1 const char filters_rcs[] = "$Id: filters.c,v 1.18 2001/06/29 13:27:38 oes Exp $";
2 /*********************************************************************
4 * File : $Source: /cvsroot/ijbswa/current/filters.c,v $
6 * Purpose : Declares functions to parse/crunch headers and pages.
7 * Functions declared include:
8 * `acl_addr', `add_stats', `block_acl', `block_imageurl',
9 * `block_url', `url_actions', `domaincmp', `dsplit',
10 * `filter_popups', `forward_url', 'redirect_url',
11 * `ij_untrusted_url', `intercept_url', `re_process_buffer',
12 * `show_proxy_args', 'ijb_send_banner', and `trust_url'
14 * Copyright : Written by and Copyright (C) 2001 the SourceForge
15 * IJBSWA team. http://ijbswa.sourceforge.net
17 * Based on the Internet Junkbuster originally written
18 * by and Copyright (C) 1997 Anonymous Coders and
19 * Junkbusters Corporation. http://www.junkbusters.com
21 * This program is free software; you can redistribute it
22 * and/or modify it under the terms of the GNU General
23 * Public License as published by the Free Software
24 * Foundation; either version 2 of the License, or (at
25 * your option) any later version.
27 * This program is distributed in the hope that it will
28 * be useful, but WITHOUT ANY WARRANTY; without even the
29 * implied warranty of MERCHANTABILITY or FITNESS FOR A
30 * PARTICULAR PURPOSE. See the GNU General Public
31 * License for more details.
33 * The GNU General Public License should be included with
34 * this file. If not, you can view it at
35 * http://www.gnu.org/copyleft/gpl.html
36 * or write to the Free Software Foundation, Inc., 59
37 * Temple Place - Suite 330, Boston, MA 02111-1307, USA.
41 * Revision 1.18 2001/06/29 13:27:38 oes
42 * - Cleaned up, renamed and reorderd functions
43 * and improved comments
46 * - Ported to CGI platform. Now delivers
47 * http_response or NULL
48 * - Unified HTML and GIF generation (moved image detection
49 * and GIF generation here from jcc.c:chat())
50 * - Fixed HTTP status to:
51 * - 403 (Forbidden) for the "blocked" HTML message
52 * - 200 (OK) for GIF answers
53 * - 302 (Redirect) for redirect to GIF
56 * - Ported to CGI platform. Now delivers
57 * http_response or NULL
58 * - Separated detection of untrusted URL into
59 * (bool)is_untrusted_url
60 * - Added enforcement of untrusted requests
62 * - Moved redirect_url() from cgi.c to here
63 * and ported it to the CGI platform
65 * - Removed logentry from cancelled commit
67 * Revision 1.17 2001/06/09 10:55:28 jongfoster
68 * Changing BUFSIZ ==> BUFFER_SIZE
70 * Revision 1.16 2001/06/07 23:10:26 jongfoster
71 * Allowing unanchored domain patterns to back off and retry
72 * if they partially match. Optimized right-anchored patterns.
73 * Moving ACL and forward files into config file.
74 * Replacing struct gateway with struct forward_spec
76 * Revision 1.15 2001/06/03 19:12:00 oes
77 * extracted-CGI relevant stuff
79 * Revision 1.14 2001/06/01 10:30:55 oes
80 * Added optional left-anchoring to domaincmp
82 * Revision 1.13 2001/05/31 21:21:30 jongfoster
83 * Permissionsfile / actions file changes:
84 * - Changed "permission" to "action" throughout
85 * - changes to file format to allow string parameters
86 * - Moved helper functions to actions.c
88 * Revision 1.12 2001/05/31 17:35:20 oes
90 * - Enhanced domain part globbing with infix and prefix asterisk
91 * matching and optional unanchored operation
93 * Revision 1.11 2001/05/29 11:53:23 oes
94 * "See why" link added to "blocked" page
96 * Revision 1.10 2001/05/29 09:50:24 jongfoster
97 * Unified blocklist/imagelist/permissionslist.
98 * File format is still under discussion, but the internal changes
101 * Also modified interceptor behaviour:
102 * - We now intercept all URLs beginning with one of the following
103 * prefixes (and *only* these prefixes):
105 * * http://ijbswa.sf.net/config/
106 * * http://ijbswa.sourceforge.net/config/
107 * - New interceptors "home page" - go to http://i.j.b/ to see it.
108 * - Internal changes so that intercepted and fast redirect pages
109 * are not replaced with an image.
110 * - Interceptors now have the option to send a binary page direct
111 * to the client. (i.e. ijb-send-banner uses this)
112 * - Implemented show-url-info interceptor. (Which is why I needed
113 * the above interceptors changes - a typical URL is
114 * "http://i.j.b/show-url-info?url=www.somesite.com/banner.gif".
115 * The previous mechanism would not have intercepted that, and
116 * if it had been intercepted then it then it would have replaced
119 * Revision 1.9 2001/05/27 22:17:04 oes
121 * - re_process_buffer no longer writes the modified buffer
122 * to the client, which was very ugly. It now returns the
123 * buffer, which it is then written by chat.
125 * - content_length now adjusts the Content-Length: header
126 * for modified documents rather than crunch()ing it.
127 * (Length info in csp->content_length, which is 0 for
128 * unmodified documents)
130 * - For this to work, sed() is called twice when filtering.
132 * Revision 1.8 2001/05/26 17:13:28 jongfoster
133 * Filled in a function comment.
135 * Revision 1.7 2001/05/26 15:26:15 jongfoster
136 * ACL feature now provides more security by immediately dropping
137 * connections from untrusted hosts.
139 * Revision 1.6 2001/05/26 00:28:36 jongfoster
140 * Automatic reloading of config file.
141 * Removed obsolete SIGHUP support (Unix) and Reload menu option (Win32).
142 * Most of the global variables have been moved to a new
143 * struct configuration_spec, accessed through csp->config->globalname
144 * Most of the globals remaining are used by the Win32 GUI.
146 * Revision 1.5 2001/05/25 22:34:30 jongfoster
149 * Revision 1.4 2001/05/22 18:46:04 oes
151 * - Enabled filtering banners by size rather than URL
152 * by adding patterns that replace all standard banner
153 * sizes with the "Junkbuster" gif to the re_filterfile
155 * - Enabled filtering WebBugs by providing a pattern
156 * which kills all 1x1 images
158 * - Added support for PCRE_UNGREEDY behaviour to pcrs,
159 * which is selected by the (nonstandard and therefore
160 * capital) letter 'U' in the option string.
161 * It causes the quantifiers to be ungreedy by default.
162 * Appending a ? turns back to greedy (!).
164 * - Added a new interceptor ijb-send-banner, which
165 * sends back the "Junkbuster" gif. Without imagelist or
166 * MSIE detection support, or if tinygif = 1, or the
167 * URL isn't recognized as an imageurl, a lame HTML
168 * explanation is sent instead.
170 * - Added new feature, which permits blocking remote
171 * script redirects and firing back a local redirect
173 * The feature is conditionally compiled, i.e. it
174 * can be disabled with --disable-fast-redirects,
175 * plus it must be activated by a "fast-redirects"
176 * line in the config file, has its own log level
177 * and of course wants to be displayed by show-proxy-args
178 * Note: Boy, all the #ifdefs in 1001 locations and
179 * all the fumbling with configure.in and acconfig.h
180 * were *way* more work than the feature itself :-(
182 * - Because a generic redirect template was needed for
183 * this, tinygif = 3 now uses the same.
185 * - Moved GIFs, and other static HTTP response templates
190 * - Removed some >400 CRs again (Jon, you really worked
193 * Revision 1.3 2001/05/20 16:44:47 jongfoster
194 * Removing last hardcoded JunkBusters.com URLs.
196 * Revision 1.2 2001/05/20 01:21:20 jongfoster
197 * Version 2.9.4 checkin.
198 * - Merged popupfile and cookiefile, and added control over PCRS
199 * filtering, in new "permissionsfile".
200 * - Implemented LOG_LEVEL_FATAL, so that if there is a configuration
201 * file error you now get a message box (in the Win32 GUI) rather
202 * than the program exiting with no explanation.
203 * - Made killpopup use the PCRS MIME-type checking and HTTP-header
205 * - Removed tabs from "config"
206 * - Moved duplicated url parsing code in "loaders.c" to a new funcition.
207 * - Bumped up version number.
209 * Revision 1.1.1.1 2001/05/15 13:58:52 oes
210 * Initial import of version 2.9.3 source tree
213 *********************************************************************/
219 #include <sys/types.h>
226 #include <netinet/in.h>
228 #include <winsock2.h>
235 #include "showargs.h"
239 #include "jbsockets.h"
241 #include "jbsockets.h"
242 #include "miscutil.h"
251 const char filters_h_rcs[] = FILTERS_H_VERSION;
253 /* Fix a problem with Solaris. There should be no effect on other
255 * Solaris's isspace() is a macro which uses it's argument directly
256 * as an array index. Therefore we need to make sure that high-bit
257 * characters generate +ve values, and ideally we also want to make
258 * the argument match the declared parameter type of "int".
260 #define ijb_isdigit(__X) isdigit((int)(unsigned char)(__X))
264 /*********************************************************************
266 * Function : block_acl
268 * Description : Block this request?
269 * Decide yes or no based on ACL file.
272 * 1 : dst = The proxy or gateway address this is going to.
273 * Or NULL to check all possible targets.
274 * 2 : csp = Current client state (buffers, headers, etc...)
275 * Also includes the client IP address.
277 * Returns : 0 = FALSE (don't block) and 1 = TRUE (do block)
279 *********************************************************************/
280 int block_acl(struct access_control_addr *dst, struct client_state *csp)
282 struct access_control_list *acl = csp->config->acl;
284 /* if not using an access control list, then permit the connection */
290 /* search the list */
293 if ((csp->ip_addr_long & acl->src->mask) == acl->src->addr)
297 /* Just want to check if they have any access */
298 if (acl->action == ACL_PERMIT)
303 else if ( ((dst->addr & acl->dst->mask) == acl->dst->addr)
304 && ((dst->port == acl->dst->port) || (acl->dst->port == 0)))
306 if (acl->action == ACL_PERMIT)
324 /*********************************************************************
326 * Function : acl_addr
328 * Description : Called from `load_aclfile' to parse an ACL address.
331 * 1 : aspec = String specifying ACL address.
332 * 2 : aca = struct access_control_addr to fill in.
334 * Returns : 0 => Ok, everything else is an error.
336 *********************************************************************/
337 int acl_addr(char *aspec, struct access_control_addr *aca)
339 int i, masklength, port;
345 if ((p = strchr(aspec, '/')))
349 if (ijb_isdigit(*p) == 0)
353 masklength = atoi(p);
356 if ((masklength < 0) || (masklength > 32))
361 if ((p = strchr(aspec, ':')))
365 if (ijb_isdigit(*p) == 0)
374 aca->addr = ntohl(resolve_hostname_to_ip(aspec));
378 log_error(LOG_LEVEL_ERROR, "can't resolve address for %s", aspec);
382 /* build the netmask */
384 for (i=1; i <= masklength ; i++)
386 aca->mask |= (1 << (32 - i));
389 /* now mask off the host portion of the ip address
390 * (i.e. save on the network portion of the address).
392 aca->addr = aca->addr & aca->mask;
397 #endif /* def ACL_FILES */
400 /*********************************************************************
402 * Function : block_url
404 * Description : Called from `chat'. Check to see if we need to block this.
407 * 1 : csp = Current client state (buffers, headers, etc...)
409 * Returns : NULL => unblocked, else HTTP block response
411 *********************************************************************/
412 struct http_response *block_url(struct client_state *csp)
415 struct http_response *rsp;
416 struct map *exports = NULL;
419 * If it's not blocked, don't block it ;-)
421 if ((csp->action->flags & ACTION_BLOCK) == 0)
427 * Else, prepare a response
429 if (NULL == ( rsp = (struct http_response *)zalloc(sizeof(*rsp))))
435 * If it's an image-url, send back an image or redirect
436 * as specified by the relevant +image action
438 #ifdef IMAGE_BLOCKING
439 if (((csp->action->flags & ACTION_IMAGE_BLOCKER) != 0)
442 /* determine HOW images should be blocked */
443 p = csp->action->string[ACTION_STRING_IMAGE_BLOCKER];
445 /* and handle accordingly: */
446 if ((p == NULL) || (0 == strcmpic(p, "logo")))
448 rsp->body = bindup(JBGIF, sizeof(JBGIF));
449 rsp->content_length = sizeof(JBGIF);
450 enlist_unique_header(rsp->headers, "Content-Type", "image/gif");
453 else if (0 == strcmpic(p, "blank"))
455 rsp->body = bindup(BLANKGIF, sizeof(BLANKGIF));
456 rsp->content_length = sizeof(BLANKGIF);
457 enlist_unique_header(rsp->headers, "Content-Type", "image/gif");
462 rsp->status = strdup("302 Local Redirect from Junkbuster");
463 enlist_unique_header(rsp->headers, "Location", p);
467 #endif /* def IMAGE_BLOCKING */
470 * Else, generate an HTML "blocked" message:
474 exports = default_exports(csp, NULL);
476 exports = map(exports, "force-prefix", 1, FORCE_PREFIX, 1);
478 exports = map_block_killer(exports, "force-support");
479 #endif /* ndef FORCE_LOAD */
481 exports = map(exports, "hostport", 1, csp->http->hostport, 1);
482 exports = map(exports, "hostport-html", 1, html_encode(csp->http->hostport), 0);
483 exports = map(exports, "path", 1, csp->http->path, 1);
484 exports = map(exports, "path-html", 1, html_encode(csp->http->path), 0);
486 rsp->body = fill_template(csp, "blocked", exports);
489 rsp->status = strdup("403 Request for blocked URL");
492 return(finish_http_response(rsp));
498 /*********************************************************************
500 * Function : trust_url FIXME: I should be called distrust_url
502 * Description : Calls is_untrusted_url to determine if the URL is trusted
503 * and if not, returns a HTTP 304 response with a reject message.
506 * 1 : csp = Current client state (buffers, headers, etc...)
508 * Returns : NULL => trusted, else http_response.
510 *********************************************************************/
511 struct http_response *trust_url(struct client_state *csp)
513 struct http_response *rsp;
514 struct map *exports = NULL;
515 char buf[BUFFER_SIZE], *p = NULL;
516 struct url_spec **tl, *t;
519 * Don't bother to work on trusted URLs
521 if (!is_untrusted_url(csp))
527 * Else, prepare a response:
529 if (NULL == ( rsp = (struct http_response *)zalloc(sizeof(*rsp))))
533 exports = default_exports(csp, NULL);
536 * Export the host, port, and referrer information
538 exports = map(exports, "hostport", 1, csp->http->hostport, 1);
539 exports = map(exports, "path", 1, csp->http->path, 1);
540 exports = map(exports, "hostport-html", 1, html_encode(csp->http->hostport), 0);
541 exports = map(exports, "path-html", 1, html_encode(csp->http->path), 0);
543 if (csp->referrer && strlen(csp->referrer) > 9)
545 exports = map(exports, "referrer", 1, csp->referrer + 9, 1);
546 exports = map(exports, "referrer-html", 1, html_encode(csp->referrer + 9), 0);
550 exports = map(exports, "referrer", 1, "unknown", 1);
551 exports = map(exports, "referrer-html", 1, "unknown", 1);
555 * Export the trust list
557 for (tl = csp->config->trust_list; (t = *tl) ; tl++)
559 sprintf(buf, "<li>%s</li>\n", t->spec);
562 exports = map(exports, "trusted-referrers", 1, p, 0);
566 * Export the trust info, if available
568 if (csp->config->trust_info->next)
572 for (l = csp->config->trust_info->next; l ; l = l->next)
574 sprintf(buf, "<li> <a href=%s>%s</a><br>\n",l->str, l->str);
577 exports = map(exports, "trust-info", 1, p, 0);
581 exports = map_block_killer(exports, "have-trust-info");
585 * Export the force prefix or the force conditional block killer
588 exports = map(exports, "force-prefix", 1, FORCE_PREFIX, 1);
590 exports = map_block_killer(exports, "force-support");
591 #endif /* ndef FORCE_LOAD */
596 rsp->body = fill_template(csp, "untrusted", exports);
599 return(finish_http_response(rsp));
602 #endif /* def TRUST_FILES */
605 #ifdef FAST_REDIRECTS
606 /*********************************************************************
608 * Function : redirect_url
610 * Description : Checks for redirection URLs and returns a HTTP redirect
611 * to the destination URL, if necessary
614 * 1 : csp = Current client state (buffers, headers, etc...)
616 * Returns : NULL if URL was clean, HTTP redirect otherwise.
618 *********************************************************************/
619 struct http_response *redirect_url(struct client_state *csp)
622 struct http_response *rsp;
624 p = q = csp->http->path;
625 log_error(LOG_LEVEL_REDIRECTS, "checking path for redirects: %s", p);
628 * find the last URL encoded in the request
630 while (p = strstr(p, "http://"))
636 * if there was any, generate and return a HTTP redirect
638 if (q != csp->http->path)
640 log_error(LOG_LEVEL_REDIRECTS, "redirecting to: %s", q);
642 if (NULL == ( rsp = zalloc(sizeof(*rsp))))
647 rsp->status = strdup("302 Local Redirect from Junkbuster");
648 enlist_unique_header(rsp->headers, "Location", q);
650 return(finish_http_response(rsp));
658 #endif /* def FAST_REDIRECTS */
661 #ifdef IMAGE_BLOCKING
662 /*********************************************************************
664 * Function : is_imageurl
666 * Description : Given a URL, decide whether it is an image or not,
667 * using either the info from a previous +image action
668 * or, #ifdef DETECT_MSIE_IMAGES, the info from the
669 * browser's accept header.
672 * 1 : csp = Current client state (buffers, headers, etc...)
674 * Returns : True (nonzero) if URL is an image, false (0)
677 *********************************************************************/
678 int is_imageurl(struct client_state *csp)
680 #ifdef DETECT_MSIE_IMAGES
681 if ((csp->accept_types
682 & (ACCEPT_TYPE_IS_MSIE|ACCEPT_TYPE_MSIE_IMAGE|ACCEPT_TYPE_MSIE_HTML))
683 == (ACCEPT_TYPE_IS_MSIE|ACCEPT_TYPE_MSIE_IMAGE))
687 else if ((csp->accept_types
688 & (ACCEPT_TYPE_IS_MSIE|ACCEPT_TYPE_MSIE_IMAGE|ACCEPT_TYPE_MSIE_HTML))
689 == (ACCEPT_TYPE_IS_MSIE|ACCEPT_TYPE_MSIE_HTML))
695 return ((csp->action->flags & ACTION_IMAGE) != 0);
698 #endif /* def IMAGE_BLOCKING */
702 /*********************************************************************
704 * Function : is_untrusted_url
706 * Description : Should we "distrust" this URL (and block it)?
708 * Yes if it matches a line in the trustfile, or if the
709 * referrer matches a line starting with "+" in the
714 * 1 : csp = Current client state (buffers, headers, etc...)
716 * Returns : 0 => trusted, 1 => untrusted
718 *********************************************************************/
719 int is_untrusted_url(struct client_state *csp)
721 struct file_list *fl;
722 struct block_spec *b;
723 struct url_spec url[1], **tl, *t;
724 struct http_request rhttp[1];
727 if (((fl = csp->tlist) == NULL) || ((b = fl->f) == NULL))
732 *url = dsplit(csp->http->host);
734 /* if splitting the domain fails, punt */
735 if (url->dbuf == NULL) return(0);
737 memset(rhttp, '\0', sizeof(*rhttp));
739 for (b = b->next; b ; b = b->next)
741 if ((b->url->port == 0) || (b->url->port == csp->http->port))
743 if ((b->url->domain[0] == '\0') || (domaincmp(b->url, url) == 0))
745 if ((b->url->path == NULL) ||
747 (regexec(b->url->preg, csp->http->path, 0, NULL, 0) == 0)
749 (strncmp(b->url->path, csp->http->path, b->url->pathlen) == 0)
756 if (b->reject == 0) return(0);
767 if ((csp->referrer == NULL)|| (strlen(csp->referrer) <= 9))
769 /* no referrer was supplied */
773 /* forge a URL from the referrer so we can use
774 * convert_url() to parse it into its components.
778 p = strsav(p, "GET ");
779 p = strsav(p, csp->referrer + 9); /* skip over "Referer: " */
780 p = strsav(p, " HTTP/1.0");
782 parse_http_request(p, rhttp, csp);
785 if (rhttp->cmd == NULL)
790 *url = dsplit(rhttp->host);
792 /* if splitting the domain fails, punt */
793 if (url->dbuf == NULL) return(1);
795 for (tl = csp->config->trust_list; (t = *tl) ; tl++)
797 if ((t->port == 0) || (t->port == rhttp->port))
799 if ((t->domain[0] == '\0') || domaincmp(t, url) == 0)
801 if ((t->path == NULL) ||
803 (regexec(t->preg, rhttp->path, 0, NULL, 0) == 0)
805 (strncmp(t->path, rhttp->path, t->pathlen) == 0)
809 /* if the URL's referrer is from a trusted referrer, then
810 * add the target spec to the trustfile as an unblocked
811 * domain and return NULL (which means it's OK).
819 if ((fp = fopen(csp->config->trustfile, "a")))
824 h = strsav(h, csp->http->hostport);
830 /* since this path points into a user's home space
831 * be sure to include this spec in the trustfile.
833 if ((p = strchr(p, '/')))
836 h = strsav(h, csp->http->path); /* FIXME: p?! */
841 fprintf(fp, "%s\n", h);
852 #endif /* def TRUST_FILES */
856 /*********************************************************************
858 * Function : re_process_buffer
860 * Description : Apply all the pcrs jobs from the joblist (re_filterfile)
861 * to the text buffer that's been accumulated in
862 * csp->iob->buf and set csp->content_length to the modified
866 * 1 : csp = Current client state (buffers, headers, etc...)
868 * Returns : a pointer to the (newly allocated) modified buffer.
869 * or an empty string in case something went wrong
871 *********************************************************************/
872 char *re_process_buffer(struct client_state *csp)
875 int size = csp->iob->eod - csp->iob->cur;
877 char *old = csp->iob->cur, *new = NULL;
880 struct file_list *fl;
881 struct re_filterfile_spec *b;
883 /* Sanity first ;-) */
889 if ( ( NULL == (fl = csp->rlist) ) || ( NULL == (b = fl->f) ) )
891 log_error(LOG_LEVEL_ERROR, "Unable to get current state of regexp filtering.");
895 log_error(LOG_LEVEL_RE_FILTER, "re_filtering %s%s (size %d) ...",
896 csp->http->hostport, csp->http->path, size);
898 /* Apply all jobs from the joblist */
899 for (job = b->joblist; NULL != job; job = job->next)
901 hits += pcrs_execute(job, old, size, &new, &size);
902 if (old != csp->iob->cur) free(old);
906 log_error(LOG_LEVEL_RE_FILTER, " produced %d hits (new size %d).", hits, size);
908 csp->content_length = size;
910 /* fwiw, reset the iob */
915 #endif /* def PCRS */
918 /*********************************************************************
920 * Function : url_actions
922 * Description : Gets the actions for this URL.
925 * 1 : http = http_request request for blocked URLs
926 * 2 : csp = Current client state (buffers, headers, etc...)
930 *********************************************************************/
931 void url_actions(struct http_request *http,
932 struct client_state *csp)
934 struct file_list *fl;
935 struct url_actions *b;
937 init_current_action(csp->action);
939 if (((fl = csp->actions_list) == NULL) || ((b = fl->f) == NULL))
944 apply_url_actions(csp->action, http, b);
949 /*********************************************************************
951 * Function : apply_url_actions
953 * Description : Applies a list of URL actions.
956 * 1 : action = Destination.
957 * 2 : http = Current URL
958 * 3 : b = list of URL actions to apply
962 *********************************************************************/
963 void apply_url_actions(struct current_action_spec *action,
964 struct http_request *http,
965 struct url_actions *b)
967 struct url_spec url[1];
971 /* Should never happen */
975 *url = dsplit(http->host);
977 /* if splitting the domain fails, punt */
978 if (url->dbuf == NULL)
983 for (b = b->next; NULL != b; b = b->next)
985 if ((b->url->port == 0) || (b->url->port == http->port))
987 if ((b->url->domain[0] == '\0') || (domaincmp(b->url, url) == 0))
989 if ((b->url->path == NULL) ||
991 (regexec(b->url->preg, http->path, 0, NULL, 0) == 0)
993 (strncmp(b->url->path, http->path, b->url->pathlen) == 0)
997 merge_current_action(action, b->action);
1008 /*********************************************************************
1010 * Function : forward_url
1012 * Description : Should we forward this to another proxy?
1015 * 1 : http = http_request request for current URL
1016 * 2 : csp = Current client state (buffers, headers, etc...)
1018 * Returns : Pointer to forwarding information.
1020 *********************************************************************/
1021 const struct forward_spec * forward_url(struct http_request *http,
1022 struct client_state *csp)
1024 static const struct forward_spec fwd_default[1] = { 0 }; /* All zeroes */
1025 struct forward_spec *fwd = csp->config->forward;
1026 struct url_spec url[1];
1030 return(fwd_default);
1033 *url = dsplit(http->host);
1035 /* if splitting the domain fails, punt */
1036 if (url->dbuf == NULL)
1038 return(fwd_default);
1043 if ((fwd->url->port == 0) || (fwd->url->port == http->port))
1045 if ((fwd->url->domain[0] == '\0') || (domaincmp(fwd->url, url) == 0))
1047 if ((fwd->url->path == NULL) ||
1049 (regexec(fwd->url->preg, http->path, 0, NULL, 0) == 0)
1051 (strncmp(fwd->url->path, http->path, fwd->url->pathlen) == 0)
1066 return(fwd_default);
1071 /*********************************************************************
1075 * Description : Takes a domain and returns a pointer to a url_spec
1076 * structure populated with dbuf, dcnt and dvec. The
1077 * other fields in the structure that is returned are zero.
1080 * 1 : domain = a URL address
1082 * Returns : url_spec structure populated with dbuf, dcnt and dvec.
1084 *********************************************************************/
1085 struct url_spec dsplit(char *domain)
1087 struct url_spec ret[1];
1088 char *v[BUFFER_SIZE];
1092 memset(ret, '\0', sizeof(*ret));
1094 if (domain[strlen(domain) - 1] == '.')
1096 ret->unanchored |= ANCHOR_RIGHT;
1099 if (domain[0] == '.')
1101 ret->unanchored |= ANCHOR_LEFT;
1104 ret->dbuf = strdup(domain);
1106 /* map to lower case */
1107 for (p = ret->dbuf; *p ; p++) *p = tolower(*p);
1109 /* split the domain name into components */
1110 ret->dcnt = ssplit(ret->dbuf, ".", v, SZ(v), 1, 1);
1114 memset(ret, '\0', sizeof(ret));
1118 /* save a copy of the pointers in dvec */
1119 size = ret->dcnt * sizeof(*ret->dvec);
1121 if ((ret->dvec = (char **)malloc(size)))
1123 memcpy(ret->dvec, v, size);
1131 /*********************************************************************
1133 * Function : simple_domaincmp
1135 * Description : Domain-wise Compare fqdn's. The comparison is
1136 * both left- and right-anchored. The individual
1137 * domain names are compared with simplematch().
1138 * This is only used by domaincmp.
1141 * 1 : pv = array of patterns to compare
1142 * 2 : fv = array of domain components to compare
1143 * 3 : len = length of the arrays (both arrays are the
1144 * same length - if they weren't, it couldn't
1145 * possibly be a match).
1147 * Returns : 0 => domains are equivalent, else no match.
1149 *********************************************************************/
1150 static int simple_domaincmp(char **pv, char **fv, int len)
1154 for (n = 0; n < len; n++)
1156 if (simplematch(pv[n], fv[n]))
1167 /*********************************************************************
1169 * Function : domaincmp
1171 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
1172 * pattern->unachored, the comparison is un-, left-,
1173 * right-anchored, or both.
1174 * The individual domain names are compared with
1178 * 1 : pattern = a domain that may contain a '*' as a wildcard.
1179 * 2 : fqdn = domain name against which the patterns are compared.
1181 * Returns : 0 => domains are equivalent, else no match.
1183 *********************************************************************/
1184 int domaincmp(struct url_spec *pattern, struct url_spec *fqdn)
1186 char **pv, **fv; /* vectors */
1188 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
1190 plen = pattern->dcnt;
1195 /* fqdn is too short to match this pattern */
1202 if (unanchored == ANCHOR_LEFT)
1207 * Convert this into a fully anchored pattern with
1208 * the fqdn and pattern the same length
1210 fv += (flen - plen); /* flen - plen >= 0 due to check above */
1211 return simple_domaincmp(pv, fv, plen);
1213 else if (unanchored == 0)
1215 /* Fully anchored, check length */
1220 return simple_domaincmp(pv, fv, plen);
1222 else if (unanchored == ANCHOR_RIGHT)
1224 /* Left anchored, ignore all extra in fqdn */
1225 return simple_domaincmp(pv, fv, plen);
1231 int maxn = flen - plen;
1232 for (n = 0; n <= maxn; n++)
1234 if (!simple_domaincmp(pv, fv, plen))
1239 * Doesn't match from start of fqdn
1240 * Try skipping first part of fqdn