1 const char urlmatch_rcs[] = "$Id: urlmatch.c,v 2.0 2002/06/04 14:34:21 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 2.0 2002/06/04 14:34:21 jongfoster
37 * Moving source files to src/
39 * Revision 1.10 2002/05/12 21:40:37 jongfoster
40 * - Removing some unused code
42 * Revision 1.9 2002/04/04 00:36:36 gliptak
43 * always use pcre for matching
45 * Revision 1.8 2002/04/03 23:32:47 jongfoster
46 * Fixing memory leak on error
48 * Revision 1.7 2002/03/26 22:29:55 swa
49 * we have a new homepage!
51 * Revision 1.6 2002/03/24 13:25:43 swa
52 * name change related issues
54 * Revision 1.5 2002/03/13 00:27:05 jongfoster
57 * Revision 1.4 2002/03/07 03:46:17 oes
58 * Fixed compiler warnings
60 * Revision 1.3 2002/03/03 14:51:11 oes
61 * Fixed CLF logging: Added ocmd member for client's request to struct http_request
63 * Revision 1.2 2002/01/21 00:14:09 jongfoster
64 * Correcting comment style
65 * Fixing an uninitialized memory bug in create_url_spec()
67 * Revision 1.1 2002/01/17 20:53:46 jongfoster
68 * Moving all our URL and URL pattern parsing code to the same file - it
69 * was scattered around in filters.c, loaders.c and parsers.c.
71 * Providing a single, simple url_match(pattern,url) function - rather than
72 * the 3-line match routine which was repeated all over the place.
74 * Renaming free_url to free_url_spec, since it frees a struct url_spec.
76 * Providing parse_http_url() so that URLs can be parsed without faking a
77 * HTTP request line for parse_http_request() or repeating the parsing
78 * code (both of which were techniques that were actually in use).
80 * Standardizing that struct http_request is used to represent a URL, and
81 * struct url_spec is used to represent a URL pattern. (Before, URLs were
82 * represented as seperate variables and a partially-filled-in url_spec).
85 *********************************************************************/
92 #include <sys/types.h>
100 #if !defined(_WIN32) && !defined(__OS2__)
105 #include "urlmatch.h"
107 #include "miscutil.h"
110 const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
113 /*********************************************************************
115 * Function : free_http_request
117 * Description : Freez a http_request structure
120 * 1 : http = points to a http_request structure to free
124 *********************************************************************/
125 void free_http_request(struct http_request *http)
134 freez(http->hostport);
137 freez(http->host_ip_addr_str);
138 freez(http->dbuffer);
144 /*********************************************************************
146 * Function : parse_http_url
148 * Description : Parse out the host and port from the URL. Find the
149 * hostname & path, port (if ':'), and/or password (if '@')
152 * 1 : url = URL (or is it URI?) to break down
153 * 2 : http = pointer to the http structure to hold elements.
154 * Will be zeroed before use. Note that this
155 * function sets the http->gpc and http->ver
157 * 3 : csp = Current client state (buffers, headers, etc...)
159 * Returns : JB_ERR_OK on success
160 * JB_ERR_MEMORY on out of memory
161 * JB_ERR_CGI_PARAMS on malformed command/URL
162 * or >100 domains deep.
164 *********************************************************************/
165 jb_err parse_http_url(const char * url,
166 struct http_request *http,
167 struct client_state *csp)
170 * Zero out the results structure
172 memset(http, '\0', sizeof(*http));
176 * Save our initial URL
178 http->url = strdup(url);
179 if (http->url == NULL)
181 return JB_ERR_MEMORY;
186 * Split URL into protocol,hostport,path.
196 return JB_ERR_MEMORY;
199 /* Find the start of the URL in our scratch space */
201 if (strncmpic(url_noproto, "http://", 7) == 0)
206 else if (strncmpic(url_noproto, "https://", 8) == 0)
216 url_path = strchr(url_noproto, '/');
217 if (url_path != NULL)
222 * NOTE: The following line ignores the path for HTTPS URLS.
223 * This means that you get consistent behaviour if you type a
224 * https URL in and it's parsed by the function. (When the
225 * URL is actually retrieved, SSL hides the path part).
227 http->path = strdup(http->ssl ? "/" : url_path);
229 http->hostport = strdup(url_noproto);
234 * Repair broken HTTP requests that don't contain a path,
235 * or CONNECT requests
237 http->path = strdup("/");
238 http->hostport = strdup(url_noproto);
243 if ( (http->path == NULL)
244 || (http->hostport == NULL))
247 free_http_request(http);
248 return JB_ERR_MEMORY;
254 * Split hostport into user/password (ignored), host, port.
261 buf = strdup(http->hostport);
264 free_http_request(http);
265 return JB_ERR_MEMORY;
268 /* check if url contains username and/or password */
269 host = strchr(buf, '@');
272 /* Contains username/password, skip it and the @ sign. */
277 /* No username or password. */
281 /* check if url contains port */
282 port = strchr(host, ':');
286 /* Terminate hostname and point to start of port string */
288 http->port = atoi(port);
292 /* No port specified. */
293 http->port = (http->ssl ? 143 : 80);
296 http->host = strdup(host);
300 if (http->host == NULL)
302 free_http_request(http);
303 return JB_ERR_MEMORY;
309 * Split domain name so we can compare it against wildcards
312 char *vec[BUFFER_SIZE];
316 http->dbuffer = strdup(http->host);
317 if (NULL == http->dbuffer)
319 free_http_request(http);
320 return JB_ERR_MEMORY;
323 /* map to lower case */
324 for (p = http->dbuffer; *p ; p++)
326 *p = tolower((int)(unsigned char)*p);
329 /* split the domain name into components */
330 http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec), 1, 1);
332 if (http->dcount <= 0)
335 * Error: More than SZ(vec) components in domain
336 * or: no components in domain
338 free_http_request(http);
342 /* save a copy of the pointers in dvec */
343 size = http->dcount * sizeof(*http->dvec);
345 http->dvec = (char **)malloc(size);
346 if (NULL == http->dvec)
348 free_http_request(http);
349 return JB_ERR_MEMORY;
352 memcpy(http->dvec, vec, size);
360 /*********************************************************************
362 * Function : parse_http_request
364 * Description : Parse out the host and port from the URL. Find the
365 * hostname & path, port (if ':'), and/or password (if '@')
368 * 1 : req = HTTP request line to break down
369 * 2 : http = pointer to the http structure to hold elements
370 * 3 : csp = Current client state (buffers, headers, etc...)
372 * Returns : JB_ERR_OK on success
373 * JB_ERR_MEMORY on out of memory
374 * JB_ERR_CGI_PARAMS on malformed command/URL
375 * or >100 domains deep.
377 *********************************************************************/
378 jb_err parse_http_request(const char *req,
379 struct http_request *http,
380 struct client_state *csp)
388 memset(http, '\0', sizeof(*http));
393 return JB_ERR_MEMORY;
396 n = ssplit(buf, " \r\n", v, SZ(v), 1, 1);
403 /* this could be a CONNECT request */
404 if (strcmpic(v[0], "connect") == 0)
409 /* or it could be any other basic HTTP request type */
410 else if ((0 == strcmpic(v[0], "get"))
411 || (0 == strcmpic(v[0], "head"))
412 || (0 == strcmpic(v[0], "post"))
413 || (0 == strcmpic(v[0], "put"))
414 || (0 == strcmpic(v[0], "delete"))
416 /* or a webDAV extension (RFC2518) */
417 || (0 == strcmpic(v[0], "propfind"))
418 || (0 == strcmpic(v[0], "proppatch"))
419 || (0 == strcmpic(v[0], "move"))
420 || (0 == strcmpic(v[0], "copy"))
421 || (0 == strcmpic(v[0], "mkcol"))
422 || (0 == strcmpic(v[0], "lock"))
423 || (0 == strcmpic(v[0], "unlock"))
425 /* Or a Microsoft webDAV extension for Exchange 2000. See: */
426 /* http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html */
427 /* http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp */
428 || (0 == strcmpic(v[0], "bcopy"))
429 || (0 == strcmpic(v[0], "bmove"))
430 || (0 == strcmpic(v[0], "bdelete"))
431 || (0 == strcmpic(v[0], "bpropfind"))
432 || (0 == strcmpic(v[0], "bproppatch"))
434 /* Or another Microsoft webDAV extension for Exchange 2000. See: */
435 /* http://systems.cs.colorado.edu/grunwald/MobileComputing/Papers/draft-cohen-gena-p-base-00.txt */
436 /* http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html */
437 /* http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp */
438 || (0 == strcmpic(v[0], "subscribe"))
439 || (0 == strcmpic(v[0], "unsubscribe"))
440 || (0 == strcmpic(v[0], "notify"))
441 || (0 == strcmpic(v[0], "poll"))
449 /* Unknown HTTP method */
450 /* Maybe one of: trace options purge */
455 err = parse_http_url(v[1], http, csp);
463 * Copy the details into the structure
465 http->ssl = is_connect;
466 http->cmd = strdup(req);
467 http->gpc = strdup(v[0]);
468 http->ver = strdup(v[2]);
470 if ( (http->cmd == NULL)
471 || (http->gpc == NULL)
472 || (http->ver == NULL) )
475 free_http_request(http);
476 return JB_ERR_MEMORY;
483 /*********************************************************************
485 * Function : simple_domaincmp
487 * Description : Domain-wise Compare fqdn's. The comparison is
488 * both left- and right-anchored. The individual
489 * domain names are compared with simplematch().
490 * This is only used by domain_match.
493 * 1 : pv = array of patterns to compare
494 * 2 : fv = array of domain components to compare
495 * 3 : len = length of the arrays (both arrays are the
496 * same length - if they weren't, it couldn't
497 * possibly be a match).
499 * Returns : 0 => domains are equivalent, else no match.
501 *********************************************************************/
502 static int simple_domaincmp(char **pv, char **fv, int len)
506 for (n = 0; n < len; n++)
508 if (simplematch(pv[n], fv[n]))
519 /*********************************************************************
521 * Function : domain_match
523 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
524 * pattern->unachored, the comparison is un-, left-,
525 * right-anchored, or both.
526 * The individual domain names are compared with
530 * 1 : pattern = a domain that may contain a '*' as a wildcard.
531 * 2 : fqdn = domain name against which the patterns are compared.
533 * Returns : 0 => domains are equivalent, else no match.
535 *********************************************************************/
536 static int domain_match(const struct url_spec *pattern, const struct http_request *fqdn)
538 char **pv, **fv; /* vectors */
540 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
542 plen = pattern->dcount;
547 /* fqdn is too short to match this pattern */
554 if (unanchored == ANCHOR_LEFT)
559 * Convert this into a fully anchored pattern with
560 * the fqdn and pattern the same length
562 fv += (flen - plen); /* flen - plen >= 0 due to check above */
563 return simple_domaincmp(pv, fv, plen);
565 else if (unanchored == 0)
567 /* Fully anchored, check length */
572 return simple_domaincmp(pv, fv, plen);
574 else if (unanchored == ANCHOR_RIGHT)
576 /* Left anchored, ignore all extra in fqdn */
577 return simple_domaincmp(pv, fv, plen);
583 int maxn = flen - plen;
584 for (n = 0; n <= maxn; n++)
586 if (!simple_domaincmp(pv, fv, plen))
591 * Doesn't match from start of fqdn
592 * Try skipping first part of fqdn
602 /*********************************************************************
604 * Function : create_url_spec
606 * Description : Creates a "url_spec" structure from a string.
607 * When finished, free with unload_url().
610 * 1 : url = Target url_spec to be filled in. Will be
612 * 2 : buf = Source pattern, null terminated. NOTE: The
613 * contents of this buffer are destroyed by this
614 * function. If this function succeeds, the
615 * buffer is copied to url->spec. If this
616 * function fails, the contents of the buffer
619 * Returns : JB_ERR_OK - Success
620 * JB_ERR_MEMORY - Out of memory
621 * JB_ERR_PARSE - Cannot parse regex (Detailed message
622 * written to system log)
624 *********************************************************************/
625 jb_err create_url_spec(struct url_spec * url, const char * buf)
633 memset(url, '\0', sizeof(*url));
635 /* save a copy of the orignal specification */
636 if ((url->spec = strdup(buf)) == NULL)
638 return JB_ERR_MEMORY;
641 if ((p = strchr(buf, '/')) != NULL)
643 if (NULL == (url->path = strdup(p)))
646 return JB_ERR_MEMORY;
648 url->pathlen = strlen(url->path);
659 char rebuf[BUFFER_SIZE];
661 if (NULL == (url->preg = zalloc(sizeof(*url->preg))))
665 return JB_ERR_MEMORY;
668 sprintf(rebuf, "^(%s)", url->path);
670 errcode = regcomp(url->preg, rebuf,
671 (REG_EXTENDED|REG_NOSUB|REG_ICASE));
674 size_t errlen = regerror(errcode,
675 url->preg, rebuf, sizeof(rebuf));
677 if (errlen > (sizeof(rebuf) - (size_t)1))
679 errlen = sizeof(rebuf) - (size_t)1;
681 rebuf[errlen] = '\0';
683 log_error(LOG_LEVEL_ERROR, "error compiling %s: %s",
694 if ((p = strchr(buf, ':')) == NULL)
709 /* Parse domain part */
710 if (buf[strlen(buf) - 1] == '.')
712 url->unanchored |= ANCHOR_RIGHT;
716 url->unanchored |= ANCHOR_LEFT;
719 /* split domain into components */
721 url->dbuffer = strdup(buf);
722 if (NULL == url->dbuffer)
728 return JB_ERR_MEMORY;
731 /* map to lower case */
732 for (p = url->dbuffer; *p ; p++)
734 *p = tolower((int)(unsigned char)*p);
737 /* split the domain name into components */
738 url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1);
748 return JB_ERR_MEMORY;
750 else if (url->dcount != 0)
753 /* save a copy of the pointers in dvec */
754 size = url->dcount * sizeof(*url->dvec);
756 url->dvec = (char **)malloc(size);
757 if (NULL == url->dvec)
765 return JB_ERR_MEMORY;
768 memcpy(url->dvec, v, size);
777 /*********************************************************************
779 * Function : free_url_spec
781 * Description : Called from the "unloaders". Freez the url
782 * structure elements.
785 * 1 : url = pointer to a url_spec structure.
789 *********************************************************************/
790 void free_url_spec(struct url_spec *url)
792 if (url == NULL) return;
806 /*********************************************************************
808 * Function : url_match
810 * Description : Compare a URL against a URL pattern.
813 * 1 : pattern = a URL pattern
814 * 2 : url = URL to match
816 * Returns : 0 iff the URL matches the pattern, else nonzero.
818 *********************************************************************/
819 int url_match(const struct url_spec *pattern,
820 const struct http_request *url)
822 return ((pattern->port == 0) || (pattern->port == url->port))
823 && ((pattern->dbuffer == NULL) || (domain_match(pattern, url) == 0))
824 && ((pattern->path == NULL) ||
825 (regexec(pattern->preg, url->path, 0, NULL, 0) == 0)