1 const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.44 2008/05/04 16:18:32 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) 2001-2003, 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 1.44 2008/05/04 16:18:32 fabiankeil
37 * Provide parse_http_url() with a third parameter to specify
38 * whether or not URLs without protocol are acceptable.
40 * Revision 1.43 2008/05/04 13:30:55 fabiankeil
41 * Streamline parse_http_url()'s prototype.
43 * Revision 1.42 2008/05/04 13:24:16 fabiankeil
44 * If the method isn't CONNECT, reject URLs without protocol.
46 * Revision 1.41 2008/05/02 09:51:34 fabiankeil
47 * In parse_http_url(), don't muck around with values
48 * that are none of its business: require an initialized
49 * http structure and never unset http->ssl.
51 * Revision 1.40 2008/04/23 16:12:28 fabiankeil
54 * Revision 1.39 2008/04/22 16:27:42 fabiankeil
55 * In parse_http_request(), remove a pointless
56 * temporary variable and free the buffer earlier.
58 * Revision 1.38 2008/04/18 05:17:18 fabiankeil
59 * Mark simplematch()'s parameters as immutable.
61 * Revision 1.37 2008/04/17 14:53:29 fabiankeil
62 * Move simplematch() into urlmatch.c as it's only
63 * used to match (old-school) domain patterns.
65 * Revision 1.36 2008/04/14 18:19:48 fabiankeil
66 * Remove now-pointless cast in create_url_spec().
68 * Revision 1.35 2008/04/14 18:11:21 fabiankeil
69 * The compiler might not notice it, but the buffer passed to
70 * create_url_spec() is modified later on and thus shouldn't
71 * be declared immutable.
73 * Revision 1.34 2008/04/13 13:32:07 fabiankeil
74 * Factor URL pattern compilation out of create_url_spec().
76 * Revision 1.33 2008/04/12 14:03:13 fabiankeil
77 * Remove an obvious comment and improve another one.
79 * Revision 1.32 2008/04/12 12:38:06 fabiankeil
80 * Factor out duplicated code to compile host, path and tag patterns.
82 * Revision 1.31 2008/04/10 14:41:04 fabiankeil
83 * Ditch url_spec's path member now that it's no longer used.
85 * Revision 1.30 2008/04/10 04:24:24 fabiankeil
86 * Stop duplicating the plain text representation of the path regex
87 * (and keeping the copy around). Once the regex is compiled it's no
90 * Revision 1.29 2008/04/10 04:17:56 fabiankeil
91 * In url_match(), check the right member for NULL when determining
92 * whether there's a path regex to execute. Looking for a plain-text
93 * representation works as well, but it looks "interesting" and that
94 * member will be removed soonish anyway.
96 * Revision 1.28 2008/04/08 16:07:39 fabiankeil
97 * Make it harder to mistake url_match()'s
98 * second parameter for an url_spec.
100 * Revision 1.27 2008/04/08 15:44:33 fabiankeil
101 * Save a bit of memory (and a few cpu cycles) by not bothering to
102 * compile slash-only path regexes that don't affect the result.
104 * Revision 1.26 2008/04/07 16:57:18 fabiankeil
105 * - Use free_url_spec() more consistently.
106 * - Let it reset url->dcount just in case.
108 * Revision 1.25 2008/04/06 15:18:38 fabiankeil
109 * Oh well, rename the --enable-pcre-host-patterns option to
110 * --enable-extended-host-patterns as it's not really PCRE syntax.
112 * Revision 1.24 2008/04/06 14:54:26 fabiankeil
113 * Use PCRE syntax in host patterns when configured
114 * with --enable-pcre-host-patterns.
116 * Revision 1.23 2008/04/05 12:19:20 fabiankeil
117 * Factor compile_host_pattern() out of create_url_spec().
119 * Revision 1.22 2008/03/30 15:02:32 fabiankeil
120 * SZitify unknown_method().
122 * Revision 1.21 2007/12/24 16:34:23 fabiankeil
123 * Band-aid (and micro-optimization) that makes it less likely to run out of
124 * stack space with overly-complex path patterns. Probably masks the problem
125 * reported by Lee in #1856679. Hohoho.
127 * Revision 1.20 2007/09/02 15:31:20 fabiankeil
128 * Move match_portlist() from filter.c to urlmatch.c.
129 * It's used for url matching, not for filtering.
131 * Revision 1.19 2007/09/02 13:42:11 fabiankeil
132 * - Allow port lists in url patterns.
133 * - Ditch unused url_spec member pathlen.
135 * Revision 1.18 2007/07/30 16:42:21 fabiankeil
136 * Move the method check into unknown_method()
137 * and loop through the known methods instead
138 * of using a screen-long OR chain.
140 * Revision 1.17 2007/04/15 16:39:21 fabiankeil
141 * Introduce tags as alternative way to specify which
142 * actions apply to a request. At the moment tags can be
143 * created based on client and server headers.
145 * Revision 1.16 2007/02/13 13:59:24 fabiankeil
146 * Remove redundant log message.
148 * Revision 1.15 2007/01/28 16:11:23 fabiankeil
149 * Accept WebDAV methods for subversion
150 * in parse_http_request(). Closes FR 1581425.
152 * Revision 1.14 2007/01/06 14:23:56 fabiankeil
153 * Fix gcc43 warnings. Mark *csp as immutable
154 * for parse_http_url() and url_match().
155 * Replace a sprintf call with snprintf.
157 * Revision 1.13 2006/12/06 19:50:54 fabiankeil
158 * parse_http_url() now handles intercepted
159 * HTTP request lines as well. Moved parts
160 * of parse_http_url()'s code into
161 * init_domain_components() so that it can
162 * be reused in chat().
164 * Revision 1.12 2006/07/18 14:48:47 david__schmidt
165 * Reorganizing the repository: swapping out what was HEAD (the old 3.1 branch)
166 * with what was really the latest development (the v_3_0_branch branch)
168 * Revision 1.10.2.7 2003/05/17 15:57:24 oes
169 * - parse_http_url now checks memory allocation failure for
170 * duplication of "*" URL and rejects "*something" URLs
172 * - Added a comment to what might look like a bug in
173 * create_url_spec (see !bug #736931)
174 * - Comment cosmetics
176 * Revision 1.10.2.6 2003/05/07 12:39:48 oes
177 * Fix typo: Default port for https URLs is 443, not 143.
178 * Thanks to Scott Tregear for spotting this one.
180 * Revision 1.10.2.5 2003/02/28 13:09:29 oes
181 * Fixed a rare double free condition as per Bug #694713
183 * Revision 1.10.2.4 2003/02/28 12:57:44 oes
184 * Moved freeing of http request structure to its owner
185 * as per Dan Price's observations in Bug #694713
187 * Revision 1.10.2.3 2002/11/12 16:50:40 oes
188 * Fixed memory leak in parse_http_request() reported by Oliver Stoeneberg. Fixes bug #637073
190 * Revision 1.10.2.2 2002/09/25 14:53:15 oes
191 * Added basic support for OPTIONS and TRACE HTTP methods:
192 * parse_http_url now recognizes the "*" URI as well as
193 * the OPTIONS and TRACE method keywords.
195 * Revision 1.10.2.1 2002/06/06 19:06:44 jongfoster
196 * Adding support for proprietary Microsoft WebDAV extensions
198 * Revision 1.10 2002/05/12 21:40:37 jongfoster
199 * - Removing some unused code
201 * Revision 1.9 2002/04/04 00:36:36 gliptak
202 * always use pcre for matching
204 * Revision 1.8 2002/04/03 23:32:47 jongfoster
205 * Fixing memory leak on error
207 * Revision 1.7 2002/03/26 22:29:55 swa
208 * we have a new homepage!
210 * Revision 1.6 2002/03/24 13:25:43 swa
211 * name change related issues
213 * Revision 1.5 2002/03/13 00:27:05 jongfoster
216 * Revision 1.4 2002/03/07 03:46:17 oes
217 * Fixed compiler warnings
219 * Revision 1.3 2002/03/03 14:51:11 oes
220 * Fixed CLF logging: Added ocmd member for client's request to struct http_request
222 * Revision 1.2 2002/01/21 00:14:09 jongfoster
223 * Correcting comment style
224 * Fixing an uninitialized memory bug in create_url_spec()
226 * Revision 1.1 2002/01/17 20:53:46 jongfoster
227 * Moving all our URL and URL pattern parsing code to the same file - it
228 * was scattered around in filters.c, loaders.c and parsers.c.
230 * Providing a single, simple url_match(pattern,url) function - rather than
231 * the 3-line match routine which was repeated all over the place.
233 * Renaming free_url to free_url_spec, since it frees a struct url_spec.
235 * Providing parse_http_url() so that URLs can be parsed without faking a
236 * HTTP request line for parse_http_request() or repeating the parsing
237 * code (both of which were techniques that were actually in use).
239 * Standardizing that struct http_request is used to represent a URL, and
240 * struct url_spec is used to represent a URL pattern. (Before, URLs were
241 * represented as seperate variables and a partially-filled-in url_spec).
244 *********************************************************************/
251 #include <sys/types.h>
259 #if !defined(_WIN32) && !defined(__OS2__)
264 #include "urlmatch.h"
266 #include "miscutil.h"
269 const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
271 enum regex_anchoring {NO_ANCHORING, LEFT_ANCHORED, RIGHT_ANCHORED};
272 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern);
274 /*********************************************************************
276 * Function : free_http_request
278 * Description : Freez a http_request structure
281 * 1 : http = points to a http_request structure to free
285 *********************************************************************/
286 void free_http_request(struct http_request *http)
295 freez(http->hostport);
298 freez(http->host_ip_addr_str);
299 freez(http->dbuffer);
305 /*********************************************************************
307 * Function : init_domain_components
309 * Description : Splits the domain name so we can compare it
310 * against wildcards. It used to be part of
311 * parse_http_url, but was separated because the
312 * same code is required in chat in case of
313 * intercepted requests.
316 * 1 : http = pointer to the http structure to hold elements.
318 * Returns : JB_ERR_OK on success
319 * JB_ERR_MEMORY on out of memory
320 * JB_ERR_PARSE on malformed command/URL
321 * or >100 domains deep.
323 *********************************************************************/
324 jb_err init_domain_components(struct http_request *http)
326 char *vec[BUFFER_SIZE];
330 http->dbuffer = strdup(http->host);
331 if (NULL == http->dbuffer)
333 return JB_ERR_MEMORY;
336 /* map to lower case */
337 for (p = http->dbuffer; *p ; p++)
339 *p = (char)tolower((int)(unsigned char)*p);
342 /* split the domain name into components */
343 http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec), 1, 1);
345 if (http->dcount <= 0)
348 * Error: More than SZ(vec) components in domain
349 * or: no components in domain
351 log_error(LOG_LEVEL_ERROR, "More than SZ(vec) components in domain or none at all.");
355 /* save a copy of the pointers in dvec */
356 size = (size_t)http->dcount * sizeof(*http->dvec);
358 http->dvec = (char **)malloc(size);
359 if (NULL == http->dvec)
361 return JB_ERR_MEMORY;
364 memcpy(http->dvec, vec, size);
370 /*********************************************************************
372 * Function : parse_http_url
374 * Description : Parse out the host and port from the URL. Find the
375 * hostname & path, port (if ':'), and/or password (if '@')
378 * 1 : url = URL (or is it URI?) to break down
379 * 2 : http = pointer to the http structure to hold elements.
380 * Must be initialized with valid values (like NULLs).
381 * 3 : require_protocol = Whether or not URLs without
382 * protocol are acceptable.
384 * Returns : JB_ERR_OK on success
385 * JB_ERR_MEMORY on out of memory
386 * JB_ERR_PARSE on malformed command/URL
387 * or >100 domains deep.
389 *********************************************************************/
390 jb_err parse_http_url(const char *url, struct http_request *http, int require_protocol)
392 int host_available = 1; /* A proxy can dream. */
395 * Save our initial URL
397 http->url = strdup(url);
398 if (http->url == NULL)
400 return JB_ERR_MEMORY;
405 * Check for * URI. If found, we're done.
407 if (*http->url == '*')
409 if ( NULL == (http->path = strdup("*"))
410 || NULL == (http->hostport = strdup("")) )
412 return JB_ERR_MEMORY;
414 if (http->url[1] != '\0')
423 * Split URL into protocol,hostport,path.
433 return JB_ERR_MEMORY;
436 /* Find the start of the URL in our scratch space */
438 if (strncmpic(url_noproto, "http://", 7) == 0)
442 else if (strncmpic(url_noproto, "https://", 8) == 0)
445 * Should only happen when called from cgi_show_url_info().
450 else if (*url_noproto == '/')
453 * Short request line without protocol and host.
454 * Most likely because the client's request
455 * was intercepted and redirected into Privoxy.
460 else if (require_protocol)
466 url_path = strchr(url_noproto, '/');
467 if (url_path != NULL)
472 * NOTE: The following line ignores the path for HTTPS URLS.
473 * This means that you get consistent behaviour if you type a
474 * https URL in and it's parsed by the function. (When the
475 * URL is actually retrieved, SSL hides the path part).
477 http->path = strdup(http->ssl ? "/" : url_path);
479 http->hostport = strdup(url_noproto);
484 * Repair broken HTTP requests that don't contain a path,
485 * or CONNECT requests
487 http->path = strdup("/");
488 http->hostport = strdup(url_noproto);
493 if ( (http->path == NULL)
494 || (http->hostport == NULL))
496 return JB_ERR_MEMORY;
502 /* Without host, there is nothing left to do here */
507 * Split hostport into user/password (ignored), host, port.
514 buf = strdup(http->hostport);
517 return JB_ERR_MEMORY;
520 /* check if url contains username and/or password */
521 host = strchr(buf, '@');
524 /* Contains username/password, skip it and the @ sign. */
529 /* No username or password. */
533 /* check if url contains port */
534 port = strchr(host, ':');
538 /* Terminate hostname and point to start of port string */
540 http->port = atoi(port);
544 /* No port specified. */
545 http->port = (http->ssl ? 443 : 80);
548 http->host = strdup(host);
552 if (http->host == NULL)
554 return JB_ERR_MEMORY;
559 * Split domain name so we can compare it against wildcards
561 return init_domain_components(http);
566 /*********************************************************************
568 * Function : unknown_method
570 * Description : Checks whether a method is unknown.
573 * 1 : method = points to a http method
575 * Returns : TRUE if it's unknown, FALSE otherwise.
577 *********************************************************************/
578 static int unknown_method(const char *method)
580 static const char *known_http_methods[] = {
581 /* Basic HTTP request type */
582 "GET", "HEAD", "POST", "PUT", "DELETE", "OPTIONS", "TRACE", "CONNECT",
583 /* webDAV extensions (RFC2518) */
584 "PROPFIND", "PROPPATCH", "MOVE", "COPY", "MKCOL", "LOCK", "UNLOCK",
586 * Microsoft webDAV extension for Exchange 2000. See:
587 * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
588 * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
590 "BCOPY", "BMOVE", "BDELETE", "BPROPFIND", "BPROPPATCH",
592 * Another Microsoft webDAV extension for Exchange 2000. See:
593 * http://systems.cs.colorado.edu/grunwald/MobileComputing/Papers/draft-cohen-gena-p-base-00.txt
594 * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
595 * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
597 "SUBSCRIBE", "UNSUBSCRIBE", "NOTIFY", "POLL",
599 * Yet another WebDAV extension, this time for
600 * Web Distributed Authoring and Versioning (RFC3253)
602 "VERSION-CONTROL", "REPORT", "CHECKOUT", "CHECKIN", "UNCHECKOUT",
603 "MKWORKSPACE", "UPDATE", "LABEL", "MERGE", "BASELINE-CONTROL", "MKACTIVITY",
607 for (i = 0; i < SZ(known_http_methods); i++)
609 if (0 == strcmpic(method, known_http_methods[i]))
620 /*********************************************************************
622 * Function : parse_http_request
624 * Description : Parse out the host and port from the URL. Find the
625 * hostname & path, port (if ':'), and/or password (if '@')
628 * 1 : req = HTTP request line to break down
629 * 2 : http = pointer to the http structure to hold elements
630 * 3 : csp = Current client state (buffers, headers, etc...)
632 * Returns : JB_ERR_OK on success
633 * JB_ERR_MEMORY on out of memory
634 * JB_ERR_CGI_PARAMS on malformed command/URL
635 * or >100 domains deep.
637 *********************************************************************/
638 jb_err parse_http_request(const char *req,
639 struct http_request *http,
640 const struct client_state *csp)
643 char *v[10]; /* XXX: Why 10? We should only need three. */
647 memset(http, '\0', sizeof(*http));
652 return JB_ERR_MEMORY;
655 n = ssplit(buf, " \r\n", v, SZ(v), 1, 1);
663 * Fail in case of unknown methods
664 * which we might not handle correctly.
666 * XXX: There should be a config option
667 * to forward requests with unknown methods
668 * anyway. Most of them don't need special
671 if (unknown_method(v[0]))
673 log_error(LOG_LEVEL_ERROR, "Unknown HTTP method detected: %s", v[0]);
678 http->ssl = !strcmpic(v[0], "CONNECT");
680 err = parse_http_url(v[1], http, !http->ssl);
688 * Copy the details into the structure
690 http->cmd = strdup(req);
691 http->gpc = strdup(v[0]);
692 http->ver = strdup(v[2]);
696 if ( (http->cmd == NULL)
697 || (http->gpc == NULL)
698 || (http->ver == NULL) )
700 return JB_ERR_MEMORY;
708 /*********************************************************************
710 * Function : compile_pattern
712 * Description : Compiles a host, domain or TAG pattern.
715 * 1 : pattern = The pattern to compile.
716 * 2 : anchoring = How the regex should be anchored.
717 * Can be either one of NO_ANCHORING,
718 * LEFT_ANCHORED or RIGHT_ANCHORED.
719 * 3 : url = In case of failures, the spec member is
720 * logged and the structure freed.
721 * 4 : regex = Where the compiled regex should be stored.
723 * Returns : JB_ERR_OK - Success
724 * JB_ERR_MEMORY - Out of memory
725 * JB_ERR_PARSE - Cannot parse regex
727 *********************************************************************/
728 static jb_err compile_pattern(const char *pattern, enum regex_anchoring anchoring,
729 struct url_spec *url, regex_t **regex)
732 char rebuf[BUFFER_SIZE];
733 const char *fmt = NULL;
736 assert(strlen(pattern) < sizeof(rebuf) - 2);
738 if (pattern[0] == '\0')
756 log_error(LOG_LEVEL_FATAL,
757 "Invalid anchoring in compile_pattern %d", anchoring);
760 *regex = zalloc(sizeof(**regex));
764 return JB_ERR_MEMORY;
767 snprintf(rebuf, sizeof(rebuf), fmt, pattern);
769 errcode = regcomp(*regex, rebuf, (REG_EXTENDED|REG_NOSUB|REG_ICASE));
773 size_t errlen = regerror(errcode, *regex, rebuf, sizeof(rebuf));
774 if (errlen > (sizeof(rebuf) - (size_t)1))
776 errlen = sizeof(rebuf) - (size_t)1;
778 rebuf[errlen] = '\0';
779 log_error(LOG_LEVEL_ERROR, "error compiling %s from %s: %s",
780 pattern, url->spec, rebuf);
791 /*********************************************************************
793 * Function : compile_url_pattern
795 * Description : Compiles the three parts of an URL pattern.
798 * 1 : url = Target url_spec to be filled in.
799 * 2 : buf = The url pattern to compile. Will be messed up.
801 * Returns : JB_ERR_OK - Success
802 * JB_ERR_MEMORY - Out of memory
803 * JB_ERR_PARSE - Cannot parse regex
805 *********************************************************************/
806 static jb_err compile_url_pattern(struct url_spec *url, char *buf)
810 p = strchr(buf, '/');
814 * Only compile the regex if it consists of more than
815 * a single slash, otherwise it wouldn't affect the result.
820 * XXX: does it make sense to compile the slash at the beginning?
822 jb_err err = compile_pattern(p, LEFT_ANCHORED, url, &url->preg);
824 if (JB_ERR_OK != err)
832 p = strchr(buf, ':');
836 url->port_list = strdup(p);
837 if (NULL == url->port_list)
839 return JB_ERR_MEMORY;
844 url->port_list = NULL;
849 return compile_host_pattern(url, buf);
857 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
858 /*********************************************************************
860 * Function : compile_host_pattern
862 * Description : Parses and compiles a host pattern..
865 * 1 : url = Target url_spec to be filled in.
866 * 2 : host_pattern = Host pattern to compile.
868 * Returns : JB_ERR_OK - Success
869 * JB_ERR_MEMORY - Out of memory
870 * JB_ERR_PARSE - Cannot parse regex
872 *********************************************************************/
873 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
875 return compile_pattern(host_pattern, RIGHT_ANCHORED, url, &url->host_regex);
880 /*********************************************************************
882 * Function : compile_host_pattern
884 * Description : Parses and "compiles" an old-school host pattern.
887 * 1 : url = Target url_spec to be filled in.
888 * 2 : host_pattern = Host pattern to parse.
890 * Returns : JB_ERR_OK - Success
891 * JB_ERR_MEMORY - Out of memory
892 * JB_ERR_PARSE - Cannot parse regex
894 *********************************************************************/
895 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
904 if (host_pattern[strlen(host_pattern) - 1] == '.')
906 url->unanchored |= ANCHOR_RIGHT;
908 if (host_pattern[0] == '.')
910 url->unanchored |= ANCHOR_LEFT;
914 * Split domain into components
916 url->dbuffer = strdup(host_pattern);
917 if (NULL == url->dbuffer)
920 return JB_ERR_MEMORY;
926 for (p = url->dbuffer; *p ; p++)
928 *p = (char)tolower((int)(unsigned char)*p);
932 * Split the domain name into components
934 url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1);
939 return JB_ERR_MEMORY;
941 else if (url->dcount != 0)
944 * Save a copy of the pointers in dvec
946 size = (size_t)url->dcount * sizeof(*url->dvec);
948 url->dvec = (char **)malloc(size);
949 if (NULL == url->dvec)
952 return JB_ERR_MEMORY;
955 memcpy(url->dvec, v, size);
958 * else dcount == 0 in which case we needn't do anything,
959 * since dvec will never be accessed and the pattern will
966 /*********************************************************************
968 * Function : simplematch
970 * Description : String matching, with a (greedy) '*' wildcard that
971 * stands for zero or more arbitrary characters and
972 * character classes in [], which take both enumerations
976 * 1 : pattern = pattern for matching
977 * 2 : text = text to be matched
979 * Returns : 0 if match, else nonzero
981 *********************************************************************/
982 static int simplematch(const char *pattern, const char *text)
984 const unsigned char *pat = (const unsigned char *)pattern;
985 const unsigned char *txt = (const unsigned char *)text;
986 const unsigned char *fallback = pat;
989 unsigned char lastchar = 'a';
991 unsigned char charmap[32];
996 /* EOF pattern but !EOF text? */
1009 /* '*' in the pattern? */
1013 /* The pattern ends afterwards? Speed up the return. */
1019 /* Else, set wildcard mode and remember position after '*' */
1024 /* Character range specification? */
1027 memset(charmap, '\0', sizeof(charmap));
1029 while (*++pat != ']')
1035 else if (*pat == '-')
1037 if ((*++pat == ']') || *pat == '\0')
1041 for (i = lastchar; i <= *pat; i++)
1043 charmap[i / 8] |= (unsigned char)(1 << (i % 8));
1048 charmap[*pat / 8] |= (unsigned char)(1 << (*pat % 8));
1052 } /* -END- if Character range specification */
1056 * Char match, or char range match?
1060 || ((*pat == ']') && (charmap[*txt / 8] & (1 << (*txt % 8)))) )
1070 * No match && no wildcard: No luck
1074 else if (pat != fallback)
1077 * Increment text pointer if in char range matching
1084 * Wildcard mode && nonmatch beyond fallback: Rewind pattern
1088 * Restart matching from current text pointer
1095 /* Cut off extra '*'s */
1096 if(*pat == '*') pat++;
1098 /* If this is the pattern's end, fine! */
1104 /*********************************************************************
1106 * Function : simple_domaincmp
1108 * Description : Domain-wise Compare fqdn's. The comparison is
1109 * both left- and right-anchored. The individual
1110 * domain names are compared with simplematch().
1111 * This is only used by domain_match.
1114 * 1 : pv = array of patterns to compare
1115 * 2 : fv = array of domain components to compare
1116 * 3 : len = length of the arrays (both arrays are the
1117 * same length - if they weren't, it couldn't
1118 * possibly be a match).
1120 * Returns : 0 => domains are equivalent, else no match.
1122 *********************************************************************/
1123 static int simple_domaincmp(char **pv, char **fv, int len)
1127 for (n = 0; n < len; n++)
1129 if (simplematch(pv[n], fv[n]))
1140 /*********************************************************************
1142 * Function : domain_match
1144 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
1145 * pattern->unachored, the comparison is un-, left-,
1146 * right-anchored, or both.
1147 * The individual domain names are compared with
1151 * 1 : pattern = a domain that may contain a '*' as a wildcard.
1152 * 2 : fqdn = domain name against which the patterns are compared.
1154 * Returns : 0 => domains are equivalent, else no match.
1156 *********************************************************************/
1157 static int domain_match(const struct url_spec *pattern, const struct http_request *fqdn)
1159 char **pv, **fv; /* vectors */
1161 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
1163 plen = pattern->dcount;
1164 flen = fqdn->dcount;
1168 /* fqdn is too short to match this pattern */
1175 if (unanchored == ANCHOR_LEFT)
1180 * Convert this into a fully anchored pattern with
1181 * the fqdn and pattern the same length
1183 fv += (flen - plen); /* flen - plen >= 0 due to check above */
1184 return simple_domaincmp(pv, fv, plen);
1186 else if (unanchored == 0)
1188 /* Fully anchored, check length */
1193 return simple_domaincmp(pv, fv, plen);
1195 else if (unanchored == ANCHOR_RIGHT)
1197 /* Left anchored, ignore all extra in fqdn */
1198 return simple_domaincmp(pv, fv, plen);
1204 int maxn = flen - plen;
1205 for (n = 0; n <= maxn; n++)
1207 if (!simple_domaincmp(pv, fv, plen))
1212 * Doesn't match from start of fqdn
1213 * Try skipping first part of fqdn
1221 #endif /* def FEATURE_EXTENDED_HOST_PATTERNS */
1224 /*********************************************************************
1226 * Function : create_url_spec
1228 * Description : Creates a "url_spec" structure from a string.
1229 * When finished, free with free_url_spec().
1232 * 1 : url = Target url_spec to be filled in. Will be
1233 * zeroed before use.
1234 * 2 : buf = Source pattern, null terminated. NOTE: The
1235 * contents of this buffer are destroyed by this
1236 * function. If this function succeeds, the
1237 * buffer is copied to url->spec. If this
1238 * function fails, the contents of the buffer
1241 * Returns : JB_ERR_OK - Success
1242 * JB_ERR_MEMORY - Out of memory
1243 * JB_ERR_PARSE - Cannot parse regex (Detailed message
1244 * written to system log)
1246 *********************************************************************/
1247 jb_err create_url_spec(struct url_spec *url, char *buf)
1252 memset(url, '\0', sizeof(*url));
1254 /* Remember the original specification for the CGI pages. */
1255 url->spec = strdup(buf);
1256 if (NULL == url->spec)
1258 return JB_ERR_MEMORY;
1261 /* Is it tag pattern? */
1262 if (0 == strncmpic("TAG:", url->spec, 4))
1264 /* The pattern starts with the first character after "TAG:" */
1265 const char *tag_pattern = buf + 4;
1266 return compile_pattern(tag_pattern, NO_ANCHORING, url, &url->tag_regex);
1269 /* If it isn't a tag pattern it must be a URL pattern. */
1270 return compile_url_pattern(url, buf);
1274 /*********************************************************************
1276 * Function : free_url_spec
1278 * Description : Called from the "unloaders". Freez the url
1279 * structure elements.
1282 * 1 : url = pointer to a url_spec structure.
1286 *********************************************************************/
1287 void free_url_spec(struct url_spec *url)
1289 if (url == NULL) return;
1292 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
1293 if (url->host_regex)
1295 regfree(url->host_regex);
1296 freez(url->host_regex);
1299 freez(url->dbuffer);
1302 #endif /* ndef FEATURE_EXTENDED_HOST_PATTERNS */
1303 freez(url->port_list);
1311 regfree(url->tag_regex);
1312 freez(url->tag_regex);
1317 /*********************************************************************
1319 * Function : url_match
1321 * Description : Compare a URL against a URL pattern.
1324 * 1 : pattern = a URL pattern
1325 * 2 : url = URL to match
1327 * Returns : Nonzero if the URL matches the pattern, else 0.
1329 *********************************************************************/
1330 int url_match(const struct url_spec *pattern,
1331 const struct http_request *http)
1333 /* XXX: these should probably be functions. */
1334 #define PORT_MATCHES ((NULL == pattern->port_list) || match_portlist(pattern->port_list, http->port))
1335 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
1336 #define DOMAIN_MATCHES ((NULL == pattern->host_regex) || (0 == regexec(pattern->host_regex, http->host, 0, NULL, 0)))
1338 #define DOMAIN_MATCHES ((NULL == pattern->dbuffer) || (0 == domain_match(pattern, http)))
1340 #define PATH_MATCHES ((NULL == pattern->preg) || (0 == regexec(pattern->preg, http->path, 0, NULL, 0)))
1342 if (pattern->tag_regex != NULL)
1344 /* It's a tag pattern and shouldn't be matched against URLs */
1348 return (PORT_MATCHES && DOMAIN_MATCHES && PATH_MATCHES);
1353 /*********************************************************************
1355 * Function : match_portlist
1357 * Description : Check if a given number is covered by a comma
1358 * separated list of numbers and ranges (a,b-c,d,..)
1361 * 1 : portlist = String with list
1362 * 2 : port = port to check
1364 * Returns : 0 => no match
1367 *********************************************************************/
1368 int match_portlist(const char *portlist, int port)
1370 char *min, *max, *next, *portlist_copy;
1372 min = next = portlist_copy = strdup(portlist);
1375 * Zero-terminate first item and remember offset for next
1377 if (NULL != (next = strchr(portlist_copy, (int) ',')))
1383 * Loop through all items, checking for match
1387 if (NULL == (max = strchr(min, (int) '-')))
1390 * No dash, check for equality
1392 if (port == atoi(min))
1394 freez(portlist_copy);
1401 * This is a range, so check if between min and max,
1402 * or, if max was omitted, between min and 65K
1405 if(port >= atoi(min) && port <= (atoi(max) ? atoi(max) : 65535))
1407 freez(portlist_copy);
1419 * Zero-terminate next item and remember offset for n+1
1421 if ((NULL != next) && (NULL != (next = strchr(next, (int) ','))))
1427 freez(portlist_copy);