1 const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.46 2009/02/11 19:31:32 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) 2001-2009
10 * the 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 1.46 2009/02/11 19:31:32 fabiankeil
37 * Reject request lines that end with neither HTTP/1.0 nor HTTP/1.1.
39 * Revision 1.45 2008/06/21 21:19:18 fabiankeil
40 * Silence bogus compiler warning.
42 * Revision 1.44 2008/05/04 16:18:32 fabiankeil
43 * Provide parse_http_url() with a third parameter to specify
44 * whether or not URLs without protocol are acceptable.
46 * Revision 1.43 2008/05/04 13:30:55 fabiankeil
47 * Streamline parse_http_url()'s prototype.
49 * Revision 1.42 2008/05/04 13:24:16 fabiankeil
50 * If the method isn't CONNECT, reject URLs without protocol.
52 * Revision 1.41 2008/05/02 09:51:34 fabiankeil
53 * In parse_http_url(), don't muck around with values
54 * that are none of its business: require an initialized
55 * http structure and never unset http->ssl.
57 * Revision 1.40 2008/04/23 16:12:28 fabiankeil
60 * Revision 1.39 2008/04/22 16:27:42 fabiankeil
61 * In parse_http_request(), remove a pointless
62 * temporary variable and free the buffer earlier.
64 * Revision 1.38 2008/04/18 05:17:18 fabiankeil
65 * Mark simplematch()'s parameters as immutable.
67 * Revision 1.37 2008/04/17 14:53:29 fabiankeil
68 * Move simplematch() into urlmatch.c as it's only
69 * used to match (old-school) domain patterns.
71 * Revision 1.36 2008/04/14 18:19:48 fabiankeil
72 * Remove now-pointless cast in create_url_spec().
74 * Revision 1.35 2008/04/14 18:11:21 fabiankeil
75 * The compiler might not notice it, but the buffer passed to
76 * create_url_spec() is modified later on and thus shouldn't
77 * be declared immutable.
79 * Revision 1.34 2008/04/13 13:32:07 fabiankeil
80 * Factor URL pattern compilation out of create_url_spec().
82 * Revision 1.33 2008/04/12 14:03:13 fabiankeil
83 * Remove an obvious comment and improve another one.
85 * Revision 1.32 2008/04/12 12:38:06 fabiankeil
86 * Factor out duplicated code to compile host, path and tag patterns.
88 * Revision 1.31 2008/04/10 14:41:04 fabiankeil
89 * Ditch url_spec's path member now that it's no longer used.
91 * Revision 1.30 2008/04/10 04:24:24 fabiankeil
92 * Stop duplicating the plain text representation of the path regex
93 * (and keeping the copy around). Once the regex is compiled it's no
96 * Revision 1.29 2008/04/10 04:17:56 fabiankeil
97 * In url_match(), check the right member for NULL when determining
98 * whether there's a path regex to execute. Looking for a plain-text
99 * representation works as well, but it looks "interesting" and that
100 * member will be removed soonish anyway.
102 * Revision 1.28 2008/04/08 16:07:39 fabiankeil
103 * Make it harder to mistake url_match()'s
104 * second parameter for an url_spec.
106 * Revision 1.27 2008/04/08 15:44:33 fabiankeil
107 * Save a bit of memory (and a few cpu cycles) by not bothering to
108 * compile slash-only path regexes that don't affect the result.
110 * Revision 1.26 2008/04/07 16:57:18 fabiankeil
111 * - Use free_url_spec() more consistently.
112 * - Let it reset url->dcount just in case.
114 * Revision 1.25 2008/04/06 15:18:38 fabiankeil
115 * Oh well, rename the --enable-pcre-host-patterns option to
116 * --enable-extended-host-patterns as it's not really PCRE syntax.
118 * Revision 1.24 2008/04/06 14:54:26 fabiankeil
119 * Use PCRE syntax in host patterns when configured
120 * with --enable-pcre-host-patterns.
122 * Revision 1.23 2008/04/05 12:19:20 fabiankeil
123 * Factor compile_host_pattern() out of create_url_spec().
125 * Revision 1.22 2008/03/30 15:02:32 fabiankeil
126 * SZitify unknown_method().
128 * Revision 1.21 2007/12/24 16:34:23 fabiankeil
129 * Band-aid (and micro-optimization) that makes it less likely to run out of
130 * stack space with overly-complex path patterns. Probably masks the problem
131 * reported by Lee in #1856679. Hohoho.
133 * Revision 1.20 2007/09/02 15:31:20 fabiankeil
134 * Move match_portlist() from filter.c to urlmatch.c.
135 * It's used for url matching, not for filtering.
137 * Revision 1.19 2007/09/02 13:42:11 fabiankeil
138 * - Allow port lists in url patterns.
139 * - Ditch unused url_spec member pathlen.
141 * Revision 1.18 2007/07/30 16:42:21 fabiankeil
142 * Move the method check into unknown_method()
143 * and loop through the known methods instead
144 * of using a screen-long OR chain.
146 * Revision 1.17 2007/04/15 16:39:21 fabiankeil
147 * Introduce tags as alternative way to specify which
148 * actions apply to a request. At the moment tags can be
149 * created based on client and server headers.
151 * Revision 1.16 2007/02/13 13:59:24 fabiankeil
152 * Remove redundant log message.
154 * Revision 1.15 2007/01/28 16:11:23 fabiankeil
155 * Accept WebDAV methods for subversion
156 * in parse_http_request(). Closes FR 1581425.
158 * Revision 1.14 2007/01/06 14:23:56 fabiankeil
159 * Fix gcc43 warnings. Mark *csp as immutable
160 * for parse_http_url() and url_match().
161 * Replace a sprintf call with snprintf.
163 * Revision 1.13 2006/12/06 19:50:54 fabiankeil
164 * parse_http_url() now handles intercepted
165 * HTTP request lines as well. Moved parts
166 * of parse_http_url()'s code into
167 * init_domain_components() so that it can
168 * be reused in chat().
170 * Revision 1.12 2006/07/18 14:48:47 david__schmidt
171 * Reorganizing the repository: swapping out what was HEAD (the old 3.1 branch)
172 * with what was really the latest development (the v_3_0_branch branch)
174 * Revision 1.10.2.7 2003/05/17 15:57:24 oes
175 * - parse_http_url now checks memory allocation failure for
176 * duplication of "*" URL and rejects "*something" URLs
178 * - Added a comment to what might look like a bug in
179 * create_url_spec (see !bug #736931)
180 * - Comment cosmetics
182 * Revision 1.10.2.6 2003/05/07 12:39:48 oes
183 * Fix typo: Default port for https URLs is 443, not 143.
184 * Thanks to Scott Tregear for spotting this one.
186 * Revision 1.10.2.5 2003/02/28 13:09:29 oes
187 * Fixed a rare double free condition as per Bug #694713
189 * Revision 1.10.2.4 2003/02/28 12:57:44 oes
190 * Moved freeing of http request structure to its owner
191 * as per Dan Price's observations in Bug #694713
193 * Revision 1.10.2.3 2002/11/12 16:50:40 oes
194 * Fixed memory leak in parse_http_request() reported by Oliver Stoeneberg. Fixes bug #637073
196 * Revision 1.10.2.2 2002/09/25 14:53:15 oes
197 * Added basic support for OPTIONS and TRACE HTTP methods:
198 * parse_http_url now recognizes the "*" URI as well as
199 * the OPTIONS and TRACE method keywords.
201 * Revision 1.10.2.1 2002/06/06 19:06:44 jongfoster
202 * Adding support for proprietary Microsoft WebDAV extensions
204 * Revision 1.10 2002/05/12 21:40:37 jongfoster
205 * - Removing some unused code
207 * Revision 1.9 2002/04/04 00:36:36 gliptak
208 * always use pcre for matching
210 * Revision 1.8 2002/04/03 23:32:47 jongfoster
211 * Fixing memory leak on error
213 * Revision 1.7 2002/03/26 22:29:55 swa
214 * we have a new homepage!
216 * Revision 1.6 2002/03/24 13:25:43 swa
217 * name change related issues
219 * Revision 1.5 2002/03/13 00:27:05 jongfoster
222 * Revision 1.4 2002/03/07 03:46:17 oes
223 * Fixed compiler warnings
225 * Revision 1.3 2002/03/03 14:51:11 oes
226 * Fixed CLF logging: Added ocmd member for client's request to struct http_request
228 * Revision 1.2 2002/01/21 00:14:09 jongfoster
229 * Correcting comment style
230 * Fixing an uninitialized memory bug in create_url_spec()
232 * Revision 1.1 2002/01/17 20:53:46 jongfoster
233 * Moving all our URL and URL pattern parsing code to the same file - it
234 * was scattered around in filters.c, loaders.c and parsers.c.
236 * Providing a single, simple url_match(pattern,url) function - rather than
237 * the 3-line match routine which was repeated all over the place.
239 * Renaming free_url to free_url_spec, since it frees a struct url_spec.
241 * Providing parse_http_url() so that URLs can be parsed without faking a
242 * HTTP request line for parse_http_request() or repeating the parsing
243 * code (both of which were techniques that were actually in use).
245 * Standardizing that struct http_request is used to represent a URL, and
246 * struct url_spec is used to represent a URL pattern. (Before, URLs were
247 * represented as seperate variables and a partially-filled-in url_spec).
250 *********************************************************************/
257 #include <sys/types.h>
265 #if !defined(_WIN32) && !defined(__OS2__)
270 #include "urlmatch.h"
272 #include "miscutil.h"
275 const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
277 enum regex_anchoring {NO_ANCHORING, LEFT_ANCHORED, RIGHT_ANCHORED};
278 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern);
280 /*********************************************************************
282 * Function : free_http_request
284 * Description : Freez a http_request structure
287 * 1 : http = points to a http_request structure to free
291 *********************************************************************/
292 void free_http_request(struct http_request *http)
301 freez(http->hostport);
304 freez(http->host_ip_addr_str);
305 freez(http->dbuffer);
311 /*********************************************************************
313 * Function : init_domain_components
315 * Description : Splits the domain name so we can compare it
316 * against wildcards. It used to be part of
317 * parse_http_url, but was separated because the
318 * same code is required in chat in case of
319 * intercepted requests.
322 * 1 : http = pointer to the http structure to hold elements.
324 * Returns : JB_ERR_OK on success
325 * JB_ERR_MEMORY on out of memory
326 * JB_ERR_PARSE on malformed command/URL
327 * or >100 domains deep.
329 *********************************************************************/
330 jb_err init_domain_components(struct http_request *http)
332 char *vec[BUFFER_SIZE];
336 http->dbuffer = strdup(http->host);
337 if (NULL == http->dbuffer)
339 return JB_ERR_MEMORY;
342 /* map to lower case */
343 for (p = http->dbuffer; *p ; p++)
345 *p = (char)tolower((int)(unsigned char)*p);
348 /* split the domain name into components */
349 http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec), 1, 1);
351 if (http->dcount <= 0)
354 * Error: More than SZ(vec) components in domain
355 * or: no components in domain
357 log_error(LOG_LEVEL_ERROR, "More than SZ(vec) components in domain or none at all.");
361 /* save a copy of the pointers in dvec */
362 size = (size_t)http->dcount * sizeof(*http->dvec);
364 http->dvec = (char **)malloc(size);
365 if (NULL == http->dvec)
367 return JB_ERR_MEMORY;
370 memcpy(http->dvec, vec, size);
376 /*********************************************************************
378 * Function : parse_http_url
380 * Description : Parse out the host and port from the URL. Find the
381 * hostname & path, port (if ':'), and/or password (if '@')
384 * 1 : url = URL (or is it URI?) to break down
385 * 2 : http = pointer to the http structure to hold elements.
386 * Must be initialized with valid values (like NULLs).
387 * 3 : require_protocol = Whether or not URLs without
388 * protocol are acceptable.
390 * Returns : JB_ERR_OK on success
391 * JB_ERR_MEMORY on out of memory
392 * JB_ERR_PARSE on malformed command/URL
393 * or >100 domains deep.
395 *********************************************************************/
396 jb_err parse_http_url(const char *url, struct http_request *http, int require_protocol)
398 int host_available = 1; /* A proxy can dream. */
401 * Save our initial URL
403 http->url = strdup(url);
404 if (http->url == NULL)
406 return JB_ERR_MEMORY;
411 * Check for * URI. If found, we're done.
413 if (*http->url == '*')
415 if ( NULL == (http->path = strdup("*"))
416 || NULL == (http->hostport = strdup("")) )
418 return JB_ERR_MEMORY;
420 if (http->url[1] != '\0')
429 * Split URL into protocol,hostport,path.
439 return JB_ERR_MEMORY;
442 /* Find the start of the URL in our scratch space */
444 if (strncmpic(url_noproto, "http://", 7) == 0)
448 else if (strncmpic(url_noproto, "https://", 8) == 0)
451 * Should only happen when called from cgi_show_url_info().
456 else if (*url_noproto == '/')
459 * Short request line without protocol and host.
460 * Most likely because the client's request
461 * was intercepted and redirected into Privoxy.
466 else if (require_protocol)
472 url_path = strchr(url_noproto, '/');
473 if (url_path != NULL)
478 * NOTE: The following line ignores the path for HTTPS URLS.
479 * This means that you get consistent behaviour if you type a
480 * https URL in and it's parsed by the function. (When the
481 * URL is actually retrieved, SSL hides the path part).
483 http->path = strdup(http->ssl ? "/" : url_path);
485 http->hostport = strdup(url_noproto);
490 * Repair broken HTTP requests that don't contain a path,
491 * or CONNECT requests
493 http->path = strdup("/");
494 http->hostport = strdup(url_noproto);
499 if ( (http->path == NULL)
500 || (http->hostport == NULL))
502 return JB_ERR_MEMORY;
508 /* Without host, there is nothing left to do here */
513 * Split hostport into user/password (ignored), host, port.
520 buf = strdup(http->hostport);
523 return JB_ERR_MEMORY;
526 /* check if url contains username and/or password */
527 host = strchr(buf, '@');
530 /* Contains username/password, skip it and the @ sign. */
535 /* No username or password. */
539 /* check if url contains port */
540 port = strchr(host, ':');
544 /* Terminate hostname and point to start of port string */
546 http->port = atoi(port);
550 /* No port specified. */
551 http->port = (http->ssl ? 443 : 80);
554 http->host = strdup(host);
558 if (http->host == NULL)
560 return JB_ERR_MEMORY;
565 * Split domain name so we can compare it against wildcards
567 return init_domain_components(http);
572 /*********************************************************************
574 * Function : unknown_method
576 * Description : Checks whether a method is unknown.
579 * 1 : method = points to a http method
581 * Returns : TRUE if it's unknown, FALSE otherwise.
583 *********************************************************************/
584 static int unknown_method(const char *method)
586 static const char *known_http_methods[] = {
587 /* Basic HTTP request type */
588 "GET", "HEAD", "POST", "PUT", "DELETE", "OPTIONS", "TRACE", "CONNECT",
589 /* webDAV extensions (RFC2518) */
590 "PROPFIND", "PROPPATCH", "MOVE", "COPY", "MKCOL", "LOCK", "UNLOCK",
592 * Microsoft webDAV extension for Exchange 2000. See:
593 * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
594 * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
596 "BCOPY", "BMOVE", "BDELETE", "BPROPFIND", "BPROPPATCH",
598 * Another Microsoft webDAV extension for Exchange 2000. See:
599 * http://systems.cs.colorado.edu/grunwald/MobileComputing/Papers/draft-cohen-gena-p-base-00.txt
600 * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
601 * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
603 "SUBSCRIBE", "UNSUBSCRIBE", "NOTIFY", "POLL",
605 * Yet another WebDAV extension, this time for
606 * Web Distributed Authoring and Versioning (RFC3253)
608 "VERSION-CONTROL", "REPORT", "CHECKOUT", "CHECKIN", "UNCHECKOUT",
609 "MKWORKSPACE", "UPDATE", "LABEL", "MERGE", "BASELINE-CONTROL", "MKACTIVITY",
613 for (i = 0; i < SZ(known_http_methods); i++)
615 if (0 == strcmpic(method, known_http_methods[i]))
626 /*********************************************************************
628 * Function : parse_http_request
630 * Description : Parse out the host and port from the URL. Find the
631 * hostname & path, port (if ':'), and/or password (if '@')
634 * 1 : req = HTTP request line to break down
635 * 2 : http = pointer to the http structure to hold elements
637 * Returns : JB_ERR_OK on success
638 * JB_ERR_MEMORY on out of memory
639 * JB_ERR_CGI_PARAMS on malformed command/URL
640 * or >100 domains deep.
642 *********************************************************************/
643 jb_err parse_http_request(const char *req, struct http_request *http)
646 char *v[10]; /* XXX: Why 10? We should only need three. */
650 memset(http, '\0', sizeof(*http));
655 return JB_ERR_MEMORY;
658 n = ssplit(buf, " \r\n", v, SZ(v), 1, 1);
666 * Fail in case of unknown methods
667 * which we might not handle correctly.
669 * XXX: There should be a config option
670 * to forward requests with unknown methods
671 * anyway. Most of them don't need special
674 if (unknown_method(v[0]))
676 log_error(LOG_LEVEL_ERROR, "Unknown HTTP method detected: %s", v[0]);
681 if (strcmpic(v[2], "HTTP/1.1") && strcmpic(v[2], "HTTP/1.0"))
683 log_error(LOG_LEVEL_ERROR, "The only supported HTTP "
684 "versions are 1.0 and 1.1. This rules out: %s", v[2]);
689 http->ssl = !strcmpic(v[0], "CONNECT");
691 err = parse_http_url(v[1], http, !http->ssl);
699 * Copy the details into the structure
701 http->cmd = strdup(req);
702 http->gpc = strdup(v[0]);
703 http->ver = strdup(v[2]);
707 if ( (http->cmd == NULL)
708 || (http->gpc == NULL)
709 || (http->ver == NULL) )
711 return JB_ERR_MEMORY;
719 /*********************************************************************
721 * Function : compile_pattern
723 * Description : Compiles a host, domain or TAG pattern.
726 * 1 : pattern = The pattern to compile.
727 * 2 : anchoring = How the regex should be anchored.
728 * Can be either one of NO_ANCHORING,
729 * LEFT_ANCHORED or RIGHT_ANCHORED.
730 * 3 : url = In case of failures, the spec member is
731 * logged and the structure freed.
732 * 4 : regex = Where the compiled regex should be stored.
734 * Returns : JB_ERR_OK - Success
735 * JB_ERR_MEMORY - Out of memory
736 * JB_ERR_PARSE - Cannot parse regex
738 *********************************************************************/
739 static jb_err compile_pattern(const char *pattern, enum regex_anchoring anchoring,
740 struct url_spec *url, regex_t **regex)
743 char rebuf[BUFFER_SIZE];
744 const char *fmt = NULL;
747 assert(strlen(pattern) < sizeof(rebuf) - 2);
749 if (pattern[0] == '\0')
767 log_error(LOG_LEVEL_FATAL,
768 "Invalid anchoring in compile_pattern %d", anchoring);
771 *regex = zalloc(sizeof(**regex));
775 return JB_ERR_MEMORY;
778 snprintf(rebuf, sizeof(rebuf), fmt, pattern);
780 errcode = regcomp(*regex, rebuf, (REG_EXTENDED|REG_NOSUB|REG_ICASE));
784 size_t errlen = regerror(errcode, *regex, rebuf, sizeof(rebuf));
785 if (errlen > (sizeof(rebuf) - (size_t)1))
787 errlen = sizeof(rebuf) - (size_t)1;
789 rebuf[errlen] = '\0';
790 log_error(LOG_LEVEL_ERROR, "error compiling %s from %s: %s",
791 pattern, url->spec, rebuf);
802 /*********************************************************************
804 * Function : compile_url_pattern
806 * Description : Compiles the three parts of an URL pattern.
809 * 1 : url = Target url_spec to be filled in.
810 * 2 : buf = The url pattern to compile. Will be messed up.
812 * Returns : JB_ERR_OK - Success
813 * JB_ERR_MEMORY - Out of memory
814 * JB_ERR_PARSE - Cannot parse regex
816 *********************************************************************/
817 static jb_err compile_url_pattern(struct url_spec *url, char *buf)
821 p = strchr(buf, '/');
825 * Only compile the regex if it consists of more than
826 * a single slash, otherwise it wouldn't affect the result.
831 * XXX: does it make sense to compile the slash at the beginning?
833 jb_err err = compile_pattern(p, LEFT_ANCHORED, url, &url->preg);
835 if (JB_ERR_OK != err)
843 p = strchr(buf, ':');
847 url->port_list = strdup(p);
848 if (NULL == url->port_list)
850 return JB_ERR_MEMORY;
855 url->port_list = NULL;
860 return compile_host_pattern(url, buf);
868 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
869 /*********************************************************************
871 * Function : compile_host_pattern
873 * Description : Parses and compiles a host pattern..
876 * 1 : url = Target url_spec to be filled in.
877 * 2 : host_pattern = Host pattern to compile.
879 * Returns : JB_ERR_OK - Success
880 * JB_ERR_MEMORY - Out of memory
881 * JB_ERR_PARSE - Cannot parse regex
883 *********************************************************************/
884 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
886 return compile_pattern(host_pattern, RIGHT_ANCHORED, url, &url->host_regex);
891 /*********************************************************************
893 * Function : compile_host_pattern
895 * Description : Parses and "compiles" an old-school host pattern.
898 * 1 : url = Target url_spec to be filled in.
899 * 2 : host_pattern = Host pattern to parse.
901 * Returns : JB_ERR_OK - Success
902 * JB_ERR_MEMORY - Out of memory
903 * JB_ERR_PARSE - Cannot parse regex
905 *********************************************************************/
906 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
915 if (host_pattern[strlen(host_pattern) - 1] == '.')
917 url->unanchored |= ANCHOR_RIGHT;
919 if (host_pattern[0] == '.')
921 url->unanchored |= ANCHOR_LEFT;
925 * Split domain into components
927 url->dbuffer = strdup(host_pattern);
928 if (NULL == url->dbuffer)
931 return JB_ERR_MEMORY;
937 for (p = url->dbuffer; *p ; p++)
939 *p = (char)tolower((int)(unsigned char)*p);
943 * Split the domain name into components
945 url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1);
950 return JB_ERR_MEMORY;
952 else if (url->dcount != 0)
955 * Save a copy of the pointers in dvec
957 size = (size_t)url->dcount * sizeof(*url->dvec);
959 url->dvec = (char **)malloc(size);
960 if (NULL == url->dvec)
963 return JB_ERR_MEMORY;
966 memcpy(url->dvec, v, size);
969 * else dcount == 0 in which case we needn't do anything,
970 * since dvec will never be accessed and the pattern will
977 /*********************************************************************
979 * Function : simplematch
981 * Description : String matching, with a (greedy) '*' wildcard that
982 * stands for zero or more arbitrary characters and
983 * character classes in [], which take both enumerations
987 * 1 : pattern = pattern for matching
988 * 2 : text = text to be matched
990 * Returns : 0 if match, else nonzero
992 *********************************************************************/
993 static int simplematch(const char *pattern, const char *text)
995 const unsigned char *pat = (const unsigned char *)pattern;
996 const unsigned char *txt = (const unsigned char *)text;
997 const unsigned char *fallback = pat;
1000 unsigned char lastchar = 'a';
1002 unsigned char charmap[32];
1007 /* EOF pattern but !EOF text? */
1020 /* '*' in the pattern? */
1024 /* The pattern ends afterwards? Speed up the return. */
1030 /* Else, set wildcard mode and remember position after '*' */
1035 /* Character range specification? */
1038 memset(charmap, '\0', sizeof(charmap));
1040 while (*++pat != ']')
1046 else if (*pat == '-')
1048 if ((*++pat == ']') || *pat == '\0')
1052 for (i = lastchar; i <= *pat; i++)
1054 charmap[i / 8] |= (unsigned char)(1 << (i % 8));
1059 charmap[*pat / 8] |= (unsigned char)(1 << (*pat % 8));
1063 } /* -END- if Character range specification */
1067 * Char match, or char range match?
1071 || ((*pat == ']') && (charmap[*txt / 8] & (1 << (*txt % 8)))) )
1081 * No match && no wildcard: No luck
1085 else if (pat != fallback)
1088 * Increment text pointer if in char range matching
1095 * Wildcard mode && nonmatch beyond fallback: Rewind pattern
1099 * Restart matching from current text pointer
1106 /* Cut off extra '*'s */
1107 if(*pat == '*') pat++;
1109 /* If this is the pattern's end, fine! */
1115 /*********************************************************************
1117 * Function : simple_domaincmp
1119 * Description : Domain-wise Compare fqdn's. The comparison is
1120 * both left- and right-anchored. The individual
1121 * domain names are compared with simplematch().
1122 * This is only used by domain_match.
1125 * 1 : pv = array of patterns to compare
1126 * 2 : fv = array of domain components to compare
1127 * 3 : len = length of the arrays (both arrays are the
1128 * same length - if they weren't, it couldn't
1129 * possibly be a match).
1131 * Returns : 0 => domains are equivalent, else no match.
1133 *********************************************************************/
1134 static int simple_domaincmp(char **pv, char **fv, int len)
1138 for (n = 0; n < len; n++)
1140 if (simplematch(pv[n], fv[n]))
1151 /*********************************************************************
1153 * Function : domain_match
1155 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
1156 * pattern->unachored, the comparison is un-, left-,
1157 * right-anchored, or both.
1158 * The individual domain names are compared with
1162 * 1 : pattern = a domain that may contain a '*' as a wildcard.
1163 * 2 : fqdn = domain name against which the patterns are compared.
1165 * Returns : 0 => domains are equivalent, else no match.
1167 *********************************************************************/
1168 static int domain_match(const struct url_spec *pattern, const struct http_request *fqdn)
1170 char **pv, **fv; /* vectors */
1172 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
1174 plen = pattern->dcount;
1175 flen = fqdn->dcount;
1179 /* fqdn is too short to match this pattern */
1186 if (unanchored == ANCHOR_LEFT)
1191 * Convert this into a fully anchored pattern with
1192 * the fqdn and pattern the same length
1194 fv += (flen - plen); /* flen - plen >= 0 due to check above */
1195 return simple_domaincmp(pv, fv, plen);
1197 else if (unanchored == 0)
1199 /* Fully anchored, check length */
1204 return simple_domaincmp(pv, fv, plen);
1206 else if (unanchored == ANCHOR_RIGHT)
1208 /* Left anchored, ignore all extra in fqdn */
1209 return simple_domaincmp(pv, fv, plen);
1215 int maxn = flen - plen;
1216 for (n = 0; n <= maxn; n++)
1218 if (!simple_domaincmp(pv, fv, plen))
1223 * Doesn't match from start of fqdn
1224 * Try skipping first part of fqdn
1232 #endif /* def FEATURE_EXTENDED_HOST_PATTERNS */
1235 /*********************************************************************
1237 * Function : create_url_spec
1239 * Description : Creates a "url_spec" structure from a string.
1240 * When finished, free with free_url_spec().
1243 * 1 : url = Target url_spec to be filled in. Will be
1244 * zeroed before use.
1245 * 2 : buf = Source pattern, null terminated. NOTE: The
1246 * contents of this buffer are destroyed by this
1247 * function. If this function succeeds, the
1248 * buffer is copied to url->spec. If this
1249 * function fails, the contents of the buffer
1252 * Returns : JB_ERR_OK - Success
1253 * JB_ERR_MEMORY - Out of memory
1254 * JB_ERR_PARSE - Cannot parse regex (Detailed message
1255 * written to system log)
1257 *********************************************************************/
1258 jb_err create_url_spec(struct url_spec *url, char *buf)
1263 memset(url, '\0', sizeof(*url));
1265 /* Remember the original specification for the CGI pages. */
1266 url->spec = strdup(buf);
1267 if (NULL == url->spec)
1269 return JB_ERR_MEMORY;
1272 /* Is it tag pattern? */
1273 if (0 == strncmpic("TAG:", url->spec, 4))
1275 /* The pattern starts with the first character after "TAG:" */
1276 const char *tag_pattern = buf + 4;
1277 return compile_pattern(tag_pattern, NO_ANCHORING, url, &url->tag_regex);
1280 /* If it isn't a tag pattern it must be a URL pattern. */
1281 return compile_url_pattern(url, buf);
1285 /*********************************************************************
1287 * Function : free_url_spec
1289 * Description : Called from the "unloaders". Freez the url
1290 * structure elements.
1293 * 1 : url = pointer to a url_spec structure.
1297 *********************************************************************/
1298 void free_url_spec(struct url_spec *url)
1300 if (url == NULL) return;
1303 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
1304 if (url->host_regex)
1306 regfree(url->host_regex);
1307 freez(url->host_regex);
1310 freez(url->dbuffer);
1313 #endif /* ndef FEATURE_EXTENDED_HOST_PATTERNS */
1314 freez(url->port_list);
1322 regfree(url->tag_regex);
1323 freez(url->tag_regex);
1328 /*********************************************************************
1330 * Function : url_match
1332 * Description : Compare a URL against a URL pattern.
1335 * 1 : pattern = a URL pattern
1336 * 2 : url = URL to match
1338 * Returns : Nonzero if the URL matches the pattern, else 0.
1340 *********************************************************************/
1341 int url_match(const struct url_spec *pattern,
1342 const struct http_request *http)
1344 /* XXX: these should probably be functions. */
1345 #define PORT_MATCHES ((NULL == pattern->port_list) || match_portlist(pattern->port_list, http->port))
1346 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
1347 #define DOMAIN_MATCHES ((NULL == pattern->host_regex) || (0 == regexec(pattern->host_regex, http->host, 0, NULL, 0)))
1349 #define DOMAIN_MATCHES ((NULL == pattern->dbuffer) || (0 == domain_match(pattern, http)))
1351 #define PATH_MATCHES ((NULL == pattern->preg) || (0 == regexec(pattern->preg, http->path, 0, NULL, 0)))
1353 if (pattern->tag_regex != NULL)
1355 /* It's a tag pattern and shouldn't be matched against URLs */
1359 return (PORT_MATCHES && DOMAIN_MATCHES && PATH_MATCHES);
1364 /*********************************************************************
1366 * Function : match_portlist
1368 * Description : Check if a given number is covered by a comma
1369 * separated list of numbers and ranges (a,b-c,d,..)
1372 * 1 : portlist = String with list
1373 * 2 : port = port to check
1375 * Returns : 0 => no match
1378 *********************************************************************/
1379 int match_portlist(const char *portlist, int port)
1381 char *min, *max, *next, *portlist_copy;
1383 min = next = portlist_copy = strdup(portlist);
1386 * Zero-terminate first item and remember offset for next
1388 if (NULL != (next = strchr(portlist_copy, (int) ',')))
1394 * Loop through all items, checking for match
1398 if (NULL == (max = strchr(min, (int) '-')))
1401 * No dash, check for equality
1403 if (port == atoi(min))
1405 freez(portlist_copy);
1412 * This is a range, so check if between min and max,
1413 * or, if max was omitted, between min and 65K
1416 if(port >= atoi(min) && port <= (atoi(max) ? atoi(max) : 65535))
1418 freez(portlist_copy);
1430 * Zero-terminate next item and remember offset for n+1
1432 if ((NULL != next) && (NULL != (next = strchr(next, (int) ','))))
1438 freez(portlist_copy);