1 const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.40 2008/04/23 16:12:28 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 1.40 2008/04/23 16:12:28 fabiankeil
39 * Revision 1.39 2008/04/22 16:27:42 fabiankeil
40 * In parse_http_request(), remove a pointless
41 * temporary variable and free the buffer earlier.
43 * Revision 1.38 2008/04/18 05:17:18 fabiankeil
44 * Mark simplematch()'s parameters as immutable.
46 * Revision 1.37 2008/04/17 14:53:29 fabiankeil
47 * Move simplematch() into urlmatch.c as it's only
48 * used to match (old-school) domain patterns.
50 * Revision 1.36 2008/04/14 18:19:48 fabiankeil
51 * Remove now-pointless cast in create_url_spec().
53 * Revision 1.35 2008/04/14 18:11:21 fabiankeil
54 * The compiler might not notice it, but the buffer passed to
55 * create_url_spec() is modified later on and thus shouldn't
56 * be declared immutable.
58 * Revision 1.34 2008/04/13 13:32:07 fabiankeil
59 * Factor URL pattern compilation out of create_url_spec().
61 * Revision 1.33 2008/04/12 14:03:13 fabiankeil
62 * Remove an obvious comment and improve another one.
64 * Revision 1.32 2008/04/12 12:38:06 fabiankeil
65 * Factor out duplicated code to compile host, path and tag patterns.
67 * Revision 1.31 2008/04/10 14:41:04 fabiankeil
68 * Ditch url_spec's path member now that it's no longer used.
70 * Revision 1.30 2008/04/10 04:24:24 fabiankeil
71 * Stop duplicating the plain text representation of the path regex
72 * (and keeping the copy around). Once the regex is compiled it's no
75 * Revision 1.29 2008/04/10 04:17:56 fabiankeil
76 * In url_match(), check the right member for NULL when determining
77 * whether there's a path regex to execute. Looking for a plain-text
78 * representation works as well, but it looks "interesting" and that
79 * member will be removed soonish anyway.
81 * Revision 1.28 2008/04/08 16:07:39 fabiankeil
82 * Make it harder to mistake url_match()'s
83 * second parameter for an url_spec.
85 * Revision 1.27 2008/04/08 15:44:33 fabiankeil
86 * Save a bit of memory (and a few cpu cycles) by not bothering to
87 * compile slash-only path regexes that don't affect the result.
89 * Revision 1.26 2008/04/07 16:57:18 fabiankeil
90 * - Use free_url_spec() more consistently.
91 * - Let it reset url->dcount just in case.
93 * Revision 1.25 2008/04/06 15:18:38 fabiankeil
94 * Oh well, rename the --enable-pcre-host-patterns option to
95 * --enable-extended-host-patterns as it's not really PCRE syntax.
97 * Revision 1.24 2008/04/06 14:54:26 fabiankeil
98 * Use PCRE syntax in host patterns when configured
99 * with --enable-pcre-host-patterns.
101 * Revision 1.23 2008/04/05 12:19:20 fabiankeil
102 * Factor compile_host_pattern() out of create_url_spec().
104 * Revision 1.22 2008/03/30 15:02:32 fabiankeil
105 * SZitify unknown_method().
107 * Revision 1.21 2007/12/24 16:34:23 fabiankeil
108 * Band-aid (and micro-optimization) that makes it less likely to run out of
109 * stack space with overly-complex path patterns. Probably masks the problem
110 * reported by Lee in #1856679. Hohoho.
112 * Revision 1.20 2007/09/02 15:31:20 fabiankeil
113 * Move match_portlist() from filter.c to urlmatch.c.
114 * It's used for url matching, not for filtering.
116 * Revision 1.19 2007/09/02 13:42:11 fabiankeil
117 * - Allow port lists in url patterns.
118 * - Ditch unused url_spec member pathlen.
120 * Revision 1.18 2007/07/30 16:42:21 fabiankeil
121 * Move the method check into unknown_method()
122 * and loop through the known methods instead
123 * of using a screen-long OR chain.
125 * Revision 1.17 2007/04/15 16:39:21 fabiankeil
126 * Introduce tags as alternative way to specify which
127 * actions apply to a request. At the moment tags can be
128 * created based on client and server headers.
130 * Revision 1.16 2007/02/13 13:59:24 fabiankeil
131 * Remove redundant log message.
133 * Revision 1.15 2007/01/28 16:11:23 fabiankeil
134 * Accept WebDAV methods for subversion
135 * in parse_http_request(). Closes FR 1581425.
137 * Revision 1.14 2007/01/06 14:23:56 fabiankeil
138 * Fix gcc43 warnings. Mark *csp as immutable
139 * for parse_http_url() and url_match().
140 * Replace a sprintf call with snprintf.
142 * Revision 1.13 2006/12/06 19:50:54 fabiankeil
143 * parse_http_url() now handles intercepted
144 * HTTP request lines as well. Moved parts
145 * of parse_http_url()'s code into
146 * init_domain_components() so that it can
147 * be reused in chat().
149 * Revision 1.12 2006/07/18 14:48:47 david__schmidt
150 * Reorganizing the repository: swapping out what was HEAD (the old 3.1 branch)
151 * with what was really the latest development (the v_3_0_branch branch)
153 * Revision 1.10.2.7 2003/05/17 15:57:24 oes
154 * - parse_http_url now checks memory allocation failure for
155 * duplication of "*" URL and rejects "*something" URLs
157 * - Added a comment to what might look like a bug in
158 * create_url_spec (see !bug #736931)
159 * - Comment cosmetics
161 * Revision 1.10.2.6 2003/05/07 12:39:48 oes
162 * Fix typo: Default port for https URLs is 443, not 143.
163 * Thanks to Scott Tregear for spotting this one.
165 * Revision 1.10.2.5 2003/02/28 13:09:29 oes
166 * Fixed a rare double free condition as per Bug #694713
168 * Revision 1.10.2.4 2003/02/28 12:57:44 oes
169 * Moved freeing of http request structure to its owner
170 * as per Dan Price's observations in Bug #694713
172 * Revision 1.10.2.3 2002/11/12 16:50:40 oes
173 * Fixed memory leak in parse_http_request() reported by Oliver Stoeneberg. Fixes bug #637073
175 * Revision 1.10.2.2 2002/09/25 14:53:15 oes
176 * Added basic support for OPTIONS and TRACE HTTP methods:
177 * parse_http_url now recognizes the "*" URI as well as
178 * the OPTIONS and TRACE method keywords.
180 * Revision 1.10.2.1 2002/06/06 19:06:44 jongfoster
181 * Adding support for proprietary Microsoft WebDAV extensions
183 * Revision 1.10 2002/05/12 21:40:37 jongfoster
184 * - Removing some unused code
186 * Revision 1.9 2002/04/04 00:36:36 gliptak
187 * always use pcre for matching
189 * Revision 1.8 2002/04/03 23:32:47 jongfoster
190 * Fixing memory leak on error
192 * Revision 1.7 2002/03/26 22:29:55 swa
193 * we have a new homepage!
195 * Revision 1.6 2002/03/24 13:25:43 swa
196 * name change related issues
198 * Revision 1.5 2002/03/13 00:27:05 jongfoster
201 * Revision 1.4 2002/03/07 03:46:17 oes
202 * Fixed compiler warnings
204 * Revision 1.3 2002/03/03 14:51:11 oes
205 * Fixed CLF logging: Added ocmd member for client's request to struct http_request
207 * Revision 1.2 2002/01/21 00:14:09 jongfoster
208 * Correcting comment style
209 * Fixing an uninitialized memory bug in create_url_spec()
211 * Revision 1.1 2002/01/17 20:53:46 jongfoster
212 * Moving all our URL and URL pattern parsing code to the same file - it
213 * was scattered around in filters.c, loaders.c and parsers.c.
215 * Providing a single, simple url_match(pattern,url) function - rather than
216 * the 3-line match routine which was repeated all over the place.
218 * Renaming free_url to free_url_spec, since it frees a struct url_spec.
220 * Providing parse_http_url() so that URLs can be parsed without faking a
221 * HTTP request line for parse_http_request() or repeating the parsing
222 * code (both of which were techniques that were actually in use).
224 * Standardizing that struct http_request is used to represent a URL, and
225 * struct url_spec is used to represent a URL pattern. (Before, URLs were
226 * represented as seperate variables and a partially-filled-in url_spec).
229 *********************************************************************/
236 #include <sys/types.h>
244 #if !defined(_WIN32) && !defined(__OS2__)
249 #include "urlmatch.h"
251 #include "miscutil.h"
254 const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
256 enum regex_anchoring {NO_ANCHORING, LEFT_ANCHORED, RIGHT_ANCHORED};
257 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern);
259 /*********************************************************************
261 * Function : free_http_request
263 * Description : Freez a http_request structure
266 * 1 : http = points to a http_request structure to free
270 *********************************************************************/
271 void free_http_request(struct http_request *http)
280 freez(http->hostport);
283 freez(http->host_ip_addr_str);
284 freez(http->dbuffer);
290 /*********************************************************************
292 * Function : init_domain_components
294 * Description : Splits the domain name so we can compare it
295 * against wildcards. It used to be part of
296 * parse_http_url, but was separated because the
297 * same code is required in chat in case of
298 * intercepted requests.
301 * 1 : http = pointer to the http structure to hold elements.
303 * Returns : JB_ERR_OK on success
304 * JB_ERR_MEMORY on out of memory
305 * JB_ERR_PARSE on malformed command/URL
306 * or >100 domains deep.
308 *********************************************************************/
309 jb_err init_domain_components(struct http_request *http)
311 char *vec[BUFFER_SIZE];
315 http->dbuffer = strdup(http->host);
316 if (NULL == http->dbuffer)
318 return JB_ERR_MEMORY;
321 /* map to lower case */
322 for (p = http->dbuffer; *p ; p++)
324 *p = (char)tolower((int)(unsigned char)*p);
327 /* split the domain name into components */
328 http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec), 1, 1);
330 if (http->dcount <= 0)
333 * Error: More than SZ(vec) components in domain
334 * or: no components in domain
336 log_error(LOG_LEVEL_ERROR, "More than SZ(vec) components in domain or none at all.");
340 /* save a copy of the pointers in dvec */
341 size = (size_t)http->dcount * sizeof(*http->dvec);
343 http->dvec = (char **)malloc(size);
344 if (NULL == http->dvec)
346 return JB_ERR_MEMORY;
349 memcpy(http->dvec, vec, size);
355 /*********************************************************************
357 * Function : parse_http_url
359 * Description : Parse out the host and port from the URL. Find the
360 * hostname & path, port (if ':'), and/or password (if '@')
363 * 1 : url = URL (or is it URI?) to break down
364 * 2 : http = pointer to the http structure to hold elements.
365 * Must be initialized with valid values (like NULLs).
366 * 3 : csp = Current client state (buffers, headers, etc...)
368 * Returns : JB_ERR_OK on success
369 * JB_ERR_MEMORY on out of memory
370 * JB_ERR_PARSE on malformed command/URL
371 * or >100 domains deep.
373 *********************************************************************/
374 jb_err parse_http_url(const char * url,
375 struct http_request *http,
376 const struct client_state *csp)
378 int host_available = 1; /* A proxy can dream. */
381 * Save our initial URL
383 http->url = strdup(url);
384 if (http->url == NULL)
386 return JB_ERR_MEMORY;
391 * Check for * URI. If found, we're done.
393 if (*http->url == '*')
395 if ( NULL == (http->path = strdup("*"))
396 || NULL == (http->hostport = strdup("")) )
398 return JB_ERR_MEMORY;
400 if (http->url[1] != '\0')
409 * Split URL into protocol,hostport,path.
419 return JB_ERR_MEMORY;
422 /* Find the start of the URL in our scratch space */
424 if (strncmpic(url_noproto, "http://", 7) == 0)
428 else if (strncmpic(url_noproto, "https://", 8) == 0)
431 * Should only happen when called from cgi_show_url_info().
436 else if (*url_noproto == '/')
439 * Short request line without protocol and host.
440 * Most likely because the client's request
441 * was intercepted and redirected into Privoxy.
447 url_path = strchr(url_noproto, '/');
448 if (url_path != NULL)
453 * NOTE: The following line ignores the path for HTTPS URLS.
454 * This means that you get consistent behaviour if you type a
455 * https URL in and it's parsed by the function. (When the
456 * URL is actually retrieved, SSL hides the path part).
458 http->path = strdup(http->ssl ? "/" : url_path);
460 http->hostport = strdup(url_noproto);
465 * Repair broken HTTP requests that don't contain a path,
466 * or CONNECT requests
468 http->path = strdup("/");
469 http->hostport = strdup(url_noproto);
474 if ( (http->path == NULL)
475 || (http->hostport == NULL))
477 return JB_ERR_MEMORY;
483 /* Without host, there is nothing left to do here */
488 * Split hostport into user/password (ignored), host, port.
495 buf = strdup(http->hostport);
498 return JB_ERR_MEMORY;
501 /* check if url contains username and/or password */
502 host = strchr(buf, '@');
505 /* Contains username/password, skip it and the @ sign. */
510 /* No username or password. */
514 /* check if url contains port */
515 port = strchr(host, ':');
519 /* Terminate hostname and point to start of port string */
521 http->port = atoi(port);
525 /* No port specified. */
526 http->port = (http->ssl ? 443 : 80);
529 http->host = strdup(host);
533 if (http->host == NULL)
535 return JB_ERR_MEMORY;
540 * Split domain name so we can compare it against wildcards
542 return init_domain_components(http);
547 /*********************************************************************
549 * Function : unknown_method
551 * Description : Checks whether a method is unknown.
554 * 1 : method = points to a http method
556 * Returns : TRUE if it's unknown, FALSE otherwise.
558 *********************************************************************/
559 static int unknown_method(const char *method)
561 static const char *known_http_methods[] = {
562 /* Basic HTTP request type */
563 "GET", "HEAD", "POST", "PUT", "DELETE", "OPTIONS", "TRACE", "CONNECT",
564 /* webDAV extensions (RFC2518) */
565 "PROPFIND", "PROPPATCH", "MOVE", "COPY", "MKCOL", "LOCK", "UNLOCK",
567 * Microsoft webDAV extension for Exchange 2000. See:
568 * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
569 * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
571 "BCOPY", "BMOVE", "BDELETE", "BPROPFIND", "BPROPPATCH",
573 * Another Microsoft webDAV extension for Exchange 2000. See:
574 * http://systems.cs.colorado.edu/grunwald/MobileComputing/Papers/draft-cohen-gena-p-base-00.txt
575 * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
576 * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
578 "SUBSCRIBE", "UNSUBSCRIBE", "NOTIFY", "POLL",
580 * Yet another WebDAV extension, this time for
581 * Web Distributed Authoring and Versioning (RFC3253)
583 "VERSION-CONTROL", "REPORT", "CHECKOUT", "CHECKIN", "UNCHECKOUT",
584 "MKWORKSPACE", "UPDATE", "LABEL", "MERGE", "BASELINE-CONTROL", "MKACTIVITY",
588 for (i = 0; i < SZ(known_http_methods); i++)
590 if (0 == strcmpic(method, known_http_methods[i]))
601 /*********************************************************************
603 * Function : parse_http_request
605 * Description : Parse out the host and port from the URL. Find the
606 * hostname & path, port (if ':'), and/or password (if '@')
609 * 1 : req = HTTP request line to break down
610 * 2 : http = pointer to the http structure to hold elements
611 * 3 : csp = Current client state (buffers, headers, etc...)
613 * Returns : JB_ERR_OK on success
614 * JB_ERR_MEMORY on out of memory
615 * JB_ERR_CGI_PARAMS on malformed command/URL
616 * or >100 domains deep.
618 *********************************************************************/
619 jb_err parse_http_request(const char *req,
620 struct http_request *http,
621 const struct client_state *csp)
624 char *v[10]; /* XXX: Why 10? We should only need three. */
628 memset(http, '\0', sizeof(*http));
633 return JB_ERR_MEMORY;
636 n = ssplit(buf, " \r\n", v, SZ(v), 1, 1);
644 * Fail in case of unknown methods
645 * which we might not handle correctly.
647 * XXX: There should be a config option
648 * to forward requests with unknown methods
649 * anyway. Most of them don't need special
652 if (unknown_method(v[0]))
654 log_error(LOG_LEVEL_ERROR, "Unknown HTTP method detected: %s", v[0]);
659 err = parse_http_url(v[1], http, csp);
667 * Copy the details into the structure
669 http->ssl = !strcmpic(v[0], "CONNECT");
670 http->cmd = strdup(req);
671 http->gpc = strdup(v[0]);
672 http->ver = strdup(v[2]);
676 if ( (http->cmd == NULL)
677 || (http->gpc == NULL)
678 || (http->ver == NULL) )
680 return JB_ERR_MEMORY;
688 /*********************************************************************
690 * Function : compile_pattern
692 * Description : Compiles a host, domain or TAG pattern.
695 * 1 : pattern = The pattern to compile.
696 * 2 : anchoring = How the regex should be anchored.
697 * Can be either one of NO_ANCHORING,
698 * LEFT_ANCHORED or RIGHT_ANCHORED.
699 * 3 : url = In case of failures, the spec member is
700 * logged and the structure freed.
701 * 4 : regex = Where the compiled regex should be stored.
703 * Returns : JB_ERR_OK - Success
704 * JB_ERR_MEMORY - Out of memory
705 * JB_ERR_PARSE - Cannot parse regex
707 *********************************************************************/
708 static jb_err compile_pattern(const char *pattern, enum regex_anchoring anchoring,
709 struct url_spec *url, regex_t **regex)
712 char rebuf[BUFFER_SIZE];
716 assert(strlen(pattern) < sizeof(rebuf) - 2);
718 if (pattern[0] == '\0')
736 log_error(LOG_LEVEL_FATAL,
737 "Invalid anchoring in compile_pattern %d", anchoring);
740 *regex = zalloc(sizeof(**regex));
744 return JB_ERR_MEMORY;
747 snprintf(rebuf, sizeof(rebuf), fmt, pattern);
749 errcode = regcomp(*regex, rebuf, (REG_EXTENDED|REG_NOSUB|REG_ICASE));
753 size_t errlen = regerror(errcode, *regex, rebuf, sizeof(rebuf));
754 if (errlen > (sizeof(rebuf) - (size_t)1))
756 errlen = sizeof(rebuf) - (size_t)1;
758 rebuf[errlen] = '\0';
759 log_error(LOG_LEVEL_ERROR, "error compiling %s from %s: %s",
760 pattern, url->spec, rebuf);
771 /*********************************************************************
773 * Function : compile_url_pattern
775 * Description : Compiles the three parts of an URL pattern.
778 * 1 : url = Target url_spec to be filled in.
779 * 2 : buf = The url pattern to compile. Will be messed up.
781 * Returns : JB_ERR_OK - Success
782 * JB_ERR_MEMORY - Out of memory
783 * JB_ERR_PARSE - Cannot parse regex
785 *********************************************************************/
786 static jb_err compile_url_pattern(struct url_spec *url, char *buf)
790 p = strchr(buf, '/');
794 * Only compile the regex if it consists of more than
795 * a single slash, otherwise it wouldn't affect the result.
800 * XXX: does it make sense to compile the slash at the beginning?
802 jb_err err = compile_pattern(p, LEFT_ANCHORED, url, &url->preg);
804 if (JB_ERR_OK != err)
812 p = strchr(buf, ':');
816 url->port_list = strdup(p);
817 if (NULL == url->port_list)
819 return JB_ERR_MEMORY;
824 url->port_list = NULL;
829 return compile_host_pattern(url, buf);
837 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
838 /*********************************************************************
840 * Function : compile_host_pattern
842 * Description : Parses and compiles a host pattern..
845 * 1 : url = Target url_spec to be filled in.
846 * 2 : host_pattern = Host pattern to compile.
848 * Returns : JB_ERR_OK - Success
849 * JB_ERR_MEMORY - Out of memory
850 * JB_ERR_PARSE - Cannot parse regex
852 *********************************************************************/
853 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
855 return compile_pattern(host_pattern, RIGHT_ANCHORED, url, &url->host_regex);
860 /*********************************************************************
862 * Function : compile_host_pattern
864 * Description : Parses and "compiles" an old-school host pattern.
867 * 1 : url = Target url_spec to be filled in.
868 * 2 : host_pattern = Host pattern to parse.
870 * Returns : JB_ERR_OK - Success
871 * JB_ERR_MEMORY - Out of memory
872 * JB_ERR_PARSE - Cannot parse regex
874 *********************************************************************/
875 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
884 if (host_pattern[strlen(host_pattern) - 1] == '.')
886 url->unanchored |= ANCHOR_RIGHT;
888 if (host_pattern[0] == '.')
890 url->unanchored |= ANCHOR_LEFT;
894 * Split domain into components
896 url->dbuffer = strdup(host_pattern);
897 if (NULL == url->dbuffer)
900 return JB_ERR_MEMORY;
906 for (p = url->dbuffer; *p ; p++)
908 *p = (char)tolower((int)(unsigned char)*p);
912 * Split the domain name into components
914 url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1);
919 return JB_ERR_MEMORY;
921 else if (url->dcount != 0)
924 * Save a copy of the pointers in dvec
926 size = (size_t)url->dcount * sizeof(*url->dvec);
928 url->dvec = (char **)malloc(size);
929 if (NULL == url->dvec)
932 return JB_ERR_MEMORY;
935 memcpy(url->dvec, v, size);
938 * else dcount == 0 in which case we needn't do anything,
939 * since dvec will never be accessed and the pattern will
946 /*********************************************************************
948 * Function : simplematch
950 * Description : String matching, with a (greedy) '*' wildcard that
951 * stands for zero or more arbitrary characters and
952 * character classes in [], which take both enumerations
956 * 1 : pattern = pattern for matching
957 * 2 : text = text to be matched
959 * Returns : 0 if match, else nonzero
961 *********************************************************************/
962 static int simplematch(const char *pattern, const char *text)
964 const unsigned char *pat = (const unsigned char *)pattern;
965 const unsigned char *txt = (const unsigned char *)text;
966 const unsigned char *fallback = pat;
969 unsigned char lastchar = 'a';
971 unsigned char charmap[32];
976 /* EOF pattern but !EOF text? */
989 /* '*' in the pattern? */
993 /* The pattern ends afterwards? Speed up the return. */
999 /* Else, set wildcard mode and remember position after '*' */
1004 /* Character range specification? */
1007 memset(charmap, '\0', sizeof(charmap));
1009 while (*++pat != ']')
1015 else if (*pat == '-')
1017 if ((*++pat == ']') || *pat == '\0')
1021 for (i = lastchar; i <= *pat; i++)
1023 charmap[i / 8] |= (unsigned char)(1 << (i % 8));
1028 charmap[*pat / 8] |= (unsigned char)(1 << (*pat % 8));
1032 } /* -END- if Character range specification */
1036 * Char match, or char range match?
1040 || ((*pat == ']') && (charmap[*txt / 8] & (1 << (*txt % 8)))) )
1050 * No match && no wildcard: No luck
1054 else if (pat != fallback)
1057 * Increment text pointer if in char range matching
1064 * Wildcard mode && nonmatch beyond fallback: Rewind pattern
1068 * Restart matching from current text pointer
1075 /* Cut off extra '*'s */
1076 if(*pat == '*') pat++;
1078 /* If this is the pattern's end, fine! */
1084 /*********************************************************************
1086 * Function : simple_domaincmp
1088 * Description : Domain-wise Compare fqdn's. The comparison is
1089 * both left- and right-anchored. The individual
1090 * domain names are compared with simplematch().
1091 * This is only used by domain_match.
1094 * 1 : pv = array of patterns to compare
1095 * 2 : fv = array of domain components to compare
1096 * 3 : len = length of the arrays (both arrays are the
1097 * same length - if they weren't, it couldn't
1098 * possibly be a match).
1100 * Returns : 0 => domains are equivalent, else no match.
1102 *********************************************************************/
1103 static int simple_domaincmp(char **pv, char **fv, int len)
1107 for (n = 0; n < len; n++)
1109 if (simplematch(pv[n], fv[n]))
1120 /*********************************************************************
1122 * Function : domain_match
1124 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
1125 * pattern->unachored, the comparison is un-, left-,
1126 * right-anchored, or both.
1127 * The individual domain names are compared with
1131 * 1 : pattern = a domain that may contain a '*' as a wildcard.
1132 * 2 : fqdn = domain name against which the patterns are compared.
1134 * Returns : 0 => domains are equivalent, else no match.
1136 *********************************************************************/
1137 static int domain_match(const struct url_spec *pattern, const struct http_request *fqdn)
1139 char **pv, **fv; /* vectors */
1141 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
1143 plen = pattern->dcount;
1144 flen = fqdn->dcount;
1148 /* fqdn is too short to match this pattern */
1155 if (unanchored == ANCHOR_LEFT)
1160 * Convert this into a fully anchored pattern with
1161 * the fqdn and pattern the same length
1163 fv += (flen - plen); /* flen - plen >= 0 due to check above */
1164 return simple_domaincmp(pv, fv, plen);
1166 else if (unanchored == 0)
1168 /* Fully anchored, check length */
1173 return simple_domaincmp(pv, fv, plen);
1175 else if (unanchored == ANCHOR_RIGHT)
1177 /* Left anchored, ignore all extra in fqdn */
1178 return simple_domaincmp(pv, fv, plen);
1184 int maxn = flen - plen;
1185 for (n = 0; n <= maxn; n++)
1187 if (!simple_domaincmp(pv, fv, plen))
1192 * Doesn't match from start of fqdn
1193 * Try skipping first part of fqdn
1201 #endif /* def FEATURE_EXTENDED_HOST_PATTERNS */
1204 /*********************************************************************
1206 * Function : create_url_spec
1208 * Description : Creates a "url_spec" structure from a string.
1209 * When finished, free with free_url_spec().
1212 * 1 : url = Target url_spec to be filled in. Will be
1213 * zeroed before use.
1214 * 2 : buf = Source pattern, null terminated. NOTE: The
1215 * contents of this buffer are destroyed by this
1216 * function. If this function succeeds, the
1217 * buffer is copied to url->spec. If this
1218 * function fails, the contents of the buffer
1221 * Returns : JB_ERR_OK - Success
1222 * JB_ERR_MEMORY - Out of memory
1223 * JB_ERR_PARSE - Cannot parse regex (Detailed message
1224 * written to system log)
1226 *********************************************************************/
1227 jb_err create_url_spec(struct url_spec *url, char *buf)
1232 memset(url, '\0', sizeof(*url));
1234 /* Remember the original specification for the CGI pages. */
1235 url->spec = strdup(buf);
1236 if (NULL == url->spec)
1238 return JB_ERR_MEMORY;
1241 /* Is it tag pattern? */
1242 if (0 == strncmpic("TAG:", url->spec, 4))
1244 /* The pattern starts with the first character after "TAG:" */
1245 const char *tag_pattern = buf + 4;
1246 return compile_pattern(tag_pattern, NO_ANCHORING, url, &url->tag_regex);
1249 /* If it isn't a tag pattern it must be a URL pattern. */
1250 return compile_url_pattern(url, buf);
1254 /*********************************************************************
1256 * Function : free_url_spec
1258 * Description : Called from the "unloaders". Freez the url
1259 * structure elements.
1262 * 1 : url = pointer to a url_spec structure.
1266 *********************************************************************/
1267 void free_url_spec(struct url_spec *url)
1269 if (url == NULL) return;
1272 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
1273 if (url->host_regex)
1275 regfree(url->host_regex);
1276 freez(url->host_regex);
1279 freez(url->dbuffer);
1282 #endif /* ndef FEATURE_EXTENDED_HOST_PATTERNS */
1283 freez(url->port_list);
1291 regfree(url->tag_regex);
1292 freez(url->tag_regex);
1297 /*********************************************************************
1299 * Function : url_match
1301 * Description : Compare a URL against a URL pattern.
1304 * 1 : pattern = a URL pattern
1305 * 2 : url = URL to match
1307 * Returns : Nonzero if the URL matches the pattern, else 0.
1309 *********************************************************************/
1310 int url_match(const struct url_spec *pattern,
1311 const struct http_request *http)
1313 /* XXX: these should probably be functions. */
1314 #define PORT_MATCHES ((NULL == pattern->port_list) || match_portlist(pattern->port_list, http->port))
1315 #ifdef FEATURE_EXTENDED_HOST_PATTERNS
1316 #define DOMAIN_MATCHES ((NULL == pattern->host_regex) || (0 == regexec(pattern->host_regex, http->host, 0, NULL, 0)))
1318 #define DOMAIN_MATCHES ((NULL == pattern->dbuffer) || (0 == domain_match(pattern, http)))
1320 #define PATH_MATCHES ((NULL == pattern->preg) || (0 == regexec(pattern->preg, http->path, 0, NULL, 0)))
1322 if (pattern->tag_regex != NULL)
1324 /* It's a tag pattern and shouldn't be matched against URLs */
1328 return (PORT_MATCHES && DOMAIN_MATCHES && PATH_MATCHES);
1333 /*********************************************************************
1335 * Function : match_portlist
1337 * Description : Check if a given number is covered by a comma
1338 * separated list of numbers and ranges (a,b-c,d,..)
1341 * 1 : portlist = String with list
1342 * 2 : port = port to check
1344 * Returns : 0 => no match
1347 *********************************************************************/
1348 int match_portlist(const char *portlist, int port)
1350 char *min, *max, *next, *portlist_copy;
1352 min = next = portlist_copy = strdup(portlist);
1355 * Zero-terminate first item and remember offset for next
1357 if (NULL != (next = strchr(portlist_copy, (int) ',')))
1363 * Loop through all items, checking for match
1367 if (NULL == (max = strchr(min, (int) '-')))
1370 * No dash, check for equality
1372 if (port == atoi(min))
1374 freez(portlist_copy);
1381 * This is a range, so check if between min and max,
1382 * or, if max was omitted, between min and 65K
1385 if(port >= atoi(min) && port <= (atoi(max) ? atoi(max) : 65535))
1387 freez(portlist_copy);
1399 * Zero-terminate next item and remember offset for n+1
1401 if ((NULL != next) && (NULL != (next = strchr(next, (int) ','))))
1407 freez(portlist_copy);