1 const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.41 2008/05/02 09:51:34 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 1.41 2008/05/02 09:51:34 fabiankeil
37 * In parse_http_url(), don't muck around with values
38 * that are none of its business: require an initialized
39 * http structure and never unset http->ssl.
41 * Revision 1.40 2008/04/23 16:12:28 fabiankeil
44 * Revision 1.39 2008/04/22 16:27:42 fabiankeil
45 * In parse_http_request(), remove a pointless
46 * temporary variable and free the buffer earlier.
48 * Revision 1.38 2008/04/18 05:17:18 fabiankeil
49 * Mark simplematch()'s parameters as immutable.
51 * Revision 1.37 2008/04/17 14:53:29 fabiankeil
52 * Move simplematch() into urlmatch.c as it's only
53 * used to match (old-school) domain patterns.
55 * Revision 1.36 2008/04/14 18:19:48 fabiankeil
56 * Remove now-pointless cast in create_url_spec().
58 * Revision 1.35 2008/04/14 18:11:21 fabiankeil
59 * The compiler might not notice it, but the buffer passed to
60 * create_url_spec() is modified later on and thus shouldn't
61 * be declared immutable.
63 * Revision 1.34 2008/04/13 13:32:07 fabiankeil
64 * Factor URL pattern compilation out of create_url_spec().
66 * Revision 1.33 2008/04/12 14:03:13 fabiankeil
67 * Remove an obvious comment and improve another one.
69 * Revision 1.32 2008/04/12 12:38:06 fabiankeil
70 * Factor out duplicated code to compile host, path and tag patterns.
72 * Revision 1.31 2008/04/10 14:41:04 fabiankeil
73 * Ditch url_spec's path member now that it's no longer used.
75 * Revision 1.30 2008/04/10 04:24:24 fabiankeil
76 * Stop duplicating the plain text representation of the path regex
77 * (and keeping the copy around). Once the regex is compiled it's no
80 * Revision 1.29 2008/04/10 04:17:56 fabiankeil
81 * In url_match(), check the right member for NULL when determining
82 * whether there's a path regex to execute. Looking for a plain-text
83 * representation works as well, but it looks "interesting" and that
84 * member will be removed soonish anyway.
86 * Revision 1.28 2008/04/08 16:07:39 fabiankeil
87 * Make it harder to mistake url_match()'s
88 * second parameter for an url_spec.
90 * Revision 1.27 2008/04/08 15:44:33 fabiankeil
91 * Save a bit of memory (and a few cpu cycles) by not bothering to
92 * compile slash-only path regexes that don't affect the result.
94 * Revision 1.26 2008/04/07 16:57:18 fabiankeil
95 * - Use free_url_spec() more consistently.
96 * - Let it reset url->dcount just in case.
98 * Revision 1.25 2008/04/06 15:18:38 fabiankeil
99 * Oh well, rename the --enable-pcre-host-patterns option to
100 * --enable-extended-host-patterns as it's not really PCRE syntax.
102 * Revision 1.24 2008/04/06 14:54:26 fabiankeil
103 * Use PCRE syntax in host patterns when configured
104 * with --enable-pcre-host-patterns.
106 * Revision 1.23 2008/04/05 12:19:20 fabiankeil
107 * Factor compile_host_pattern() out of create_url_spec().
109 * Revision 1.22 2008/03/30 15:02:32 fabiankeil
110 * SZitify unknown_method().
112 * Revision 1.21 2007/12/24 16:34:23 fabiankeil
113 * Band-aid (and micro-optimization) that makes it less likely to run out of
114 * stack space with overly-complex path patterns. Probably masks the problem
115 * reported by Lee in #1856679. Hohoho.
117 * Revision 1.20 2007/09/02 15:31:20 fabiankeil
118 * Move match_portlist() from filter.c to urlmatch.c.
119 * It's used for url matching, not for filtering.
121 * Revision 1.19 2007/09/02 13:42:11 fabiankeil
122 * - Allow port lists in url patterns.
123 * - Ditch unused url_spec member pathlen.
125 * Revision 1.18 2007/07/30 16:42:21 fabiankeil
126 * Move the method check into unknown_method()
127 * and loop through the known methods instead
128 * of using a screen-long OR chain.
130 * Revision 1.17 2007/04/15 16:39:21 fabiankeil
131 * Introduce tags as alternative way to specify which
132 * actions apply to a request. At the moment tags can be
133 * created based on client and server headers.
135 * Revision 1.16 2007/02/13 13:59:24 fabiankeil
136 * Remove redundant log message.
138 * Revision 1.15 2007/01/28 16:11:23 fabiankeil
139 * Accept WebDAV methods for subversion
140 * in parse_http_request(). Closes FR 1581425.
142 * Revision 1.14 2007/01/06 14:23:56 fabiankeil
143 * Fix gcc43 warnings. Mark *csp as immutable
144 * for parse_http_url() and url_match().
145 * Replace a sprintf call with snprintf.
147 * Revision 1.13 2006/12/06 19:50:54 fabiankeil
148 * parse_http_url() now handles intercepted
149 * HTTP request lines as well. Moved parts
150 * of parse_http_url()'s code into
151 * init_domain_components() so that it can
152 * be reused in chat().
154 * Revision 1.12 2006/07/18 14:48:47 david__schmidt
155 * Reorganizing the repository: swapping out what was HEAD (the old 3.1 branch)
156 * with what was really the latest development (the v_3_0_branch branch)
158 * Revision 1.10.2.7 2003/05/17 15:57:24 oes
159 * - parse_http_url now checks memory allocation failure for
160 * duplication of "*" URL and rejects "*something" URLs
162 * - Added a comment to what might look like a bug in
163 * create_url_spec (see !bug #736931)
164 * - Comment cosmetics
166 * Revision 1.10.2.6 2003/05/07 12:39:48 oes
167 * Fix typo: Default port for https URLs is 443, not 143.
168 * Thanks to Scott Tregear for spotting this one.
170 * Revision 1.10.2.5 2003/02/28 13:09:29 oes
171 * Fixed a rare double free condition as per Bug #694713
173 * Revision 1.10.2.4 2003/02/28 12:57:44 oes
174 * Moved freeing of http request structure to its owner
175 * as per Dan Price's observations in Bug #694713
177 * Revision 1.10.2.3 2002/11/12 16:50:40 oes
178 * Fixed memory leak in parse_http_request() reported by Oliver Stoeneberg. Fixes bug #637073
180 * Revision 1.10.2.2 2002/09/25 14:53:15 oes
181 * Added basic support for OPTIONS and TRACE HTTP methods:
182 * parse_http_url now recognizes the "*" URI as well as
183 * the OPTIONS and TRACE method keywords.
185 * Revision 1.10.2.1 2002/06/06 19:06:44 jongfoster
186 * Adding support for proprietary Microsoft WebDAV extensions
188 * Revision 1.10 2002/05/12 21:40:37 jongfoster
189 * - Removing some unused code
191 * Revision 1.9 2002/04/04 00:36:36 gliptak
192 * always use pcre for matching
194 * Revision 1.8 2002/04/03 23:32:47 jongfoster
195 * Fixing memory leak on error
197 * Revision 1.7 2002/03/26 22:29:55 swa
198 * we have a new homepage!
200 * Revision 1.6 2002/03/24 13:25:43 swa
201 * name change related issues
203 * Revision 1.5 2002/03/13 00:27:05 jongfoster
206 * Revision 1.4 2002/03/07 03:46:17 oes
207 * Fixed compiler warnings
209 * Revision 1.3 2002/03/03 14:51:11 oes
210 * Fixed CLF logging: Added ocmd member for client's request to struct http_request
212 * Revision 1.2 2002/01/21 00:14:09 jongfoster
213 * Correcting comment style
214 * Fixing an uninitialized memory bug in create_url_spec()
216 * Revision 1.1 2002/01/17 20:53:46 jongfoster
217 * Moving all our URL and URL pattern parsing code to the same file - it
218 * was scattered around in filters.c, loaders.c and parsers.c.
220 * Providing a single, simple url_match(pattern,url) function - rather than
221 * the 3-line match routine which was repeated all over the place.
223 * Renaming free_url to free_url_spec, since it frees a struct url_spec.
225 * Providing parse_http_url() so that URLs can be parsed without faking a
226 * HTTP request line for parse_http_request() or repeating the parsing
227 * code (both of which were techniques that were actually in use).
229 * Standardizing that struct http_request is used to represent a URL, and
230 * struct url_spec is used to represent a URL pattern. (Before, URLs were
231 * represented as seperate variables and a partially-filled-in url_spec).
234 *********************************************************************/
241 #include <sys/types.h>
249 #if !defined(_WIN32) && !defined(__OS2__)
254 #include "urlmatch.h"
256 #include "miscutil.h"
259 const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
261 enum regex_anchoring {NO_ANCHORING, LEFT_ANCHORED, RIGHT_ANCHORED};
262 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern);
264 /*********************************************************************
266 * Function : free_http_request
268 * Description : Freez a http_request structure
271 * 1 : http = points to a http_request structure to free
275 *********************************************************************/
276 void free_http_request(struct http_request *http)
285 freez(http->hostport);
288 freez(http->host_ip_addr_str);
289 freez(http->dbuffer);
295 /*********************************************************************
297 * Function : init_domain_components
299 * Description : Splits the domain name so we can compare it
300 * against wildcards. It used to be part of
301 * parse_http_url, but was separated because the
302 * same code is required in chat in case of
303 * intercepted requests.
306 * 1 : http = pointer to the http structure to hold elements.
308 * Returns : JB_ERR_OK on success
309 * JB_ERR_MEMORY on out of memory
310 * JB_ERR_PARSE on malformed command/URL
311 * or >100 domains deep.
313 *********************************************************************/
314 jb_err init_domain_components(struct http_request *http)
316 char *vec[BUFFER_SIZE];
320 http->dbuffer = strdup(http->host);
321 if (NULL == http->dbuffer)
323 return JB_ERR_MEMORY;
326 /* map to lower case */
327 for (p = http->dbuffer; *p ; p++)
329 *p = (char)tolower((int)(unsigned char)*p);
332 /* split the domain name into components */
333 http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec), 1, 1);
335 if (http->dcount <= 0)
338 * Error: More than SZ(vec) components in domain
339 * or: no components in domain
341 log_error(LOG_LEVEL_ERROR, "More than SZ(vec) components in domain or none at all.");
345 /* save a copy of the pointers in dvec */
346 size = (size_t)http->dcount * sizeof(*http->dvec);
348 http->dvec = (char **)malloc(size);
349 if (NULL == http->dvec)
351 return JB_ERR_MEMORY;
354 memcpy(http->dvec, vec, size);
360 /*********************************************************************
362 * Function : parse_http_url
364 * Description : Parse out the host and port from the URL. Find the
365 * hostname & path, port (if ':'), and/or password (if '@')
368 * 1 : url = URL (or is it URI?) to break down
369 * 2 : http = pointer to the http structure to hold elements.
370 * Must be initialized with valid values (like NULLs).
371 * 3 : csp = Current client state (buffers, headers, etc...)
373 * Returns : JB_ERR_OK on success
374 * JB_ERR_MEMORY on out of memory
375 * JB_ERR_PARSE on malformed command/URL
376 * or >100 domains deep.
378 *********************************************************************/
379 jb_err parse_http_url(const char * url,
380 struct http_request *http,
381 const struct client_state *csp)
383 int host_available = 1; /* A proxy can dream. */
386 * Save our initial URL
388 http->url = strdup(url);
389 if (http->url == NULL)
391 return JB_ERR_MEMORY;
396 * Check for * URI. If found, we're done.
398 if (*http->url == '*')
400 if ( NULL == (http->path = strdup("*"))
401 || NULL == (http->hostport = strdup("")) )
403 return JB_ERR_MEMORY;
405 if (http->url[1] != '\0')
414 * Split URL into protocol,hostport,path.
424 return JB_ERR_MEMORY;
427 /* Find the start of the URL in our scratch space */
429 if (strncmpic(url_noproto, "http://", 7) == 0)
433 else if (strncmpic(url_noproto, "https://", 8) == 0)
436 * Should only happen when called from cgi_show_url_info().
441 else if (*url_noproto == '/')
444 * Short request line without protocol and host.
445 * Most likely because the client's request
446 * was intercepted and redirected into Privoxy.
457 url_path = strchr(url_noproto, '/');
458 if (url_path != NULL)
463 * NOTE: The following line ignores the path for HTTPS URLS.
464 * This means that you get consistent behaviour if you type a
465 * https URL in and it's parsed by the function. (When the
466 * URL is actually retrieved, SSL hides the path part).
468 http->path = strdup(http->ssl ? "/" : url_path);
470 http->hostport = strdup(url_noproto);
475 * Repair broken HTTP requests that don't contain a path,
476 * or CONNECT requests
478 http->path = strdup("/");
479 http->hostport = strdup(url_noproto);
484 if ( (http->path == NULL)
485 || (http->hostport == NULL))
487 return JB_ERR_MEMORY;
493 /* Without host, there is nothing left to do here */
498 * Split hostport into user/password (ignored), host, port.
505 buf = strdup(http->hostport);
508 return JB_ERR_MEMORY;
511 /* check if url contains username and/or password */
512 host = strchr(buf, '@');
515 /* Contains username/password, skip it and the @ sign. */
520 /* No username or password. */
524 /* check if url contains port */
525 port = strchr(host, ':');
529 /* Terminate hostname and point to start of port string */
531 http->port = atoi(port);
535 /* No port specified. */
536 http->port = (http->ssl ? 443 : 80);
539 http->host = strdup(host);
543 if (http->host == NULL)
545 return JB_ERR_MEMORY;
550 * Split domain name so we can compare it against wildcards
552 return init_domain_components(http);
557 /*********************************************************************
559 * Function : unknown_method
561 * Description : Checks whether a method is unknown.
564 * 1 : method = points to a http method
566 * Returns : TRUE if it's unknown, FALSE otherwise.
568 *********************************************************************/
569 static int unknown_method(const char *method)
571 static const char *known_http_methods[] = {
572 /* Basic HTTP request type */
573 "GET", "HEAD", "POST", "PUT", "DELETE", "OPTIONS", "TRACE", "CONNECT",
574 /* webDAV extensions (RFC2518) */
575 "PROPFIND", "PROPPATCH", "MOVE", "COPY", "MKCOL", "LOCK", "UNLOCK",
577 * Microsoft webDAV extension for Exchange 2000. See:
578 * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
579 * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
581 "BCOPY", "BMOVE", "BDELETE", "BPROPFIND", "BPROPPATCH",
583 * Another Microsoft webDAV extension for Exchange 2000. See:
584 * http://systems.cs.colorado.edu/grunwald/MobileComputing/Papers/draft-cohen-gena-p-base-00.txt
585 * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
586 * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
588 "SUBSCRIBE", "UNSUBSCRIBE", "NOTIFY", "POLL",
590 * Yet another WebDAV extension, this time for
591 * Web Distributed Authoring and Versioning (RFC3253)
593 "VERSION-CONTROL", "REPORT", "CHECKOUT", "CHECKIN", "UNCHECKOUT",
594 "MKWORKSPACE", "UPDATE", "LABEL", "MERGE", "BASELINE-CONTROL", "MKACTIVITY",
598 for (i = 0; i < SZ(known_http_methods); i++)
600 if (0 == strcmpic(method, known_http_methods[i]))
611 /*********************************************************************
613 * Function : parse_http_request
615 * Description : Parse out the host and port from the URL. Find the
616 * hostname & path, port (if ':'), and/or password (if '@')
619 * 1 : req = HTTP request line to break down
620 * 2 : http = pointer to the http structure to hold elements
621 * 3 : csp = Current client state (buffers, headers, etc...)
623 * Returns : JB_ERR_OK on success
624 * JB_ERR_MEMORY on out of memory
625 * JB_ERR_CGI_PARAMS on malformed command/URL
626 * or >100 domains deep.
628 *********************************************************************/
629 jb_err parse_http_request(const char *req,
630 struct http_request *http,
631 const struct client_state *csp)
634 char *v[10]; /* XXX: Why 10? We should only need three. */
638 memset(http, '\0', sizeof(*http));
643 return JB_ERR_MEMORY;
646 n = ssplit(buf, " \r\n", v, SZ(v), 1, 1);
654 * Fail in case of unknown methods
655 * which we might not handle correctly.
657 * XXX: There should be a config option
658 * to forward requests with unknown methods
659 * anyway. Most of them don't need special
662 if (unknown_method(v[0]))
664 log_error(LOG_LEVEL_ERROR, "Unknown HTTP method detected: %s", v[0]);
669 http->ssl = !strcmpic(v[0], "CONNECT");
671 err = parse_http_url(v[1], http, csp);
679 * Copy the details into the structure
681 http->cmd = strdup(req);
682 http->gpc = strdup(v[0]);
683 http->ver = strdup(v[2]);
687 if ( (http->cmd == NULL)
688 || (http->gpc == NULL)
689 || (http->ver == NULL) )
691 return JB_ERR_MEMORY;
699 /*********************************************************************
701 * Function : compile_pattern
703 * Description : Compiles a host, domain or TAG pattern.
706 * 1 : pattern = The pattern to compile.
707 * 2 : anchoring = How the regex should be anchored.
708 * Can be either one of NO_ANCHORING,
709 * LEFT_ANCHORED or RIGHT_ANCHORED.
710 * 3 : url = In case of failures, the spec member is
711 * logged and the structure freed.
712 * 4 : regex = Where the compiled regex should be stored.
714 * Returns : JB_ERR_OK - Success
715 * JB_ERR_MEMORY - Out of memory
716 * JB_ERR_PARSE - Cannot parse regex
718 *********************************************************************/
719 static jb_err compile_pattern(const char *pattern, enum regex_anchoring anchoring,
720 struct url_spec *url, regex_t **regex)
723 char rebuf[BUFFER_SIZE];
727 assert(strlen(pattern) < sizeof(rebuf) - 2);
729 if (pattern[0] == '\0')
747 log_error(LOG_LEVEL_FATAL,
748 "Invalid anchoring in compile_pattern %d", anchoring);
751 *regex = zalloc(sizeof(**regex));
755 return JB_ERR_MEMORY;
758 snprintf(rebuf, sizeof(rebuf), fmt, pattern);
760 errcode = regcomp(*regex, rebuf, (REG_EXTENDED|REG_NOSUB|REG_ICASE));
764 size_t errlen = regerror(errcode, *regex, rebuf, sizeof(rebuf));
765 if (errlen > (sizeof(rebuf) - (size_t)1))
767 errlen = sizeof(rebuf) - (size_t)1;
769 rebuf[errlen] = '\0';
770 log_error(LOG_LEVEL_ERROR, "error compiling %s from %s: %s",
771 pattern, url->spec, rebuf);
782 /*********************************************************************
784 * Function : compile_url_pattern
786 * Description : Compiles the three parts of an URL pattern.
789 * 1 : url = Target url_spec to be filled in.
790 * 2 : buf = The url pattern to compile. Will be messed up.
792 * Returns : JB_ERR_OK - Success
793 * JB_ERR_MEMORY - Out of memory
794 * JB_ERR_PARSE - Cannot parse regex
796 *********************************************************************/
797 static jb_err compile_url_pattern(struct url_spec *url, char *buf)
801 p = strchr(buf, '/');
805 * Only compile the regex if it consists of more than
806 * a single slash, otherwise it wouldn't affect the result.
811 * XXX: does it make sense to compile the slash at the beginning?
813 jb_err err = compile_pattern(p, LEFT_ANCHORED, url, &url->preg);
815 if (JB_ERR_OK != err)
823 p = strchr(buf, ':');
827 url->port_list = strdup(p);
828 if (NULL == url->port_list)
830 return JB_ERR_MEMORY;
835 url->port_list = NULL;
840 return compile_host_pattern(url, buf);
848 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
849 /*********************************************************************
851 * Function : compile_host_pattern
853 * Description : Parses and compiles a host pattern..
856 * 1 : url = Target url_spec to be filled in.
857 * 2 : host_pattern = Host pattern to compile.
859 * Returns : JB_ERR_OK - Success
860 * JB_ERR_MEMORY - Out of memory
861 * JB_ERR_PARSE - Cannot parse regex
863 *********************************************************************/
864 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
866 return compile_pattern(host_pattern, RIGHT_ANCHORED, url, &url->host_regex);
871 /*********************************************************************
873 * Function : compile_host_pattern
875 * Description : Parses and "compiles" an old-school host pattern.
878 * 1 : url = Target url_spec to be filled in.
879 * 2 : host_pattern = Host pattern to parse.
881 * Returns : JB_ERR_OK - Success
882 * JB_ERR_MEMORY - Out of memory
883 * JB_ERR_PARSE - Cannot parse regex
885 *********************************************************************/
886 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
895 if (host_pattern[strlen(host_pattern) - 1] == '.')
897 url->unanchored |= ANCHOR_RIGHT;
899 if (host_pattern[0] == '.')
901 url->unanchored |= ANCHOR_LEFT;
905 * Split domain into components
907 url->dbuffer = strdup(host_pattern);
908 if (NULL == url->dbuffer)
911 return JB_ERR_MEMORY;
917 for (p = url->dbuffer; *p ; p++)
919 *p = (char)tolower((int)(unsigned char)*p);
923 * Split the domain name into components
925 url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1);
930 return JB_ERR_MEMORY;
932 else if (url->dcount != 0)
935 * Save a copy of the pointers in dvec
937 size = (size_t)url->dcount * sizeof(*url->dvec);
939 url->dvec = (char **)malloc(size);
940 if (NULL == url->dvec)
943 return JB_ERR_MEMORY;
946 memcpy(url->dvec, v, size);
949 * else dcount == 0 in which case we needn't do anything,
950 * since dvec will never be accessed and the pattern will
957 /*********************************************************************
959 * Function : simplematch
961 * Description : String matching, with a (greedy) '*' wildcard that
962 * stands for zero or more arbitrary characters and
963 * character classes in [], which take both enumerations
967 * 1 : pattern = pattern for matching
968 * 2 : text = text to be matched
970 * Returns : 0 if match, else nonzero
972 *********************************************************************/
973 static int simplematch(const char *pattern, const char *text)
975 const unsigned char *pat = (const unsigned char *)pattern;
976 const unsigned char *txt = (const unsigned char *)text;
977 const unsigned char *fallback = pat;
980 unsigned char lastchar = 'a';
982 unsigned char charmap[32];
987 /* EOF pattern but !EOF text? */
1000 /* '*' in the pattern? */
1004 /* The pattern ends afterwards? Speed up the return. */
1010 /* Else, set wildcard mode and remember position after '*' */
1015 /* Character range specification? */
1018 memset(charmap, '\0', sizeof(charmap));
1020 while (*++pat != ']')
1026 else if (*pat == '-')
1028 if ((*++pat == ']') || *pat == '\0')
1032 for (i = lastchar; i <= *pat; i++)
1034 charmap[i / 8] |= (unsigned char)(1 << (i % 8));
1039 charmap[*pat / 8] |= (unsigned char)(1 << (*pat % 8));
1043 } /* -END- if Character range specification */
1047 * Char match, or char range match?
1051 || ((*pat == ']') && (charmap[*txt / 8] & (1 << (*txt % 8)))) )
1061 * No match && no wildcard: No luck
1065 else if (pat != fallback)
1068 * Increment text pointer if in char range matching
1075 * Wildcard mode && nonmatch beyond fallback: Rewind pattern
1079 * Restart matching from current text pointer
1086 /* Cut off extra '*'s */
1087 if(*pat == '*') pat++;
1089 /* If this is the pattern's end, fine! */
1095 /*********************************************************************
1097 * Function : simple_domaincmp
1099 * Description : Domain-wise Compare fqdn's. The comparison is
1100 * both left- and right-anchored. The individual
1101 * domain names are compared with simplematch().
1102 * This is only used by domain_match.
1105 * 1 : pv = array of patterns to compare
1106 * 2 : fv = array of domain components to compare
1107 * 3 : len = length of the arrays (both arrays are the
1108 * same length - if they weren't, it couldn't
1109 * possibly be a match).
1111 * Returns : 0 => domains are equivalent, else no match.
1113 *********************************************************************/
1114 static int simple_domaincmp(char **pv, char **fv, int len)
1118 for (n = 0; n < len; n++)
1120 if (simplematch(pv[n], fv[n]))
1131 /*********************************************************************
1133 * Function : domain_match
1135 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
1136 * pattern->unachored, the comparison is un-, left-,
1137 * right-anchored, or both.
1138 * The individual domain names are compared with
1142 * 1 : pattern = a domain that may contain a '*' as a wildcard.
1143 * 2 : fqdn = domain name against which the patterns are compared.
1145 * Returns : 0 => domains are equivalent, else no match.
1147 *********************************************************************/
1148 static int domain_match(const struct url_spec *pattern, const struct http_request *fqdn)
1150 char **pv, **fv; /* vectors */
1152 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
1154 plen = pattern->dcount;
1155 flen = fqdn->dcount;
1159 /* fqdn is too short to match this pattern */
1166 if (unanchored == ANCHOR_LEFT)
1171 * Convert this into a fully anchored pattern with
1172 * the fqdn and pattern the same length
1174 fv += (flen - plen); /* flen - plen >= 0 due to check above */
1175 return simple_domaincmp(pv, fv, plen);
1177 else if (unanchored == 0)
1179 /* Fully anchored, check length */
1184 return simple_domaincmp(pv, fv, plen);
1186 else if (unanchored == ANCHOR_RIGHT)
1188 /* Left anchored, ignore all extra in fqdn */
1189 return simple_domaincmp(pv, fv, plen);
1195 int maxn = flen - plen;
1196 for (n = 0; n <= maxn; n++)
1198 if (!simple_domaincmp(pv, fv, plen))
1203 * Doesn't match from start of fqdn
1204 * Try skipping first part of fqdn
1212 #endif /* def FEATURE_EXTENDED_HOST_PATTERNS */
1215 /*********************************************************************
1217 * Function : create_url_spec
1219 * Description : Creates a "url_spec" structure from a string.
1220 * When finished, free with free_url_spec().
1223 * 1 : url = Target url_spec to be filled in. Will be
1224 * zeroed before use.
1225 * 2 : buf = Source pattern, null terminated. NOTE: The
1226 * contents of this buffer are destroyed by this
1227 * function. If this function succeeds, the
1228 * buffer is copied to url->spec. If this
1229 * function fails, the contents of the buffer
1232 * Returns : JB_ERR_OK - Success
1233 * JB_ERR_MEMORY - Out of memory
1234 * JB_ERR_PARSE - Cannot parse regex (Detailed message
1235 * written to system log)
1237 *********************************************************************/
1238 jb_err create_url_spec(struct url_spec *url, char *buf)
1243 memset(url, '\0', sizeof(*url));
1245 /* Remember the original specification for the CGI pages. */
1246 url->spec = strdup(buf);
1247 if (NULL == url->spec)
1249 return JB_ERR_MEMORY;
1252 /* Is it tag pattern? */
1253 if (0 == strncmpic("TAG:", url->spec, 4))
1255 /* The pattern starts with the first character after "TAG:" */
1256 const char *tag_pattern = buf + 4;
1257 return compile_pattern(tag_pattern, NO_ANCHORING, url, &url->tag_regex);
1260 /* If it isn't a tag pattern it must be a URL pattern. */
1261 return compile_url_pattern(url, buf);
1265 /*********************************************************************
1267 * Function : free_url_spec
1269 * Description : Called from the "unloaders". Freez the url
1270 * structure elements.
1273 * 1 : url = pointer to a url_spec structure.
1277 *********************************************************************/
1278 void free_url_spec(struct url_spec *url)
1280 if (url == NULL) return;
1283 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
1284 if (url->host_regex)
1286 regfree(url->host_regex);
1287 freez(url->host_regex);
1290 freez(url->dbuffer);
1293 #endif /* ndef FEATURE_EXTENDED_HOST_PATTERNS */
1294 freez(url->port_list);
1302 regfree(url->tag_regex);
1303 freez(url->tag_regex);
1308 /*********************************************************************
1310 * Function : url_match
1312 * Description : Compare a URL against a URL pattern.
1315 * 1 : pattern = a URL pattern
1316 * 2 : url = URL to match
1318 * Returns : Nonzero if the URL matches the pattern, else 0.
1320 *********************************************************************/
1321 int url_match(const struct url_spec *pattern,
1322 const struct http_request *http)
1324 /* XXX: these should probably be functions. */
1325 #define PORT_MATCHES ((NULL == pattern->port_list) || match_portlist(pattern->port_list, http->port))
1326 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
1327 #define DOMAIN_MATCHES ((NULL == pattern->host_regex) || (0 == regexec(pattern->host_regex, http->host, 0, NULL, 0)))
1329 #define DOMAIN_MATCHES ((NULL == pattern->dbuffer) || (0 == domain_match(pattern, http)))
1331 #define PATH_MATCHES ((NULL == pattern->preg) || (0 == regexec(pattern->preg, http->path, 0, NULL, 0)))
1333 if (pattern->tag_regex != NULL)
1335 /* It's a tag pattern and shouldn't be matched against URLs */
1339 return (PORT_MATCHES && DOMAIN_MATCHES && PATH_MATCHES);
1344 /*********************************************************************
1346 * Function : match_portlist
1348 * Description : Check if a given number is covered by a comma
1349 * separated list of numbers and ranges (a,b-c,d,..)
1352 * 1 : portlist = String with list
1353 * 2 : port = port to check
1355 * Returns : 0 => no match
1358 *********************************************************************/
1359 int match_portlist(const char *portlist, int port)
1361 char *min, *max, *next, *portlist_copy;
1363 min = next = portlist_copy = strdup(portlist);
1366 * Zero-terminate first item and remember offset for next
1368 if (NULL != (next = strchr(portlist_copy, (int) ',')))
1374 * Loop through all items, checking for match
1378 if (NULL == (max = strchr(min, (int) '-')))
1381 * No dash, check for equality
1383 if (port == atoi(min))
1385 freez(portlist_copy);
1392 * This is a range, so check if between min and max,
1393 * or, if max was omitted, between min and 65K
1396 if(port >= atoi(min) && port <= (atoi(max) ? atoi(max) : 65535))
1398 freez(portlist_copy);
1410 * Zero-terminate next item and remember offset for n+1
1412 if ((NULL != next) && (NULL != (next = strchr(next, (int) ','))))
1418 freez(portlist_copy);