1 const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.33 2008/04/12 14:03:13 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 1.33 2008/04/12 14:03:13 fabiankeil
37 * Remove an obvious comment and improve another one.
39 * Revision 1.32 2008/04/12 12:38:06 fabiankeil
40 * Factor out duplicated code to compile host, path and tag patterns.
42 * Revision 1.31 2008/04/10 14:41:04 fabiankeil
43 * Ditch url_spec's path member now that it's no longer used.
45 * Revision 1.30 2008/04/10 04:24:24 fabiankeil
46 * Stop duplicating the plain text representation of the path regex
47 * (and keeping the copy around). Once the regex is compiled it's no
50 * Revision 1.29 2008/04/10 04:17:56 fabiankeil
51 * In url_match(), check the right member for NULL when determining
52 * whether there's a path regex to execute. Looking for a plain-text
53 * representation works as well, but it looks "interesting" and that
54 * member will be removed soonish anyway.
56 * Revision 1.28 2008/04/08 16:07:39 fabiankeil
57 * Make it harder to mistake url_match()'s
58 * second parameter for an url_spec.
60 * Revision 1.27 2008/04/08 15:44:33 fabiankeil
61 * Save a bit of memory (and a few cpu cycles) by not bothering to
62 * compile slash-only path regexes that don't affect the result.
64 * Revision 1.26 2008/04/07 16:57:18 fabiankeil
65 * - Use free_url_spec() more consistently.
66 * - Let it reset url->dcount just in case.
68 * Revision 1.25 2008/04/06 15:18:38 fabiankeil
69 * Oh well, rename the --enable-pcre-host-patterns option to
70 * --enable-extended-host-patterns as it's not really PCRE syntax.
72 * Revision 1.24 2008/04/06 14:54:26 fabiankeil
73 * Use PCRE syntax in host patterns when configured
74 * with --enable-pcre-host-patterns.
76 * Revision 1.23 2008/04/05 12:19:20 fabiankeil
77 * Factor compile_host_pattern() out of create_url_spec().
79 * Revision 1.22 2008/03/30 15:02:32 fabiankeil
80 * SZitify unknown_method().
82 * Revision 1.21 2007/12/24 16:34:23 fabiankeil
83 * Band-aid (and micro-optimization) that makes it less likely to run out of
84 * stack space with overly-complex path patterns. Probably masks the problem
85 * reported by Lee in #1856679. Hohoho.
87 * Revision 1.20 2007/09/02 15:31:20 fabiankeil
88 * Move match_portlist() from filter.c to urlmatch.c.
89 * It's used for url matching, not for filtering.
91 * Revision 1.19 2007/09/02 13:42:11 fabiankeil
92 * - Allow port lists in url patterns.
93 * - Ditch unused url_spec member pathlen.
95 * Revision 1.18 2007/07/30 16:42:21 fabiankeil
96 * Move the method check into unknown_method()
97 * and loop through the known methods instead
98 * of using a screen-long OR chain.
100 * Revision 1.17 2007/04/15 16:39:21 fabiankeil
101 * Introduce tags as alternative way to specify which
102 * actions apply to a request. At the moment tags can be
103 * created based on client and server headers.
105 * Revision 1.16 2007/02/13 13:59:24 fabiankeil
106 * Remove redundant log message.
108 * Revision 1.15 2007/01/28 16:11:23 fabiankeil
109 * Accept WebDAV methods for subversion
110 * in parse_http_request(). Closes FR 1581425.
112 * Revision 1.14 2007/01/06 14:23:56 fabiankeil
113 * Fix gcc43 warnings. Mark *csp as immutable
114 * for parse_http_url() and url_match().
115 * Replace a sprintf call with snprintf.
117 * Revision 1.13 2006/12/06 19:50:54 fabiankeil
118 * parse_http_url() now handles intercepted
119 * HTTP request lines as well. Moved parts
120 * of parse_http_url()'s code into
121 * init_domain_components() so that it can
122 * be reused in chat().
124 * Revision 1.12 2006/07/18 14:48:47 david__schmidt
125 * Reorganizing the repository: swapping out what was HEAD (the old 3.1 branch)
126 * with what was really the latest development (the v_3_0_branch branch)
128 * Revision 1.10.2.7 2003/05/17 15:57:24 oes
129 * - parse_http_url now checks memory allocation failure for
130 * duplication of "*" URL and rejects "*something" URLs
132 * - Added a comment to what might look like a bug in
133 * create_url_spec (see !bug #736931)
134 * - Comment cosmetics
136 * Revision 1.10.2.6 2003/05/07 12:39:48 oes
137 * Fix typo: Default port for https URLs is 443, not 143.
138 * Thanks to Scott Tregear for spotting this one.
140 * Revision 1.10.2.5 2003/02/28 13:09:29 oes
141 * Fixed a rare double free condition as per Bug #694713
143 * Revision 1.10.2.4 2003/02/28 12:57:44 oes
144 * Moved freeing of http request structure to its owner
145 * as per Dan Price's observations in Bug #694713
147 * Revision 1.10.2.3 2002/11/12 16:50:40 oes
148 * Fixed memory leak in parse_http_request() reported by Oliver Stoeneberg. Fixes bug #637073
150 * Revision 1.10.2.2 2002/09/25 14:53:15 oes
151 * Added basic support for OPTIONS and TRACE HTTP methods:
152 * parse_http_url now recognizes the "*" URI as well as
153 * the OPTIONS and TRACE method keywords.
155 * Revision 1.10.2.1 2002/06/06 19:06:44 jongfoster
156 * Adding support for proprietary Microsoft WebDAV extensions
158 * Revision 1.10 2002/05/12 21:40:37 jongfoster
159 * - Removing some unused code
161 * Revision 1.9 2002/04/04 00:36:36 gliptak
162 * always use pcre for matching
164 * Revision 1.8 2002/04/03 23:32:47 jongfoster
165 * Fixing memory leak on error
167 * Revision 1.7 2002/03/26 22:29:55 swa
168 * we have a new homepage!
170 * Revision 1.6 2002/03/24 13:25:43 swa
171 * name change related issues
173 * Revision 1.5 2002/03/13 00:27:05 jongfoster
176 * Revision 1.4 2002/03/07 03:46:17 oes
177 * Fixed compiler warnings
179 * Revision 1.3 2002/03/03 14:51:11 oes
180 * Fixed CLF logging: Added ocmd member for client's request to struct http_request
182 * Revision 1.2 2002/01/21 00:14:09 jongfoster
183 * Correcting comment style
184 * Fixing an uninitialized memory bug in create_url_spec()
186 * Revision 1.1 2002/01/17 20:53:46 jongfoster
187 * Moving all our URL and URL pattern parsing code to the same file - it
188 * was scattered around in filters.c, loaders.c and parsers.c.
190 * Providing a single, simple url_match(pattern,url) function - rather than
191 * the 3-line match routine which was repeated all over the place.
193 * Renaming free_url to free_url_spec, since it frees a struct url_spec.
195 * Providing parse_http_url() so that URLs can be parsed without faking a
196 * HTTP request line for parse_http_request() or repeating the parsing
197 * code (both of which were techniques that were actually in use).
199 * Standardizing that struct http_request is used to represent a URL, and
200 * struct url_spec is used to represent a URL pattern. (Before, URLs were
201 * represented as seperate variables and a partially-filled-in url_spec).
204 *********************************************************************/
211 #include <sys/types.h>
219 #if !defined(_WIN32) && !defined(__OS2__)
224 #include "urlmatch.h"
226 #include "miscutil.h"
229 const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
231 enum regex_anchoring {NO_ANCHORING, LEFT_ANCHORED, RIGHT_ANCHORED};
232 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern);
234 /*********************************************************************
236 * Function : free_http_request
238 * Description : Freez a http_request structure
241 * 1 : http = points to a http_request structure to free
245 *********************************************************************/
246 void free_http_request(struct http_request *http)
255 freez(http->hostport);
258 freez(http->host_ip_addr_str);
259 freez(http->dbuffer);
265 /*********************************************************************
267 * Function : init_domain_components
269 * Description : Splits the domain name so we can compare it
270 * against wildcards. It used to be part of
271 * parse_http_url, but was separated because the
272 * same code is required in chat in case of
273 * intercepted requests.
276 * 1 : http = pointer to the http structure to hold elements.
278 * Returns : JB_ERR_OK on success
279 * JB_ERR_MEMORY on out of memory
280 * JB_ERR_PARSE on malformed command/URL
281 * or >100 domains deep.
283 *********************************************************************/
284 jb_err init_domain_components(struct http_request *http)
286 char *vec[BUFFER_SIZE];
290 http->dbuffer = strdup(http->host);
291 if (NULL == http->dbuffer)
293 return JB_ERR_MEMORY;
296 /* map to lower case */
297 for (p = http->dbuffer; *p ; p++)
299 *p = (char)tolower((int)(unsigned char)*p);
302 /* split the domain name into components */
303 http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec), 1, 1);
305 if (http->dcount <= 0)
308 * Error: More than SZ(vec) components in domain
309 * or: no components in domain
311 log_error(LOG_LEVEL_ERROR, "More than SZ(vec) components in domain or none at all.");
315 /* save a copy of the pointers in dvec */
316 size = (size_t)http->dcount * sizeof(*http->dvec);
318 http->dvec = (char **)malloc(size);
319 if (NULL == http->dvec)
321 return JB_ERR_MEMORY;
324 memcpy(http->dvec, vec, size);
330 /*********************************************************************
332 * Function : parse_http_url
334 * Description : Parse out the host and port from the URL. Find the
335 * hostname & path, port (if ':'), and/or password (if '@')
338 * 1 : url = URL (or is it URI?) to break down
339 * 2 : http = pointer to the http structure to hold elements.
340 * Will be zeroed before use. Note that this
341 * function sets the http->gpc and http->ver
343 * 3 : csp = Current client state (buffers, headers, etc...)
345 * Returns : JB_ERR_OK on success
346 * JB_ERR_MEMORY on out of memory
347 * JB_ERR_PARSE on malformed command/URL
348 * or >100 domains deep.
350 *********************************************************************/
351 jb_err parse_http_url(const char * url,
352 struct http_request *http,
353 const struct client_state *csp)
355 int host_available = 1; /* A proxy can dream. */
358 * Zero out the results structure
360 memset(http, '\0', sizeof(*http));
364 * Save our initial URL
366 http->url = strdup(url);
367 if (http->url == NULL)
369 return JB_ERR_MEMORY;
374 * Check for * URI. If found, we're done.
376 if (*http->url == '*')
378 if ( NULL == (http->path = strdup("*"))
379 || NULL == (http->hostport = strdup("")) )
381 return JB_ERR_MEMORY;
383 if (http->url[1] != '\0')
392 * Split URL into protocol,hostport,path.
402 return JB_ERR_MEMORY;
405 /* Find the start of the URL in our scratch space */
407 if (strncmpic(url_noproto, "http://", 7) == 0)
412 else if (strncmpic(url_noproto, "https://", 8) == 0)
417 else if (*url_noproto == '/')
420 * Short request line without protocol and host.
421 * Most likely because the client's request
422 * was intercepted and redirected into Privoxy.
433 url_path = strchr(url_noproto, '/');
434 if (url_path != NULL)
439 * NOTE: The following line ignores the path for HTTPS URLS.
440 * This means that you get consistent behaviour if you type a
441 * https URL in and it's parsed by the function. (When the
442 * URL is actually retrieved, SSL hides the path part).
444 http->path = strdup(http->ssl ? "/" : url_path);
446 http->hostport = strdup(url_noproto);
451 * Repair broken HTTP requests that don't contain a path,
452 * or CONNECT requests
454 http->path = strdup("/");
455 http->hostport = strdup(url_noproto);
460 if ( (http->path == NULL)
461 || (http->hostport == NULL))
463 return JB_ERR_MEMORY;
469 /* Without host, there is nothing left to do here */
474 * Split hostport into user/password (ignored), host, port.
481 buf = strdup(http->hostport);
484 return JB_ERR_MEMORY;
487 /* check if url contains username and/or password */
488 host = strchr(buf, '@');
491 /* Contains username/password, skip it and the @ sign. */
496 /* No username or password. */
500 /* check if url contains port */
501 port = strchr(host, ':');
505 /* Terminate hostname and point to start of port string */
507 http->port = atoi(port);
511 /* No port specified. */
512 http->port = (http->ssl ? 443 : 80);
515 http->host = strdup(host);
519 if (http->host == NULL)
521 return JB_ERR_MEMORY;
526 * Split domain name so we can compare it against wildcards
528 return init_domain_components(http);
533 /*********************************************************************
535 * Function : unknown_method
537 * Description : Checks whether a method is unknown.
540 * 1 : method = points to a http method
542 * Returns : TRUE if it's unknown, FALSE otherwise.
544 *********************************************************************/
545 static int unknown_method(const char *method)
547 static const char *known_http_methods[] = {
548 /* Basic HTTP request type */
549 "GET", "HEAD", "POST", "PUT", "DELETE", "OPTIONS", "TRACE", "CONNECT",
550 /* webDAV extensions (RFC2518) */
551 "PROPFIND", "PROPPATCH", "MOVE", "COPY", "MKCOL", "LOCK", "UNLOCK",
553 * Microsoft webDAV extension for Exchange 2000. See:
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 "BCOPY", "BMOVE", "BDELETE", "BPROPFIND", "BPROPPATCH",
559 * Another Microsoft webDAV extension for Exchange 2000. See:
560 * http://systems.cs.colorado.edu/grunwald/MobileComputing/Papers/draft-cohen-gena-p-base-00.txt
561 * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
562 * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
564 "SUBSCRIBE", "UNSUBSCRIBE", "NOTIFY", "POLL",
566 * Yet another WebDAV extension, this time for
567 * Web Distributed Authoring and Versioning (RFC3253)
569 "VERSION-CONTROL", "REPORT", "CHECKOUT", "CHECKIN", "UNCHECKOUT",
570 "MKWORKSPACE", "UPDATE", "LABEL", "MERGE", "BASELINE-CONTROL", "MKACTIVITY",
574 for (i = 0; i < SZ(known_http_methods); i++)
576 if (0 == strcmpic(method, known_http_methods[i]))
587 /*********************************************************************
589 * Function : parse_http_request
591 * Description : Parse out the host and port from the URL. Find the
592 * hostname & path, port (if ':'), and/or password (if '@')
595 * 1 : req = HTTP request line to break down
596 * 2 : http = pointer to the http structure to hold elements
597 * 3 : csp = Current client state (buffers, headers, etc...)
599 * Returns : JB_ERR_OK on success
600 * JB_ERR_MEMORY on out of memory
601 * JB_ERR_CGI_PARAMS on malformed command/URL
602 * or >100 domains deep.
604 *********************************************************************/
605 jb_err parse_http_request(const char *req,
606 struct http_request *http,
607 const struct client_state *csp)
610 char *v[10]; /* XXX: Why 10? We should only need three. */
615 memset(http, '\0', sizeof(*http));
620 return JB_ERR_MEMORY;
623 n = ssplit(buf, " \r\n", v, SZ(v), 1, 1);
631 * Fail in case of unknown methods
632 * which we might not handle correctly.
634 * XXX: There should be a config option
635 * to forward requests with unknown methods
636 * anyway. Most of them don't need special
639 if (unknown_method(v[0]))
641 log_error(LOG_LEVEL_ERROR, "Unknown HTTP method detected: %s", v[0]);
646 if (strcmpic(v[0], "CONNECT") == 0)
651 err = parse_http_url(v[1], http, csp);
659 * Copy the details into the structure
661 http->ssl = is_connect;
662 http->cmd = strdup(req);
663 http->gpc = strdup(v[0]);
664 http->ver = strdup(v[2]);
666 if ( (http->cmd == NULL)
667 || (http->gpc == NULL)
668 || (http->ver == NULL) )
671 return JB_ERR_MEMORY;
680 /*********************************************************************
682 * Function : compile_pattern
684 * Description : Compiles a host, domain or TAG pattern.
687 * 1 : pattern = The pattern to compile.
688 * 2 : anchoring = How the regex should be anchored.
689 * Can be either one of NO_ANCHORING,
690 * LEFT_ANCHORED or RIGHT_ANCHORED.
691 * 3 : url = In case of failures, the spec member is
692 * logged and the structure freed.
693 * 4 : regex = Where the compiled regex should be stored.
695 * Returns : JB_ERR_OK - Success
696 * JB_ERR_MEMORY - Out of memory
697 * JB_ERR_PARSE - Cannot parse regex
699 *********************************************************************/
700 static jb_err compile_pattern(const char *pattern, enum regex_anchoring anchoring,
701 struct url_spec *url, regex_t **regex)
704 char rebuf[BUFFER_SIZE];
708 assert(strlen(pattern) < sizeof(rebuf) - 2);
710 if (pattern[0] == '\0')
728 log_error(LOG_LEVEL_FATAL,
729 "Invalid anchoring in compile_pattern %d", anchoring);
732 *regex = zalloc(sizeof(**regex));
736 return JB_ERR_MEMORY;
739 snprintf(rebuf, sizeof(rebuf), fmt, pattern);
741 errcode = regcomp(*regex, rebuf, (REG_EXTENDED|REG_NOSUB|REG_ICASE));
745 size_t errlen = regerror(errcode, *regex, rebuf, sizeof(rebuf));
746 if (errlen > (sizeof(rebuf) - (size_t)1))
748 errlen = sizeof(rebuf) - (size_t)1;
750 rebuf[errlen] = '\0';
751 log_error(LOG_LEVEL_ERROR, "error compiling %s from %s: %s",
752 pattern, url->spec, rebuf);
763 /*********************************************************************
765 * Function : compile_url_pattern
767 * Description : Compiles the three parts of an URL pattern.
770 * 1 : url = Target url_spec to be filled in.
771 * 2 : buf = The url pattern to compile. Will be messed up.
773 * Returns : JB_ERR_OK - Success
774 * JB_ERR_MEMORY - Out of memory
775 * JB_ERR_PARSE - Cannot parse regex
777 *********************************************************************/
778 static jb_err compile_url_pattern(struct url_spec *url, char *buf)
782 p = strchr(buf, '/');
786 * Only compile the regex if it consists of more than
787 * a single slash, otherwise it wouldn't affect the result.
792 * XXX: does it make sense to compile the slash at the beginning?
794 jb_err err = compile_pattern(p, LEFT_ANCHORED, url, &url->preg);
796 if (JB_ERR_OK != err)
804 p = strchr(buf, ':');
808 url->port_list = strdup(p);
809 if (NULL == url->port_list)
811 return JB_ERR_MEMORY;
816 url->port_list = NULL;
821 return compile_host_pattern(url, buf);
829 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
830 /*********************************************************************
832 * Function : compile_host_pattern
834 * Description : Parses and compiles a host pattern..
837 * 1 : url = Target url_spec to be filled in.
838 * 2 : host_pattern = Host pattern to compile.
840 * Returns : JB_ERR_OK - Success
841 * JB_ERR_MEMORY - Out of memory
842 * JB_ERR_PARSE - Cannot parse regex
844 *********************************************************************/
845 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
847 return compile_pattern(host_pattern, RIGHT_ANCHORED, url, &url->host_regex);
852 /*********************************************************************
854 * Function : compile_host_pattern
856 * Description : Parses and "compiles" an old-school host pattern.
859 * 1 : url = Target url_spec to be filled in.
860 * 2 : host_pattern = Host pattern to parse.
862 * Returns : JB_ERR_OK - Success
863 * JB_ERR_MEMORY - Out of memory
864 * JB_ERR_PARSE - Cannot parse regex
866 *********************************************************************/
867 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
876 if (host_pattern[strlen(host_pattern) - 1] == '.')
878 url->unanchored |= ANCHOR_RIGHT;
880 if (host_pattern[0] == '.')
882 url->unanchored |= ANCHOR_LEFT;
886 * Split domain into components
888 url->dbuffer = strdup(host_pattern);
889 if (NULL == url->dbuffer)
892 return JB_ERR_MEMORY;
898 for (p = url->dbuffer; *p ; p++)
900 *p = (char)tolower((int)(unsigned char)*p);
904 * Split the domain name into components
906 url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1);
911 return JB_ERR_MEMORY;
913 else if (url->dcount != 0)
916 * Save a copy of the pointers in dvec
918 size = (size_t)url->dcount * sizeof(*url->dvec);
920 url->dvec = (char **)malloc(size);
921 if (NULL == url->dvec)
924 return JB_ERR_MEMORY;
927 memcpy(url->dvec, v, size);
930 * else dcount == 0 in which case we needn't do anything,
931 * since dvec will never be accessed and the pattern will
938 /*********************************************************************
940 * Function : simple_domaincmp
942 * Description : Domain-wise Compare fqdn's. The comparison is
943 * both left- and right-anchored. The individual
944 * domain names are compared with simplematch().
945 * This is only used by domain_match.
948 * 1 : pv = array of patterns to compare
949 * 2 : fv = array of domain components to compare
950 * 3 : len = length of the arrays (both arrays are the
951 * same length - if they weren't, it couldn't
952 * possibly be a match).
954 * Returns : 0 => domains are equivalent, else no match.
956 *********************************************************************/
957 static int simple_domaincmp(char **pv, char **fv, int len)
961 for (n = 0; n < len; n++)
963 if (simplematch(pv[n], fv[n]))
974 /*********************************************************************
976 * Function : domain_match
978 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
979 * pattern->unachored, the comparison is un-, left-,
980 * right-anchored, or both.
981 * The individual domain names are compared with
985 * 1 : pattern = a domain that may contain a '*' as a wildcard.
986 * 2 : fqdn = domain name against which the patterns are compared.
988 * Returns : 0 => domains are equivalent, else no match.
990 *********************************************************************/
991 static int domain_match(const struct url_spec *pattern, const struct http_request *fqdn)
993 char **pv, **fv; /* vectors */
995 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
997 plen = pattern->dcount;
1002 /* fqdn is too short to match this pattern */
1009 if (unanchored == ANCHOR_LEFT)
1014 * Convert this into a fully anchored pattern with
1015 * the fqdn and pattern the same length
1017 fv += (flen - plen); /* flen - plen >= 0 due to check above */
1018 return simple_domaincmp(pv, fv, plen);
1020 else if (unanchored == 0)
1022 /* Fully anchored, check length */
1027 return simple_domaincmp(pv, fv, plen);
1029 else if (unanchored == ANCHOR_RIGHT)
1031 /* Left anchored, ignore all extra in fqdn */
1032 return simple_domaincmp(pv, fv, plen);
1038 int maxn = flen - plen;
1039 for (n = 0; n <= maxn; n++)
1041 if (!simple_domaincmp(pv, fv, plen))
1046 * Doesn't match from start of fqdn
1047 * Try skipping first part of fqdn
1055 #endif /* def FEATURE_EXTENDED_HOST_PATTERNS */
1058 /*********************************************************************
1060 * Function : create_url_spec
1062 * Description : Creates a "url_spec" structure from a string.
1063 * When finished, free with free_url_spec().
1066 * 1 : url = Target url_spec to be filled in. Will be
1067 * zeroed before use.
1068 * 2 : buf = Source pattern, null terminated. NOTE: The
1069 * contents of this buffer are destroyed by this
1070 * function. If this function succeeds, the
1071 * buffer is copied to url->spec. If this
1072 * function fails, the contents of the buffer
1073 * are lost forever. XXX: Why is this const?
1075 * Returns : JB_ERR_OK - Success
1076 * JB_ERR_MEMORY - Out of memory
1077 * JB_ERR_PARSE - Cannot parse regex (Detailed message
1078 * written to system log)
1080 *********************************************************************/
1081 jb_err create_url_spec(struct url_spec * url, const char * buf)
1086 memset(url, '\0', sizeof(*url));
1088 /* Remember the original specification for the CGI pages. */
1089 url->spec = strdup(buf);
1090 if (NULL == url->spec)
1092 return JB_ERR_MEMORY;
1095 /* Is it tag pattern? */
1096 if (0 == strncmpic("TAG:", url->spec, 4))
1098 /* The pattern starts with the first character after "TAG:" */
1099 const char *tag_pattern = buf + 4;
1100 return compile_pattern(tag_pattern, NO_ANCHORING, url, &url->tag_regex);
1103 /* If it isn't a tag pattern it must be a URL pattern. */
1104 return compile_url_pattern(url, (char *)buf);
1108 /*********************************************************************
1110 * Function : free_url_spec
1112 * Description : Called from the "unloaders". Freez the url
1113 * structure elements.
1116 * 1 : url = pointer to a url_spec structure.
1120 *********************************************************************/
1121 void free_url_spec(struct url_spec *url)
1123 if (url == NULL) return;
1126 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
1127 if (url->host_regex)
1129 regfree(url->host_regex);
1130 freez(url->host_regex);
1133 freez(url->dbuffer);
1136 #endif /* ndef FEATURE_EXTENDED_HOST_PATTERNS */
1137 freez(url->port_list);
1145 regfree(url->tag_regex);
1146 freez(url->tag_regex);
1151 /*********************************************************************
1153 * Function : url_match
1155 * Description : Compare a URL against a URL pattern.
1158 * 1 : pattern = a URL pattern
1159 * 2 : url = URL to match
1161 * Returns : Nonzero if the URL matches the pattern, else 0.
1163 *********************************************************************/
1164 int url_match(const struct url_spec *pattern,
1165 const struct http_request *http)
1167 /* XXX: these should probably be functions. */
1168 #define PORT_MATCHES ((NULL == pattern->port_list) || match_portlist(pattern->port_list, http->port))
1169 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
1170 #define DOMAIN_MATCHES ((NULL == pattern->host_regex) || (0 == regexec(pattern->host_regex, http->host, 0, NULL, 0)))
1172 #define DOMAIN_MATCHES ((NULL == pattern->dbuffer) || (0 == domain_match(pattern, http)))
1174 #define PATH_MATCHES ((NULL == pattern->preg) || (0 == regexec(pattern->preg, http->path, 0, NULL, 0)))
1176 if (pattern->tag_regex != NULL)
1178 /* It's a tag pattern and shouldn't be matched against URLs */
1182 return (PORT_MATCHES && DOMAIN_MATCHES && PATH_MATCHES);
1187 /*********************************************************************
1189 * Function : match_portlist
1191 * Description : Check if a given number is covered by a comma
1192 * separated list of numbers and ranges (a,b-c,d,..)
1195 * 1 : portlist = String with list
1196 * 2 : port = port to check
1198 * Returns : 0 => no match
1201 *********************************************************************/
1202 int match_portlist(const char *portlist, int port)
1204 char *min, *max, *next, *portlist_copy;
1206 min = next = portlist_copy = strdup(portlist);
1209 * Zero-terminate first item and remember offset for next
1211 if (NULL != (next = strchr(portlist_copy, (int) ',')))
1217 * Loop through all items, checking for match
1221 if (NULL == (max = strchr(min, (int) '-')))
1224 * No dash, check for equality
1226 if (port == atoi(min))
1228 free(portlist_copy);
1235 * This is a range, so check if between min and max,
1236 * or, if max was omitted, between min and 65K
1239 if(port >= atoi(min) && port <= (atoi(max) ? atoi(max) : 65535))
1241 free(portlist_copy);
1253 * Zero-terminate next item and remember offset for n+1
1255 if ((NULL != next) && (NULL != (next = strchr(next, (int) ','))))
1261 free(portlist_copy);