1 const char urlmatch_rcs[] = "$Id: urlmatch.c,v 2.1 2002/06/06 19:03:29 jongfoster Exp $";
2 /*********************************************************************
4 * File : $Source: /cvsroot/ijbswa//current/src/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 2.1 2002/06/06 19:03:29 jongfoster
37 * Adding support for proprietary Microsoft WebDAV extensions
39 * Revision 2.0 2002/06/04 14:34:21 jongfoster
40 * Moving source files to src/
42 * Revision 1.10 2002/05/12 21:40:37 jongfoster
43 * - Removing some unused code
45 * Revision 1.9 2002/04/04 00:36:36 gliptak
46 * always use pcre for matching
48 * Revision 1.8 2002/04/03 23:32:47 jongfoster
49 * Fixing memory leak on error
51 * Revision 1.7 2002/03/26 22:29:55 swa
52 * we have a new homepage!
54 * Revision 1.6 2002/03/24 13:25:43 swa
55 * name change related issues
57 * Revision 1.5 2002/03/13 00:27:05 jongfoster
60 * Revision 1.4 2002/03/07 03:46:17 oes
61 * Fixed compiler warnings
63 * Revision 1.3 2002/03/03 14:51:11 oes
64 * Fixed CLF logging: Added ocmd member for client's request to struct http_request
66 * Revision 1.2 2002/01/21 00:14:09 jongfoster
67 * Correcting comment style
68 * Fixing an uninitialized memory bug in create_url_spec()
70 * Revision 1.1 2002/01/17 20:53:46 jongfoster
71 * Moving all our URL and URL pattern parsing code to the same file - it
72 * was scattered around in filters.c, loaders.c and parsers.c.
74 * Providing a single, simple url_match(pattern,url) function - rather than
75 * the 3-line match routine which was repeated all over the place.
77 * Renaming free_url to free_url_spec, since it frees a struct url_spec.
79 * Providing parse_http_url() so that URLs can be parsed without faking a
80 * HTTP request line for parse_http_request() or repeating the parsing
81 * code (both of which were techniques that were actually in use).
83 * Standardizing that struct http_request is used to represent a URL, and
84 * struct url_spec is used to represent a URL pattern. (Before, URLs were
85 * represented as seperate variables and a partially-filled-in url_spec).
88 *********************************************************************/
95 #include <sys/types.h>
103 #if !defined(_WIN32) && !defined(__OS2__)
108 #include "urlmatch.h"
110 #include "miscutil.h"
113 const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
116 /*********************************************************************
118 * Function : free_http_request
120 * Description : Freez a http_request structure
123 * 1 : http = points to a http_request structure to free
127 *********************************************************************/
128 void free_http_request(struct http_request *http)
137 freez(http->hostport);
140 freez(http->host_ip_addr_str);
141 freez(http->dbuffer);
147 /*********************************************************************
149 * Function : parse_http_url
151 * Description : Parse out the host and port from the URL. Find the
152 * hostname & path, port (if ':'), and/or password (if '@')
155 * 1 : url = URL (or is it URI?) to break down
156 * 2 : http = pointer to the http structure to hold elements.
157 * Will be zeroed before use. Note that this
158 * function sets the http->gpc and http->ver
160 * 3 : csp = Current client state (buffers, headers, etc...)
162 * Returns : JB_ERR_OK on success
163 * JB_ERR_MEMORY on out of memory
164 * JB_ERR_CGI_PARAMS on malformed command/URL
165 * or >100 domains deep.
167 *********************************************************************/
168 jb_err parse_http_url(const char * url,
169 struct http_request *http,
170 struct client_state *csp)
173 * Zero out the results structure
175 memset(http, '\0', sizeof(*http));
179 * Save our initial URL
181 http->url = strdup(url);
182 if (http->url == NULL)
184 return JB_ERR_MEMORY;
189 * Split URL into protocol,hostport,path.
199 return JB_ERR_MEMORY;
202 /* Find the start of the URL in our scratch space */
204 if (strncmpic(url_noproto, "http://", 7) == 0)
209 else if (strncmpic(url_noproto, "https://", 8) == 0)
219 url_path = strchr(url_noproto, '/');
220 if (url_path != NULL)
225 * NOTE: The following line ignores the path for HTTPS URLS.
226 * This means that you get consistent behaviour if you type a
227 * https URL in and it's parsed by the function. (When the
228 * URL is actually retrieved, SSL hides the path part).
230 http->path = strdup(http->ssl ? "/" : url_path);
232 http->hostport = strdup(url_noproto);
237 * Repair broken HTTP requests that don't contain a path,
238 * or CONNECT requests
240 http->path = strdup("/");
241 http->hostport = strdup(url_noproto);
246 if ( (http->path == NULL)
247 || (http->hostport == NULL))
250 free_http_request(http);
251 return JB_ERR_MEMORY;
257 * Split hostport into user/password (ignored), host, port.
264 buf = strdup(http->hostport);
267 free_http_request(http);
268 return JB_ERR_MEMORY;
271 /* check if url contains username and/or password */
272 host = strchr(buf, '@');
275 /* Contains username/password, skip it and the @ sign. */
280 /* No username or password. */
284 /* check if url contains port */
285 port = strchr(host, ':');
289 /* Terminate hostname and point to start of port string */
291 http->port = atoi(port);
295 /* No port specified. */
296 http->port = (http->ssl ? 143 : 80);
299 http->host = strdup(host);
303 if (http->host == NULL)
305 free_http_request(http);
306 return JB_ERR_MEMORY;
312 * Split domain name so we can compare it against wildcards
315 char *vec[BUFFER_SIZE];
319 http->dbuffer = strdup(http->host);
320 if (NULL == http->dbuffer)
322 free_http_request(http);
323 return JB_ERR_MEMORY;
326 /* map to lower case */
327 for (p = http->dbuffer; *p ; p++)
329 *p = tolower((int)(unsigned char)*p);
332 /* split the domain name into components */
333 http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec), 1, 1);
335 if (http->dcount <= 0)
338 * Error: More than SZ(vec) components in domain
339 * or: no components in domain
341 free_http_request(http);
345 /* save a copy of the pointers in dvec */
346 size = http->dcount * sizeof(*http->dvec);
348 http->dvec = (char **)malloc(size);
349 if (NULL == http->dvec)
351 free_http_request(http);
352 return JB_ERR_MEMORY;
355 memcpy(http->dvec, vec, size);
363 /*********************************************************************
365 * Function : parse_http_request
367 * Description : Parse out the host and port from the URL. Find the
368 * hostname & path, port (if ':'), and/or password (if '@')
371 * 1 : req = HTTP request line to break down
372 * 2 : http = pointer to the http structure to hold elements
373 * 3 : csp = Current client state (buffers, headers, etc...)
375 * Returns : JB_ERR_OK on success
376 * JB_ERR_MEMORY on out of memory
377 * JB_ERR_CGI_PARAMS on malformed command/URL
378 * or >100 domains deep.
380 *********************************************************************/
381 jb_err parse_http_request(const char *req,
382 struct http_request *http,
383 struct client_state *csp)
391 memset(http, '\0', sizeof(*http));
396 return JB_ERR_MEMORY;
399 n = ssplit(buf, " \r\n", v, SZ(v), 1, 1);
406 /* this could be a CONNECT request */
407 if (strcmpic(v[0], "connect") == 0)
412 /* or it could be any other basic HTTP request type */
413 else if ((0 == strcmpic(v[0], "get"))
414 || (0 == strcmpic(v[0], "head"))
415 || (0 == strcmpic(v[0], "post"))
416 || (0 == strcmpic(v[0], "put"))
417 || (0 == strcmpic(v[0], "delete"))
419 /* or a webDAV extension (RFC2518) */
420 || (0 == strcmpic(v[0], "propfind"))
421 || (0 == strcmpic(v[0], "proppatch"))
422 || (0 == strcmpic(v[0], "move"))
423 || (0 == strcmpic(v[0], "copy"))
424 || (0 == strcmpic(v[0], "mkcol"))
425 || (0 == strcmpic(v[0], "lock"))
426 || (0 == strcmpic(v[0], "unlock"))
428 /* Or a Microsoft webDAV extension for Exchange 2000. See: */
429 /* http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html */
430 /* http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp */
431 || (0 == strcmpic(v[0], "bcopy"))
432 || (0 == strcmpic(v[0], "bmove"))
433 || (0 == strcmpic(v[0], "bdelete"))
434 || (0 == strcmpic(v[0], "bpropfind"))
435 || (0 == strcmpic(v[0], "bproppatch"))
437 /* Or another Microsoft webDAV extension for Exchange 2000. See: */
438 /* http://systems.cs.colorado.edu/grunwald/MobileComputing/Papers/draft-cohen-gena-p-base-00.txt */
439 /* http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html */
440 /* http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp */
441 || (0 == strcmpic(v[0], "subscribe"))
442 || (0 == strcmpic(v[0], "unsubscribe"))
443 || (0 == strcmpic(v[0], "notify"))
444 || (0 == strcmpic(v[0], "poll"))
452 /* Unknown HTTP method */
453 /* Maybe one of: trace options purge */
458 err = parse_http_url(v[1], http, csp);
466 * Copy the details into the structure
468 http->ssl = is_connect;
469 http->cmd = strdup(req);
470 http->gpc = strdup(v[0]);
471 http->ver = strdup(v[2]);
473 if ( (http->cmd == NULL)
474 || (http->gpc == NULL)
475 || (http->ver == NULL) )
478 free_http_request(http);
479 return JB_ERR_MEMORY;
487 /*********************************************************************
489 * Function : simple_domaincmp
491 * Description : Domain-wise Compare fqdn's. The comparison is
492 * both left- and right-anchored. The individual
493 * domain names are compared with simplematch().
494 * This is only used by domain_match.
497 * 1 : pv = array of patterns to compare
498 * 2 : fv = array of domain components to compare
499 * 3 : len = length of the arrays (both arrays are the
500 * same length - if they weren't, it couldn't
501 * possibly be a match).
503 * Returns : 0 => domains are equivalent, else no match.
505 *********************************************************************/
506 static int simple_domaincmp(char **pv, char **fv, int len)
510 for (n = 0; n < len; n++)
512 if (simplematch(pv[n], fv[n]))
523 /*********************************************************************
525 * Function : domain_match
527 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
528 * pattern->unachored, the comparison is un-, left-,
529 * right-anchored, or both.
530 * The individual domain names are compared with
534 * 1 : pattern = a domain that may contain a '*' as a wildcard.
535 * 2 : fqdn = domain name against which the patterns are compared.
537 * Returns : 0 => domains are equivalent, else no match.
539 *********************************************************************/
540 static int domain_match(const struct url_spec *pattern, const struct http_request *fqdn)
542 char **pv, **fv; /* vectors */
544 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
546 plen = pattern->dcount;
551 /* fqdn is too short to match this pattern */
558 if (unanchored == ANCHOR_LEFT)
563 * Convert this into a fully anchored pattern with
564 * the fqdn and pattern the same length
566 fv += (flen - plen); /* flen - plen >= 0 due to check above */
567 return simple_domaincmp(pv, fv, plen);
569 else if (unanchored == 0)
571 /* Fully anchored, check length */
576 return simple_domaincmp(pv, fv, plen);
578 else if (unanchored == ANCHOR_RIGHT)
580 /* Left anchored, ignore all extra in fqdn */
581 return simple_domaincmp(pv, fv, plen);
587 int maxn = flen - plen;
588 for (n = 0; n <= maxn; n++)
590 if (!simple_domaincmp(pv, fv, plen))
595 * Doesn't match from start of fqdn
596 * Try skipping first part of fqdn
606 /*********************************************************************
608 * Function : create_url_spec
610 * Description : Creates a "url_spec" structure from a string.
611 * When finished, free with unload_url().
614 * 1 : url = Target url_spec to be filled in. Will be
616 * 2 : buf = Source pattern, null terminated. NOTE: The
617 * contents of this buffer are destroyed by this
618 * function. If this function succeeds, the
619 * buffer is copied to url->spec. If this
620 * function fails, the contents of the buffer
623 * Returns : JB_ERR_OK - Success
624 * JB_ERR_MEMORY - Out of memory
625 * JB_ERR_PARSE - Cannot parse regex (Detailed message
626 * written to system log)
628 *********************************************************************/
629 jb_err create_url_spec(struct url_spec * url, const char * buf)
637 memset(url, '\0', sizeof(*url));
639 /* save a copy of the orignal specification */
640 if ((url->spec = strdup(buf)) == NULL)
642 return JB_ERR_MEMORY;
645 if ((p = strchr(buf, '/')) != NULL)
647 if (NULL == (url->path = strdup(p)))
650 return JB_ERR_MEMORY;
652 url->pathlen = strlen(url->path);
663 char rebuf[BUFFER_SIZE];
665 if (NULL == (url->preg = zalloc(sizeof(*url->preg))))
669 return JB_ERR_MEMORY;
672 sprintf(rebuf, "^(%s)", url->path);
674 errcode = regcomp(url->preg, rebuf,
675 (REG_EXTENDED|REG_NOSUB|REG_ICASE));
678 size_t errlen = regerror(errcode,
679 url->preg, rebuf, sizeof(rebuf));
681 if (errlen > (sizeof(rebuf) - (size_t)1))
683 errlen = sizeof(rebuf) - (size_t)1;
685 rebuf[errlen] = '\0';
687 log_error(LOG_LEVEL_ERROR, "error compiling %s: %s",
698 if ((p = strchr(buf, ':')) == NULL)
713 /* Parse domain part */
714 if (buf[strlen(buf) - 1] == '.')
716 url->unanchored |= ANCHOR_RIGHT;
720 url->unanchored |= ANCHOR_LEFT;
723 /* split domain into components */
725 url->dbuffer = strdup(buf);
726 if (NULL == url->dbuffer)
732 return JB_ERR_MEMORY;
735 /* map to lower case */
736 for (p = url->dbuffer; *p ; p++)
738 *p = tolower((int)(unsigned char)*p);
741 /* split the domain name into components */
742 url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1);
752 return JB_ERR_MEMORY;
754 else if (url->dcount != 0)
757 /* save a copy of the pointers in dvec */
758 size = url->dcount * sizeof(*url->dvec);
760 url->dvec = (char **)malloc(size);
761 if (NULL == url->dvec)
769 return JB_ERR_MEMORY;
772 memcpy(url->dvec, v, size);
781 /*********************************************************************
783 * Function : free_url_spec
785 * Description : Called from the "unloaders". Freez the url
786 * structure elements.
789 * 1 : url = pointer to a url_spec structure.
793 *********************************************************************/
794 void free_url_spec(struct url_spec *url)
796 if (url == NULL) return;
810 /*********************************************************************
812 * Function : url_match
814 * Description : Compare a URL against a URL pattern.
817 * 1 : pattern = a URL pattern
818 * 2 : url = URL to match
820 * Returns : 0 iff the URL matches the pattern, else nonzero.
822 *********************************************************************/
823 int url_match(const struct url_spec *pattern,
824 const struct http_request *url)
826 return ((pattern->port == 0) || (pattern->port == url->port))
827 && ((pattern->dbuffer == NULL) || (domain_match(pattern, url) == 0))
828 && ((pattern->path == NULL) ||
829 (regexec(pattern->preg, url->path, 0, NULL, 0) == 0)