1 const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.10.2.6 2003/05/07 12:39:48 oes Exp $";
2 /*********************************************************************
4 * File : $Source: /cvsroot/ijbswa/current/Attic/urlmatch.c,v $
6 * Purpose : Declares functions to match URLs against URL
9 * Copyright : Written by and Copyright (C) 2001 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 1.10.2.6 2003/05/07 12:39:48 oes
37 * Fix typo: Default port for https URLs is 443, not 143.
38 * Thanks to Scott Tregear for spotting this one.
40 * Revision 1.10.2.5 2003/02/28 13:09:29 oes
41 * Fixed a rare double free condition as per Bug #694713
43 * Revision 1.10.2.4 2003/02/28 12:57:44 oes
44 * Moved freeing of http request structure to its owner
45 * as per Dan Price's observations in Bug #694713
47 * Revision 1.10.2.3 2002/11/12 16:50:40 oes
48 * Fixed memory leak in parse_http_request() reported by Oliver Stoeneberg. Fixes bug #637073
50 * Revision 1.10.2.2 2002/09/25 14:53:15 oes
51 * Added basic support for OPTIONS and TRACE HTTP methods:
52 * parse_http_url now recognizes the "*" URI as well as
53 * the OPTIONS and TRACE method keywords.
55 * Revision 1.10.2.1 2002/06/06 19:06:44 jongfoster
56 * Adding support for proprietary Microsoft WebDAV extensions
58 * Revision 1.10 2002/05/12 21:40:37 jongfoster
59 * - Removing some unused code
61 * Revision 1.9 2002/04/04 00:36:36 gliptak
62 * always use pcre for matching
64 * Revision 1.8 2002/04/03 23:32:47 jongfoster
65 * Fixing memory leak on error
67 * Revision 1.7 2002/03/26 22:29:55 swa
68 * we have a new homepage!
70 * Revision 1.6 2002/03/24 13:25:43 swa
71 * name change related issues
73 * Revision 1.5 2002/03/13 00:27:05 jongfoster
76 * Revision 1.4 2002/03/07 03:46:17 oes
77 * Fixed compiler warnings
79 * Revision 1.3 2002/03/03 14:51:11 oes
80 * Fixed CLF logging: Added ocmd member for client's request to struct http_request
82 * Revision 1.2 2002/01/21 00:14:09 jongfoster
83 * Correcting comment style
84 * Fixing an uninitialized memory bug in create_url_spec()
86 * Revision 1.1 2002/01/17 20:53:46 jongfoster
87 * Moving all our URL and URL pattern parsing code to the same file - it
88 * was scattered around in filters.c, loaders.c and parsers.c.
90 * Providing a single, simple url_match(pattern,url) function - rather than
91 * the 3-line match routine which was repeated all over the place.
93 * Renaming free_url to free_url_spec, since it frees a struct url_spec.
95 * Providing parse_http_url() so that URLs can be parsed without faking a
96 * HTTP request line for parse_http_request() or repeating the parsing
97 * code (both of which were techniques that were actually in use).
99 * Standardizing that struct http_request is used to represent a URL, and
100 * struct url_spec is used to represent a URL pattern. (Before, URLs were
101 * represented as seperate variables and a partially-filled-in url_spec).
104 *********************************************************************/
111 #include <sys/types.h>
119 #if !defined(_WIN32) && !defined(__OS2__)
124 #include "urlmatch.h"
126 #include "miscutil.h"
129 const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
132 /*********************************************************************
134 * Function : free_http_request
136 * Description : Freez a http_request structure
139 * 1 : http = points to a http_request structure to free
143 *********************************************************************/
144 void free_http_request(struct http_request *http)
153 freez(http->hostport);
156 freez(http->host_ip_addr_str);
157 freez(http->dbuffer);
163 /*********************************************************************
165 * Function : parse_http_url
167 * Description : Parse out the host and port from the URL. Find the
168 * hostname & path, port (if ':'), and/or password (if '@')
171 * 1 : url = URL (or is it URI?) to break down
172 * 2 : http = pointer to the http structure to hold elements.
173 * Will be zeroed before use. Note that this
174 * function sets the http->gpc and http->ver
176 * 3 : csp = Current client state (buffers, headers, etc...)
178 * Returns : JB_ERR_OK on success
179 * JB_ERR_MEMORY on out of memory
180 * JB_ERR_PARSE on malformed command/URL
181 * or >100 domains deep.
183 *********************************************************************/
184 jb_err parse_http_url(const char * url,
185 struct http_request *http,
186 struct client_state *csp)
189 * Zero out the results structure
191 memset(http, '\0', sizeof(*http));
195 * Save our initial URL
197 http->url = strdup(url);
198 if (http->url == NULL)
200 return JB_ERR_MEMORY;
205 * Check for * URI. If found, we're done.
207 if (*http->url == '*')
209 if ( NULL == (http->path = strdup("*"))
210 || NULL == (http->hostport = strdup("")) )
212 return JB_ERR_MEMORY;
214 if (http->url[1] != '\0')
223 * Split URL into protocol,hostport,path.
233 return JB_ERR_MEMORY;
236 /* Find the start of the URL in our scratch space */
238 if (strncmpic(url_noproto, "http://", 7) == 0)
243 else if (strncmpic(url_noproto, "https://", 8) == 0)
253 url_path = strchr(url_noproto, '/');
254 if (url_path != NULL)
259 * NOTE: The following line ignores the path for HTTPS URLS.
260 * This means that you get consistent behaviour if you type a
261 * https URL in and it's parsed by the function. (When the
262 * URL is actually retrieved, SSL hides the path part).
264 http->path = strdup(http->ssl ? "/" : url_path);
266 http->hostport = strdup(url_noproto);
271 * Repair broken HTTP requests that don't contain a path,
272 * or CONNECT requests
274 http->path = strdup("/");
275 http->hostport = strdup(url_noproto);
280 if ( (http->path == NULL)
281 || (http->hostport == NULL))
283 return JB_ERR_MEMORY;
289 * Split hostport into user/password (ignored), host, port.
296 buf = strdup(http->hostport);
299 return JB_ERR_MEMORY;
302 /* check if url contains username and/or password */
303 host = strchr(buf, '@');
306 /* Contains username/password, skip it and the @ sign. */
311 /* No username or password. */
315 /* check if url contains port */
316 port = strchr(host, ':');
320 /* Terminate hostname and point to start of port string */
322 http->port = atoi(port);
326 /* No port specified. */
327 http->port = (http->ssl ? 443 : 80);
330 http->host = strdup(host);
334 if (http->host == NULL)
336 return JB_ERR_MEMORY;
341 * Split domain name so we can compare it against wildcards
345 char *vec[BUFFER_SIZE];
349 http->dbuffer = strdup(http->host);
350 if (NULL == http->dbuffer)
352 return JB_ERR_MEMORY;
355 /* map to lower case */
356 for (p = http->dbuffer; *p ; p++)
358 *p = tolower((int)(unsigned char)*p);
361 /* split the domain name into components */
362 http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec), 1, 1);
364 if (http->dcount <= 0)
367 * Error: More than SZ(vec) components in domain
368 * or: no components in domain
373 /* save a copy of the pointers in dvec */
374 size = http->dcount * sizeof(*http->dvec);
376 http->dvec = (char **)malloc(size);
377 if (NULL == http->dvec)
379 return JB_ERR_MEMORY;
382 memcpy(http->dvec, vec, size);
390 /*********************************************************************
392 * Function : parse_http_request
394 * Description : Parse out the host and port from the URL. Find the
395 * hostname & path, port (if ':'), and/or password (if '@')
398 * 1 : req = HTTP request line to break down
399 * 2 : http = pointer to the http structure to hold elements
400 * 3 : csp = Current client state (buffers, headers, etc...)
402 * Returns : JB_ERR_OK on success
403 * JB_ERR_MEMORY on out of memory
404 * JB_ERR_CGI_PARAMS on malformed command/URL
405 * or >100 domains deep.
407 *********************************************************************/
408 jb_err parse_http_request(const char *req,
409 struct http_request *http,
410 struct client_state *csp)
418 memset(http, '\0', sizeof(*http));
423 return JB_ERR_MEMORY;
426 n = ssplit(buf, " \r\n", v, SZ(v), 1, 1);
433 /* this could be a CONNECT request */
434 if (strcmpic(v[0], "connect") == 0)
439 /* or it could be any other basic HTTP request type */
440 else if ((0 == strcmpic(v[0], "get"))
441 || (0 == strcmpic(v[0], "head"))
442 || (0 == strcmpic(v[0], "post"))
443 || (0 == strcmpic(v[0], "put"))
444 || (0 == strcmpic(v[0], "delete"))
445 || (0 == strcmpic(v[0], "options"))
446 || (0 == strcmpic(v[0], "trace"))
448 /* or a webDAV extension (RFC2518) */
449 || (0 == strcmpic(v[0], "propfind"))
450 || (0 == strcmpic(v[0], "proppatch"))
451 || (0 == strcmpic(v[0], "move"))
452 || (0 == strcmpic(v[0], "copy"))
453 || (0 == strcmpic(v[0], "mkcol"))
454 || (0 == strcmpic(v[0], "lock"))
455 || (0 == strcmpic(v[0], "unlock"))
457 /* Or a Microsoft webDAV extension for Exchange 2000. See: */
458 /* http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html */
459 /* http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp */
460 || (0 == strcmpic(v[0], "bcopy"))
461 || (0 == strcmpic(v[0], "bmove"))
462 || (0 == strcmpic(v[0], "bdelete"))
463 || (0 == strcmpic(v[0], "bpropfind"))
464 || (0 == strcmpic(v[0], "bproppatch"))
466 /* Or another Microsoft webDAV extension for Exchange 2000. See: */
467 /* http://systems.cs.colorado.edu/grunwald/MobileComputing/Papers/draft-cohen-gena-p-base-00.txt */
468 /* http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html */
469 /* http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp */
470 || (0 == strcmpic(v[0], "subscribe"))
471 || (0 == strcmpic(v[0], "unsubscribe"))
472 || (0 == strcmpic(v[0], "notify"))
473 || (0 == strcmpic(v[0], "poll"))
481 /* Unknown HTTP method */
486 err = parse_http_url(v[1], http, csp);
494 * Copy the details into the structure
496 http->ssl = is_connect;
497 http->cmd = strdup(req);
498 http->gpc = strdup(v[0]);
499 http->ver = strdup(v[2]);
501 if ( (http->cmd == NULL)
502 || (http->gpc == NULL)
503 || (http->ver == NULL) )
506 return JB_ERR_MEMORY;
515 /*********************************************************************
517 * Function : simple_domaincmp
519 * Description : Domain-wise Compare fqdn's. The comparison is
520 * both left- and right-anchored. The individual
521 * domain names are compared with simplematch().
522 * This is only used by domain_match.
525 * 1 : pv = array of patterns to compare
526 * 2 : fv = array of domain components to compare
527 * 3 : len = length of the arrays (both arrays are the
528 * same length - if they weren't, it couldn't
529 * possibly be a match).
531 * Returns : 0 => domains are equivalent, else no match.
533 *********************************************************************/
534 static int simple_domaincmp(char **pv, char **fv, int len)
538 for (n = 0; n < len; n++)
540 if (simplematch(pv[n], fv[n]))
551 /*********************************************************************
553 * Function : domain_match
555 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
556 * pattern->unachored, the comparison is un-, left-,
557 * right-anchored, or both.
558 * The individual domain names are compared with
562 * 1 : pattern = a domain that may contain a '*' as a wildcard.
563 * 2 : fqdn = domain name against which the patterns are compared.
565 * Returns : 0 => domains are equivalent, else no match.
567 *********************************************************************/
568 static int domain_match(const struct url_spec *pattern, const struct http_request *fqdn)
570 char **pv, **fv; /* vectors */
572 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
574 plen = pattern->dcount;
579 /* fqdn is too short to match this pattern */
586 if (unanchored == ANCHOR_LEFT)
591 * Convert this into a fully anchored pattern with
592 * the fqdn and pattern the same length
594 fv += (flen - plen); /* flen - plen >= 0 due to check above */
595 return simple_domaincmp(pv, fv, plen);
597 else if (unanchored == 0)
599 /* Fully anchored, check length */
604 return simple_domaincmp(pv, fv, plen);
606 else if (unanchored == ANCHOR_RIGHT)
608 /* Left anchored, ignore all extra in fqdn */
609 return simple_domaincmp(pv, fv, plen);
615 int maxn = flen - plen;
616 for (n = 0; n <= maxn; n++)
618 if (!simple_domaincmp(pv, fv, plen))
623 * Doesn't match from start of fqdn
624 * Try skipping first part of fqdn
634 /*********************************************************************
636 * Function : create_url_spec
638 * Description : Creates a "url_spec" structure from a string.
639 * When finished, free with free_url_spec().
642 * 1 : url = Target url_spec to be filled in. Will be
644 * 2 : buf = Source pattern, null terminated. NOTE: The
645 * contents of this buffer are destroyed by this
646 * function. If this function succeeds, the
647 * buffer is copied to url->spec. If this
648 * function fails, the contents of the buffer
651 * Returns : JB_ERR_OK - Success
652 * JB_ERR_MEMORY - Out of memory
653 * JB_ERR_PARSE - Cannot parse regex (Detailed message
654 * written to system log)
656 *********************************************************************/
657 jb_err create_url_spec(struct url_spec * url, const char * buf)
667 memset(url, '\0', sizeof(*url));
670 * Save a copy of the orignal specification
672 if ((url->spec = strdup(buf)) == NULL)
674 return JB_ERR_MEMORY;
677 if ((p = strchr(buf, '/')) != NULL)
679 if (NULL == (url->path = strdup(p)))
682 return JB_ERR_MEMORY;
684 url->pathlen = strlen(url->path);
695 char rebuf[BUFFER_SIZE];
697 if (NULL == (url->preg = zalloc(sizeof(*url->preg))))
701 return JB_ERR_MEMORY;
704 sprintf(rebuf, "^(%s)", url->path);
706 errcode = regcomp(url->preg, rebuf,
707 (REG_EXTENDED|REG_NOSUB|REG_ICASE));
710 size_t errlen = regerror(errcode,
711 url->preg, rebuf, sizeof(rebuf));
713 if (errlen > (sizeof(rebuf) - (size_t)1))
715 errlen = sizeof(rebuf) - (size_t)1;
717 rebuf[errlen] = '\0';
719 log_error(LOG_LEVEL_ERROR, "error compiling %s: %s",
730 if ((p = strchr(buf, ':')) == NULL)
748 if (buf[strlen(buf) - 1] == '.')
750 url->unanchored |= ANCHOR_RIGHT;
754 url->unanchored |= ANCHOR_LEFT;
758 * Split domain into components
760 url->dbuffer = strdup(buf);
761 if (NULL == url->dbuffer)
767 return JB_ERR_MEMORY;
773 for (p = url->dbuffer; *p ; p++)
775 *p = tolower((int)(unsigned char)*p);
779 * Split the domain name into components
781 url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1);
791 return JB_ERR_MEMORY;
793 else if (url->dcount != 0)
797 * Save a copy of the pointers in dvec
799 size = url->dcount * sizeof(*url->dvec);
801 url->dvec = (char **)malloc(size);
802 if (NULL == url->dvec)
810 return JB_ERR_MEMORY;
813 memcpy(url->dvec, v, size);
816 * else dcount == 0 in which case we needn't do anything,
817 * since dvec will never be accessed and the pattern will
827 /*********************************************************************
829 * Function : free_url_spec
831 * Description : Called from the "unloaders". Freez the url
832 * structure elements.
835 * 1 : url = pointer to a url_spec structure.
839 *********************************************************************/
840 void free_url_spec(struct url_spec *url)
842 if (url == NULL) return;
856 /*********************************************************************
858 * Function : url_match
860 * Description : Compare a URL against a URL pattern.
863 * 1 : pattern = a URL pattern
864 * 2 : url = URL to match
866 * Returns : 0 iff the URL matches the pattern, else nonzero.
868 *********************************************************************/
869 int url_match(const struct url_spec *pattern,
870 const struct http_request *url)
872 return ((pattern->port == 0) || (pattern->port == url->port))
873 && ((pattern->dbuffer == NULL) || (domain_match(pattern, url) == 0))
874 && ((pattern->path == NULL) ||
875 (regexec(pattern->preg, url->path, 0, NULL, 0) == 0)