, pattern);
737 errcode = regcomp(*regex, rebuf, (REG_EXTENDED|REG_NOSUB|REG_ICASE));
741 size_t errlen = regerror(errcode, *regex, rebuf, sizeof(rebuf));
742 if (errlen > (sizeof(rebuf) - (size_t)1))
744 errlen = sizeof(rebuf) - (size_t)1;
746 rebuf[errlen] = '\0';
747 log_error(LOG_LEVEL_ERROR, "error compiling %s from %s: %s",
748 pattern, url->spec, rebuf);
758 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
759 /*********************************************************************
761 * Function : compile_host_pattern
763 * Description : Parses and compiles a host pattern..
766 * 1 : url = Target url_spec to be filled in.
767 * 2 : host_pattern = Host pattern to compile.
769 * Returns : JB_ERR_OK - Success
770 * JB_ERR_MEMORY - Out of memory
771 * JB_ERR_PARSE - Cannot parse regex
773 *********************************************************************/
774 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
776 return compile_pattern(host_pattern, RIGHT_ANCHORED, url, &url->host_regex);
781 /*********************************************************************
783 * Function : compile_host_pattern
785 * Description : Parses and "compiles" an old-school host pattern.
788 * 1 : url = Target url_spec to be filled in.
789 * 2 : host_pattern = Host pattern to parse.
791 * Returns : JB_ERR_OK - Success
792 * JB_ERR_MEMORY - Out of memory
793 * JB_ERR_PARSE - Cannot parse regex
795 *********************************************************************/
796 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
805 if (host_pattern[strlen(host_pattern) - 1] == '.')
807 url->unanchored |= ANCHOR_RIGHT;
809 if (host_pattern[0] == '.')
811 url->unanchored |= ANCHOR_LEFT;
815 * Split domain into components
817 url->dbuffer = strdup(host_pattern);
818 if (NULL == url->dbuffer)
821 return JB_ERR_MEMORY;
827 for (p = url->dbuffer; *p ; p++)
829 *p = (char)tolower((int)(unsigned char)*p);
833 * Split the domain name into components
835 url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1);
840 return JB_ERR_MEMORY;
842 else if (url->dcount != 0)
845 * Save a copy of the pointers in dvec
847 size = (size_t)url->dcount * sizeof(*url->dvec);
849 url->dvec = (char **)malloc(size);
850 if (NULL == url->dvec)
853 return JB_ERR_MEMORY;
856 memcpy(url->dvec, v, size);
859 * else dcount == 0 in which case we needn't do anything,
860 * since dvec will never be accessed and the pattern will
867 /*********************************************************************
869 * Function : simple_domaincmp
871 * Description : Domain-wise Compare fqdn's. The comparison is
872 * both left- and right-anchored. The individual
873 * domain names are compared with simplematch().
874 * This is only used by domain_match.
877 * 1 : pv = array of patterns to compare
878 * 2 : fv = array of domain components to compare
879 * 3 : len = length of the arrays (both arrays are the
880 * same length - if they weren't, it couldn't
881 * possibly be a match).
883 * Returns : 0 => domains are equivalent, else no match.
885 *********************************************************************/
886 static int simple_domaincmp(char **pv, char **fv, int len)
890 for (n = 0; n < len; n++)
892 if (simplematch(pv[n], fv[n]))
903 /*********************************************************************
905 * Function : domain_match
907 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
908 * pattern->unachored, the comparison is un-, left-,
909 * right-anchored, or both.
910 * The individual domain names are compared with
914 * 1 : pattern = a domain that may contain a '*' as a wildcard.
915 * 2 : fqdn = domain name against which the patterns are compared.
917 * Returns : 0 => domains are equivalent, else no match.
919 *********************************************************************/
920 static int domain_match(const struct url_spec *pattern, const struct http_request *fqdn)
922 char **pv, **fv; /* vectors */
924 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
926 plen = pattern->dcount;
931 /* fqdn is too short to match this pattern */
938 if (unanchored == ANCHOR_LEFT)
943 * Convert this into a fully anchored pattern with
944 * the fqdn and pattern the same length
946 fv += (flen - plen); /* flen - plen >= 0 due to check above */
947 return simple_domaincmp(pv, fv, plen);
949 else if (unanchored == 0)
951 /* Fully anchored, check length */
956 return simple_domaincmp(pv, fv, plen);
958 else if (unanchored == ANCHOR_RIGHT)
960 /* Left anchored, ignore all extra in fqdn */
961 return simple_domaincmp(pv, fv, plen);
967 int maxn = flen - plen;
968 for (n = 0; n <= maxn; n++)
970 if (!simple_domaincmp(pv, fv, plen))
975 * Doesn't match from start of fqdn
976 * Try skipping first part of fqdn
984 #endif /* def FEATURE_EXTENDED_HOST_PATTERNS */
987 /*********************************************************************
989 * Function : create_url_spec
991 * Description : Creates a "url_spec" structure from a string.
992 * When finished, free with free_url_spec().
995 * 1 : url = Target url_spec to be filled in. Will be
997 * 2 : buf = Source pattern, null terminated. NOTE: The
998 * contents of this buffer are destroyed by this
999 * function. If this function succeeds, the
1000 * buffer is copied to url->spec. If this
1001 * function fails, the contents of the buffer
1004 * Returns : JB_ERR_OK - Success
1005 * JB_ERR_MEMORY - Out of memory
1006 * JB_ERR_PARSE - Cannot parse regex (Detailed message
1007 * written to system log)
1009 *********************************************************************/
1010 jb_err create_url_spec(struct url_spec * url, const char * buf)
1017 memset(url, '\0', sizeof(*url));
1019 /* Remember the original specification for the CGI pages. */
1020 url->spec = strdup(buf);
1021 if (NULL == url->spec)
1023 return JB_ERR_MEMORY;
1026 /* Is it tag pattern? */
1027 if (0 == strncmpic("TAG:", url->spec, 4))
1029 /* The pattern starts with the first character after "TAG:" */
1030 const char *tag_pattern = buf + 4;
1031 return compile_pattern(tag_pattern, NO_ANCHORING, url, &url->tag_regex);
1034 /* Only reached for URL patterns. XXX: should be factored out. */
1035 p = strchr(buf, '/');
1039 * Only compile the regex if it consists of more than
1040 * a single slash, otherwise it wouldn't affect the result.
1045 * XXX: does it make sense to compile the slash at the beginning?
1047 jb_err err = compile_pattern(p, LEFT_ANCHORED, url, &url->preg);
1049 if (JB_ERR_OK != err)
1057 p = strchr(buf, ':');
1061 url->port_list = strdup(p);
1062 if (NULL == url->port_list)
1064 return JB_ERR_MEMORY;
1069 url->port_list = NULL;
1074 return compile_host_pattern(url, buf);
1082 /*********************************************************************
1084 * Function : free_url_spec
1086 * Description : Called from the "unloaders". Freez the url
1087 * structure elements.
1090 * 1 : url = pointer to a url_spec structure.
1094 *********************************************************************/
1095 void free_url_spec(struct url_spec *url)
1097 if (url == NULL) return;
1100 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
1101 if (url->host_regex)
1103 regfree(url->host_regex);
1104 freez(url->host_regex);
1107 freez(url->dbuffer);
1110 #endif /* ndef FEATURE_EXTENDED_HOST_PATTERNS */
1111 freez(url->port_list);
1119 regfree(url->tag_regex);
1120 freez(url->tag_regex);
1125 /*********************************************************************
1127 * Function : url_match
1129 * Description : Compare a URL against a URL pattern.
1132 * 1 : pattern = a URL pattern
1133 * 2 : url = URL to match
1135 * Returns : Nonzero if the URL matches the pattern, else 0.
1137 *********************************************************************/
1138 int url_match(const struct url_spec *pattern,
1139 const struct http_request *http)
1141 /* XXX: these should probably be functions. */
1142 #define PORT_MATCHES ((NULL == pattern->port_list) || match_portlist(pattern->port_list, http->port))
1143 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
1144 #define DOMAIN_MATCHES ((NULL == pattern->host_regex) || (0 == regexec(pattern->host_regex, http->host, 0, NULL, 0)))
1146 #define DOMAIN_MATCHES ((NULL == pattern->dbuffer) || (0 == domain_match(pattern, http)))
1148 #define PATH_MATCHES ((NULL == pattern->preg) || (0 == regexec(pattern->preg, http->path, 0, NULL, 0)))
1150 if (pattern->tag_regex != NULL)
1152 /* It's a tag pattern and shouldn't be matched against URLs */
1156 return (PORT_MATCHES && DOMAIN_MATCHES && PATH_MATCHES);
1161 /*********************************************************************
1163 * Function : match_portlist
1165 * Description : Check if a given number is covered by a comma
1166 * separated list of numbers and ranges (a,b-c,d,..)
1169 * 1 : portlist = String with list
1170 * 2 : port = port to check
1172 * Returns : 0 => no match
1175 *********************************************************************/
1176 int match_portlist(const char *portlist, int port)
1178 char *min, *max, *next, *portlist_copy;
1180 min = next = portlist_copy = strdup(portlist);
1183 * Zero-terminate first item and remember offset for next
1185 if (NULL != (next = strchr(portlist_copy, (int) ',')))
1191 * Loop through all items, checking for match
1195 if (NULL == (max = strchr(min, (int) '-')))
1198 * No dash, check for equality
1200 if (port == atoi(min))
1202 free(portlist_copy);
1209 * This is a range, so check if between min and max,
1210 * or, if max was omitted, between min and 65K
1213 if(port >= atoi(min) && port <= (atoi(max) ? atoi(max) : 65535))
1215 free(portlist_copy);
1227 * Zero-terminate next item and remember offset for n+1
1229 if ((NULL != next) && (NULL != (next = strchr(next, (int) ','))))
1235 free(portlist_copy);