1 const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.31 2008/04/10 14:41:04 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 1.31 2008/04/10 14:41:04 fabiankeil
37 * Ditch url_spec's path member now that it's no longer used.
39 * Revision 1.30 2008/04/10 04:24:24 fabiankeil
40 * Stop duplicating the plain text representation of the path regex
41 * (and keeping the copy around). Once the regex is compiled it's no
44 * Revision 1.29 2008/04/10 04:17:56 fabiankeil
45 * In url_match(), check the right member for NULL when determining
46 * whether there's a path regex to execute. Looking for a plain-text
47 * representation works as well, but it looks "interesting" and that
48 * member will be removed soonish anyway.
50 * Revision 1.28 2008/04/08 16:07:39 fabiankeil
51 * Make it harder to mistake url_match()'s
52 * second parameter for an url_spec.
54 * Revision 1.27 2008/04/08 15:44:33 fabiankeil
55 * Save a bit of memory (and a few cpu cycles) by not bothering to
56 * compile slash-only path regexes that don't affect the result.
58 * Revision 1.26 2008/04/07 16:57:18 fabiankeil
59 * - Use free_url_spec() more consistently.
60 * - Let it reset url->dcount just in case.
62 * Revision 1.25 2008/04/06 15:18:38 fabiankeil
63 * Oh well, rename the --enable-pcre-host-patterns option to
64 * --enable-extended-host-patterns as it's not really PCRE syntax.
66 * Revision 1.24 2008/04/06 14:54:26 fabiankeil
67 * Use PCRE syntax in host patterns when configured
68 * with --enable-pcre-host-patterns.
70 * Revision 1.23 2008/04/05 12:19:20 fabiankeil
71 * Factor compile_host_pattern() out of create_url_spec().
73 * Revision 1.22 2008/03/30 15:02:32 fabiankeil
74 * SZitify unknown_method().
76 * Revision 1.21 2007/12/24 16:34:23 fabiankeil
77 * Band-aid (and micro-optimization) that makes it less likely to run out of
78 * stack space with overly-complex path patterns. Probably masks the problem
79 * reported by Lee in #1856679. Hohoho.
81 * Revision 1.20 2007/09/02 15:31:20 fabiankeil
82 * Move match_portlist() from filter.c to urlmatch.c.
83 * It's used for url matching, not for filtering.
85 * Revision 1.19 2007/09/02 13:42:11 fabiankeil
86 * - Allow port lists in url patterns.
87 * - Ditch unused url_spec member pathlen.
89 * Revision 1.18 2007/07/30 16:42:21 fabiankeil
90 * Move the method check into unknown_method()
91 * and loop through the known methods instead
92 * of using a screen-long OR chain.
94 * Revision 1.17 2007/04/15 16:39:21 fabiankeil
95 * Introduce tags as alternative way to specify which
96 * actions apply to a request. At the moment tags can be
97 * created based on client and server headers.
99 * Revision 1.16 2007/02/13 13:59:24 fabiankeil
100 * Remove redundant log message.
102 * Revision 1.15 2007/01/28 16:11:23 fabiankeil
103 * Accept WebDAV methods for subversion
104 * in parse_http_request(). Closes FR 1581425.
106 * Revision 1.14 2007/01/06 14:23:56 fabiankeil
107 * Fix gcc43 warnings. Mark *csp as immutable
108 * for parse_http_url() and url_match().
109 * Replace a sprintf call with snprintf.
111 * Revision 1.13 2006/12/06 19:50:54 fabiankeil
112 * parse_http_url() now handles intercepted
113 * HTTP request lines as well. Moved parts
114 * of parse_http_url()'s code into
115 * init_domain_components() so that it can
116 * be reused in chat().
118 * Revision 1.12 2006/07/18 14:48:47 david__schmidt
119 * Reorganizing the repository: swapping out what was HEAD (the old 3.1 branch)
120 * with what was really the latest development (the v_3_0_branch branch)
122 * Revision 1.10.2.7 2003/05/17 15:57:24 oes
123 * - parse_http_url now checks memory allocation failure for
124 * duplication of "*" URL and rejects "*something" URLs
126 * - Added a comment to what might look like a bug in
127 * create_url_spec (see !bug #736931)
128 * - Comment cosmetics
130 * Revision 1.10.2.6 2003/05/07 12:39:48 oes
131 * Fix typo: Default port for https URLs is 443, not 143.
132 * Thanks to Scott Tregear for spotting this one.
134 * Revision 1.10.2.5 2003/02/28 13:09:29 oes
135 * Fixed a rare double free condition as per Bug #694713
137 * Revision 1.10.2.4 2003/02/28 12:57:44 oes
138 * Moved freeing of http request structure to its owner
139 * as per Dan Price's observations in Bug #694713
141 * Revision 1.10.2.3 2002/11/12 16:50:40 oes
142 * Fixed memory leak in parse_http_request() reported by Oliver Stoeneberg. Fixes bug #637073
144 * Revision 1.10.2.2 2002/09/25 14:53:15 oes
145 * Added basic support for OPTIONS and TRACE HTTP methods:
146 * parse_http_url now recognizes the "*" URI as well as
147 * the OPTIONS and TRACE method keywords.
149 * Revision 1.10.2.1 2002/06/06 19:06:44 jongfoster
150 * Adding support for proprietary Microsoft WebDAV extensions
152 * Revision 1.10 2002/05/12 21:40:37 jongfoster
153 * - Removing some unused code
155 * Revision 1.9 2002/04/04 00:36:36 gliptak
156 * always use pcre for matching
158 * Revision 1.8 2002/04/03 23:32:47 jongfoster
159 * Fixing memory leak on error
161 * Revision 1.7 2002/03/26 22:29:55 swa
162 * we have a new homepage!
164 * Revision 1.6 2002/03/24 13:25:43 swa
165 * name change related issues
167 * Revision 1.5 2002/03/13 00:27:05 jongfoster
170 * Revision 1.4 2002/03/07 03:46:17 oes
171 * Fixed compiler warnings
173 * Revision 1.3 2002/03/03 14:51:11 oes
174 * Fixed CLF logging: Added ocmd member for client's request to struct http_request
176 * Revision 1.2 2002/01/21 00:14:09 jongfoster
177 * Correcting comment style
178 * Fixing an uninitialized memory bug in create_url_spec()
180 * Revision 1.1 2002/01/17 20:53:46 jongfoster
181 * Moving all our URL and URL pattern parsing code to the same file - it
182 * was scattered around in filters.c, loaders.c and parsers.c.
184 * Providing a single, simple url_match(pattern,url) function - rather than
185 * the 3-line match routine which was repeated all over the place.
187 * Renaming free_url to free_url_spec, since it frees a struct url_spec.
189 * Providing parse_http_url() so that URLs can be parsed without faking a
190 * HTTP request line for parse_http_request() or repeating the parsing
191 * code (both of which were techniques that were actually in use).
193 * Standardizing that struct http_request is used to represent a URL, and
194 * struct url_spec is used to represent a URL pattern. (Before, URLs were
195 * represented as seperate variables and a partially-filled-in url_spec).
198 *********************************************************************/
205 #include <sys/types.h>
213 #if !defined(_WIN32) && !defined(__OS2__)
218 #include "urlmatch.h"
220 #include "miscutil.h"
223 const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
225 enum regex_anchoring {NO_ANCHORING, LEFT_ANCHORED, RIGHT_ANCHORED};
227 /*********************************************************************
229 * Function : free_http_request
231 * Description : Freez a http_request structure
234 * 1 : http = points to a http_request structure to free
238 *********************************************************************/
239 void free_http_request(struct http_request *http)
248 freez(http->hostport);
251 freez(http->host_ip_addr_str);
252 freez(http->dbuffer);
258 /*********************************************************************
260 * Function : init_domain_components
262 * Description : Splits the domain name so we can compare it
263 * against wildcards. It used to be part of
264 * parse_http_url, but was separated because the
265 * same code is required in chat in case of
266 * intercepted requests.
269 * 1 : http = pointer to the http structure to hold elements.
271 * Returns : JB_ERR_OK on success
272 * JB_ERR_MEMORY on out of memory
273 * JB_ERR_PARSE on malformed command/URL
274 * or >100 domains deep.
276 *********************************************************************/
277 jb_err init_domain_components(struct http_request *http)
279 char *vec[BUFFER_SIZE];
283 http->dbuffer = strdup(http->host);
284 if (NULL == http->dbuffer)
286 return JB_ERR_MEMORY;
289 /* map to lower case */
290 for (p = http->dbuffer; *p ; p++)
292 *p = (char)tolower((int)(unsigned char)*p);
295 /* split the domain name into components */
296 http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec), 1, 1);
298 if (http->dcount <= 0)
301 * Error: More than SZ(vec) components in domain
302 * or: no components in domain
304 log_error(LOG_LEVEL_ERROR, "More than SZ(vec) components in domain or none at all.");
308 /* save a copy of the pointers in dvec */
309 size = (size_t)http->dcount * sizeof(*http->dvec);
311 http->dvec = (char **)malloc(size);
312 if (NULL == http->dvec)
314 return JB_ERR_MEMORY;
317 memcpy(http->dvec, vec, size);
323 /*********************************************************************
325 * Function : parse_http_url
327 * Description : Parse out the host and port from the URL. Find the
328 * hostname & path, port (if ':'), and/or password (if '@')
331 * 1 : url = URL (or is it URI?) to break down
332 * 2 : http = pointer to the http structure to hold elements.
333 * Will be zeroed before use. Note that this
334 * function sets the http->gpc and http->ver
336 * 3 : csp = Current client state (buffers, headers, etc...)
338 * Returns : JB_ERR_OK on success
339 * JB_ERR_MEMORY on out of memory
340 * JB_ERR_PARSE on malformed command/URL
341 * or >100 domains deep.
343 *********************************************************************/
344 jb_err parse_http_url(const char * url,
345 struct http_request *http,
346 const struct client_state *csp)
348 int host_available = 1; /* A proxy can dream. */
351 * Zero out the results structure
353 memset(http, '\0', sizeof(*http));
357 * Save our initial URL
359 http->url = strdup(url);
360 if (http->url == NULL)
362 return JB_ERR_MEMORY;
367 * Check for * URI. If found, we're done.
369 if (*http->url == '*')
371 if ( NULL == (http->path = strdup("*"))
372 || NULL == (http->hostport = strdup("")) )
374 return JB_ERR_MEMORY;
376 if (http->url[1] != '\0')
385 * Split URL into protocol,hostport,path.
395 return JB_ERR_MEMORY;
398 /* Find the start of the URL in our scratch space */
400 if (strncmpic(url_noproto, "http://", 7) == 0)
405 else if (strncmpic(url_noproto, "https://", 8) == 0)
410 else if (*url_noproto == '/')
413 * Short request line without protocol and host.
414 * Most likely because the client's request
415 * was intercepted and redirected into Privoxy.
426 url_path = strchr(url_noproto, '/');
427 if (url_path != NULL)
432 * NOTE: The following line ignores the path for HTTPS URLS.
433 * This means that you get consistent behaviour if you type a
434 * https URL in and it's parsed by the function. (When the
435 * URL is actually retrieved, SSL hides the path part).
437 http->path = strdup(http->ssl ? "/" : url_path);
439 http->hostport = strdup(url_noproto);
444 * Repair broken HTTP requests that don't contain a path,
445 * or CONNECT requests
447 http->path = strdup("/");
448 http->hostport = strdup(url_noproto);
453 if ( (http->path == NULL)
454 || (http->hostport == NULL))
456 return JB_ERR_MEMORY;
462 /* Without host, there is nothing left to do here */
467 * Split hostport into user/password (ignored), host, port.
474 buf = strdup(http->hostport);
477 return JB_ERR_MEMORY;
480 /* check if url contains username and/or password */
481 host = strchr(buf, '@');
484 /* Contains username/password, skip it and the @ sign. */
489 /* No username or password. */
493 /* check if url contains port */
494 port = strchr(host, ':');
498 /* Terminate hostname and point to start of port string */
500 http->port = atoi(port);
504 /* No port specified. */
505 http->port = (http->ssl ? 443 : 80);
508 http->host = strdup(host);
512 if (http->host == NULL)
514 return JB_ERR_MEMORY;
519 * Split domain name so we can compare it against wildcards
521 return init_domain_components(http);
526 /*********************************************************************
528 * Function : unknown_method
530 * Description : Checks whether a method is unknown.
533 * 1 : method = points to a http method
535 * Returns : TRUE if it's unknown, FALSE otherwise.
537 *********************************************************************/
538 static int unknown_method(const char *method)
540 static const char *known_http_methods[] = {
541 /* Basic HTTP request type */
542 "GET", "HEAD", "POST", "PUT", "DELETE", "OPTIONS", "TRACE", "CONNECT",
543 /* webDAV extensions (RFC2518) */
544 "PROPFIND", "PROPPATCH", "MOVE", "COPY", "MKCOL", "LOCK", "UNLOCK",
546 * Microsoft webDAV extension for Exchange 2000. See:
547 * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
548 * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
550 "BCOPY", "BMOVE", "BDELETE", "BPROPFIND", "BPROPPATCH",
552 * Another Microsoft webDAV extension for Exchange 2000. See:
553 * http://systems.cs.colorado.edu/grunwald/MobileComputing/Papers/draft-cohen-gena-p-base-00.txt
554 * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
555 * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
557 "SUBSCRIBE", "UNSUBSCRIBE", "NOTIFY", "POLL",
559 * Yet another WebDAV extension, this time for
560 * Web Distributed Authoring and Versioning (RFC3253)
562 "VERSION-CONTROL", "REPORT", "CHECKOUT", "CHECKIN", "UNCHECKOUT",
563 "MKWORKSPACE", "UPDATE", "LABEL", "MERGE", "BASELINE-CONTROL", "MKACTIVITY",
567 for (i = 0; i < SZ(known_http_methods); i++)
569 if (0 == strcmpic(method, known_http_methods[i]))
580 /*********************************************************************
582 * Function : parse_http_request
584 * Description : Parse out the host and port from the URL. Find the
585 * hostname & path, port (if ':'), and/or password (if '@')
588 * 1 : req = HTTP request line to break down
589 * 2 : http = pointer to the http structure to hold elements
590 * 3 : csp = Current client state (buffers, headers, etc...)
592 * Returns : JB_ERR_OK on success
593 * JB_ERR_MEMORY on out of memory
594 * JB_ERR_CGI_PARAMS on malformed command/URL
595 * or >100 domains deep.
597 *********************************************************************/
598 jb_err parse_http_request(const char *req,
599 struct http_request *http,
600 const struct client_state *csp)
603 char *v[10]; /* XXX: Why 10? We should only need three. */
608 memset(http, '\0', sizeof(*http));
613 return JB_ERR_MEMORY;
616 n = ssplit(buf, " \r\n", v, SZ(v), 1, 1);
624 * Fail in case of unknown methods
625 * which we might not handle correctly.
627 * XXX: There should be a config option
628 * to forward requests with unknown methods
629 * anyway. Most of them don't need special
632 if (unknown_method(v[0]))
634 log_error(LOG_LEVEL_ERROR, "Unknown HTTP method detected: %s", v[0]);
639 if (strcmpic(v[0], "CONNECT") == 0)
644 err = parse_http_url(v[1], http, csp);
652 * Copy the details into the structure
654 http->ssl = is_connect;
655 http->cmd = strdup(req);
656 http->gpc = strdup(v[0]);
657 http->ver = strdup(v[2]);
659 if ( (http->cmd == NULL)
660 || (http->gpc == NULL)
661 || (http->ver == NULL) )
664 return JB_ERR_MEMORY;
673 /*********************************************************************
675 * Function : compile_pattern
677 * Description : Compiles a host, domain or TAG pattern.
680 * 1 : pattern = The pattern to compile.
681 * 2 : anchoring = How the regex should be anchored.
682 * Can be either one of NO_ANCHORING,
683 * LEFT_ANCHORED or RIGHT_ANCHORED.
684 * 3 : url = In case of failures, the spec member is
685 * logged and the structure freed.
686 * 4 : regex = Where the compiled regex should be stored.
688 * Returns : JB_ERR_OK - Success
689 * JB_ERR_MEMORY - Out of memory
690 * JB_ERR_PARSE - Cannot parse regex
692 *********************************************************************/
693 static jb_err compile_pattern(const char *pattern, enum regex_anchoring anchoring,
694 struct url_spec *url, regex_t **regex)
697 char rebuf[BUFFER_SIZE];
701 assert(strlen(pattern) < sizeof(rebuf) - 2);
703 if (pattern[0] == '\0')
721 log_error(LOG_LEVEL_FATAL,
722 "Invalid anchoring in compile_pattern %d", anchoring);
725 *regex = zalloc(sizeof(**regex));
729 return JB_ERR_MEMORY;
732 snprintf(rebuf, sizeof(rebuf), fmt, pattern);
734 errcode = regcomp(*regex, rebuf, (REG_EXTENDED|REG_NOSUB|REG_ICASE));
738 size_t errlen = regerror(errcode, *regex, rebuf, sizeof(rebuf));
739 if (errlen > (sizeof(rebuf) - (size_t)1))
741 errlen = sizeof(rebuf) - (size_t)1;
743 rebuf[errlen] = '\0';
744 log_error(LOG_LEVEL_ERROR, "error compiling %s from %s: %s",
745 pattern, url->spec, rebuf);
755 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
756 /*********************************************************************
758 * Function : compile_host_pattern
760 * Description : Parses and compiles a host pattern..
763 * 1 : url = Target url_spec to be filled in.
764 * 2 : host_pattern = Host pattern to compile.
766 * Returns : JB_ERR_OK - Success
767 * JB_ERR_MEMORY - Out of memory
768 * JB_ERR_PARSE - Cannot parse regex
770 *********************************************************************/
771 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
773 return compile_pattern(host_pattern, RIGHT_ANCHORED, url, &url->host_regex);
778 /*********************************************************************
780 * Function : compile_host_pattern
782 * Description : Parses and "compiles" an old-school host pattern.
785 * 1 : url = Target url_spec to be filled in.
786 * 2 : host_pattern = Host pattern to parse.
788 * Returns : JB_ERR_OK - Success
789 * JB_ERR_MEMORY - Out of memory
790 * JB_ERR_PARSE - Cannot parse regex
792 *********************************************************************/
793 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
802 if (host_pattern[strlen(host_pattern) - 1] == '.')
804 url->unanchored |= ANCHOR_RIGHT;
806 if (host_pattern[0] == '.')
808 url->unanchored |= ANCHOR_LEFT;
812 * Split domain into components
814 url->dbuffer = strdup(host_pattern);
815 if (NULL == url->dbuffer)
818 return JB_ERR_MEMORY;
824 for (p = url->dbuffer; *p ; p++)
826 *p = (char)tolower((int)(unsigned char)*p);
830 * Split the domain name into components
832 url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1);
837 return JB_ERR_MEMORY;
839 else if (url->dcount != 0)
842 * Save a copy of the pointers in dvec
844 size = (size_t)url->dcount * sizeof(*url->dvec);
846 url->dvec = (char **)malloc(size);
847 if (NULL == url->dvec)
850 return JB_ERR_MEMORY;
853 memcpy(url->dvec, v, size);
856 * else dcount == 0 in which case we needn't do anything,
857 * since dvec will never be accessed and the pattern will
864 /*********************************************************************
866 * Function : simple_domaincmp
868 * Description : Domain-wise Compare fqdn's. The comparison is
869 * both left- and right-anchored. The individual
870 * domain names are compared with simplematch().
871 * This is only used by domain_match.
874 * 1 : pv = array of patterns to compare
875 * 2 : fv = array of domain components to compare
876 * 3 : len = length of the arrays (both arrays are the
877 * same length - if they weren't, it couldn't
878 * possibly be a match).
880 * Returns : 0 => domains are equivalent, else no match.
882 *********************************************************************/
883 static int simple_domaincmp(char **pv, char **fv, int len)
887 for (n = 0; n < len; n++)
889 if (simplematch(pv[n], fv[n]))
900 /*********************************************************************
902 * Function : domain_match
904 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
905 * pattern->unachored, the comparison is un-, left-,
906 * right-anchored, or both.
907 * The individual domain names are compared with
911 * 1 : pattern = a domain that may contain a '*' as a wildcard.
912 * 2 : fqdn = domain name against which the patterns are compared.
914 * Returns : 0 => domains are equivalent, else no match.
916 *********************************************************************/
917 static int domain_match(const struct url_spec *pattern, const struct http_request *fqdn)
919 char **pv, **fv; /* vectors */
921 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
923 plen = pattern->dcount;
928 /* fqdn is too short to match this pattern */
935 if (unanchored == ANCHOR_LEFT)
940 * Convert this into a fully anchored pattern with
941 * the fqdn and pattern the same length
943 fv += (flen - plen); /* flen - plen >= 0 due to check above */
944 return simple_domaincmp(pv, fv, plen);
946 else if (unanchored == 0)
948 /* Fully anchored, check length */
953 return simple_domaincmp(pv, fv, plen);
955 else if (unanchored == ANCHOR_RIGHT)
957 /* Left anchored, ignore all extra in fqdn */
958 return simple_domaincmp(pv, fv, plen);
964 int maxn = flen - plen;
965 for (n = 0; n <= maxn; n++)
967 if (!simple_domaincmp(pv, fv, plen))
972 * Doesn't match from start of fqdn
973 * Try skipping first part of fqdn
981 #endif /* def FEATURE_EXTENDED_HOST_PATTERNS */
984 /*********************************************************************
986 * Function : create_url_spec
988 * Description : Creates a "url_spec" structure from a string.
989 * When finished, free with free_url_spec().
992 * 1 : url = Target url_spec to be filled in. Will be
994 * 2 : buf = Source pattern, null terminated. NOTE: The
995 * contents of this buffer are destroyed by this
996 * function. If this function succeeds, the
997 * buffer is copied to url->spec. If this
998 * function fails, the contents of the buffer
1001 * Returns : JB_ERR_OK - Success
1002 * JB_ERR_MEMORY - Out of memory
1003 * JB_ERR_PARSE - Cannot parse regex (Detailed message
1004 * written to system log)
1006 *********************************************************************/
1007 jb_err create_url_spec(struct url_spec * url, const char * buf)
1017 memset(url, '\0', sizeof(*url));
1020 * Save a copy of the orignal specification
1022 if ((url->spec = strdup(buf)) == NULL)
1024 return JB_ERR_MEMORY;
1027 /* Is it tag pattern? */
1028 if (0 == strncmpic("TAG:", url->spec, 4))
1030 /* The pattern starts with the first character after "TAG:" */
1031 const char *tag_pattern = buf + 4;
1032 return compile_pattern(tag_pattern, NO_ANCHORING, url, &url->tag_regex);
1035 /* Only reached for URL patterns. XXX: should be factored out. */
1036 p = strchr(buf, '/');
1040 * Only compile the regex if it consists of more than
1041 * a single slash, otherwise it wouldn't affect the result.
1046 * XXX: does it make sense to compile the slash at the beginning?
1048 jb_err err = compile_pattern(p, LEFT_ANCHORED, url, &url->preg);
1050 if (JB_ERR_OK != err)
1058 p = strchr(buf, ':');
1062 url->port_list = strdup(p);
1063 if (NULL == url->port_list)
1065 return JB_ERR_MEMORY;
1070 url->port_list = NULL;
1075 return compile_host_pattern(url, buf);
1083 /*********************************************************************
1085 * Function : free_url_spec
1087 * Description : Called from the "unloaders". Freez the url
1088 * structure elements.
1091 * 1 : url = pointer to a url_spec structure.
1095 *********************************************************************/
1096 void free_url_spec(struct url_spec *url)
1098 if (url == NULL) return;
1101 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
1102 if (url->host_regex)
1104 regfree(url->host_regex);
1105 freez(url->host_regex);
1108 freez(url->dbuffer);
1111 #endif /* ndef FEATURE_EXTENDED_HOST_PATTERNS */
1112 freez(url->port_list);
1120 regfree(url->tag_regex);
1121 freez(url->tag_regex);
1126 /*********************************************************************
1128 * Function : url_match
1130 * Description : Compare a URL against a URL pattern.
1133 * 1 : pattern = a URL pattern
1134 * 2 : url = URL to match
1136 * Returns : Nonzero if the URL matches the pattern, else 0.
1138 *********************************************************************/
1139 int url_match(const struct url_spec *pattern,
1140 const struct http_request *http)
1142 /* XXX: these should probably be functions. */
1143 #define PORT_MATCHES ((NULL == pattern->port_list) || match_portlist(pattern->port_list, http->port))
1144 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
1145 #define DOMAIN_MATCHES ((NULL == pattern->host_regex) || (0 == regexec(pattern->host_regex, http->host, 0, NULL, 0)))
1147 #define DOMAIN_MATCHES ((NULL == pattern->dbuffer) || (0 == domain_match(pattern, http)))
1149 #define PATH_MATCHES ((NULL == pattern->preg) || (0 == regexec(pattern->preg, http->path, 0, NULL, 0)))
1151 if (pattern->tag_regex != NULL)
1153 /* It's a tag pattern and shouldn't be matched against URLs */
1157 return (PORT_MATCHES && DOMAIN_MATCHES && PATH_MATCHES);
1162 /*********************************************************************
1164 * Function : match_portlist
1166 * Description : Check if a given number is covered by a comma
1167 * separated list of numbers and ranges (a,b-c,d,..)
1170 * 1 : portlist = String with list
1171 * 2 : port = port to check
1173 * Returns : 0 => no match
1176 *********************************************************************/
1177 int match_portlist(const char *portlist, int port)
1179 char *min, *max, *next, *portlist_copy;
1181 min = next = portlist_copy = strdup(portlist);
1184 * Zero-terminate first item and remember offset for next
1186 if (NULL != (next = strchr(portlist_copy, (int) ',')))
1192 * Loop through all items, checking for match
1196 if (NULL == (max = strchr(min, (int) '-')))
1199 * No dash, check for equality
1201 if (port == atoi(min))
1203 free(portlist_copy);
1210 * This is a range, so check if between min and max,
1211 * or, if max was omitted, between min and 65K
1214 if(port >= atoi(min) && port <= (atoi(max) ? atoi(max) : 65535))
1216 free(portlist_copy);
1228 * Zero-terminate next item and remember offset for n+1
1230 if ((NULL != next) && (NULL != (next = strchr(next, (int) ','))))
1236 free(portlist_copy);