1 const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.37 2008/04/17 14:53:29 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 1.37 2008/04/17 14:53:29 fabiankeil
37 * Move simplematch() into urlmatch.c as it's only
38 * used to match (old-school) domain patterns.
40 * Revision 1.36 2008/04/14 18:19:48 fabiankeil
41 * Remove now-pointless cast in create_url_spec().
43 * Revision 1.35 2008/04/14 18:11:21 fabiankeil
44 * The compiler might not notice it, but the buffer passed to
45 * create_url_spec() is modified later on and thus shouldn't
46 * be declared immutable.
48 * Revision 1.34 2008/04/13 13:32:07 fabiankeil
49 * Factor URL pattern compilation out of create_url_spec().
51 * Revision 1.33 2008/04/12 14:03:13 fabiankeil
52 * Remove an obvious comment and improve another one.
54 * Revision 1.32 2008/04/12 12:38:06 fabiankeil
55 * Factor out duplicated code to compile host, path and tag patterns.
57 * Revision 1.31 2008/04/10 14:41:04 fabiankeil
58 * Ditch url_spec's path member now that it's no longer used.
60 * Revision 1.30 2008/04/10 04:24:24 fabiankeil
61 * Stop duplicating the plain text representation of the path regex
62 * (and keeping the copy around). Once the regex is compiled it's no
65 * Revision 1.29 2008/04/10 04:17:56 fabiankeil
66 * In url_match(), check the right member for NULL when determining
67 * whether there's a path regex to execute. Looking for a plain-text
68 * representation works as well, but it looks "interesting" and that
69 * member will be removed soonish anyway.
71 * Revision 1.28 2008/04/08 16:07:39 fabiankeil
72 * Make it harder to mistake url_match()'s
73 * second parameter for an url_spec.
75 * Revision 1.27 2008/04/08 15:44:33 fabiankeil
76 * Save a bit of memory (and a few cpu cycles) by not bothering to
77 * compile slash-only path regexes that don't affect the result.
79 * Revision 1.26 2008/04/07 16:57:18 fabiankeil
80 * - Use free_url_spec() more consistently.
81 * - Let it reset url->dcount just in case.
83 * Revision 1.25 2008/04/06 15:18:38 fabiankeil
84 * Oh well, rename the --enable-pcre-host-patterns option to
85 * --enable-extended-host-patterns as it's not really PCRE syntax.
87 * Revision 1.24 2008/04/06 14:54:26 fabiankeil
88 * Use PCRE syntax in host patterns when configured
89 * with --enable-pcre-host-patterns.
91 * Revision 1.23 2008/04/05 12:19:20 fabiankeil
92 * Factor compile_host_pattern() out of create_url_spec().
94 * Revision 1.22 2008/03/30 15:02:32 fabiankeil
95 * SZitify unknown_method().
97 * Revision 1.21 2007/12/24 16:34:23 fabiankeil
98 * Band-aid (and micro-optimization) that makes it less likely to run out of
99 * stack space with overly-complex path patterns. Probably masks the problem
100 * reported by Lee in #1856679. Hohoho.
102 * Revision 1.20 2007/09/02 15:31:20 fabiankeil
103 * Move match_portlist() from filter.c to urlmatch.c.
104 * It's used for url matching, not for filtering.
106 * Revision 1.19 2007/09/02 13:42:11 fabiankeil
107 * - Allow port lists in url patterns.
108 * - Ditch unused url_spec member pathlen.
110 * Revision 1.18 2007/07/30 16:42:21 fabiankeil
111 * Move the method check into unknown_method()
112 * and loop through the known methods instead
113 * of using a screen-long OR chain.
115 * Revision 1.17 2007/04/15 16:39:21 fabiankeil
116 * Introduce tags as alternative way to specify which
117 * actions apply to a request. At the moment tags can be
118 * created based on client and server headers.
120 * Revision 1.16 2007/02/13 13:59:24 fabiankeil
121 * Remove redundant log message.
123 * Revision 1.15 2007/01/28 16:11:23 fabiankeil
124 * Accept WebDAV methods for subversion
125 * in parse_http_request(). Closes FR 1581425.
127 * Revision 1.14 2007/01/06 14:23:56 fabiankeil
128 * Fix gcc43 warnings. Mark *csp as immutable
129 * for parse_http_url() and url_match().
130 * Replace a sprintf call with snprintf.
132 * Revision 1.13 2006/12/06 19:50:54 fabiankeil
133 * parse_http_url() now handles intercepted
134 * HTTP request lines as well. Moved parts
135 * of parse_http_url()'s code into
136 * init_domain_components() so that it can
137 * be reused in chat().
139 * Revision 1.12 2006/07/18 14:48:47 david__schmidt
140 * Reorganizing the repository: swapping out what was HEAD (the old 3.1 branch)
141 * with what was really the latest development (the v_3_0_branch branch)
143 * Revision 1.10.2.7 2003/05/17 15:57:24 oes
144 * - parse_http_url now checks memory allocation failure for
145 * duplication of "*" URL and rejects "*something" URLs
147 * - Added a comment to what might look like a bug in
148 * create_url_spec (see !bug #736931)
149 * - Comment cosmetics
151 * Revision 1.10.2.6 2003/05/07 12:39:48 oes
152 * Fix typo: Default port for https URLs is 443, not 143.
153 * Thanks to Scott Tregear for spotting this one.
155 * Revision 1.10.2.5 2003/02/28 13:09:29 oes
156 * Fixed a rare double free condition as per Bug #694713
158 * Revision 1.10.2.4 2003/02/28 12:57:44 oes
159 * Moved freeing of http request structure to its owner
160 * as per Dan Price's observations in Bug #694713
162 * Revision 1.10.2.3 2002/11/12 16:50:40 oes
163 * Fixed memory leak in parse_http_request() reported by Oliver Stoeneberg. Fixes bug #637073
165 * Revision 1.10.2.2 2002/09/25 14:53:15 oes
166 * Added basic support for OPTIONS and TRACE HTTP methods:
167 * parse_http_url now recognizes the "*" URI as well as
168 * the OPTIONS and TRACE method keywords.
170 * Revision 1.10.2.1 2002/06/06 19:06:44 jongfoster
171 * Adding support for proprietary Microsoft WebDAV extensions
173 * Revision 1.10 2002/05/12 21:40:37 jongfoster
174 * - Removing some unused code
176 * Revision 1.9 2002/04/04 00:36:36 gliptak
177 * always use pcre for matching
179 * Revision 1.8 2002/04/03 23:32:47 jongfoster
180 * Fixing memory leak on error
182 * Revision 1.7 2002/03/26 22:29:55 swa
183 * we have a new homepage!
185 * Revision 1.6 2002/03/24 13:25:43 swa
186 * name change related issues
188 * Revision 1.5 2002/03/13 00:27:05 jongfoster
191 * Revision 1.4 2002/03/07 03:46:17 oes
192 * Fixed compiler warnings
194 * Revision 1.3 2002/03/03 14:51:11 oes
195 * Fixed CLF logging: Added ocmd member for client's request to struct http_request
197 * Revision 1.2 2002/01/21 00:14:09 jongfoster
198 * Correcting comment style
199 * Fixing an uninitialized memory bug in create_url_spec()
201 * Revision 1.1 2002/01/17 20:53:46 jongfoster
202 * Moving all our URL and URL pattern parsing code to the same file - it
203 * was scattered around in filters.c, loaders.c and parsers.c.
205 * Providing a single, simple url_match(pattern,url) function - rather than
206 * the 3-line match routine which was repeated all over the place.
208 * Renaming free_url to free_url_spec, since it frees a struct url_spec.
210 * Providing parse_http_url() so that URLs can be parsed without faking a
211 * HTTP request line for parse_http_request() or repeating the parsing
212 * code (both of which were techniques that were actually in use).
214 * Standardizing that struct http_request is used to represent a URL, and
215 * struct url_spec is used to represent a URL pattern. (Before, URLs were
216 * represented as seperate variables and a partially-filled-in url_spec).
219 *********************************************************************/
226 #include <sys/types.h>
234 #if !defined(_WIN32) && !defined(__OS2__)
239 #include "urlmatch.h"
241 #include "miscutil.h"
244 const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
246 enum regex_anchoring {NO_ANCHORING, LEFT_ANCHORED, RIGHT_ANCHORED};
247 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern);
249 /*********************************************************************
251 * Function : free_http_request
253 * Description : Freez a http_request structure
256 * 1 : http = points to a http_request structure to free
260 *********************************************************************/
261 void free_http_request(struct http_request *http)
270 freez(http->hostport);
273 freez(http->host_ip_addr_str);
274 freez(http->dbuffer);
280 /*********************************************************************
282 * Function : init_domain_components
284 * Description : Splits the domain name so we can compare it
285 * against wildcards. It used to be part of
286 * parse_http_url, but was separated because the
287 * same code is required in chat in case of
288 * intercepted requests.
291 * 1 : http = pointer to the http structure to hold elements.
293 * Returns : JB_ERR_OK on success
294 * JB_ERR_MEMORY on out of memory
295 * JB_ERR_PARSE on malformed command/URL
296 * or >100 domains deep.
298 *********************************************************************/
299 jb_err init_domain_components(struct http_request *http)
301 char *vec[BUFFER_SIZE];
305 http->dbuffer = strdup(http->host);
306 if (NULL == http->dbuffer)
308 return JB_ERR_MEMORY;
311 /* map to lower case */
312 for (p = http->dbuffer; *p ; p++)
314 *p = (char)tolower((int)(unsigned char)*p);
317 /* split the domain name into components */
318 http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec), 1, 1);
320 if (http->dcount <= 0)
323 * Error: More than SZ(vec) components in domain
324 * or: no components in domain
326 log_error(LOG_LEVEL_ERROR, "More than SZ(vec) components in domain or none at all.");
330 /* save a copy of the pointers in dvec */
331 size = (size_t)http->dcount * sizeof(*http->dvec);
333 http->dvec = (char **)malloc(size);
334 if (NULL == http->dvec)
336 return JB_ERR_MEMORY;
339 memcpy(http->dvec, vec, size);
345 /*********************************************************************
347 * Function : parse_http_url
349 * Description : Parse out the host and port from the URL. Find the
350 * hostname & path, port (if ':'), and/or password (if '@')
353 * 1 : url = URL (or is it URI?) to break down
354 * 2 : http = pointer to the http structure to hold elements.
355 * Will be zeroed before use. Note that this
356 * function sets the http->gpc and http->ver
358 * 3 : csp = Current client state (buffers, headers, etc...)
360 * Returns : JB_ERR_OK on success
361 * JB_ERR_MEMORY on out of memory
362 * JB_ERR_PARSE on malformed command/URL
363 * or >100 domains deep.
365 *********************************************************************/
366 jb_err parse_http_url(const char * url,
367 struct http_request *http,
368 const struct client_state *csp)
370 int host_available = 1; /* A proxy can dream. */
373 * Zero out the results structure
375 memset(http, '\0', sizeof(*http));
379 * Save our initial URL
381 http->url = strdup(url);
382 if (http->url == NULL)
384 return JB_ERR_MEMORY;
389 * Check for * URI. If found, we're done.
391 if (*http->url == '*')
393 if ( NULL == (http->path = strdup("*"))
394 || NULL == (http->hostport = strdup("")) )
396 return JB_ERR_MEMORY;
398 if (http->url[1] != '\0')
407 * Split URL into protocol,hostport,path.
417 return JB_ERR_MEMORY;
420 /* Find the start of the URL in our scratch space */
422 if (strncmpic(url_noproto, "http://", 7) == 0)
427 else if (strncmpic(url_noproto, "https://", 8) == 0)
432 else if (*url_noproto == '/')
435 * Short request line without protocol and host.
436 * Most likely because the client's request
437 * was intercepted and redirected into Privoxy.
448 url_path = strchr(url_noproto, '/');
449 if (url_path != NULL)
454 * NOTE: The following line ignores the path for HTTPS URLS.
455 * This means that you get consistent behaviour if you type a
456 * https URL in and it's parsed by the function. (When the
457 * URL is actually retrieved, SSL hides the path part).
459 http->path = strdup(http->ssl ? "/" : url_path);
461 http->hostport = strdup(url_noproto);
466 * Repair broken HTTP requests that don't contain a path,
467 * or CONNECT requests
469 http->path = strdup("/");
470 http->hostport = strdup(url_noproto);
475 if ( (http->path == NULL)
476 || (http->hostport == NULL))
478 return JB_ERR_MEMORY;
484 /* Without host, there is nothing left to do here */
489 * Split hostport into user/password (ignored), host, port.
496 buf = strdup(http->hostport);
499 return JB_ERR_MEMORY;
502 /* check if url contains username and/or password */
503 host = strchr(buf, '@');
506 /* Contains username/password, skip it and the @ sign. */
511 /* No username or password. */
515 /* check if url contains port */
516 port = strchr(host, ':');
520 /* Terminate hostname and point to start of port string */
522 http->port = atoi(port);
526 /* No port specified. */
527 http->port = (http->ssl ? 443 : 80);
530 http->host = strdup(host);
534 if (http->host == NULL)
536 return JB_ERR_MEMORY;
541 * Split domain name so we can compare it against wildcards
543 return init_domain_components(http);
548 /*********************************************************************
550 * Function : unknown_method
552 * Description : Checks whether a method is unknown.
555 * 1 : method = points to a http method
557 * Returns : TRUE if it's unknown, FALSE otherwise.
559 *********************************************************************/
560 static int unknown_method(const char *method)
562 static const char *known_http_methods[] = {
563 /* Basic HTTP request type */
564 "GET", "HEAD", "POST", "PUT", "DELETE", "OPTIONS", "TRACE", "CONNECT",
565 /* webDAV extensions (RFC2518) */
566 "PROPFIND", "PROPPATCH", "MOVE", "COPY", "MKCOL", "LOCK", "UNLOCK",
568 * Microsoft webDAV extension for Exchange 2000. See:
569 * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
570 * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
572 "BCOPY", "BMOVE", "BDELETE", "BPROPFIND", "BPROPPATCH",
574 * Another Microsoft webDAV extension for Exchange 2000. See:
575 * http://systems.cs.colorado.edu/grunwald/MobileComputing/Papers/draft-cohen-gena-p-base-00.txt
576 * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
577 * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
579 "SUBSCRIBE", "UNSUBSCRIBE", "NOTIFY", "POLL",
581 * Yet another WebDAV extension, this time for
582 * Web Distributed Authoring and Versioning (RFC3253)
584 "VERSION-CONTROL", "REPORT", "CHECKOUT", "CHECKIN", "UNCHECKOUT",
585 "MKWORKSPACE", "UPDATE", "LABEL", "MERGE", "BASELINE-CONTROL", "MKACTIVITY",
589 for (i = 0; i < SZ(known_http_methods); i++)
591 if (0 == strcmpic(method, known_http_methods[i]))
602 /*********************************************************************
604 * Function : parse_http_request
606 * Description : Parse out the host and port from the URL. Find the
607 * hostname & path, port (if ':'), and/or password (if '@')
610 * 1 : req = HTTP request line to break down
611 * 2 : http = pointer to the http structure to hold elements
612 * 3 : csp = Current client state (buffers, headers, etc...)
614 * Returns : JB_ERR_OK on success
615 * JB_ERR_MEMORY on out of memory
616 * JB_ERR_CGI_PARAMS on malformed command/URL
617 * or >100 domains deep.
619 *********************************************************************/
620 jb_err parse_http_request(const char *req,
621 struct http_request *http,
622 const struct client_state *csp)
625 char *v[10]; /* XXX: Why 10? We should only need three. */
630 memset(http, '\0', sizeof(*http));
635 return JB_ERR_MEMORY;
638 n = ssplit(buf, " \r\n", v, SZ(v), 1, 1);
646 * Fail in case of unknown methods
647 * which we might not handle correctly.
649 * XXX: There should be a config option
650 * to forward requests with unknown methods
651 * anyway. Most of them don't need special
654 if (unknown_method(v[0]))
656 log_error(LOG_LEVEL_ERROR, "Unknown HTTP method detected: %s", v[0]);
661 if (strcmpic(v[0], "CONNECT") == 0)
666 err = parse_http_url(v[1], http, csp);
674 * Copy the details into the structure
676 http->ssl = is_connect;
677 http->cmd = strdup(req);
678 http->gpc = strdup(v[0]);
679 http->ver = strdup(v[2]);
681 if ( (http->cmd == NULL)
682 || (http->gpc == NULL)
683 || (http->ver == NULL) )
686 return JB_ERR_MEMORY;
695 /*********************************************************************
697 * Function : compile_pattern
699 * Description : Compiles a host, domain or TAG pattern.
702 * 1 : pattern = The pattern to compile.
703 * 2 : anchoring = How the regex should be anchored.
704 * Can be either one of NO_ANCHORING,
705 * LEFT_ANCHORED or RIGHT_ANCHORED.
706 * 3 : url = In case of failures, the spec member is
707 * logged and the structure freed.
708 * 4 : regex = Where the compiled regex should be stored.
710 * Returns : JB_ERR_OK - Success
711 * JB_ERR_MEMORY - Out of memory
712 * JB_ERR_PARSE - Cannot parse regex
714 *********************************************************************/
715 static jb_err compile_pattern(const char *pattern, enum regex_anchoring anchoring,
716 struct url_spec *url, regex_t **regex)
719 char rebuf[BUFFER_SIZE];
723 assert(strlen(pattern) < sizeof(rebuf) - 2);
725 if (pattern[0] == '\0')
743 log_error(LOG_LEVEL_FATAL,
744 "Invalid anchoring in compile_pattern %d", anchoring);
747 *regex = zalloc(sizeof(**regex));
751 return JB_ERR_MEMORY;
754 snprintf(rebuf, sizeof(rebuf), fmt, pattern);
756 errcode = regcomp(*regex, rebuf, (REG_EXTENDED|REG_NOSUB|REG_ICASE));
760 size_t errlen = regerror(errcode, *regex, rebuf, sizeof(rebuf));
761 if (errlen > (sizeof(rebuf) - (size_t)1))
763 errlen = sizeof(rebuf) - (size_t)1;
765 rebuf[errlen] = '\0';
766 log_error(LOG_LEVEL_ERROR, "error compiling %s from %s: %s",
767 pattern, url->spec, rebuf);
778 /*********************************************************************
780 * Function : compile_url_pattern
782 * Description : Compiles the three parts of an URL pattern.
785 * 1 : url = Target url_spec to be filled in.
786 * 2 : buf = The url pattern to compile. Will be messed up.
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_url_pattern(struct url_spec *url, char *buf)
797 p = strchr(buf, '/');
801 * Only compile the regex if it consists of more than
802 * a single slash, otherwise it wouldn't affect the result.
807 * XXX: does it make sense to compile the slash at the beginning?
809 jb_err err = compile_pattern(p, LEFT_ANCHORED, url, &url->preg);
811 if (JB_ERR_OK != err)
819 p = strchr(buf, ':');
823 url->port_list = strdup(p);
824 if (NULL == url->port_list)
826 return JB_ERR_MEMORY;
831 url->port_list = NULL;
836 return compile_host_pattern(url, buf);
844 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
845 /*********************************************************************
847 * Function : compile_host_pattern
849 * Description : Parses and compiles a host pattern..
852 * 1 : url = Target url_spec to be filled in.
853 * 2 : host_pattern = Host pattern to compile.
855 * Returns : JB_ERR_OK - Success
856 * JB_ERR_MEMORY - Out of memory
857 * JB_ERR_PARSE - Cannot parse regex
859 *********************************************************************/
860 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
862 return compile_pattern(host_pattern, RIGHT_ANCHORED, url, &url->host_regex);
867 /*********************************************************************
869 * Function : compile_host_pattern
871 * Description : Parses and "compiles" an old-school host pattern.
874 * 1 : url = Target url_spec to be filled in.
875 * 2 : host_pattern = Host pattern to parse.
877 * Returns : JB_ERR_OK - Success
878 * JB_ERR_MEMORY - Out of memory
879 * JB_ERR_PARSE - Cannot parse regex
881 *********************************************************************/
882 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
891 if (host_pattern[strlen(host_pattern) - 1] == '.')
893 url->unanchored |= ANCHOR_RIGHT;
895 if (host_pattern[0] == '.')
897 url->unanchored |= ANCHOR_LEFT;
901 * Split domain into components
903 url->dbuffer = strdup(host_pattern);
904 if (NULL == url->dbuffer)
907 return JB_ERR_MEMORY;
913 for (p = url->dbuffer; *p ; p++)
915 *p = (char)tolower((int)(unsigned char)*p);
919 * Split the domain name into components
921 url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1);
926 return JB_ERR_MEMORY;
928 else if (url->dcount != 0)
931 * Save a copy of the pointers in dvec
933 size = (size_t)url->dcount * sizeof(*url->dvec);
935 url->dvec = (char **)malloc(size);
936 if (NULL == url->dvec)
939 return JB_ERR_MEMORY;
942 memcpy(url->dvec, v, size);
945 * else dcount == 0 in which case we needn't do anything,
946 * since dvec will never be accessed and the pattern will
953 /*********************************************************************
955 * Function : simplematch
957 * Description : String matching, with a (greedy) '*' wildcard that
958 * stands for zero or more arbitrary characters and
959 * character classes in [], which take both enumerations
963 * 1 : pattern = pattern for matching
964 * 2 : text = text to be matched
966 * Returns : 0 if match, else nonzero
968 *********************************************************************/
969 static int simplematch(const char *pattern, const char *text)
971 const unsigned char *pat = (const unsigned char *)pattern;
972 const unsigned char *txt = (const unsigned char *)text;
973 const unsigned char *fallback = pat;
976 unsigned char lastchar = 'a';
978 unsigned char charmap[32];
983 /* EOF pattern but !EOF text? */
996 /* '*' in the pattern? */
1000 /* The pattern ends afterwards? Speed up the return. */
1006 /* Else, set wildcard mode and remember position after '*' */
1011 /* Character range specification? */
1014 memset(charmap, '\0', sizeof(charmap));
1016 while (*++pat != ']')
1022 else if (*pat == '-')
1024 if ((*++pat == ']') || *pat == '\0')
1028 for (i = lastchar; i <= *pat; i++)
1030 charmap[i / 8] |= (unsigned char)(1 << (i % 8));
1035 charmap[*pat / 8] |= (unsigned char)(1 << (*pat % 8));
1039 } /* -END- if Character range specification */
1043 * Char match, or char range match?
1047 || ((*pat == ']') && (charmap[*txt / 8] & (1 << (*txt % 8)))) )
1057 * No match && no wildcard: No luck
1061 else if (pat != fallback)
1064 * Increment text pointer if in char range matching
1071 * Wildcard mode && nonmatch beyond fallback: Rewind pattern
1075 * Restart matching from current text pointer
1082 /* Cut off extra '*'s */
1083 if(*pat == '*') pat++;
1085 /* If this is the pattern's end, fine! */
1091 /*********************************************************************
1093 * Function : simple_domaincmp
1095 * Description : Domain-wise Compare fqdn's. The comparison is
1096 * both left- and right-anchored. The individual
1097 * domain names are compared with simplematch().
1098 * This is only used by domain_match.
1101 * 1 : pv = array of patterns to compare
1102 * 2 : fv = array of domain components to compare
1103 * 3 : len = length of the arrays (both arrays are the
1104 * same length - if they weren't, it couldn't
1105 * possibly be a match).
1107 * Returns : 0 => domains are equivalent, else no match.
1109 *********************************************************************/
1110 static int simple_domaincmp(char **pv, char **fv, int len)
1114 for (n = 0; n < len; n++)
1116 if (simplematch(pv[n], fv[n]))
1127 /*********************************************************************
1129 * Function : domain_match
1131 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
1132 * pattern->unachored, the comparison is un-, left-,
1133 * right-anchored, or both.
1134 * The individual domain names are compared with
1138 * 1 : pattern = a domain that may contain a '*' as a wildcard.
1139 * 2 : fqdn = domain name against which the patterns are compared.
1141 * Returns : 0 => domains are equivalent, else no match.
1143 *********************************************************************/
1144 static int domain_match(const struct url_spec *pattern, const struct http_request *fqdn)
1146 char **pv, **fv; /* vectors */
1148 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
1150 plen = pattern->dcount;
1151 flen = fqdn->dcount;
1155 /* fqdn is too short to match this pattern */
1162 if (unanchored == ANCHOR_LEFT)
1167 * Convert this into a fully anchored pattern with
1168 * the fqdn and pattern the same length
1170 fv += (flen - plen); /* flen - plen >= 0 due to check above */
1171 return simple_domaincmp(pv, fv, plen);
1173 else if (unanchored == 0)
1175 /* Fully anchored, check length */
1180 return simple_domaincmp(pv, fv, plen);
1182 else if (unanchored == ANCHOR_RIGHT)
1184 /* Left anchored, ignore all extra in fqdn */
1185 return simple_domaincmp(pv, fv, plen);
1191 int maxn = flen - plen;
1192 for (n = 0; n <= maxn; n++)
1194 if (!simple_domaincmp(pv, fv, plen))
1199 * Doesn't match from start of fqdn
1200 * Try skipping first part of fqdn
1208 #endif /* def FEATURE_EXTENDED_HOST_PATTERNS */
1211 /*********************************************************************
1213 * Function : create_url_spec
1215 * Description : Creates a "url_spec" structure from a string.
1216 * When finished, free with free_url_spec().
1219 * 1 : url = Target url_spec to be filled in. Will be
1220 * zeroed before use.
1221 * 2 : buf = Source pattern, null terminated. NOTE: The
1222 * contents of this buffer are destroyed by this
1223 * function. If this function succeeds, the
1224 * buffer is copied to url->spec. If this
1225 * function fails, the contents of the buffer
1228 * Returns : JB_ERR_OK - Success
1229 * JB_ERR_MEMORY - Out of memory
1230 * JB_ERR_PARSE - Cannot parse regex (Detailed message
1231 * written to system log)
1233 *********************************************************************/
1234 jb_err create_url_spec(struct url_spec *url, char *buf)
1239 memset(url, '\0', sizeof(*url));
1241 /* Remember the original specification for the CGI pages. */
1242 url->spec = strdup(buf);
1243 if (NULL == url->spec)
1245 return JB_ERR_MEMORY;
1248 /* Is it tag pattern? */
1249 if (0 == strncmpic("TAG:", url->spec, 4))
1251 /* The pattern starts with the first character after "TAG:" */
1252 const char *tag_pattern = buf + 4;
1253 return compile_pattern(tag_pattern, NO_ANCHORING, url, &url->tag_regex);
1256 /* If it isn't a tag pattern it must be a URL pattern. */
1257 return compile_url_pattern(url, buf);
1261 /*********************************************************************
1263 * Function : free_url_spec
1265 * Description : Called from the "unloaders". Freez the url
1266 * structure elements.
1269 * 1 : url = pointer to a url_spec structure.
1273 *********************************************************************/
1274 void free_url_spec(struct url_spec *url)
1276 if (url == NULL) return;
1279 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
1280 if (url->host_regex)
1282 regfree(url->host_regex);
1283 freez(url->host_regex);
1286 freez(url->dbuffer);
1289 #endif /* ndef FEATURE_EXTENDED_HOST_PATTERNS */
1290 freez(url->port_list);
1298 regfree(url->tag_regex);
1299 freez(url->tag_regex);
1304 /*********************************************************************
1306 * Function : url_match
1308 * Description : Compare a URL against a URL pattern.
1311 * 1 : pattern = a URL pattern
1312 * 2 : url = URL to match
1314 * Returns : Nonzero if the URL matches the pattern, else 0.
1316 *********************************************************************/
1317 int url_match(const struct url_spec *pattern,
1318 const struct http_request *http)
1320 /* XXX: these should probably be functions. */
1321 #define PORT_MATCHES ((NULL == pattern->port_list) || match_portlist(pattern->port_list, http->port))
1322 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
1323 #define DOMAIN_MATCHES ((NULL == pattern->host_regex) || (0 == regexec(pattern->host_regex, http->host, 0, NULL, 0)))
1325 #define DOMAIN_MATCHES ((NULL == pattern->dbuffer) || (0 == domain_match(pattern, http)))
1327 #define PATH_MATCHES ((NULL == pattern->preg) || (0 == regexec(pattern->preg, http->path, 0, NULL, 0)))
1329 if (pattern->tag_regex != NULL)
1331 /* It's a tag pattern and shouldn't be matched against URLs */
1335 return (PORT_MATCHES && DOMAIN_MATCHES && PATH_MATCHES);
1340 /*********************************************************************
1342 * Function : match_portlist
1344 * Description : Check if a given number is covered by a comma
1345 * separated list of numbers and ranges (a,b-c,d,..)
1348 * 1 : portlist = String with list
1349 * 2 : port = port to check
1351 * Returns : 0 => no match
1354 *********************************************************************/
1355 int match_portlist(const char *portlist, int port)
1357 char *min, *max, *next, *portlist_copy;
1359 min = next = portlist_copy = strdup(portlist);
1362 * Zero-terminate first item and remember offset for next
1364 if (NULL != (next = strchr(portlist_copy, (int) ',')))
1370 * Loop through all items, checking for match
1374 if (NULL == (max = strchr(min, (int) '-')))
1377 * No dash, check for equality
1379 if (port == atoi(min))
1381 free(portlist_copy);
1388 * This is a range, so check if between min and max,
1389 * or, if max was omitted, between min and 65K
1392 if(port >= atoi(min) && port <= (atoi(max) ? atoi(max) : 65535))
1394 free(portlist_copy);
1406 * Zero-terminate next item and remember offset for n+1
1408 if ((NULL != next) && (NULL != (next = strchr(next, (int) ','))))
1414 free(portlist_copy);