1 const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.10.2.2 2002/09/25 14:53:15 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 1.10.2.2 2002/09/25 14:53:15 oes
37 * Added basic support for OPTIONS and TRACE HTTP methods:
38 * parse_http_url now recognizes the "*" URI as well as
39 * the OPTIONS and TRACE method keywords.
41 * Revision 1.10.2.1 2002/06/06 19:06:44 jongfoster
42 * Adding support for proprietary Microsoft WebDAV extensions
44 * Revision 1.10 2002/05/12 21:40:37 jongfoster
45 * - Removing some unused code
47 * Revision 1.9 2002/04/04 00:36:36 gliptak
48 * always use pcre for matching
50 * Revision 1.8 2002/04/03 23:32:47 jongfoster
51 * Fixing memory leak on error
53 * Revision 1.7 2002/03/26 22:29:55 swa
54 * we have a new homepage!
56 * Revision 1.6 2002/03/24 13:25:43 swa
57 * name change related issues
59 * Revision 1.5 2002/03/13 00:27:05 jongfoster
62 * Revision 1.4 2002/03/07 03:46:17 oes
63 * Fixed compiler warnings
65 * Revision 1.3 2002/03/03 14:51:11 oes
66 * Fixed CLF logging: Added ocmd member for client's request to struct http_request
68 * Revision 1.2 2002/01/21 00:14:09 jongfoster
69 * Correcting comment style
70 * Fixing an uninitialized memory bug in create_url_spec()
72 * Revision 1.1 2002/01/17 20:53:46 jongfoster
73 * Moving all our URL and URL pattern parsing code to the same file - it
74 * was scattered around in filters.c, loaders.c and parsers.c.
76 * Providing a single, simple url_match(pattern,url) function - rather than
77 * the 3-line match routine which was repeated all over the place.
79 * Renaming free_url to free_url_spec, since it frees a struct url_spec.
81 * Providing parse_http_url() so that URLs can be parsed without faking a
82 * HTTP request line for parse_http_request() or repeating the parsing
83 * code (both of which were techniques that were actually in use).
85 * Standardizing that struct http_request is used to represent a URL, and
86 * struct url_spec is used to represent a URL pattern. (Before, URLs were
87 * represented as seperate variables and a partially-filled-in url_spec).
90 *********************************************************************/
97 #include <sys/types.h>
105 #if !defined(_WIN32) && !defined(__OS2__)
110 #include "urlmatch.h"
112 #include "miscutil.h"
115 const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
118 /*********************************************************************
120 * Function : free_http_request
122 * Description : Freez a http_request structure
125 * 1 : http = points to a http_request structure to free
129 *********************************************************************/
130 void free_http_request(struct http_request *http)
139 freez(http->hostport);
142 freez(http->host_ip_addr_str);
143 freez(http->dbuffer);
149 /*********************************************************************
151 * Function : parse_http_url
153 * Description : Parse out the host and port from the URL. Find the
154 * hostname & path, port (if ':'), and/or password (if '@')
157 * 1 : url = URL (or is it URI?) to break down
158 * 2 : http = pointer to the http structure to hold elements.
159 * Will be zeroed before use. Note that this
160 * function sets the http->gpc and http->ver
162 * 3 : csp = Current client state (buffers, headers, etc...)
164 * Returns : JB_ERR_OK on success
165 * JB_ERR_MEMORY on out of memory
166 * JB_ERR_PARSE on malformed command/URL
167 * or >100 domains deep.
169 *********************************************************************/
170 jb_err parse_http_url(const char * url,
171 struct http_request *http,
172 struct client_state *csp)
175 * Zero out the results structure
177 memset(http, '\0', sizeof(*http));
181 * Save our initial URL
183 http->url = strdup(url);
184 if (http->url == NULL)
186 return JB_ERR_MEMORY;
191 * Check for * URI. If found, we're done.
193 if (*http->url == '*')
195 http->path = strdup("*");
196 http->hostport = strdup("");
202 * Split URL into protocol,hostport,path.
212 return JB_ERR_MEMORY;
215 /* Find the start of the URL in our scratch space */
217 if (strncmpic(url_noproto, "http://", 7) == 0)
222 else if (strncmpic(url_noproto, "https://", 8) == 0)
232 url_path = strchr(url_noproto, '/');
233 if (url_path != NULL)
238 * NOTE: The following line ignores the path for HTTPS URLS.
239 * This means that you get consistent behaviour if you type a
240 * https URL in and it's parsed by the function. (When the
241 * URL is actually retrieved, SSL hides the path part).
243 http->path = strdup(http->ssl ? "/" : url_path);
245 http->hostport = strdup(url_noproto);
250 * Repair broken HTTP requests that don't contain a path,
251 * or CONNECT requests
253 http->path = strdup("/");
254 http->hostport = strdup(url_noproto);
259 if ( (http->path == NULL)
260 || (http->hostport == NULL))
263 free_http_request(http);
264 return JB_ERR_MEMORY;
270 * Split hostport into user/password (ignored), host, port.
277 buf = strdup(http->hostport);
280 free_http_request(http);
281 return JB_ERR_MEMORY;
284 /* check if url contains username and/or password */
285 host = strchr(buf, '@');
288 /* Contains username/password, skip it and the @ sign. */
293 /* No username or password. */
297 /* check if url contains port */
298 port = strchr(host, ':');
302 /* Terminate hostname and point to start of port string */
304 http->port = atoi(port);
308 /* No port specified. */
309 http->port = (http->ssl ? 143 : 80);
312 http->host = strdup(host);
316 if (http->host == NULL)
318 free_http_request(http);
319 return JB_ERR_MEMORY;
324 * Split domain name so we can compare it against wildcards
328 char *vec[BUFFER_SIZE];
332 http->dbuffer = strdup(http->host);
333 if (NULL == http->dbuffer)
335 free_http_request(http);
336 return JB_ERR_MEMORY;
339 /* map to lower case */
340 for (p = http->dbuffer; *p ; p++)
342 *p = tolower((int)(unsigned char)*p);
345 /* split the domain name into components */
346 http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec), 1, 1);
348 if (http->dcount <= 0)
351 * Error: More than SZ(vec) components in domain
352 * or: no components in domain
354 free_http_request(http);
358 /* save a copy of the pointers in dvec */
359 size = http->dcount * sizeof(*http->dvec);
361 http->dvec = (char **)malloc(size);
362 if (NULL == http->dvec)
364 free_http_request(http);
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 free_http_request(http);
493 return JB_ERR_MEMORY;
502 /*********************************************************************
504 * Function : simple_domaincmp
506 * Description : Domain-wise Compare fqdn's. The comparison is
507 * both left- and right-anchored. The individual
508 * domain names are compared with simplematch().
509 * This is only used by domain_match.
512 * 1 : pv = array of patterns to compare
513 * 2 : fv = array of domain components to compare
514 * 3 : len = length of the arrays (both arrays are the
515 * same length - if they weren't, it couldn't
516 * possibly be a match).
518 * Returns : 0 => domains are equivalent, else no match.
520 *********************************************************************/
521 static int simple_domaincmp(char **pv, char **fv, int len)
525 for (n = 0; n < len; n++)
527 if (simplematch(pv[n], fv[n]))
538 /*********************************************************************
540 * Function : domain_match
542 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
543 * pattern->unachored, the comparison is un-, left-,
544 * right-anchored, or both.
545 * The individual domain names are compared with
549 * 1 : pattern = a domain that may contain a '*' as a wildcard.
550 * 2 : fqdn = domain name against which the patterns are compared.
552 * Returns : 0 => domains are equivalent, else no match.
554 *********************************************************************/
555 static int domain_match(const struct url_spec *pattern, const struct http_request *fqdn)
557 char **pv, **fv; /* vectors */
559 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
561 plen = pattern->dcount;
566 /* fqdn is too short to match this pattern */
573 if (unanchored == ANCHOR_LEFT)
578 * Convert this into a fully anchored pattern with
579 * the fqdn and pattern the same length
581 fv += (flen - plen); /* flen - plen >= 0 due to check above */
582 return simple_domaincmp(pv, fv, plen);
584 else if (unanchored == 0)
586 /* Fully anchored, check length */
591 return simple_domaincmp(pv, fv, plen);
593 else if (unanchored == ANCHOR_RIGHT)
595 /* Left anchored, ignore all extra in fqdn */
596 return simple_domaincmp(pv, fv, plen);
602 int maxn = flen - plen;
603 for (n = 0; n <= maxn; n++)
605 if (!simple_domaincmp(pv, fv, plen))
610 * Doesn't match from start of fqdn
611 * Try skipping first part of fqdn
621 /*********************************************************************
623 * Function : create_url_spec
625 * Description : Creates a "url_spec" structure from a string.
626 * When finished, free with unload_url().
629 * 1 : url = Target url_spec to be filled in. Will be
631 * 2 : buf = Source pattern, null terminated. NOTE: The
632 * contents of this buffer are destroyed by this
633 * function. If this function succeeds, the
634 * buffer is copied to url->spec. If this
635 * function fails, the contents of the buffer
638 * Returns : JB_ERR_OK - Success
639 * JB_ERR_MEMORY - Out of memory
640 * JB_ERR_PARSE - Cannot parse regex (Detailed message
641 * written to system log)
643 *********************************************************************/
644 jb_err create_url_spec(struct url_spec * url, const char * buf)
652 memset(url, '\0', sizeof(*url));
654 /* save a copy of the orignal specification */
655 if ((url->spec = strdup(buf)) == NULL)
657 return JB_ERR_MEMORY;
660 if ((p = strchr(buf, '/')) != NULL)
662 if (NULL == (url->path = strdup(p)))
665 return JB_ERR_MEMORY;
667 url->pathlen = strlen(url->path);
678 char rebuf[BUFFER_SIZE];
680 if (NULL == (url->preg = zalloc(sizeof(*url->preg))))
684 return JB_ERR_MEMORY;
687 sprintf(rebuf, "^(%s)", url->path);
689 errcode = regcomp(url->preg, rebuf,
690 (REG_EXTENDED|REG_NOSUB|REG_ICASE));
693 size_t errlen = regerror(errcode,
694 url->preg, rebuf, sizeof(rebuf));
696 if (errlen > (sizeof(rebuf) - (size_t)1))
698 errlen = sizeof(rebuf) - (size_t)1;
700 rebuf[errlen] = '\0';
702 log_error(LOG_LEVEL_ERROR, "error compiling %s: %s",
713 if ((p = strchr(buf, ':')) == NULL)
728 /* Parse domain part */
729 if (buf[strlen(buf) - 1] == '.')
731 url->unanchored |= ANCHOR_RIGHT;
735 url->unanchored |= ANCHOR_LEFT;
738 /* split domain into components */
740 url->dbuffer = strdup(buf);
741 if (NULL == url->dbuffer)
747 return JB_ERR_MEMORY;
750 /* map to lower case */
751 for (p = url->dbuffer; *p ; p++)
753 *p = tolower((int)(unsigned char)*p);
756 /* split the domain name into components */
757 url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1);
767 return JB_ERR_MEMORY;
769 else if (url->dcount != 0)
772 /* save a copy of the pointers in dvec */
773 size = url->dcount * sizeof(*url->dvec);
775 url->dvec = (char **)malloc(size);
776 if (NULL == url->dvec)
784 return JB_ERR_MEMORY;
787 memcpy(url->dvec, v, size);
796 /*********************************************************************
798 * Function : free_url_spec
800 * Description : Called from the "unloaders". Freez the url
801 * structure elements.
804 * 1 : url = pointer to a url_spec structure.
808 *********************************************************************/
809 void free_url_spec(struct url_spec *url)
811 if (url == NULL) return;
825 /*********************************************************************
827 * Function : url_match
829 * Description : Compare a URL against a URL pattern.
832 * 1 : pattern = a URL pattern
833 * 2 : url = URL to match
835 * Returns : 0 iff the URL matches the pattern, else nonzero.
837 *********************************************************************/
838 int url_match(const struct url_spec *pattern,
839 const struct http_request *url)
841 return ((pattern->port == 0) || (pattern->port == url->port))
842 && ((pattern->dbuffer == NULL) || (domain_match(pattern, url) == 0))
843 && ((pattern->path == NULL) ||
844 (regexec(pattern->preg, url->path, 0, NULL, 0) == 0)