1 const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.43 2008/05/04 13:30:55 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 1.43 2008/05/04 13:30:55 fabiankeil
37 * Streamline parse_http_url()'s prototype.
39 * Revision 1.42 2008/05/04 13:24:16 fabiankeil
40 * If the method isn't CONNECT, reject URLs without protocol.
42 * Revision 1.41 2008/05/02 09:51:34 fabiankeil
43 * In parse_http_url(), don't muck around with values
44 * that are none of its business: require an initialized
45 * http structure and never unset http->ssl.
47 * Revision 1.40 2008/04/23 16:12:28 fabiankeil
50 * Revision 1.39 2008/04/22 16:27:42 fabiankeil
51 * In parse_http_request(), remove a pointless
52 * temporary variable and free the buffer earlier.
54 * Revision 1.38 2008/04/18 05:17:18 fabiankeil
55 * Mark simplematch()'s parameters as immutable.
57 * Revision 1.37 2008/04/17 14:53:29 fabiankeil
58 * Move simplematch() into urlmatch.c as it's only
59 * used to match (old-school) domain patterns.
61 * Revision 1.36 2008/04/14 18:19:48 fabiankeil
62 * Remove now-pointless cast in create_url_spec().
64 * Revision 1.35 2008/04/14 18:11:21 fabiankeil
65 * The compiler might not notice it, but the buffer passed to
66 * create_url_spec() is modified later on and thus shouldn't
67 * be declared immutable.
69 * Revision 1.34 2008/04/13 13:32:07 fabiankeil
70 * Factor URL pattern compilation out of create_url_spec().
72 * Revision 1.33 2008/04/12 14:03:13 fabiankeil
73 * Remove an obvious comment and improve another one.
75 * Revision 1.32 2008/04/12 12:38:06 fabiankeil
76 * Factor out duplicated code to compile host, path and tag patterns.
78 * Revision 1.31 2008/04/10 14:41:04 fabiankeil
79 * Ditch url_spec's path member now that it's no longer used.
81 * Revision 1.30 2008/04/10 04:24:24 fabiankeil
82 * Stop duplicating the plain text representation of the path regex
83 * (and keeping the copy around). Once the regex is compiled it's no
86 * Revision 1.29 2008/04/10 04:17:56 fabiankeil
87 * In url_match(), check the right member for NULL when determining
88 * whether there's a path regex to execute. Looking for a plain-text
89 * representation works as well, but it looks "interesting" and that
90 * member will be removed soonish anyway.
92 * Revision 1.28 2008/04/08 16:07:39 fabiankeil
93 * Make it harder to mistake url_match()'s
94 * second parameter for an url_spec.
96 * Revision 1.27 2008/04/08 15:44:33 fabiankeil
97 * Save a bit of memory (and a few cpu cycles) by not bothering to
98 * compile slash-only path regexes that don't affect the result.
100 * Revision 1.26 2008/04/07 16:57:18 fabiankeil
101 * - Use free_url_spec() more consistently.
102 * - Let it reset url->dcount just in case.
104 * Revision 1.25 2008/04/06 15:18:38 fabiankeil
105 * Oh well, rename the --enable-pcre-host-patterns option to
106 * --enable-extended-host-patterns as it's not really PCRE syntax.
108 * Revision 1.24 2008/04/06 14:54:26 fabiankeil
109 * Use PCRE syntax in host patterns when configured
110 * with --enable-pcre-host-patterns.
112 * Revision 1.23 2008/04/05 12:19:20 fabiankeil
113 * Factor compile_host_pattern() out of create_url_spec().
115 * Revision 1.22 2008/03/30 15:02:32 fabiankeil
116 * SZitify unknown_method().
118 * Revision 1.21 2007/12/24 16:34:23 fabiankeil
119 * Band-aid (and micro-optimization) that makes it less likely to run out of
120 * stack space with overly-complex path patterns. Probably masks the problem
121 * reported by Lee in #1856679. Hohoho.
123 * Revision 1.20 2007/09/02 15:31:20 fabiankeil
124 * Move match_portlist() from filter.c to urlmatch.c.
125 * It's used for url matching, not for filtering.
127 * Revision 1.19 2007/09/02 13:42:11 fabiankeil
128 * - Allow port lists in url patterns.
129 * - Ditch unused url_spec member pathlen.
131 * Revision 1.18 2007/07/30 16:42:21 fabiankeil
132 * Move the method check into unknown_method()
133 * and loop through the known methods instead
134 * of using a screen-long OR chain.
136 * Revision 1.17 2007/04/15 16:39:21 fabiankeil
137 * Introduce tags as alternative way to specify which
138 * actions apply to a request. At the moment tags can be
139 * created based on client and server headers.
141 * Revision 1.16 2007/02/13 13:59:24 fabiankeil
142 * Remove redundant log message.
144 * Revision 1.15 2007/01/28 16:11:23 fabiankeil
145 * Accept WebDAV methods for subversion
146 * in parse_http_request(). Closes FR 1581425.
148 * Revision 1.14 2007/01/06 14:23:56 fabiankeil
149 * Fix gcc43 warnings. Mark *csp as immutable
150 * for parse_http_url() and url_match().
151 * Replace a sprintf call with snprintf.
153 * Revision 1.13 2006/12/06 19:50:54 fabiankeil
154 * parse_http_url() now handles intercepted
155 * HTTP request lines as well. Moved parts
156 * of parse_http_url()'s code into
157 * init_domain_components() so that it can
158 * be reused in chat().
160 * Revision 1.12 2006/07/18 14:48:47 david__schmidt
161 * Reorganizing the repository: swapping out what was HEAD (the old 3.1 branch)
162 * with what was really the latest development (the v_3_0_branch branch)
164 * Revision 1.10.2.7 2003/05/17 15:57:24 oes
165 * - parse_http_url now checks memory allocation failure for
166 * duplication of "*" URL and rejects "*something" URLs
168 * - Added a comment to what might look like a bug in
169 * create_url_spec (see !bug #736931)
170 * - Comment cosmetics
172 * Revision 1.10.2.6 2003/05/07 12:39:48 oes
173 * Fix typo: Default port for https URLs is 443, not 143.
174 * Thanks to Scott Tregear for spotting this one.
176 * Revision 1.10.2.5 2003/02/28 13:09:29 oes
177 * Fixed a rare double free condition as per Bug #694713
179 * Revision 1.10.2.4 2003/02/28 12:57:44 oes
180 * Moved freeing of http request structure to its owner
181 * as per Dan Price's observations in Bug #694713
183 * Revision 1.10.2.3 2002/11/12 16:50:40 oes
184 * Fixed memory leak in parse_http_request() reported by Oliver Stoeneberg. Fixes bug #637073
186 * Revision 1.10.2.2 2002/09/25 14:53:15 oes
187 * Added basic support for OPTIONS and TRACE HTTP methods:
188 * parse_http_url now recognizes the "*" URI as well as
189 * the OPTIONS and TRACE method keywords.
191 * Revision 1.10.2.1 2002/06/06 19:06:44 jongfoster
192 * Adding support for proprietary Microsoft WebDAV extensions
194 * Revision 1.10 2002/05/12 21:40:37 jongfoster
195 * - Removing some unused code
197 * Revision 1.9 2002/04/04 00:36:36 gliptak
198 * always use pcre for matching
200 * Revision 1.8 2002/04/03 23:32:47 jongfoster
201 * Fixing memory leak on error
203 * Revision 1.7 2002/03/26 22:29:55 swa
204 * we have a new homepage!
206 * Revision 1.6 2002/03/24 13:25:43 swa
207 * name change related issues
209 * Revision 1.5 2002/03/13 00:27:05 jongfoster
212 * Revision 1.4 2002/03/07 03:46:17 oes
213 * Fixed compiler warnings
215 * Revision 1.3 2002/03/03 14:51:11 oes
216 * Fixed CLF logging: Added ocmd member for client's request to struct http_request
218 * Revision 1.2 2002/01/21 00:14:09 jongfoster
219 * Correcting comment style
220 * Fixing an uninitialized memory bug in create_url_spec()
222 * Revision 1.1 2002/01/17 20:53:46 jongfoster
223 * Moving all our URL and URL pattern parsing code to the same file - it
224 * was scattered around in filters.c, loaders.c and parsers.c.
226 * Providing a single, simple url_match(pattern,url) function - rather than
227 * the 3-line match routine which was repeated all over the place.
229 * Renaming free_url to free_url_spec, since it frees a struct url_spec.
231 * Providing parse_http_url() so that URLs can be parsed without faking a
232 * HTTP request line for parse_http_request() or repeating the parsing
233 * code (both of which were techniques that were actually in use).
235 * Standardizing that struct http_request is used to represent a URL, and
236 * struct url_spec is used to represent a URL pattern. (Before, URLs were
237 * represented as seperate variables and a partially-filled-in url_spec).
240 *********************************************************************/
247 #include <sys/types.h>
255 #if !defined(_WIN32) && !defined(__OS2__)
260 #include "urlmatch.h"
262 #include "miscutil.h"
265 const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
267 enum regex_anchoring {NO_ANCHORING, LEFT_ANCHORED, RIGHT_ANCHORED};
268 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern);
270 /*********************************************************************
272 * Function : free_http_request
274 * Description : Freez a http_request structure
277 * 1 : http = points to a http_request structure to free
281 *********************************************************************/
282 void free_http_request(struct http_request *http)
291 freez(http->hostport);
294 freez(http->host_ip_addr_str);
295 freez(http->dbuffer);
301 /*********************************************************************
303 * Function : init_domain_components
305 * Description : Splits the domain name so we can compare it
306 * against wildcards. It used to be part of
307 * parse_http_url, but was separated because the
308 * same code is required in chat in case of
309 * intercepted requests.
312 * 1 : http = pointer to the http structure to hold elements.
314 * Returns : JB_ERR_OK on success
315 * JB_ERR_MEMORY on out of memory
316 * JB_ERR_PARSE on malformed command/URL
317 * or >100 domains deep.
319 *********************************************************************/
320 jb_err init_domain_components(struct http_request *http)
322 char *vec[BUFFER_SIZE];
326 http->dbuffer = strdup(http->host);
327 if (NULL == http->dbuffer)
329 return JB_ERR_MEMORY;
332 /* map to lower case */
333 for (p = http->dbuffer; *p ; p++)
335 *p = (char)tolower((int)(unsigned char)*p);
338 /* split the domain name into components */
339 http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec), 1, 1);
341 if (http->dcount <= 0)
344 * Error: More than SZ(vec) components in domain
345 * or: no components in domain
347 log_error(LOG_LEVEL_ERROR, "More than SZ(vec) components in domain or none at all.");
351 /* save a copy of the pointers in dvec */
352 size = (size_t)http->dcount * sizeof(*http->dvec);
354 http->dvec = (char **)malloc(size);
355 if (NULL == http->dvec)
357 return JB_ERR_MEMORY;
360 memcpy(http->dvec, vec, size);
366 /*********************************************************************
368 * Function : parse_http_url
370 * Description : Parse out the host and port from the URL. Find the
371 * hostname & path, port (if ':'), and/or password (if '@')
374 * 1 : url = URL (or is it URI?) to break down
375 * 2 : http = pointer to the http structure to hold elements.
376 * Must be initialized with valid values (like NULLs).
377 * 3 : require_protocol = Whether or not URLs without
378 * protocol are acceptable.
380 * Returns : JB_ERR_OK on success
381 * JB_ERR_MEMORY on out of memory
382 * JB_ERR_PARSE on malformed command/URL
383 * or >100 domains deep.
385 *********************************************************************/
386 jb_err parse_http_url(const char *url, struct http_request *http, int require_protocol)
388 int host_available = 1; /* A proxy can dream. */
391 * Save our initial URL
393 http->url = strdup(url);
394 if (http->url == NULL)
396 return JB_ERR_MEMORY;
401 * Check for * URI. If found, we're done.
403 if (*http->url == '*')
405 if ( NULL == (http->path = strdup("*"))
406 || NULL == (http->hostport = strdup("")) )
408 return JB_ERR_MEMORY;
410 if (http->url[1] != '\0')
419 * Split URL into protocol,hostport,path.
429 return JB_ERR_MEMORY;
432 /* Find the start of the URL in our scratch space */
434 if (strncmpic(url_noproto, "http://", 7) == 0)
438 else if (strncmpic(url_noproto, "https://", 8) == 0)
441 * Should only happen when called from cgi_show_url_info().
446 else if (*url_noproto == '/')
449 * Short request line without protocol and host.
450 * Most likely because the client's request
451 * was intercepted and redirected into Privoxy.
456 else if (require_protocol)
462 url_path = strchr(url_noproto, '/');
463 if (url_path != NULL)
468 * NOTE: The following line ignores the path for HTTPS URLS.
469 * This means that you get consistent behaviour if you type a
470 * https URL in and it's parsed by the function. (When the
471 * URL is actually retrieved, SSL hides the path part).
473 http->path = strdup(http->ssl ? "/" : url_path);
475 http->hostport = strdup(url_noproto);
480 * Repair broken HTTP requests that don't contain a path,
481 * or CONNECT requests
483 http->path = strdup("/");
484 http->hostport = strdup(url_noproto);
489 if ( (http->path == NULL)
490 || (http->hostport == NULL))
492 return JB_ERR_MEMORY;
498 /* Without host, there is nothing left to do here */
503 * Split hostport into user/password (ignored), host, port.
510 buf = strdup(http->hostport);
513 return JB_ERR_MEMORY;
516 /* check if url contains username and/or password */
517 host = strchr(buf, '@');
520 /* Contains username/password, skip it and the @ sign. */
525 /* No username or password. */
529 /* check if url contains port */
530 port = strchr(host, ':');
534 /* Terminate hostname and point to start of port string */
536 http->port = atoi(port);
540 /* No port specified. */
541 http->port = (http->ssl ? 443 : 80);
544 http->host = strdup(host);
548 if (http->host == NULL)
550 return JB_ERR_MEMORY;
555 * Split domain name so we can compare it against wildcards
557 return init_domain_components(http);
562 /*********************************************************************
564 * Function : unknown_method
566 * Description : Checks whether a method is unknown.
569 * 1 : method = points to a http method
571 * Returns : TRUE if it's unknown, FALSE otherwise.
573 *********************************************************************/
574 static int unknown_method(const char *method)
576 static const char *known_http_methods[] = {
577 /* Basic HTTP request type */
578 "GET", "HEAD", "POST", "PUT", "DELETE", "OPTIONS", "TRACE", "CONNECT",
579 /* webDAV extensions (RFC2518) */
580 "PROPFIND", "PROPPATCH", "MOVE", "COPY", "MKCOL", "LOCK", "UNLOCK",
582 * Microsoft webDAV extension for Exchange 2000. See:
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 "BCOPY", "BMOVE", "BDELETE", "BPROPFIND", "BPROPPATCH",
588 * Another Microsoft webDAV extension for Exchange 2000. See:
589 * http://systems.cs.colorado.edu/grunwald/MobileComputing/Papers/draft-cohen-gena-p-base-00.txt
590 * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
591 * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
593 "SUBSCRIBE", "UNSUBSCRIBE", "NOTIFY", "POLL",
595 * Yet another WebDAV extension, this time for
596 * Web Distributed Authoring and Versioning (RFC3253)
598 "VERSION-CONTROL", "REPORT", "CHECKOUT", "CHECKIN", "UNCHECKOUT",
599 "MKWORKSPACE", "UPDATE", "LABEL", "MERGE", "BASELINE-CONTROL", "MKACTIVITY",
603 for (i = 0; i < SZ(known_http_methods); i++)
605 if (0 == strcmpic(method, known_http_methods[i]))
616 /*********************************************************************
618 * Function : parse_http_request
620 * Description : Parse out the host and port from the URL. Find the
621 * hostname & path, port (if ':'), and/or password (if '@')
624 * 1 : req = HTTP request line to break down
625 * 2 : http = pointer to the http structure to hold elements
626 * 3 : csp = Current client state (buffers, headers, etc...)
628 * Returns : JB_ERR_OK on success
629 * JB_ERR_MEMORY on out of memory
630 * JB_ERR_CGI_PARAMS on malformed command/URL
631 * or >100 domains deep.
633 *********************************************************************/
634 jb_err parse_http_request(const char *req,
635 struct http_request *http,
636 const struct client_state *csp)
639 char *v[10]; /* XXX: Why 10? We should only need three. */
643 memset(http, '\0', sizeof(*http));
648 return JB_ERR_MEMORY;
651 n = ssplit(buf, " \r\n", v, SZ(v), 1, 1);
659 * Fail in case of unknown methods
660 * which we might not handle correctly.
662 * XXX: There should be a config option
663 * to forward requests with unknown methods
664 * anyway. Most of them don't need special
667 if (unknown_method(v[0]))
669 log_error(LOG_LEVEL_ERROR, "Unknown HTTP method detected: %s", v[0]);
674 http->ssl = !strcmpic(v[0], "CONNECT");
676 err = parse_http_url(v[1], http, !http->ssl);
684 * Copy the details into the structure
686 http->cmd = strdup(req);
687 http->gpc = strdup(v[0]);
688 http->ver = strdup(v[2]);
692 if ( (http->cmd == NULL)
693 || (http->gpc == NULL)
694 || (http->ver == NULL) )
696 return JB_ERR_MEMORY;
704 /*********************************************************************
706 * Function : compile_pattern
708 * Description : Compiles a host, domain or TAG pattern.
711 * 1 : pattern = The pattern to compile.
712 * 2 : anchoring = How the regex should be anchored.
713 * Can be either one of NO_ANCHORING,
714 * LEFT_ANCHORED or RIGHT_ANCHORED.
715 * 3 : url = In case of failures, the spec member is
716 * logged and the structure freed.
717 * 4 : regex = Where the compiled regex should be stored.
719 * Returns : JB_ERR_OK - Success
720 * JB_ERR_MEMORY - Out of memory
721 * JB_ERR_PARSE - Cannot parse regex
723 *********************************************************************/
724 static jb_err compile_pattern(const char *pattern, enum regex_anchoring anchoring,
725 struct url_spec *url, regex_t **regex)
728 char rebuf[BUFFER_SIZE];
732 assert(strlen(pattern) < sizeof(rebuf) - 2);
734 if (pattern[0] == '\0')
752 log_error(LOG_LEVEL_FATAL,
753 "Invalid anchoring in compile_pattern %d", anchoring);
756 *regex = zalloc(sizeof(**regex));
760 return JB_ERR_MEMORY;
763 snprintf(rebuf, sizeof(rebuf), fmt, pattern);
765 errcode = regcomp(*regex, rebuf, (REG_EXTENDED|REG_NOSUB|REG_ICASE));
769 size_t errlen = regerror(errcode, *regex, rebuf, sizeof(rebuf));
770 if (errlen > (sizeof(rebuf) - (size_t)1))
772 errlen = sizeof(rebuf) - (size_t)1;
774 rebuf[errlen] = '\0';
775 log_error(LOG_LEVEL_ERROR, "error compiling %s from %s: %s",
776 pattern, url->spec, rebuf);
787 /*********************************************************************
789 * Function : compile_url_pattern
791 * Description : Compiles the three parts of an URL pattern.
794 * 1 : url = Target url_spec to be filled in.
795 * 2 : buf = The url pattern to compile. Will be messed up.
797 * Returns : JB_ERR_OK - Success
798 * JB_ERR_MEMORY - Out of memory
799 * JB_ERR_PARSE - Cannot parse regex
801 *********************************************************************/
802 static jb_err compile_url_pattern(struct url_spec *url, char *buf)
806 p = strchr(buf, '/');
810 * Only compile the regex if it consists of more than
811 * a single slash, otherwise it wouldn't affect the result.
816 * XXX: does it make sense to compile the slash at the beginning?
818 jb_err err = compile_pattern(p, LEFT_ANCHORED, url, &url->preg);
820 if (JB_ERR_OK != err)
828 p = strchr(buf, ':');
832 url->port_list = strdup(p);
833 if (NULL == url->port_list)
835 return JB_ERR_MEMORY;
840 url->port_list = NULL;
845 return compile_host_pattern(url, buf);
853 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
854 /*********************************************************************
856 * Function : compile_host_pattern
858 * Description : Parses and compiles a host pattern..
861 * 1 : url = Target url_spec to be filled in.
862 * 2 : host_pattern = Host pattern to compile.
864 * Returns : JB_ERR_OK - Success
865 * JB_ERR_MEMORY - Out of memory
866 * JB_ERR_PARSE - Cannot parse regex
868 *********************************************************************/
869 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
871 return compile_pattern(host_pattern, RIGHT_ANCHORED, url, &url->host_regex);
876 /*********************************************************************
878 * Function : compile_host_pattern
880 * Description : Parses and "compiles" an old-school host pattern.
883 * 1 : url = Target url_spec to be filled in.
884 * 2 : host_pattern = Host pattern to parse.
886 * Returns : JB_ERR_OK - Success
887 * JB_ERR_MEMORY - Out of memory
888 * JB_ERR_PARSE - Cannot parse regex
890 *********************************************************************/
891 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
900 if (host_pattern[strlen(host_pattern) - 1] == '.')
902 url->unanchored |= ANCHOR_RIGHT;
904 if (host_pattern[0] == '.')
906 url->unanchored |= ANCHOR_LEFT;
910 * Split domain into components
912 url->dbuffer = strdup(host_pattern);
913 if (NULL == url->dbuffer)
916 return JB_ERR_MEMORY;
922 for (p = url->dbuffer; *p ; p++)
924 *p = (char)tolower((int)(unsigned char)*p);
928 * Split the domain name into components
930 url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1);
935 return JB_ERR_MEMORY;
937 else if (url->dcount != 0)
940 * Save a copy of the pointers in dvec
942 size = (size_t)url->dcount * sizeof(*url->dvec);
944 url->dvec = (char **)malloc(size);
945 if (NULL == url->dvec)
948 return JB_ERR_MEMORY;
951 memcpy(url->dvec, v, size);
954 * else dcount == 0 in which case we needn't do anything,
955 * since dvec will never be accessed and the pattern will
962 /*********************************************************************
964 * Function : simplematch
966 * Description : String matching, with a (greedy) '*' wildcard that
967 * stands for zero or more arbitrary characters and
968 * character classes in [], which take both enumerations
972 * 1 : pattern = pattern for matching
973 * 2 : text = text to be matched
975 * Returns : 0 if match, else nonzero
977 *********************************************************************/
978 static int simplematch(const char *pattern, const char *text)
980 const unsigned char *pat = (const unsigned char *)pattern;
981 const unsigned char *txt = (const unsigned char *)text;
982 const unsigned char *fallback = pat;
985 unsigned char lastchar = 'a';
987 unsigned char charmap[32];
992 /* EOF pattern but !EOF text? */
1005 /* '*' in the pattern? */
1009 /* The pattern ends afterwards? Speed up the return. */
1015 /* Else, set wildcard mode and remember position after '*' */
1020 /* Character range specification? */
1023 memset(charmap, '\0', sizeof(charmap));
1025 while (*++pat != ']')
1031 else if (*pat == '-')
1033 if ((*++pat == ']') || *pat == '\0')
1037 for (i = lastchar; i <= *pat; i++)
1039 charmap[i / 8] |= (unsigned char)(1 << (i % 8));
1044 charmap[*pat / 8] |= (unsigned char)(1 << (*pat % 8));
1048 } /* -END- if Character range specification */
1052 * Char match, or char range match?
1056 || ((*pat == ']') && (charmap[*txt / 8] & (1 << (*txt % 8)))) )
1066 * No match && no wildcard: No luck
1070 else if (pat != fallback)
1073 * Increment text pointer if in char range matching
1080 * Wildcard mode && nonmatch beyond fallback: Rewind pattern
1084 * Restart matching from current text pointer
1091 /* Cut off extra '*'s */
1092 if(*pat == '*') pat++;
1094 /* If this is the pattern's end, fine! */
1100 /*********************************************************************
1102 * Function : simple_domaincmp
1104 * Description : Domain-wise Compare fqdn's. The comparison is
1105 * both left- and right-anchored. The individual
1106 * domain names are compared with simplematch().
1107 * This is only used by domain_match.
1110 * 1 : pv = array of patterns to compare
1111 * 2 : fv = array of domain components to compare
1112 * 3 : len = length of the arrays (both arrays are the
1113 * same length - if they weren't, it couldn't
1114 * possibly be a match).
1116 * Returns : 0 => domains are equivalent, else no match.
1118 *********************************************************************/
1119 static int simple_domaincmp(char **pv, char **fv, int len)
1123 for (n = 0; n < len; n++)
1125 if (simplematch(pv[n], fv[n]))
1136 /*********************************************************************
1138 * Function : domain_match
1140 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
1141 * pattern->unachored, the comparison is un-, left-,
1142 * right-anchored, or both.
1143 * The individual domain names are compared with
1147 * 1 : pattern = a domain that may contain a '*' as a wildcard.
1148 * 2 : fqdn = domain name against which the patterns are compared.
1150 * Returns : 0 => domains are equivalent, else no match.
1152 *********************************************************************/
1153 static int domain_match(const struct url_spec *pattern, const struct http_request *fqdn)
1155 char **pv, **fv; /* vectors */
1157 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
1159 plen = pattern->dcount;
1160 flen = fqdn->dcount;
1164 /* fqdn is too short to match this pattern */
1171 if (unanchored == ANCHOR_LEFT)
1176 * Convert this into a fully anchored pattern with
1177 * the fqdn and pattern the same length
1179 fv += (flen - plen); /* flen - plen >= 0 due to check above */
1180 return simple_domaincmp(pv, fv, plen);
1182 else if (unanchored == 0)
1184 /* Fully anchored, check length */
1189 return simple_domaincmp(pv, fv, plen);
1191 else if (unanchored == ANCHOR_RIGHT)
1193 /* Left anchored, ignore all extra in fqdn */
1194 return simple_domaincmp(pv, fv, plen);
1200 int maxn = flen - plen;
1201 for (n = 0; n <= maxn; n++)
1203 if (!simple_domaincmp(pv, fv, plen))
1208 * Doesn't match from start of fqdn
1209 * Try skipping first part of fqdn
1217 #endif /* def FEATURE_EXTENDED_HOST_PATTERNS */
1220 /*********************************************************************
1222 * Function : create_url_spec
1224 * Description : Creates a "url_spec" structure from a string.
1225 * When finished, free with free_url_spec().
1228 * 1 : url = Target url_spec to be filled in. Will be
1229 * zeroed before use.
1230 * 2 : buf = Source pattern, null terminated. NOTE: The
1231 * contents of this buffer are destroyed by this
1232 * function. If this function succeeds, the
1233 * buffer is copied to url->spec. If this
1234 * function fails, the contents of the buffer
1237 * Returns : JB_ERR_OK - Success
1238 * JB_ERR_MEMORY - Out of memory
1239 * JB_ERR_PARSE - Cannot parse regex (Detailed message
1240 * written to system log)
1242 *********************************************************************/
1243 jb_err create_url_spec(struct url_spec *url, char *buf)
1248 memset(url, '\0', sizeof(*url));
1250 /* Remember the original specification for the CGI pages. */
1251 url->spec = strdup(buf);
1252 if (NULL == url->spec)
1254 return JB_ERR_MEMORY;
1257 /* Is it tag pattern? */
1258 if (0 == strncmpic("TAG:", url->spec, 4))
1260 /* The pattern starts with the first character after "TAG:" */
1261 const char *tag_pattern = buf + 4;
1262 return compile_pattern(tag_pattern, NO_ANCHORING, url, &url->tag_regex);
1265 /* If it isn't a tag pattern it must be a URL pattern. */
1266 return compile_url_pattern(url, buf);
1270 /*********************************************************************
1272 * Function : free_url_spec
1274 * Description : Called from the "unloaders". Freez the url
1275 * structure elements.
1278 * 1 : url = pointer to a url_spec structure.
1282 *********************************************************************/
1283 void free_url_spec(struct url_spec *url)
1285 if (url == NULL) return;
1288 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
1289 if (url->host_regex)
1291 regfree(url->host_regex);
1292 freez(url->host_regex);
1295 freez(url->dbuffer);
1298 #endif /* ndef FEATURE_EXTENDED_HOST_PATTERNS */
1299 freez(url->port_list);
1307 regfree(url->tag_regex);
1308 freez(url->tag_regex);
1313 /*********************************************************************
1315 * Function : url_match
1317 * Description : Compare a URL against a URL pattern.
1320 * 1 : pattern = a URL pattern
1321 * 2 : url = URL to match
1323 * Returns : Nonzero if the URL matches the pattern, else 0.
1325 *********************************************************************/
1326 int url_match(const struct url_spec *pattern,
1327 const struct http_request *http)
1329 /* XXX: these should probably be functions. */
1330 #define PORT_MATCHES ((NULL == pattern->port_list) || match_portlist(pattern->port_list, http->port))
1331 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
1332 #define DOMAIN_MATCHES ((NULL == pattern->host_regex) || (0 == regexec(pattern->host_regex, http->host, 0, NULL, 0)))
1334 #define DOMAIN_MATCHES ((NULL == pattern->dbuffer) || (0 == domain_match(pattern, http)))
1336 #define PATH_MATCHES ((NULL == pattern->preg) || (0 == regexec(pattern->preg, http->path, 0, NULL, 0)))
1338 if (pattern->tag_regex != NULL)
1340 /* It's a tag pattern and shouldn't be matched against URLs */
1344 return (PORT_MATCHES && DOMAIN_MATCHES && PATH_MATCHES);
1349 /*********************************************************************
1351 * Function : match_portlist
1353 * Description : Check if a given number is covered by a comma
1354 * separated list of numbers and ranges (a,b-c,d,..)
1357 * 1 : portlist = String with list
1358 * 2 : port = port to check
1360 * Returns : 0 => no match
1363 *********************************************************************/
1364 int match_portlist(const char *portlist, int port)
1366 char *min, *max, *next, *portlist_copy;
1368 min = next = portlist_copy = strdup(portlist);
1371 * Zero-terminate first item and remember offset for next
1373 if (NULL != (next = strchr(portlist_copy, (int) ',')))
1379 * Loop through all items, checking for match
1383 if (NULL == (max = strchr(min, (int) '-')))
1386 * No dash, check for equality
1388 if (port == atoi(min))
1390 freez(portlist_copy);
1397 * This is a range, so check if between min and max,
1398 * or, if max was omitted, between min and 65K
1401 if(port >= atoi(min) && port <= (atoi(max) ? atoi(max) : 65535))
1403 freez(portlist_copy);
1415 * Zero-terminate next item and remember offset for n+1
1417 if ((NULL != next) && (NULL != (next = strchr(next, (int) ','))))
1423 freez(portlist_copy);