1 const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.39 2008/04/22 16:27:42 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 1.39 2008/04/22 16:27:42 fabiankeil
37 * In parse_http_request(), remove a pointless
38 * temporary variable and free the buffer earlier.
40 * Revision 1.38 2008/04/18 05:17:18 fabiankeil
41 * Mark simplematch()'s parameters as immutable.
43 * Revision 1.37 2008/04/17 14:53:29 fabiankeil
44 * Move simplematch() into urlmatch.c as it's only
45 * used to match (old-school) domain patterns.
47 * Revision 1.36 2008/04/14 18:19:48 fabiankeil
48 * Remove now-pointless cast in create_url_spec().
50 * Revision 1.35 2008/04/14 18:11:21 fabiankeil
51 * The compiler might not notice it, but the buffer passed to
52 * create_url_spec() is modified later on and thus shouldn't
53 * be declared immutable.
55 * Revision 1.34 2008/04/13 13:32:07 fabiankeil
56 * Factor URL pattern compilation out of create_url_spec().
58 * Revision 1.33 2008/04/12 14:03:13 fabiankeil
59 * Remove an obvious comment and improve another one.
61 * Revision 1.32 2008/04/12 12:38:06 fabiankeil
62 * Factor out duplicated code to compile host, path and tag patterns.
64 * Revision 1.31 2008/04/10 14:41:04 fabiankeil
65 * Ditch url_spec's path member now that it's no longer used.
67 * Revision 1.30 2008/04/10 04:24:24 fabiankeil
68 * Stop duplicating the plain text representation of the path regex
69 * (and keeping the copy around). Once the regex is compiled it's no
72 * Revision 1.29 2008/04/10 04:17:56 fabiankeil
73 * In url_match(), check the right member for NULL when determining
74 * whether there's a path regex to execute. Looking for a plain-text
75 * representation works as well, but it looks "interesting" and that
76 * member will be removed soonish anyway.
78 * Revision 1.28 2008/04/08 16:07:39 fabiankeil
79 * Make it harder to mistake url_match()'s
80 * second parameter for an url_spec.
82 * Revision 1.27 2008/04/08 15:44:33 fabiankeil
83 * Save a bit of memory (and a few cpu cycles) by not bothering to
84 * compile slash-only path regexes that don't affect the result.
86 * Revision 1.26 2008/04/07 16:57:18 fabiankeil
87 * - Use free_url_spec() more consistently.
88 * - Let it reset url->dcount just in case.
90 * Revision 1.25 2008/04/06 15:18:38 fabiankeil
91 * Oh well, rename the --enable-pcre-host-patterns option to
92 * --enable-extended-host-patterns as it's not really PCRE syntax.
94 * Revision 1.24 2008/04/06 14:54:26 fabiankeil
95 * Use PCRE syntax in host patterns when configured
96 * with --enable-pcre-host-patterns.
98 * Revision 1.23 2008/04/05 12:19:20 fabiankeil
99 * Factor compile_host_pattern() out of create_url_spec().
101 * Revision 1.22 2008/03/30 15:02:32 fabiankeil
102 * SZitify unknown_method().
104 * Revision 1.21 2007/12/24 16:34:23 fabiankeil
105 * Band-aid (and micro-optimization) that makes it less likely to run out of
106 * stack space with overly-complex path patterns. Probably masks the problem
107 * reported by Lee in #1856679. Hohoho.
109 * Revision 1.20 2007/09/02 15:31:20 fabiankeil
110 * Move match_portlist() from filter.c to urlmatch.c.
111 * It's used for url matching, not for filtering.
113 * Revision 1.19 2007/09/02 13:42:11 fabiankeil
114 * - Allow port lists in url patterns.
115 * - Ditch unused url_spec member pathlen.
117 * Revision 1.18 2007/07/30 16:42:21 fabiankeil
118 * Move the method check into unknown_method()
119 * and loop through the known methods instead
120 * of using a screen-long OR chain.
122 * Revision 1.17 2007/04/15 16:39:21 fabiankeil
123 * Introduce tags as alternative way to specify which
124 * actions apply to a request. At the moment tags can be
125 * created based on client and server headers.
127 * Revision 1.16 2007/02/13 13:59:24 fabiankeil
128 * Remove redundant log message.
130 * Revision 1.15 2007/01/28 16:11:23 fabiankeil
131 * Accept WebDAV methods for subversion
132 * in parse_http_request(). Closes FR 1581425.
134 * Revision 1.14 2007/01/06 14:23:56 fabiankeil
135 * Fix gcc43 warnings. Mark *csp as immutable
136 * for parse_http_url() and url_match().
137 * Replace a sprintf call with snprintf.
139 * Revision 1.13 2006/12/06 19:50:54 fabiankeil
140 * parse_http_url() now handles intercepted
141 * HTTP request lines as well. Moved parts
142 * of parse_http_url()'s code into
143 * init_domain_components() so that it can
144 * be reused in chat().
146 * Revision 1.12 2006/07/18 14:48:47 david__schmidt
147 * Reorganizing the repository: swapping out what was HEAD (the old 3.1 branch)
148 * with what was really the latest development (the v_3_0_branch branch)
150 * Revision 1.10.2.7 2003/05/17 15:57:24 oes
151 * - parse_http_url now checks memory allocation failure for
152 * duplication of "*" URL and rejects "*something" URLs
154 * - Added a comment to what might look like a bug in
155 * create_url_spec (see !bug #736931)
156 * - Comment cosmetics
158 * Revision 1.10.2.6 2003/05/07 12:39:48 oes
159 * Fix typo: Default port for https URLs is 443, not 143.
160 * Thanks to Scott Tregear for spotting this one.
162 * Revision 1.10.2.5 2003/02/28 13:09:29 oes
163 * Fixed a rare double free condition as per Bug #694713
165 * Revision 1.10.2.4 2003/02/28 12:57:44 oes
166 * Moved freeing of http request structure to its owner
167 * as per Dan Price's observations in Bug #694713
169 * Revision 1.10.2.3 2002/11/12 16:50:40 oes
170 * Fixed memory leak in parse_http_request() reported by Oliver Stoeneberg. Fixes bug #637073
172 * Revision 1.10.2.2 2002/09/25 14:53:15 oes
173 * Added basic support for OPTIONS and TRACE HTTP methods:
174 * parse_http_url now recognizes the "*" URI as well as
175 * the OPTIONS and TRACE method keywords.
177 * Revision 1.10.2.1 2002/06/06 19:06:44 jongfoster
178 * Adding support for proprietary Microsoft WebDAV extensions
180 * Revision 1.10 2002/05/12 21:40:37 jongfoster
181 * - Removing some unused code
183 * Revision 1.9 2002/04/04 00:36:36 gliptak
184 * always use pcre for matching
186 * Revision 1.8 2002/04/03 23:32:47 jongfoster
187 * Fixing memory leak on error
189 * Revision 1.7 2002/03/26 22:29:55 swa
190 * we have a new homepage!
192 * Revision 1.6 2002/03/24 13:25:43 swa
193 * name change related issues
195 * Revision 1.5 2002/03/13 00:27:05 jongfoster
198 * Revision 1.4 2002/03/07 03:46:17 oes
199 * Fixed compiler warnings
201 * Revision 1.3 2002/03/03 14:51:11 oes
202 * Fixed CLF logging: Added ocmd member for client's request to struct http_request
204 * Revision 1.2 2002/01/21 00:14:09 jongfoster
205 * Correcting comment style
206 * Fixing an uninitialized memory bug in create_url_spec()
208 * Revision 1.1 2002/01/17 20:53:46 jongfoster
209 * Moving all our URL and URL pattern parsing code to the same file - it
210 * was scattered around in filters.c, loaders.c and parsers.c.
212 * Providing a single, simple url_match(pattern,url) function - rather than
213 * the 3-line match routine which was repeated all over the place.
215 * Renaming free_url to free_url_spec, since it frees a struct url_spec.
217 * Providing parse_http_url() so that URLs can be parsed without faking a
218 * HTTP request line for parse_http_request() or repeating the parsing
219 * code (both of which were techniques that were actually in use).
221 * Standardizing that struct http_request is used to represent a URL, and
222 * struct url_spec is used to represent a URL pattern. (Before, URLs were
223 * represented as seperate variables and a partially-filled-in url_spec).
226 *********************************************************************/
233 #include <sys/types.h>
241 #if !defined(_WIN32) && !defined(__OS2__)
246 #include "urlmatch.h"
248 #include "miscutil.h"
251 const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
253 enum regex_anchoring {NO_ANCHORING, LEFT_ANCHORED, RIGHT_ANCHORED};
254 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern);
256 /*********************************************************************
258 * Function : free_http_request
260 * Description : Freez a http_request structure
263 * 1 : http = points to a http_request structure to free
267 *********************************************************************/
268 void free_http_request(struct http_request *http)
277 freez(http->hostport);
280 freez(http->host_ip_addr_str);
281 freez(http->dbuffer);
287 /*********************************************************************
289 * Function : init_domain_components
291 * Description : Splits the domain name so we can compare it
292 * against wildcards. It used to be part of
293 * parse_http_url, but was separated because the
294 * same code is required in chat in case of
295 * intercepted requests.
298 * 1 : http = pointer to the http structure to hold elements.
300 * Returns : JB_ERR_OK on success
301 * JB_ERR_MEMORY on out of memory
302 * JB_ERR_PARSE on malformed command/URL
303 * or >100 domains deep.
305 *********************************************************************/
306 jb_err init_domain_components(struct http_request *http)
308 char *vec[BUFFER_SIZE];
312 http->dbuffer = strdup(http->host);
313 if (NULL == http->dbuffer)
315 return JB_ERR_MEMORY;
318 /* map to lower case */
319 for (p = http->dbuffer; *p ; p++)
321 *p = (char)tolower((int)(unsigned char)*p);
324 /* split the domain name into components */
325 http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec), 1, 1);
327 if (http->dcount <= 0)
330 * Error: More than SZ(vec) components in domain
331 * or: no components in domain
333 log_error(LOG_LEVEL_ERROR, "More than SZ(vec) components in domain or none at all.");
337 /* save a copy of the pointers in dvec */
338 size = (size_t)http->dcount * sizeof(*http->dvec);
340 http->dvec = (char **)malloc(size);
341 if (NULL == http->dvec)
343 return JB_ERR_MEMORY;
346 memcpy(http->dvec, vec, size);
352 /*********************************************************************
354 * Function : parse_http_url
356 * Description : Parse out the host and port from the URL. Find the
357 * hostname & path, port (if ':'), and/or password (if '@')
360 * 1 : url = URL (or is it URI?) to break down
361 * 2 : http = pointer to the http structure to hold elements.
362 * Will be zeroed before use. Note that this
363 * function sets the http->gpc and http->ver
365 * 3 : csp = Current client state (buffers, headers, etc...)
367 * Returns : JB_ERR_OK on success
368 * JB_ERR_MEMORY on out of memory
369 * JB_ERR_PARSE on malformed command/URL
370 * or >100 domains deep.
372 *********************************************************************/
373 jb_err parse_http_url(const char * url,
374 struct http_request *http,
375 const struct client_state *csp)
377 int host_available = 1; /* A proxy can dream. */
380 * Zero out the results structure
382 memset(http, '\0', sizeof(*http));
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)
434 else if (strncmpic(url_noproto, "https://", 8) == 0)
439 else if (*url_noproto == '/')
442 * Short request line without protocol and host.
443 * Most likely because the client's request
444 * was intercepted and redirected into Privoxy.
455 url_path = strchr(url_noproto, '/');
456 if (url_path != NULL)
461 * NOTE: The following line ignores the path for HTTPS URLS.
462 * This means that you get consistent behaviour if you type a
463 * https URL in and it's parsed by the function. (When the
464 * URL is actually retrieved, SSL hides the path part).
466 http->path = strdup(http->ssl ? "/" : url_path);
468 http->hostport = strdup(url_noproto);
473 * Repair broken HTTP requests that don't contain a path,
474 * or CONNECT requests
476 http->path = strdup("/");
477 http->hostport = strdup(url_noproto);
482 if ( (http->path == NULL)
483 || (http->hostport == NULL))
485 return JB_ERR_MEMORY;
491 /* Without host, there is nothing left to do here */
496 * Split hostport into user/password (ignored), host, port.
503 buf = strdup(http->hostport);
506 return JB_ERR_MEMORY;
509 /* check if url contains username and/or password */
510 host = strchr(buf, '@');
513 /* Contains username/password, skip it and the @ sign. */
518 /* No username or password. */
522 /* check if url contains port */
523 port = strchr(host, ':');
527 /* Terminate hostname and point to start of port string */
529 http->port = atoi(port);
533 /* No port specified. */
534 http->port = (http->ssl ? 443 : 80);
537 http->host = strdup(host);
541 if (http->host == NULL)
543 return JB_ERR_MEMORY;
548 * Split domain name so we can compare it against wildcards
550 return init_domain_components(http);
555 /*********************************************************************
557 * Function : unknown_method
559 * Description : Checks whether a method is unknown.
562 * 1 : method = points to a http method
564 * Returns : TRUE if it's unknown, FALSE otherwise.
566 *********************************************************************/
567 static int unknown_method(const char *method)
569 static const char *known_http_methods[] = {
570 /* Basic HTTP request type */
571 "GET", "HEAD", "POST", "PUT", "DELETE", "OPTIONS", "TRACE", "CONNECT",
572 /* webDAV extensions (RFC2518) */
573 "PROPFIND", "PROPPATCH", "MOVE", "COPY", "MKCOL", "LOCK", "UNLOCK",
575 * Microsoft webDAV extension for Exchange 2000. See:
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 "BCOPY", "BMOVE", "BDELETE", "BPROPFIND", "BPROPPATCH",
581 * Another Microsoft webDAV extension for Exchange 2000. See:
582 * http://systems.cs.colorado.edu/grunwald/MobileComputing/Papers/draft-cohen-gena-p-base-00.txt
583 * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
584 * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
586 "SUBSCRIBE", "UNSUBSCRIBE", "NOTIFY", "POLL",
588 * Yet another WebDAV extension, this time for
589 * Web Distributed Authoring and Versioning (RFC3253)
591 "VERSION-CONTROL", "REPORT", "CHECKOUT", "CHECKIN", "UNCHECKOUT",
592 "MKWORKSPACE", "UPDATE", "LABEL", "MERGE", "BASELINE-CONTROL", "MKACTIVITY",
596 for (i = 0; i < SZ(known_http_methods); i++)
598 if (0 == strcmpic(method, known_http_methods[i]))
609 /*********************************************************************
611 * Function : parse_http_request
613 * Description : Parse out the host and port from the URL. Find the
614 * hostname & path, port (if ':'), and/or password (if '@')
617 * 1 : req = HTTP request line to break down
618 * 2 : http = pointer to the http structure to hold elements
619 * 3 : csp = Current client state (buffers, headers, etc...)
621 * Returns : JB_ERR_OK on success
622 * JB_ERR_MEMORY on out of memory
623 * JB_ERR_CGI_PARAMS on malformed command/URL
624 * or >100 domains deep.
626 *********************************************************************/
627 jb_err parse_http_request(const char *req,
628 struct http_request *http,
629 const struct client_state *csp)
632 char *v[10]; /* XXX: Why 10? We should only need three. */
636 memset(http, '\0', sizeof(*http));
641 return JB_ERR_MEMORY;
644 n = ssplit(buf, " \r\n", v, SZ(v), 1, 1);
652 * Fail in case of unknown methods
653 * which we might not handle correctly.
655 * XXX: There should be a config option
656 * to forward requests with unknown methods
657 * anyway. Most of them don't need special
660 if (unknown_method(v[0]))
662 log_error(LOG_LEVEL_ERROR, "Unknown HTTP method detected: %s", v[0]);
667 err = parse_http_url(v[1], http, csp);
675 * Copy the details into the structure
677 http->ssl = !strcmpic(v[0], "CONNECT");
678 http->cmd = strdup(req);
679 http->gpc = strdup(v[0]);
680 http->ver = strdup(v[2]);
684 if ( (http->cmd == NULL)
685 || (http->gpc == NULL)
686 || (http->ver == NULL) )
688 return JB_ERR_MEMORY;
696 /*********************************************************************
698 * Function : compile_pattern
700 * Description : Compiles a host, domain or TAG pattern.
703 * 1 : pattern = The pattern to compile.
704 * 2 : anchoring = How the regex should be anchored.
705 * Can be either one of NO_ANCHORING,
706 * LEFT_ANCHORED or RIGHT_ANCHORED.
707 * 3 : url = In case of failures, the spec member is
708 * logged and the structure freed.
709 * 4 : regex = Where the compiled regex should be stored.
711 * Returns : JB_ERR_OK - Success
712 * JB_ERR_MEMORY - Out of memory
713 * JB_ERR_PARSE - Cannot parse regex
715 *********************************************************************/
716 static jb_err compile_pattern(const char *pattern, enum regex_anchoring anchoring,
717 struct url_spec *url, regex_t **regex)
720 char rebuf[BUFFER_SIZE];
724 assert(strlen(pattern) < sizeof(rebuf) - 2);
726 if (pattern[0] == '\0')
744 log_error(LOG_LEVEL_FATAL,
745 "Invalid anchoring in compile_pattern %d", anchoring);
748 *regex = zalloc(sizeof(**regex));
752 return JB_ERR_MEMORY;
755 snprintf(rebuf, sizeof(rebuf), fmt, pattern);
757 errcode = regcomp(*regex, rebuf, (REG_EXTENDED|REG_NOSUB|REG_ICASE));
761 size_t errlen = regerror(errcode, *regex, rebuf, sizeof(rebuf));
762 if (errlen > (sizeof(rebuf) - (size_t)1))
764 errlen = sizeof(rebuf) - (size_t)1;
766 rebuf[errlen] = '\0';
767 log_error(LOG_LEVEL_ERROR, "error compiling %s from %s: %s",
768 pattern, url->spec, rebuf);
779 /*********************************************************************
781 * Function : compile_url_pattern
783 * Description : Compiles the three parts of an URL pattern.
786 * 1 : url = Target url_spec to be filled in.
787 * 2 : buf = The url pattern to compile. Will be messed up.
789 * Returns : JB_ERR_OK - Success
790 * JB_ERR_MEMORY - Out of memory
791 * JB_ERR_PARSE - Cannot parse regex
793 *********************************************************************/
794 static jb_err compile_url_pattern(struct url_spec *url, char *buf)
798 p = strchr(buf, '/');
802 * Only compile the regex if it consists of more than
803 * a single slash, otherwise it wouldn't affect the result.
808 * XXX: does it make sense to compile the slash at the beginning?
810 jb_err err = compile_pattern(p, LEFT_ANCHORED, url, &url->preg);
812 if (JB_ERR_OK != err)
820 p = strchr(buf, ':');
824 url->port_list = strdup(p);
825 if (NULL == url->port_list)
827 return JB_ERR_MEMORY;
832 url->port_list = NULL;
837 return compile_host_pattern(url, buf);
845 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
846 /*********************************************************************
848 * Function : compile_host_pattern
850 * Description : Parses and compiles a host pattern..
853 * 1 : url = Target url_spec to be filled in.
854 * 2 : host_pattern = Host pattern to compile.
856 * Returns : JB_ERR_OK - Success
857 * JB_ERR_MEMORY - Out of memory
858 * JB_ERR_PARSE - Cannot parse regex
860 *********************************************************************/
861 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
863 return compile_pattern(host_pattern, RIGHT_ANCHORED, url, &url->host_regex);
868 /*********************************************************************
870 * Function : compile_host_pattern
872 * Description : Parses and "compiles" an old-school host pattern.
875 * 1 : url = Target url_spec to be filled in.
876 * 2 : host_pattern = Host pattern to parse.
878 * Returns : JB_ERR_OK - Success
879 * JB_ERR_MEMORY - Out of memory
880 * JB_ERR_PARSE - Cannot parse regex
882 *********************************************************************/
883 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
892 if (host_pattern[strlen(host_pattern) - 1] == '.')
894 url->unanchored |= ANCHOR_RIGHT;
896 if (host_pattern[0] == '.')
898 url->unanchored |= ANCHOR_LEFT;
902 * Split domain into components
904 url->dbuffer = strdup(host_pattern);
905 if (NULL == url->dbuffer)
908 return JB_ERR_MEMORY;
914 for (p = url->dbuffer; *p ; p++)
916 *p = (char)tolower((int)(unsigned char)*p);
920 * Split the domain name into components
922 url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1);
927 return JB_ERR_MEMORY;
929 else if (url->dcount != 0)
932 * Save a copy of the pointers in dvec
934 size = (size_t)url->dcount * sizeof(*url->dvec);
936 url->dvec = (char **)malloc(size);
937 if (NULL == url->dvec)
940 return JB_ERR_MEMORY;
943 memcpy(url->dvec, v, size);
946 * else dcount == 0 in which case we needn't do anything,
947 * since dvec will never be accessed and the pattern will
954 /*********************************************************************
956 * Function : simplematch
958 * Description : String matching, with a (greedy) '*' wildcard that
959 * stands for zero or more arbitrary characters and
960 * character classes in [], which take both enumerations
964 * 1 : pattern = pattern for matching
965 * 2 : text = text to be matched
967 * Returns : 0 if match, else nonzero
969 *********************************************************************/
970 static int simplematch(const char *pattern, const char *text)
972 const unsigned char *pat = (const unsigned char *)pattern;
973 const unsigned char *txt = (const unsigned char *)text;
974 const unsigned char *fallback = pat;
977 unsigned char lastchar = 'a';
979 unsigned char charmap[32];
984 /* EOF pattern but !EOF text? */
997 /* '*' in the pattern? */
1001 /* The pattern ends afterwards? Speed up the return. */
1007 /* Else, set wildcard mode and remember position after '*' */
1012 /* Character range specification? */
1015 memset(charmap, '\0', sizeof(charmap));
1017 while (*++pat != ']')
1023 else if (*pat == '-')
1025 if ((*++pat == ']') || *pat == '\0')
1029 for (i = lastchar; i <= *pat; i++)
1031 charmap[i / 8] |= (unsigned char)(1 << (i % 8));
1036 charmap[*pat / 8] |= (unsigned char)(1 << (*pat % 8));
1040 } /* -END- if Character range specification */
1044 * Char match, or char range match?
1048 || ((*pat == ']') && (charmap[*txt / 8] & (1 << (*txt % 8)))) )
1058 * No match && no wildcard: No luck
1062 else if (pat != fallback)
1065 * Increment text pointer if in char range matching
1072 * Wildcard mode && nonmatch beyond fallback: Rewind pattern
1076 * Restart matching from current text pointer
1083 /* Cut off extra '*'s */
1084 if(*pat == '*') pat++;
1086 /* If this is the pattern's end, fine! */
1092 /*********************************************************************
1094 * Function : simple_domaincmp
1096 * Description : Domain-wise Compare fqdn's. The comparison is
1097 * both left- and right-anchored. The individual
1098 * domain names are compared with simplematch().
1099 * This is only used by domain_match.
1102 * 1 : pv = array of patterns to compare
1103 * 2 : fv = array of domain components to compare
1104 * 3 : len = length of the arrays (both arrays are the
1105 * same length - if they weren't, it couldn't
1106 * possibly be a match).
1108 * Returns : 0 => domains are equivalent, else no match.
1110 *********************************************************************/
1111 static int simple_domaincmp(char **pv, char **fv, int len)
1115 for (n = 0; n < len; n++)
1117 if (simplematch(pv[n], fv[n]))
1128 /*********************************************************************
1130 * Function : domain_match
1132 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
1133 * pattern->unachored, the comparison is un-, left-,
1134 * right-anchored, or both.
1135 * The individual domain names are compared with
1139 * 1 : pattern = a domain that may contain a '*' as a wildcard.
1140 * 2 : fqdn = domain name against which the patterns are compared.
1142 * Returns : 0 => domains are equivalent, else no match.
1144 *********************************************************************/
1145 static int domain_match(const struct url_spec *pattern, const struct http_request *fqdn)
1147 char **pv, **fv; /* vectors */
1149 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
1151 plen = pattern->dcount;
1152 flen = fqdn->dcount;
1156 /* fqdn is too short to match this pattern */
1163 if (unanchored == ANCHOR_LEFT)
1168 * Convert this into a fully anchored pattern with
1169 * the fqdn and pattern the same length
1171 fv += (flen - plen); /* flen - plen >= 0 due to check above */
1172 return simple_domaincmp(pv, fv, plen);
1174 else if (unanchored == 0)
1176 /* Fully anchored, check length */
1181 return simple_domaincmp(pv, fv, plen);
1183 else if (unanchored == ANCHOR_RIGHT)
1185 /* Left anchored, ignore all extra in fqdn */
1186 return simple_domaincmp(pv, fv, plen);
1192 int maxn = flen - plen;
1193 for (n = 0; n <= maxn; n++)
1195 if (!simple_domaincmp(pv, fv, plen))
1200 * Doesn't match from start of fqdn
1201 * Try skipping first part of fqdn
1209 #endif /* def FEATURE_EXTENDED_HOST_PATTERNS */
1212 /*********************************************************************
1214 * Function : create_url_spec
1216 * Description : Creates a "url_spec" structure from a string.
1217 * When finished, free with free_url_spec().
1220 * 1 : url = Target url_spec to be filled in. Will be
1221 * zeroed before use.
1222 * 2 : buf = Source pattern, null terminated. NOTE: The
1223 * contents of this buffer are destroyed by this
1224 * function. If this function succeeds, the
1225 * buffer is copied to url->spec. If this
1226 * function fails, the contents of the buffer
1229 * Returns : JB_ERR_OK - Success
1230 * JB_ERR_MEMORY - Out of memory
1231 * JB_ERR_PARSE - Cannot parse regex (Detailed message
1232 * written to system log)
1234 *********************************************************************/
1235 jb_err create_url_spec(struct url_spec *url, char *buf)
1240 memset(url, '\0', sizeof(*url));
1242 /* Remember the original specification for the CGI pages. */
1243 url->spec = strdup(buf);
1244 if (NULL == url->spec)
1246 return JB_ERR_MEMORY;
1249 /* Is it tag pattern? */
1250 if (0 == strncmpic("TAG:", url->spec, 4))
1252 /* The pattern starts with the first character after "TAG:" */
1253 const char *tag_pattern = buf + 4;
1254 return compile_pattern(tag_pattern, NO_ANCHORING, url, &url->tag_regex);
1257 /* If it isn't a tag pattern it must be a URL pattern. */
1258 return compile_url_pattern(url, buf);
1262 /*********************************************************************
1264 * Function : free_url_spec
1266 * Description : Called from the "unloaders". Freez the url
1267 * structure elements.
1270 * 1 : url = pointer to a url_spec structure.
1274 *********************************************************************/
1275 void free_url_spec(struct url_spec *url)
1277 if (url == NULL) return;
1280 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
1281 if (url->host_regex)
1283 regfree(url->host_regex);
1284 freez(url->host_regex);
1287 freez(url->dbuffer);
1290 #endif /* ndef FEATURE_EXTENDED_HOST_PATTERNS */
1291 freez(url->port_list);
1299 regfree(url->tag_regex);
1300 freez(url->tag_regex);
1305 /*********************************************************************
1307 * Function : url_match
1309 * Description : Compare a URL against a URL pattern.
1312 * 1 : pattern = a URL pattern
1313 * 2 : url = URL to match
1315 * Returns : Nonzero if the URL matches the pattern, else 0.
1317 *********************************************************************/
1318 int url_match(const struct url_spec *pattern,
1319 const struct http_request *http)
1321 /* XXX: these should probably be functions. */
1322 #define PORT_MATCHES ((NULL == pattern->port_list) || match_portlist(pattern->port_list, http->port))
1323 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
1324 #define DOMAIN_MATCHES ((NULL == pattern->host_regex) || (0 == regexec(pattern->host_regex, http->host, 0, NULL, 0)))
1326 #define DOMAIN_MATCHES ((NULL == pattern->dbuffer) || (0 == domain_match(pattern, http)))
1328 #define PATH_MATCHES ((NULL == pattern->preg) || (0 == regexec(pattern->preg, http->path, 0, NULL, 0)))
1330 if (pattern->tag_regex != NULL)
1332 /* It's a tag pattern and shouldn't be matched against URLs */
1336 return (PORT_MATCHES && DOMAIN_MATCHES && PATH_MATCHES);
1341 /*********************************************************************
1343 * Function : match_portlist
1345 * Description : Check if a given number is covered by a comma
1346 * separated list of numbers and ranges (a,b-c,d,..)
1349 * 1 : portlist = String with list
1350 * 2 : port = port to check
1352 * Returns : 0 => no match
1355 *********************************************************************/
1356 int match_portlist(const char *portlist, int port)
1358 char *min, *max, *next, *portlist_copy;
1360 min = next = portlist_copy = strdup(portlist);
1363 * Zero-terminate first item and remember offset for next
1365 if (NULL != (next = strchr(portlist_copy, (int) ',')))
1371 * Loop through all items, checking for match
1375 if (NULL == (max = strchr(min, (int) '-')))
1378 * No dash, check for equality
1380 if (port == atoi(min))
1382 freez(portlist_copy);
1389 * This is a range, so check if between min and max,
1390 * or, if max was omitted, between min and 65K
1393 if(port >= atoi(min) && port <= (atoi(max) ? atoi(max) : 65535))
1395 freez(portlist_copy);
1407 * Zero-terminate next item and remember offset for n+1
1409 if ((NULL != next) && (NULL != (next = strchr(next, (int) ','))))
1415 freez(portlist_copy);