1 const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.12 2006/07/18 14:48:47 david__schmidt 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 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 1.12 2006/07/18 14:48:47 david__schmidt
37 * Reorganizing the repository: swapping out what was HEAD (the old 3.1 branch)
38 * with what was really the latest development (the v_3_0_branch branch)
40 * Revision 1.10.2.7 2003/05/17 15:57:24 oes
41 * - parse_http_url now checks memory allocation failure for
42 * duplication of "*" URL and rejects "*something" URLs
44 * - Added a comment to what might look like a bug in
45 * create_url_spec (see !bug #736931)
48 * Revision 1.10.2.6 2003/05/07 12:39:48 oes
49 * Fix typo: Default port for https URLs is 443, not 143.
50 * Thanks to Scott Tregear for spotting this one.
52 * Revision 1.10.2.5 2003/02/28 13:09:29 oes
53 * Fixed a rare double free condition as per Bug #694713
55 * Revision 1.10.2.4 2003/02/28 12:57:44 oes
56 * Moved freeing of http request structure to its owner
57 * as per Dan Price's observations in Bug #694713
59 * Revision 1.10.2.3 2002/11/12 16:50:40 oes
60 * Fixed memory leak in parse_http_request() reported by Oliver Stoeneberg. Fixes bug #637073
62 * Revision 1.10.2.2 2002/09/25 14:53:15 oes
63 * Added basic support for OPTIONS and TRACE HTTP methods:
64 * parse_http_url now recognizes the "*" URI as well as
65 * the OPTIONS and TRACE method keywords.
67 * Revision 1.10.2.1 2002/06/06 19:06:44 jongfoster
68 * Adding support for proprietary Microsoft WebDAV extensions
70 * Revision 1.10 2002/05/12 21:40:37 jongfoster
71 * - Removing some unused code
73 * Revision 1.9 2002/04/04 00:36:36 gliptak
74 * always use pcre for matching
76 * Revision 1.8 2002/04/03 23:32:47 jongfoster
77 * Fixing memory leak on error
79 * Revision 1.7 2002/03/26 22:29:55 swa
80 * we have a new homepage!
82 * Revision 1.6 2002/03/24 13:25:43 swa
83 * name change related issues
85 * Revision 1.5 2002/03/13 00:27:05 jongfoster
88 * Revision 1.4 2002/03/07 03:46:17 oes
89 * Fixed compiler warnings
91 * Revision 1.3 2002/03/03 14:51:11 oes
92 * Fixed CLF logging: Added ocmd member for client's request to struct http_request
94 * Revision 1.2 2002/01/21 00:14:09 jongfoster
95 * Correcting comment style
96 * Fixing an uninitialized memory bug in create_url_spec()
98 * Revision 1.1 2002/01/17 20:53:46 jongfoster
99 * Moving all our URL and URL pattern parsing code to the same file - it
100 * was scattered around in filters.c, loaders.c and parsers.c.
102 * Providing a single, simple url_match(pattern,url) function - rather than
103 * the 3-line match routine which was repeated all over the place.
105 * Renaming free_url to free_url_spec, since it frees a struct url_spec.
107 * Providing parse_http_url() so that URLs can be parsed without faking a
108 * HTTP request line for parse_http_request() or repeating the parsing
109 * code (both of which were techniques that were actually in use).
111 * Standardizing that struct http_request is used to represent a URL, and
112 * struct url_spec is used to represent a URL pattern. (Before, URLs were
113 * represented as seperate variables and a partially-filled-in url_spec).
116 *********************************************************************/
123 #include <sys/types.h>
131 #if !defined(_WIN32) && !defined(__OS2__)
136 #include "urlmatch.h"
138 #include "miscutil.h"
141 const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
144 /*********************************************************************
146 * Function : free_http_request
148 * Description : Freez a http_request structure
151 * 1 : http = points to a http_request structure to free
155 *********************************************************************/
156 void free_http_request(struct http_request *http)
165 freez(http->hostport);
168 freez(http->host_ip_addr_str);
169 freez(http->dbuffer);
174 /*********************************************************************
176 * Function : init_domain_components
178 * Description : Splits the domain name so we can compare it
179 * against wildcards. It used to be part of
180 * parse_http_url, but was separated because the
181 * same code is required in chat in case of
182 * intercepted requests.
185 * 1 : http = pointer to the http structure to hold elements.
187 * Returns : JB_ERR_OK on success
188 * JB_ERR_MEMORY on out of memory
189 * JB_ERR_PARSE on malformed command/URL
190 * or >100 domains deep.
192 *********************************************************************/
193 jb_err init_domain_components(struct http_request *http)
195 char *vec[BUFFER_SIZE];
199 http->dbuffer = strdup(http->host);
200 if (NULL == http->dbuffer)
202 return JB_ERR_MEMORY;
205 /* map to lower case */
206 for (p = http->dbuffer; *p ; p++)
208 *p = tolower((int)(unsigned char)*p);
211 /* split the domain name into components */
212 http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec), 1, 1);
214 if (http->dcount <= 0)
217 * Error: More than SZ(vec) components in domain
218 * or: no components in domain
220 log_error(LOG_LEVEL_ERROR, "More than SZ(vec) components in domain or none at all.");
224 /* save a copy of the pointers in dvec */
225 size = http->dcount * sizeof(*http->dvec);
227 http->dvec = (char **)malloc(size);
228 if (NULL == http->dvec)
230 return JB_ERR_MEMORY;
233 memcpy(http->dvec, vec, size);
238 /*********************************************************************
240 * Function : parse_http_url
242 * Description : Parse out the host and port from the URL. Find the
243 * hostname & path, port (if ':'), and/or password (if '@')
246 * 1 : url = URL (or is it URI?) to break down
247 * 2 : http = pointer to the http structure to hold elements.
248 * Will be zeroed before use. Note that this
249 * function sets the http->gpc and http->ver
251 * 3 : csp = Current client state (buffers, headers, etc...)
253 * Returns : JB_ERR_OK on success
254 * JB_ERR_MEMORY on out of memory
255 * JB_ERR_PARSE on malformed command/URL
256 * or >100 domains deep.
258 *********************************************************************/
259 jb_err parse_http_url(const char * url,
260 struct http_request *http,
261 struct client_state *csp)
263 int host_available = 1; /* A proxy can dream. */
266 * Zero out the results structure
268 memset(http, '\0', sizeof(*http));
272 * Save our initial URL
274 http->url = strdup(url);
275 if (http->url == NULL)
277 return JB_ERR_MEMORY;
282 * Check for * URI. If found, we're done.
284 if (*http->url == '*')
286 if ( NULL == (http->path = strdup("*"))
287 || NULL == (http->hostport = strdup("")) )
289 return JB_ERR_MEMORY;
291 if (http->url[1] != '\0')
300 * Split URL into protocol,hostport,path.
310 return JB_ERR_MEMORY;
313 /* Find the start of the URL in our scratch space */
315 if (strncmpic(url_noproto, "http://", 7) == 0)
320 else if (strncmpic(url_noproto, "https://", 8) == 0)
325 else if (*url_noproto == '/')
328 * Short request line without protocol and host.
329 * Most likely because the client's request
330 * was intercepted and redirected into Privoxy.
341 url_path = strchr(url_noproto, '/');
342 if (url_path != NULL)
347 * NOTE: The following line ignores the path for HTTPS URLS.
348 * This means that you get consistent behaviour if you type a
349 * https URL in and it's parsed by the function. (When the
350 * URL is actually retrieved, SSL hides the path part).
352 http->path = strdup(http->ssl ? "/" : url_path);
354 http->hostport = strdup(url_noproto);
359 * Repair broken HTTP requests that don't contain a path,
360 * or CONNECT requests
362 http->path = strdup("/");
363 http->hostport = strdup(url_noproto);
368 if ( (http->path == NULL)
369 || (http->hostport == NULL))
371 return JB_ERR_MEMORY;
377 /* Without host, there is nothing left to do here */
382 * Split hostport into user/password (ignored), host, port.
389 buf = strdup(http->hostport);
392 return JB_ERR_MEMORY;
395 /* check if url contains username and/or password */
396 host = strchr(buf, '@');
399 /* Contains username/password, skip it and the @ sign. */
404 /* No username or password. */
408 /* check if url contains port */
409 port = strchr(host, ':');
413 /* Terminate hostname and point to start of port string */
415 http->port = atoi(port);
419 /* No port specified. */
420 http->port = (http->ssl ? 443 : 80);
423 http->host = strdup(host);
427 if (http->host == NULL)
429 return JB_ERR_MEMORY;
434 * Split domain name so we can compare it against wildcards
436 return init_domain_components(http);
441 /*********************************************************************
443 * Function : parse_http_request
445 * Description : Parse out the host and port from the URL. Find the
446 * hostname & path, port (if ':'), and/or password (if '@')
449 * 1 : req = HTTP request line to break down
450 * 2 : http = pointer to the http structure to hold elements
451 * 3 : csp = Current client state (buffers, headers, etc...)
453 * Returns : JB_ERR_OK on success
454 * JB_ERR_MEMORY on out of memory
455 * JB_ERR_CGI_PARAMS on malformed command/URL
456 * or >100 domains deep.
458 *********************************************************************/
459 jb_err parse_http_request(const char *req,
460 struct http_request *http,
461 struct client_state *csp)
469 memset(http, '\0', sizeof(*http));
474 return JB_ERR_MEMORY;
477 n = ssplit(buf, " \r\n", v, SZ(v), 1, 1);
480 log_error(LOG_LEVEL_ERROR, "Trouble ssplitting: %s", buf);
485 /* this could be a CONNECT request */
486 if (strcmpic(v[0], "connect") == 0)
491 /* or it could be any other basic HTTP request type */
492 else if ((0 == strcmpic(v[0], "get"))
493 || (0 == strcmpic(v[0], "head"))
494 || (0 == strcmpic(v[0], "post"))
495 || (0 == strcmpic(v[0], "put"))
496 || (0 == strcmpic(v[0], "delete"))
497 || (0 == strcmpic(v[0], "options"))
498 || (0 == strcmpic(v[0], "trace"))
500 /* or a webDAV extension (RFC2518) */
501 || (0 == strcmpic(v[0], "propfind"))
502 || (0 == strcmpic(v[0], "proppatch"))
503 || (0 == strcmpic(v[0], "move"))
504 || (0 == strcmpic(v[0], "copy"))
505 || (0 == strcmpic(v[0], "mkcol"))
506 || (0 == strcmpic(v[0], "lock"))
507 || (0 == strcmpic(v[0], "unlock"))
509 /* Or a Microsoft webDAV extension for Exchange 2000. See: */
510 /* http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html */
511 /* http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp */
512 || (0 == strcmpic(v[0], "bcopy"))
513 || (0 == strcmpic(v[0], "bmove"))
514 || (0 == strcmpic(v[0], "bdelete"))
515 || (0 == strcmpic(v[0], "bpropfind"))
516 || (0 == strcmpic(v[0], "bproppatch"))
518 /* Or another Microsoft webDAV extension for Exchange 2000. See: */
519 /* http://systems.cs.colorado.edu/grunwald/MobileComputing/Papers/draft-cohen-gena-p-base-00.txt */
520 /* http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html */
521 /* http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp */
522 || (0 == strcmpic(v[0], "subscribe"))
523 || (0 == strcmpic(v[0], "unsubscribe"))
524 || (0 == strcmpic(v[0], "notify"))
525 || (0 == strcmpic(v[0], "poll"))
533 /* Unknown HTTP method */
534 log_error(LOG_LEVEL_ERROR, "Unknown HTTP method detected: %s", v[0]);
539 err = parse_http_url(v[1], http, csp);
547 * Copy the details into the structure
549 http->ssl = is_connect;
550 http->cmd = strdup(req);
551 http->gpc = strdup(v[0]);
552 http->ver = strdup(v[2]);
554 if ( (http->cmd == NULL)
555 || (http->gpc == NULL)
556 || (http->ver == NULL) )
559 return JB_ERR_MEMORY;
568 /*********************************************************************
570 * Function : simple_domaincmp
572 * Description : Domain-wise Compare fqdn's. The comparison is
573 * both left- and right-anchored. The individual
574 * domain names are compared with simplematch().
575 * This is only used by domain_match.
578 * 1 : pv = array of patterns to compare
579 * 2 : fv = array of domain components to compare
580 * 3 : len = length of the arrays (both arrays are the
581 * same length - if they weren't, it couldn't
582 * possibly be a match).
584 * Returns : 0 => domains are equivalent, else no match.
586 *********************************************************************/
587 static int simple_domaincmp(char **pv, char **fv, int len)
591 for (n = 0; n < len; n++)
593 if (simplematch(pv[n], fv[n]))
604 /*********************************************************************
606 * Function : domain_match
608 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
609 * pattern->unachored, the comparison is un-, left-,
610 * right-anchored, or both.
611 * The individual domain names are compared with
615 * 1 : pattern = a domain that may contain a '*' as a wildcard.
616 * 2 : fqdn = domain name against which the patterns are compared.
618 * Returns : 0 => domains are equivalent, else no match.
620 *********************************************************************/
621 static int domain_match(const struct url_spec *pattern, const struct http_request *fqdn)
623 char **pv, **fv; /* vectors */
625 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
627 plen = pattern->dcount;
632 /* fqdn is too short to match this pattern */
639 if (unanchored == ANCHOR_LEFT)
644 * Convert this into a fully anchored pattern with
645 * the fqdn and pattern the same length
647 fv += (flen - plen); /* flen - plen >= 0 due to check above */
648 return simple_domaincmp(pv, fv, plen);
650 else if (unanchored == 0)
652 /* Fully anchored, check length */
657 return simple_domaincmp(pv, fv, plen);
659 else if (unanchored == ANCHOR_RIGHT)
661 /* Left anchored, ignore all extra in fqdn */
662 return simple_domaincmp(pv, fv, plen);
668 int maxn = flen - plen;
669 for (n = 0; n <= maxn; n++)
671 if (!simple_domaincmp(pv, fv, plen))
676 * Doesn't match from start of fqdn
677 * Try skipping first part of fqdn
687 /*********************************************************************
689 * Function : create_url_spec
691 * Description : Creates a "url_spec" structure from a string.
692 * When finished, free with free_url_spec().
695 * 1 : url = Target url_spec to be filled in. Will be
697 * 2 : buf = Source pattern, null terminated. NOTE: The
698 * contents of this buffer are destroyed by this
699 * function. If this function succeeds, the
700 * buffer is copied to url->spec. If this
701 * function fails, the contents of the buffer
704 * Returns : JB_ERR_OK - Success
705 * JB_ERR_MEMORY - Out of memory
706 * JB_ERR_PARSE - Cannot parse regex (Detailed message
707 * written to system log)
709 *********************************************************************/
710 jb_err create_url_spec(struct url_spec * url, const char * buf)
720 memset(url, '\0', sizeof(*url));
723 * Save a copy of the orignal specification
725 if ((url->spec = strdup(buf)) == NULL)
727 return JB_ERR_MEMORY;
730 if ((p = strchr(buf, '/')) != NULL)
732 if (NULL == (url->path = strdup(p)))
735 return JB_ERR_MEMORY;
737 url->pathlen = strlen(url->path);
748 char rebuf[BUFFER_SIZE];
750 if (NULL == (url->preg = zalloc(sizeof(*url->preg))))
754 return JB_ERR_MEMORY;
757 sprintf(rebuf, "^(%s)", url->path);
759 errcode = regcomp(url->preg, rebuf,
760 (REG_EXTENDED|REG_NOSUB|REG_ICASE));
763 size_t errlen = regerror(errcode,
764 url->preg, rebuf, sizeof(rebuf));
766 if (errlen > (sizeof(rebuf) - (size_t)1))
768 errlen = sizeof(rebuf) - (size_t)1;
770 rebuf[errlen] = '\0';
772 log_error(LOG_LEVEL_ERROR, "error compiling %s: %s",
783 if ((p = strchr(buf, ':')) == NULL)
801 if (buf[strlen(buf) - 1] == '.')
803 url->unanchored |= ANCHOR_RIGHT;
807 url->unanchored |= ANCHOR_LEFT;
811 * Split domain into components
813 url->dbuffer = strdup(buf);
814 if (NULL == url->dbuffer)
820 return JB_ERR_MEMORY;
826 for (p = url->dbuffer; *p ; p++)
828 *p = tolower((int)(unsigned char)*p);
832 * Split the domain name into components
834 url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1);
844 return JB_ERR_MEMORY;
846 else if (url->dcount != 0)
850 * Save a copy of the pointers in dvec
852 size = url->dcount * sizeof(*url->dvec);
854 url->dvec = (char **)malloc(size);
855 if (NULL == url->dvec)
863 return JB_ERR_MEMORY;
866 memcpy(url->dvec, v, size);
869 * else dcount == 0 in which case we needn't do anything,
870 * since dvec will never be accessed and the pattern will
880 /*********************************************************************
882 * Function : free_url_spec
884 * Description : Called from the "unloaders". Freez the url
885 * structure elements.
888 * 1 : url = pointer to a url_spec structure.
892 *********************************************************************/
893 void free_url_spec(struct url_spec *url)
895 if (url == NULL) return;
909 /*********************************************************************
911 * Function : url_match
913 * Description : Compare a URL against a URL pattern.
916 * 1 : pattern = a URL pattern
917 * 2 : url = URL to match
919 * Returns : 0 iff the URL matches the pattern, else nonzero.
921 *********************************************************************/
922 int url_match(const struct url_spec *pattern,
923 const struct http_request *url)
925 return ((pattern->port == 0) || (pattern->port == url->port))
926 && ((pattern->dbuffer == NULL) || (domain_match(pattern, url) == 0))
927 && ((pattern->path == NULL) ||
928 (regexec(pattern->preg, url->path, 0, NULL, 0) == 0)