1 const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.10.2.4 2003/02/28 12:57:44 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 1.10.2.4 2003/02/28 12:57:44 oes
37 * Moved freeing of http request structure to its owner
38 * as per Dan Price's observations in Bug #694713
40 * Revision 1.10.2.3 2002/11/12 16:50:40 oes
41 * Fixed memory leak in parse_http_request() reported by Oliver Stoeneberg. Fixes bug #637073
43 * Revision 1.10.2.2 2002/09/25 14:53:15 oes
44 * Added basic support for OPTIONS and TRACE HTTP methods:
45 * parse_http_url now recognizes the "*" URI as well as
46 * the OPTIONS and TRACE method keywords.
48 * Revision 1.10.2.1 2002/06/06 19:06:44 jongfoster
49 * Adding support for proprietary Microsoft WebDAV extensions
51 * Revision 1.10 2002/05/12 21:40:37 jongfoster
52 * - Removing some unused code
54 * Revision 1.9 2002/04/04 00:36:36 gliptak
55 * always use pcre for matching
57 * Revision 1.8 2002/04/03 23:32:47 jongfoster
58 * Fixing memory leak on error
60 * Revision 1.7 2002/03/26 22:29:55 swa
61 * we have a new homepage!
63 * Revision 1.6 2002/03/24 13:25:43 swa
64 * name change related issues
66 * Revision 1.5 2002/03/13 00:27:05 jongfoster
69 * Revision 1.4 2002/03/07 03:46:17 oes
70 * Fixed compiler warnings
72 * Revision 1.3 2002/03/03 14:51:11 oes
73 * Fixed CLF logging: Added ocmd member for client's request to struct http_request
75 * Revision 1.2 2002/01/21 00:14:09 jongfoster
76 * Correcting comment style
77 * Fixing an uninitialized memory bug in create_url_spec()
79 * Revision 1.1 2002/01/17 20:53:46 jongfoster
80 * Moving all our URL and URL pattern parsing code to the same file - it
81 * was scattered around in filters.c, loaders.c and parsers.c.
83 * Providing a single, simple url_match(pattern,url) function - rather than
84 * the 3-line match routine which was repeated all over the place.
86 * Renaming free_url to free_url_spec, since it frees a struct url_spec.
88 * Providing parse_http_url() so that URLs can be parsed without faking a
89 * HTTP request line for parse_http_request() or repeating the parsing
90 * code (both of which were techniques that were actually in use).
92 * Standardizing that struct http_request is used to represent a URL, and
93 * struct url_spec is used to represent a URL pattern. (Before, URLs were
94 * represented as seperate variables and a partially-filled-in url_spec).
97 *********************************************************************/
104 #include <sys/types.h>
112 #if !defined(_WIN32) && !defined(__OS2__)
117 #include "urlmatch.h"
119 #include "miscutil.h"
122 const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
125 /*********************************************************************
127 * Function : free_http_request
129 * Description : Freez a http_request structure
132 * 1 : http = points to a http_request structure to free
136 *********************************************************************/
137 void free_http_request(struct http_request *http)
146 freez(http->hostport);
149 freez(http->host_ip_addr_str);
150 freez(http->dbuffer);
156 /*********************************************************************
158 * Function : parse_http_url
160 * Description : Parse out the host and port from the URL. Find the
161 * hostname & path, port (if ':'), and/or password (if '@')
164 * 1 : url = URL (or is it URI?) to break down
165 * 2 : http = pointer to the http structure to hold elements.
166 * Will be zeroed before use. Note that this
167 * function sets the http->gpc and http->ver
169 * 3 : csp = Current client state (buffers, headers, etc...)
171 * Returns : JB_ERR_OK on success
172 * JB_ERR_MEMORY on out of memory
173 * JB_ERR_PARSE on malformed command/URL
174 * or >100 domains deep.
176 *********************************************************************/
177 jb_err parse_http_url(const char * url,
178 struct http_request *http,
179 struct client_state *csp)
182 * Zero out the results structure
184 memset(http, '\0', sizeof(*http));
188 * Save our initial URL
190 http->url = strdup(url);
191 if (http->url == NULL)
193 return JB_ERR_MEMORY;
198 * Check for * URI. If found, we're done.
200 if (*http->url == '*')
202 http->path = strdup("*");
203 http->hostport = strdup("");
209 * Split URL into protocol,hostport,path.
219 return JB_ERR_MEMORY;
222 /* Find the start of the URL in our scratch space */
224 if (strncmpic(url_noproto, "http://", 7) == 0)
229 else if (strncmpic(url_noproto, "https://", 8) == 0)
239 url_path = strchr(url_noproto, '/');
240 if (url_path != NULL)
245 * NOTE: The following line ignores the path for HTTPS URLS.
246 * This means that you get consistent behaviour if you type a
247 * https URL in and it's parsed by the function. (When the
248 * URL is actually retrieved, SSL hides the path part).
250 http->path = strdup(http->ssl ? "/" : url_path);
252 http->hostport = strdup(url_noproto);
257 * Repair broken HTTP requests that don't contain a path,
258 * or CONNECT requests
260 http->path = strdup("/");
261 http->hostport = strdup(url_noproto);
266 if ( (http->path == NULL)
267 || (http->hostport == NULL))
269 return JB_ERR_MEMORY;
275 * Split hostport into user/password (ignored), host, port.
282 buf = strdup(http->hostport);
285 return JB_ERR_MEMORY;
288 /* check if url contains username and/or password */
289 host = strchr(buf, '@');
292 /* Contains username/password, skip it and the @ sign. */
297 /* No username or password. */
301 /* check if url contains port */
302 port = strchr(host, ':');
306 /* Terminate hostname and point to start of port string */
308 http->port = atoi(port);
312 /* No port specified. */
313 http->port = (http->ssl ? 143 : 80);
316 http->host = strdup(host);
320 if (http->host == NULL)
322 return JB_ERR_MEMORY;
327 * Split domain name so we can compare it against wildcards
331 char *vec[BUFFER_SIZE];
335 http->dbuffer = strdup(http->host);
336 if (NULL == http->dbuffer)
338 return JB_ERR_MEMORY;
341 /* map to lower case */
342 for (p = http->dbuffer; *p ; p++)
344 *p = tolower((int)(unsigned char)*p);
347 /* split the domain name into components */
348 http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec), 1, 1);
350 if (http->dcount <= 0)
353 * Error: More than SZ(vec) components in domain
354 * or: no components in domain
359 /* save a copy of the pointers in dvec */
360 size = http->dcount * sizeof(*http->dvec);
362 http->dvec = (char **)malloc(size);
363 if (NULL == http->dvec)
365 return JB_ERR_MEMORY;
368 memcpy(http->dvec, vec, size);
376 /*********************************************************************
378 * Function : parse_http_request
380 * Description : Parse out the host and port from the URL. Find the
381 * hostname & path, port (if ':'), and/or password (if '@')
384 * 1 : req = HTTP request line to break down
385 * 2 : http = pointer to the http structure to hold elements
386 * 3 : csp = Current client state (buffers, headers, etc...)
388 * Returns : JB_ERR_OK on success
389 * JB_ERR_MEMORY on out of memory
390 * JB_ERR_CGI_PARAMS on malformed command/URL
391 * or >100 domains deep.
393 *********************************************************************/
394 jb_err parse_http_request(const char *req,
395 struct http_request *http,
396 struct client_state *csp)
404 memset(http, '\0', sizeof(*http));
409 return JB_ERR_MEMORY;
412 n = ssplit(buf, " \r\n", v, SZ(v), 1, 1);
419 /* this could be a CONNECT request */
420 if (strcmpic(v[0], "connect") == 0)
425 /* or it could be any other basic HTTP request type */
426 else if ((0 == strcmpic(v[0], "get"))
427 || (0 == strcmpic(v[0], "head"))
428 || (0 == strcmpic(v[0], "post"))
429 || (0 == strcmpic(v[0], "put"))
430 || (0 == strcmpic(v[0], "delete"))
431 || (0 == strcmpic(v[0], "options"))
432 || (0 == strcmpic(v[0], "trace"))
434 /* or a webDAV extension (RFC2518) */
435 || (0 == strcmpic(v[0], "propfind"))
436 || (0 == strcmpic(v[0], "proppatch"))
437 || (0 == strcmpic(v[0], "move"))
438 || (0 == strcmpic(v[0], "copy"))
439 || (0 == strcmpic(v[0], "mkcol"))
440 || (0 == strcmpic(v[0], "lock"))
441 || (0 == strcmpic(v[0], "unlock"))
443 /* Or a Microsoft webDAV extension for Exchange 2000. See: */
444 /* http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html */
445 /* http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp */
446 || (0 == strcmpic(v[0], "bcopy"))
447 || (0 == strcmpic(v[0], "bmove"))
448 || (0 == strcmpic(v[0], "bdelete"))
449 || (0 == strcmpic(v[0], "bpropfind"))
450 || (0 == strcmpic(v[0], "bproppatch"))
452 /* Or another Microsoft webDAV extension for Exchange 2000. See: */
453 /* http://systems.cs.colorado.edu/grunwald/MobileComputing/Papers/draft-cohen-gena-p-base-00.txt */
454 /* http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html */
455 /* http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp */
456 || (0 == strcmpic(v[0], "subscribe"))
457 || (0 == strcmpic(v[0], "unsubscribe"))
458 || (0 == strcmpic(v[0], "notify"))
459 || (0 == strcmpic(v[0], "poll"))
467 /* Unknown HTTP method */
472 err = parse_http_url(v[1], http, csp);
480 * Copy the details into the structure
482 http->ssl = is_connect;
483 http->cmd = strdup(req);
484 http->gpc = strdup(v[0]);
485 http->ver = strdup(v[2]);
487 if ( (http->cmd == NULL)
488 || (http->gpc == NULL)
489 || (http->ver == NULL) )
492 return JB_ERR_MEMORY;
501 /*********************************************************************
503 * Function : simple_domaincmp
505 * Description : Domain-wise Compare fqdn's. The comparison is
506 * both left- and right-anchored. The individual
507 * domain names are compared with simplematch().
508 * This is only used by domain_match.
511 * 1 : pv = array of patterns to compare
512 * 2 : fv = array of domain components to compare
513 * 3 : len = length of the arrays (both arrays are the
514 * same length - if they weren't, it couldn't
515 * possibly be a match).
517 * Returns : 0 => domains are equivalent, else no match.
519 *********************************************************************/
520 static int simple_domaincmp(char **pv, char **fv, int len)
524 for (n = 0; n < len; n++)
526 if (simplematch(pv[n], fv[n]))
537 /*********************************************************************
539 * Function : domain_match
541 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
542 * pattern->unachored, the comparison is un-, left-,
543 * right-anchored, or both.
544 * The individual domain names are compared with
548 * 1 : pattern = a domain that may contain a '*' as a wildcard.
549 * 2 : fqdn = domain name against which the patterns are compared.
551 * Returns : 0 => domains are equivalent, else no match.
553 *********************************************************************/
554 static int domain_match(const struct url_spec *pattern, const struct http_request *fqdn)
556 char **pv, **fv; /* vectors */
558 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
560 plen = pattern->dcount;
565 /* fqdn is too short to match this pattern */
572 if (unanchored == ANCHOR_LEFT)
577 * Convert this into a fully anchored pattern with
578 * the fqdn and pattern the same length
580 fv += (flen - plen); /* flen - plen >= 0 due to check above */
581 return simple_domaincmp(pv, fv, plen);
583 else if (unanchored == 0)
585 /* Fully anchored, check length */
590 return simple_domaincmp(pv, fv, plen);
592 else if (unanchored == ANCHOR_RIGHT)
594 /* Left anchored, ignore all extra in fqdn */
595 return simple_domaincmp(pv, fv, plen);
601 int maxn = flen - plen;
602 for (n = 0; n <= maxn; n++)
604 if (!simple_domaincmp(pv, fv, plen))
609 * Doesn't match from start of fqdn
610 * Try skipping first part of fqdn
620 /*********************************************************************
622 * Function : create_url_spec
624 * Description : Creates a "url_spec" structure from a string.
625 * When finished, free with unload_url().
628 * 1 : url = Target url_spec to be filled in. Will be
630 * 2 : buf = Source pattern, null terminated. NOTE: The
631 * contents of this buffer are destroyed by this
632 * function. If this function succeeds, the
633 * buffer is copied to url->spec. If this
634 * function fails, the contents of the buffer
637 * Returns : JB_ERR_OK - Success
638 * JB_ERR_MEMORY - Out of memory
639 * JB_ERR_PARSE - Cannot parse regex (Detailed message
640 * written to system log)
642 *********************************************************************/
643 jb_err create_url_spec(struct url_spec * url, const char * buf)
651 memset(url, '\0', sizeof(*url));
653 /* save a copy of the orignal specification */
654 if ((url->spec = strdup(buf)) == NULL)
656 return JB_ERR_MEMORY;
659 if ((p = strchr(buf, '/')) != NULL)
661 if (NULL == (url->path = strdup(p)))
664 return JB_ERR_MEMORY;
666 url->pathlen = strlen(url->path);
677 char rebuf[BUFFER_SIZE];
679 if (NULL == (url->preg = zalloc(sizeof(*url->preg))))
683 return JB_ERR_MEMORY;
686 sprintf(rebuf, "^(%s)", url->path);
688 errcode = regcomp(url->preg, rebuf,
689 (REG_EXTENDED|REG_NOSUB|REG_ICASE));
692 size_t errlen = regerror(errcode,
693 url->preg, rebuf, sizeof(rebuf));
695 if (errlen > (sizeof(rebuf) - (size_t)1))
697 errlen = sizeof(rebuf) - (size_t)1;
699 rebuf[errlen] = '\0';
701 log_error(LOG_LEVEL_ERROR, "error compiling %s: %s",
712 if ((p = strchr(buf, ':')) == NULL)
727 /* Parse domain part */
728 if (buf[strlen(buf) - 1] == '.')
730 url->unanchored |= ANCHOR_RIGHT;
734 url->unanchored |= ANCHOR_LEFT;
737 /* split domain into components */
739 url->dbuffer = strdup(buf);
740 if (NULL == url->dbuffer)
746 return JB_ERR_MEMORY;
749 /* map to lower case */
750 for (p = url->dbuffer; *p ; p++)
752 *p = tolower((int)(unsigned char)*p);
755 /* split the domain name into components */
756 url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1);
766 return JB_ERR_MEMORY;
768 else if (url->dcount != 0)
771 /* save a copy of the pointers in dvec */
772 size = url->dcount * sizeof(*url->dvec);
774 url->dvec = (char **)malloc(size);
775 if (NULL == url->dvec)
783 return JB_ERR_MEMORY;
786 memcpy(url->dvec, v, size);
795 /*********************************************************************
797 * Function : free_url_spec
799 * Description : Called from the "unloaders". Freez the url
800 * structure elements.
803 * 1 : url = pointer to a url_spec structure.
807 *********************************************************************/
808 void free_url_spec(struct url_spec *url)
810 if (url == NULL) return;
824 /*********************************************************************
826 * Function : url_match
828 * Description : Compare a URL against a URL pattern.
831 * 1 : pattern = a URL pattern
832 * 2 : url = URL to match
834 * Returns : 0 iff the URL matches the pattern, else nonzero.
836 *********************************************************************/
837 int url_match(const struct url_spec *pattern,
838 const struct http_request *url)
840 return ((pattern->port == 0) || (pattern->port == url->port))
841 && ((pattern->dbuffer == NULL) || (domain_match(pattern, url) == 0))
842 && ((pattern->path == NULL) ||
843 (regexec(pattern->preg, url->path, 0, NULL, 0) == 0)