1 const char filters_rcs[] = "$Id: filters.c,v 1.20 2001/07/01 17:01:04 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', `pcrs_filter_respose',
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 1.20 2001/07/01 17:01:04 oes
42 * Added comments and missing return statement in is_untrusted_url()
44 * Revision 1.19 2001/06/29 21:45:41 oes
45 * Indentation, CRLF->LF, Tab-> Space
47 * Revision 1.18 2001/06/29 13:27:38 oes
48 * - Cleaned up, renamed and reorderd functions
49 * and improved comments
52 * - Ported to CGI platform. Now delivers
53 * http_response or NULL
54 * - Unified HTML and GIF generation (moved image detection
55 * and GIF generation here from jcc.c:chat())
56 * - Fixed HTTP status to:
57 * - 403 (Forbidden) for the "blocked" HTML message
58 * - 200 (OK) for GIF answers
59 * - 302 (Redirect) for redirect to GIF
62 * - Ported to CGI platform. Now delivers
63 * http_response or NULL
64 * - Separated detection of untrusted URL into
65 * (bool)is_untrusted_url
66 * - Added enforcement of untrusted requests
68 * - Moved redirect_url() from cgi.c to here
69 * and ported it to the CGI platform
71 * - Removed logentry from cancelled commit
73 * Revision 1.17 2001/06/09 10:55:28 jongfoster
74 * Changing BUFSIZ ==> BUFFER_SIZE
76 * Revision 1.16 2001/06/07 23:10:26 jongfoster
77 * Allowing unanchored domain patterns to back off and retry
78 * if they partially match. Optimized right-anchored patterns.
79 * Moving ACL and forward files into config file.
80 * Replacing struct gateway with struct forward_spec
82 * Revision 1.15 2001/06/03 19:12:00 oes
83 * extracted-CGI relevant stuff
85 * Revision 1.14 2001/06/01 10:30:55 oes
86 * Added optional left-anchoring to domaincmp
88 * Revision 1.13 2001/05/31 21:21:30 jongfoster
89 * Permissionsfile / actions file changes:
90 * - Changed "permission" to "action" throughout
91 * - changes to file format to allow string parameters
92 * - Moved helper functions to actions.c
94 * Revision 1.12 2001/05/31 17:35:20 oes
96 * - Enhanced domain part globbing with infix and prefix asterisk
97 * matching and optional unanchored operation
99 * Revision 1.11 2001/05/29 11:53:23 oes
100 * "See why" link added to "blocked" page
102 * Revision 1.10 2001/05/29 09:50:24 jongfoster
103 * Unified blocklist/imagelist/permissionslist.
104 * File format is still under discussion, but the internal changes
107 * Also modified interceptor behaviour:
108 * - We now intercept all URLs beginning with one of the following
109 * prefixes (and *only* these prefixes):
111 * * http://ijbswa.sf.net/config/
112 * * http://ijbswa.sourceforge.net/config/
113 * - New interceptors "home page" - go to http://i.j.b/ to see it.
114 * - Internal changes so that intercepted and fast redirect pages
115 * are not replaced with an image.
116 * - Interceptors now have the option to send a binary page direct
117 * to the client. (i.e. ijb-send-banner uses this)
118 * - Implemented show-url-info interceptor. (Which is why I needed
119 * the above interceptors changes - a typical URL is
120 * "http://i.j.b/show-url-info?url=www.somesite.com/banner.gif".
121 * The previous mechanism would not have intercepted that, and
122 * if it had been intercepted then it then it would have replaced
125 * Revision 1.9 2001/05/27 22:17:04 oes
127 * - re_process_buffer no longer writes the modified buffer
128 * to the client, which was very ugly. It now returns the
129 * buffer, which it is then written by chat.
131 * - content_length now adjusts the Content-Length: header
132 * for modified documents rather than crunch()ing it.
133 * (Length info in csp->content_length, which is 0 for
134 * unmodified documents)
136 * - For this to work, sed() is called twice when filtering.
138 * Revision 1.8 2001/05/26 17:13:28 jongfoster
139 * Filled in a function comment.
141 * Revision 1.7 2001/05/26 15:26:15 jongfoster
142 * ACL feature now provides more security by immediately dropping
143 * connections from untrusted hosts.
145 * Revision 1.6 2001/05/26 00:28:36 jongfoster
146 * Automatic reloading of config file.
147 * Removed obsolete SIGHUP support (Unix) and Reload menu option (Win32).
148 * Most of the global variables have been moved to a new
149 * struct configuration_spec, accessed through csp->config->globalname
150 * Most of the globals remaining are used by the Win32 GUI.
152 * Revision 1.5 2001/05/25 22:34:30 jongfoster
155 * Revision 1.4 2001/05/22 18:46:04 oes
157 * - Enabled filtering banners by size rather than URL
158 * by adding patterns that replace all standard banner
159 * sizes with the "Junkbuster" gif to the re_filterfile
161 * - Enabled filtering WebBugs by providing a pattern
162 * which kills all 1x1 images
164 * - Added support for PCRE_UNGREEDY behaviour to pcrs,
165 * which is selected by the (nonstandard and therefore
166 * capital) letter 'U' in the option string.
167 * It causes the quantifiers to be ungreedy by default.
168 * Appending a ? turns back to greedy (!).
170 * - Added a new interceptor ijb-send-banner, which
171 * sends back the "Junkbuster" gif. Without imagelist or
172 * MSIE detection support, or if tinygif = 1, or the
173 * URL isn't recognized as an imageurl, a lame HTML
174 * explanation is sent instead.
176 * - Added new feature, which permits blocking remote
177 * script redirects and firing back a local redirect
179 * The feature is conditionally compiled, i.e. it
180 * can be disabled with --disable-fast-redirects,
181 * plus it must be activated by a "fast-redirects"
182 * line in the config file, has its own log level
183 * and of course wants to be displayed by show-proxy-args
184 * Note: Boy, all the #ifdefs in 1001 locations and
185 * all the fumbling with configure.in and acconfig.h
186 * were *way* more work than the feature itself :-(
188 * - Because a generic redirect template was needed for
189 * this, tinygif = 3 now uses the same.
191 * - Moved GIFs, and other static HTTP response templates
196 * - Removed some >400 CRs again (Jon, you really worked
199 * Revision 1.3 2001/05/20 16:44:47 jongfoster
200 * Removing last hardcoded JunkBusters.com URLs.
202 * Revision 1.2 2001/05/20 01:21:20 jongfoster
203 * Version 2.9.4 checkin.
204 * - Merged popupfile and cookiefile, and added control over PCRS
205 * filtering, in new "permissionsfile".
206 * - Implemented LOG_LEVEL_FATAL, so that if there is a configuration
207 * file error you now get a message box (in the Win32 GUI) rather
208 * than the program exiting with no explanation.
209 * - Made killpopup use the PCRS MIME-type checking and HTTP-header
211 * - Removed tabs from "config"
212 * - Moved duplicated url parsing code in "loaders.c" to a new funcition.
213 * - Bumped up version number.
215 * Revision 1.1.1.1 2001/05/15 13:58:52 oes
216 * Initial import of version 2.9.3 source tree
219 *********************************************************************/
225 #include <sys/types.h>
232 #include <netinet/in.h>
234 #include <winsock2.h>
241 #include "showargs.h"
245 #include "jbsockets.h"
247 #include "jbsockets.h"
248 #include "miscutil.h"
252 #include "deanimate.h"
258 const char filters_h_rcs[] = FILTERS_H_VERSION;
260 /* Fix a problem with Solaris. There should be no effect on other
262 * Solaris's isspace() is a macro which uses it's argument directly
263 * as an array index. Therefore we need to make sure that high-bit
264 * characters generate +ve values, and ideally we also want to make
265 * the argument match the declared parameter type of "int".
267 #define ijb_isdigit(__X) isdigit((int)(unsigned char)(__X))
271 /*********************************************************************
273 * Function : block_acl
275 * Description : Block this request?
276 * Decide yes or no based on ACL file.
279 * 1 : dst = The proxy or gateway address this is going to.
280 * Or NULL to check all possible targets.
281 * 2 : csp = Current client state (buffers, headers, etc...)
282 * Also includes the client IP address.
284 * Returns : 0 = FALSE (don't block) and 1 = TRUE (do block)
286 *********************************************************************/
287 int block_acl(struct access_control_addr *dst, struct client_state *csp)
289 struct access_control_list *acl = csp->config->acl;
291 /* if not using an access control list, then permit the connection */
297 /* search the list */
300 if ((csp->ip_addr_long & acl->src->mask) == acl->src->addr)
304 /* Just want to check if they have any access */
305 if (acl->action == ACL_PERMIT)
310 else if ( ((dst->addr & acl->dst->mask) == acl->dst->addr)
311 && ((dst->port == acl->dst->port) || (acl->dst->port == 0)))
313 if (acl->action == ACL_PERMIT)
331 /*********************************************************************
333 * Function : acl_addr
335 * Description : Called from `load_aclfile' to parse an ACL address.
338 * 1 : aspec = String specifying ACL address.
339 * 2 : aca = struct access_control_addr to fill in.
341 * Returns : 0 => Ok, everything else is an error.
343 *********************************************************************/
344 int acl_addr(char *aspec, struct access_control_addr *aca)
346 int i, masklength, port;
352 if ((p = strchr(aspec, '/')))
356 if (ijb_isdigit(*p) == 0)
360 masklength = atoi(p);
363 if ((masklength < 0) || (masklength > 32))
368 if ((p = strchr(aspec, ':')))
372 if (ijb_isdigit(*p) == 0)
381 aca->addr = ntohl(resolve_hostname_to_ip(aspec));
385 log_error(LOG_LEVEL_ERROR, "can't resolve address for %s", aspec);
389 /* build the netmask */
391 for (i=1; i <= masklength ; i++)
393 aca->mask |= (1 << (32 - i));
396 /* now mask off the host portion of the ip address
397 * (i.e. save on the network portion of the address).
399 aca->addr = aca->addr & aca->mask;
404 #endif /* def ACL_FILES */
407 /*********************************************************************
409 * Function : block_url
411 * Description : Called from `chat'. Check to see if we need to block this.
414 * 1 : csp = Current client state (buffers, headers, etc...)
416 * Returns : NULL => unblocked, else HTTP block response
418 *********************************************************************/
419 struct http_response *block_url(struct client_state *csp)
422 struct http_response *rsp;
423 struct map *exports = NULL;
426 * If it's not blocked, don't block it ;-)
428 if ((csp->action->flags & ACTION_BLOCK) == 0)
434 * Else, prepare a response
436 if (NULL == ( rsp = (struct http_response *)zalloc(sizeof(*rsp))))
442 * If it's an image-url, send back an image or redirect
443 * as specified by the relevant +image action
445 #ifdef IMAGE_BLOCKING
446 if (((csp->action->flags & ACTION_IMAGE_BLOCKER) != 0)
449 /* determine HOW images should be blocked */
450 p = csp->action->string[ACTION_STRING_IMAGE_BLOCKER];
452 /* and handle accordingly: */
453 if ((p == NULL) || (0 == strcmpic(p, "logo")))
455 rsp->body = bindup(JBGIF, sizeof(JBGIF));
456 rsp->content_length = sizeof(JBGIF);
457 enlist_unique_header(rsp->headers, "Content-Type", "image/gif");
460 else if (0 == strcmpic(p, "blank"))
462 rsp->body = bindup(BLANKGIF, sizeof(BLANKGIF));
463 rsp->content_length = sizeof(BLANKGIF);
464 enlist_unique_header(rsp->headers, "Content-Type", "image/gif");
469 rsp->status = strdup("302 Local Redirect from Junkbuster");
470 enlist_unique_header(rsp->headers, "Location", p);
474 #endif /* def IMAGE_BLOCKING */
477 * Else, generate an HTML "blocked" message:
481 exports = default_exports(csp, NULL);
483 exports = map(exports, "force-prefix", 1, FORCE_PREFIX, 1);
485 exports = map_block_killer(exports, "force-support");
486 #endif /* ndef FORCE_LOAD */
488 exports = map(exports, "hostport", 1, csp->http->hostport, 1);
489 exports = map(exports, "hostport-html", 1, html_encode(csp->http->hostport), 0);
490 exports = map(exports, "path", 1, csp->http->path, 1);
491 exports = map(exports, "path-html", 1, html_encode(csp->http->path), 0);
493 rsp->body = fill_template(csp, "blocked", exports);
496 rsp->status = strdup("403 Request for blocked URL");
499 return(finish_http_response(rsp));
505 /*********************************************************************
507 * Function : trust_url FIXME: I should be called distrust_url
509 * Description : Calls is_untrusted_url to determine if the URL is trusted
510 * and if not, returns a HTTP 304 response with a reject message.
513 * 1 : csp = Current client state (buffers, headers, etc...)
515 * Returns : NULL => trusted, else http_response.
517 *********************************************************************/
518 struct http_response *trust_url(struct client_state *csp)
520 struct http_response *rsp;
521 struct map *exports = NULL;
522 char buf[BUFFER_SIZE], *p = NULL;
523 struct url_spec **tl, *t;
526 * Don't bother to work on trusted URLs
528 if (!is_untrusted_url(csp))
534 * Else, prepare a response:
536 if (NULL == ( rsp = (struct http_response *)zalloc(sizeof(*rsp))))
540 exports = default_exports(csp, NULL);
543 * Export the host, port, and referrer information
545 exports = map(exports, "hostport", 1, csp->http->hostport, 1);
546 exports = map(exports, "path", 1, csp->http->path, 1);
547 exports = map(exports, "hostport-html", 1, html_encode(csp->http->hostport), 0);
548 exports = map(exports, "path-html", 1, html_encode(csp->http->path), 0);
550 if (csp->referrer && strlen(csp->referrer) > 9)
552 exports = map(exports, "referrer", 1, csp->referrer + 9, 1);
553 exports = map(exports, "referrer-html", 1, html_encode(csp->referrer + 9), 0);
557 exports = map(exports, "referrer", 1, "unknown", 1);
558 exports = map(exports, "referrer-html", 1, "unknown", 1);
562 * Export the trust list
564 for (tl = csp->config->trust_list; (t = *tl) ; tl++)
566 sprintf(buf, "<li>%s</li>\n", t->spec);
569 exports = map(exports, "trusted-referrers", 1, p, 0);
573 * Export the trust info, if available
575 if (csp->config->trust_info->next)
579 for (l = csp->config->trust_info->next; l ; l = l->next)
581 sprintf(buf, "<li> <a href=%s>%s</a><br>\n",l->str, l->str);
584 exports = map(exports, "trust-info", 1, p, 0);
588 exports = map_block_killer(exports, "have-trust-info");
592 * Export the force prefix or the force conditional block killer
595 exports = map(exports, "force-prefix", 1, FORCE_PREFIX, 1);
597 exports = map_block_killer(exports, "force-support");
598 #endif /* ndef FORCE_LOAD */
603 rsp->body = fill_template(csp, "untrusted", exports);
606 return(finish_http_response(rsp));
609 #endif /* def TRUST_FILES */
612 #ifdef FAST_REDIRECTS
613 /*********************************************************************
615 * Function : redirect_url
617 * Description : Checks for redirection URLs and returns a HTTP redirect
618 * to the destination URL, if necessary
621 * 1 : csp = Current client state (buffers, headers, etc...)
623 * Returns : NULL if URL was clean, HTTP redirect otherwise.
625 *********************************************************************/
626 struct http_response *redirect_url(struct client_state *csp)
629 struct http_response *rsp;
631 p = q = csp->http->path;
632 log_error(LOG_LEVEL_REDIRECTS, "checking path for redirects: %s", p);
635 * find the last URL encoded in the request
637 while (p = strstr(p, "http://"))
643 * if there was any, generate and return a HTTP redirect
645 if (q != csp->http->path)
647 log_error(LOG_LEVEL_REDIRECTS, "redirecting to: %s", q);
649 if (NULL == ( rsp = zalloc(sizeof(*rsp))))
654 rsp->status = strdup("302 Local Redirect from Junkbuster");
655 enlist_unique_header(rsp->headers, "Location", q);
657 return(finish_http_response(rsp));
665 #endif /* def FAST_REDIRECTS */
668 #ifdef IMAGE_BLOCKING
669 /*********************************************************************
671 * Function : is_imageurl
673 * Description : Given a URL, decide whether it is an image or not,
674 * using either the info from a previous +image action
675 * or, #ifdef DETECT_MSIE_IMAGES, the info from the
676 * browser's accept header.
679 * 1 : csp = Current client state (buffers, headers, etc...)
681 * Returns : True (nonzero) if URL is an image, false (0)
684 *********************************************************************/
685 int is_imageurl(struct client_state *csp)
687 #ifdef DETECT_MSIE_IMAGES
688 if ((csp->accept_types
689 & (ACCEPT_TYPE_IS_MSIE|ACCEPT_TYPE_MSIE_IMAGE|ACCEPT_TYPE_MSIE_HTML))
690 == (ACCEPT_TYPE_IS_MSIE|ACCEPT_TYPE_MSIE_IMAGE))
694 else if ((csp->accept_types
695 & (ACCEPT_TYPE_IS_MSIE|ACCEPT_TYPE_MSIE_IMAGE|ACCEPT_TYPE_MSIE_HTML))
696 == (ACCEPT_TYPE_IS_MSIE|ACCEPT_TYPE_MSIE_HTML))
702 return ((csp->action->flags & ACTION_IMAGE) != 0);
705 #endif /* def IMAGE_BLOCKING */
709 /*********************************************************************
711 * Function : is_untrusted_url
713 * Description : Should we "distrust" this URL (and block it)?
715 * Yes if it matches a line in the trustfile, or if the
716 * referrer matches a line starting with "+" in the
721 * 1 : csp = Current client state (buffers, headers, etc...)
723 * Returns : 0 => trusted, 1 => untrusted
725 *********************************************************************/
726 int is_untrusted_url(struct client_state *csp)
728 struct file_list *fl;
729 struct block_spec *b;
730 struct url_spec url[1], **tl, *t;
731 struct http_request rhttp[1];
735 * If we don't have a trustlist, we trust everybody
737 if (((fl = csp->tlist) == NULL) || ((b = fl->f) == NULL))
744 * Do we trust the request URL itself?
746 *url = dsplit(csp->http->host);
748 /* if splitting the domain fails, punt */
749 if (url->dbuf == NULL) return(0);
751 memset(rhttp, '\0', sizeof(*rhttp));
753 for (b = b->next; b ; b = b->next)
755 if ((b->url->port == 0) || (b->url->port == csp->http->port))
757 if ((b->url->domain[0] == '\0') || (domaincmp(b->url, url) == 0))
759 if ((b->url->path == NULL) ||
761 (regexec(b->url->preg, csp->http->path, 0, NULL, 0) == 0)
763 (strncmp(b->url->path, csp->http->path, b->url->pathlen) == 0)
770 if (b->reject == 0) return(0);
781 if ((csp->referrer == NULL)|| (strlen(csp->referrer) <= 9))
783 /* no referrer was supplied */
787 /* forge a URL from the referrer so we can use
788 * convert_url() to parse it into its components.
792 p = strsav(p, "GET ");
793 p = strsav(p, csp->referrer + 9); /* skip over "Referer: " */
794 p = strsav(p, " HTTP/1.0");
796 parse_http_request(p, rhttp, csp);
799 if (rhttp->cmd == NULL)
806 * If not, do we maybe trust its referrer?
808 *url = dsplit(rhttp->host);
810 /* if splitting the domain fails, punt */
811 if (url->dbuf == NULL) return(1);
813 for (tl = csp->config->trust_list; (t = *tl) ; tl++)
815 if ((t->port == 0) || (t->port == rhttp->port))
817 if ((t->domain[0] == '\0') || domaincmp(t, url) == 0)
819 if ((t->path == NULL) ||
821 (regexec(t->preg, rhttp->path, 0, NULL, 0) == 0)
823 (strncmp(t->path, rhttp->path, t->pathlen) == 0)
827 /* if the URL's referrer is from a trusted referrer, then
828 * add the target spec to the trustfile as an unblocked
829 * domain and return NULL (which means it's OK).
837 if ((fp = fopen(csp->config->trustfile, "a")))
842 h = strsav(h, csp->http->hostport);
848 /* since this path points into a user's home space
849 * be sure to include this spec in the trustfile.
851 if ((p = strchr(p, '/')))
854 h = strsav(h, csp->http->path); /* FIXME: p?! */
859 fprintf(fp, "%s\n", h);
870 #endif /* def TRUST_FILES */
873 /*********************************************************************
875 * Function : pcrs_filter_response
877 * Description : Apply all the pcrs jobs from the joblist (re_filterfile)
878 * to the text buffer that's been accumulated in
879 * csp->iob->buf and set csp->content_length to the modified
883 * 1 : csp = Current client state (buffers, headers, etc...)
885 * Returns : a pointer to the (newly allocated) modified buffer.
886 * or NULL in case something went wrong
888 *********************************************************************/
889 char *pcrs_filter_response(struct client_state *csp)
892 int size = csp->iob->eod - csp->iob->cur;
894 char *old = csp->iob->cur, *new = NULL;
897 struct file_list *fl;
898 struct re_filterfile_spec *b;
900 /* Sanity first ;-) */
906 if ( ( NULL == (fl = csp->rlist) ) || ( NULL == (b = fl->f) ) )
908 log_error(LOG_LEVEL_ERROR, "Unable to get current state of regexp filtering.");
912 log_error(LOG_LEVEL_RE_FILTER, "re_filtering %s%s (size %d) ...",
913 csp->http->hostport, csp->http->path, size);
915 /* Apply all jobs from the joblist */
916 for (job = b->joblist; NULL != job; job = job->next)
918 hits += pcrs_execute(job, old, size, &new, &size);
919 if (old != csp->iob->cur) free(old);
923 log_error(LOG_LEVEL_RE_FILTER, " produced %d hits (new size %d).", hits, size);
925 csp->content_length = size;
927 /* fwiw, reset the iob */
934 /*********************************************************************
936 * Function : gif_deanimate_response
938 * Description : Deanimate the GIF image that has been accumulated in
939 * csp->iob->buf and set csp->content_length to the modified
943 * 1 : csp = Current client state (buffers, headers, etc...)
945 * Returns : a pointer to the (newly allocated) modified buffer.
946 * or NULL in case something went wrong.
948 *********************************************************************/
949 char *gif_deanimate_response(struct client_state *csp)
951 struct binbuffer *in, *out;
953 int size = csp->iob->eod - csp->iob->cur;
955 if ( (NULL == (in = (struct binbuffer *)zalloc(sizeof *in )))
956 || (NULL == (out = (struct binbuffer *)zalloc(sizeof *out))) )
958 log_error(LOG_LEVEL_DEANIMATE, "failed! (No Mem!)");
962 in->buffer = csp->iob->cur;
965 if (gif_deanimate(in, out))
967 log_error(LOG_LEVEL_DEANIMATE, "failed! (size %d)", size);
974 log_error(LOG_LEVEL_DEANIMATE, "Success! GIF shrunk from %d bytes to %d.", size, out->offset);
975 csp->content_length = out->offset;
985 /*********************************************************************
987 * Function : url_actions
989 * Description : Gets the actions for this URL.
992 * 1 : http = http_request request for blocked URLs
993 * 2 : csp = Current client state (buffers, headers, etc...)
997 *********************************************************************/
998 void url_actions(struct http_request *http,
999 struct client_state *csp)
1001 struct file_list *fl;
1002 struct url_actions *b;
1004 init_current_action(csp->action);
1006 if (((fl = csp->actions_list) == NULL) || ((b = fl->f) == NULL))
1011 apply_url_actions(csp->action, http, b);
1016 /*********************************************************************
1018 * Function : apply_url_actions
1020 * Description : Applies a list of URL actions.
1023 * 1 : action = Destination.
1024 * 2 : http = Current URL
1025 * 3 : b = list of URL actions to apply
1029 *********************************************************************/
1030 void apply_url_actions(struct current_action_spec *action,
1031 struct http_request *http,
1032 struct url_actions *b)
1034 struct url_spec url[1];
1038 /* Should never happen */
1042 *url = dsplit(http->host);
1044 /* if splitting the domain fails, punt */
1045 if (url->dbuf == NULL)
1050 for (b = b->next; NULL != b; b = b->next)
1052 if ((b->url->port == 0) || (b->url->port == http->port))
1054 if ((b->url->domain[0] == '\0') || (domaincmp(b->url, url) == 0))
1056 if ((b->url->path == NULL) ||
1058 (regexec(b->url->preg, http->path, 0, NULL, 0) == 0)
1060 (strncmp(b->url->path, http->path, b->url->pathlen) == 0)
1064 merge_current_action(action, b->action);
1075 /*********************************************************************
1077 * Function : forward_url
1079 * Description : Should we forward this to another proxy?
1082 * 1 : http = http_request request for current URL
1083 * 2 : csp = Current client state (buffers, headers, etc...)
1085 * Returns : Pointer to forwarding information.
1087 *********************************************************************/
1088 const struct forward_spec * forward_url(struct http_request *http,
1089 struct client_state *csp)
1091 static const struct forward_spec fwd_default[1] = { 0 }; /* All zeroes */
1092 struct forward_spec *fwd = csp->config->forward;
1093 struct url_spec url[1];
1097 return(fwd_default);
1100 *url = dsplit(http->host);
1102 /* if splitting the domain fails, punt */
1103 if (url->dbuf == NULL)
1105 return(fwd_default);
1110 if ((fwd->url->port == 0) || (fwd->url->port == http->port))
1112 if ((fwd->url->domain[0] == '\0') || (domaincmp(fwd->url, url) == 0))
1114 if ((fwd->url->path == NULL) ||
1116 (regexec(fwd->url->preg, http->path, 0, NULL, 0) == 0)
1118 (strncmp(fwd->url->path, http->path, fwd->url->pathlen) == 0)
1133 return(fwd_default);
1138 /*********************************************************************
1142 * Description : Takes a domain and returns a pointer to a url_spec
1143 * structure populated with dbuf, dcnt and dvec. The
1144 * other fields in the structure that is returned are zero.
1147 * 1 : domain = a URL address
1149 * Returns : url_spec structure populated with dbuf, dcnt and dvec.
1151 *********************************************************************/
1152 struct url_spec dsplit(char *domain)
1154 struct url_spec ret[1];
1155 char *v[BUFFER_SIZE];
1159 memset(ret, '\0', sizeof(*ret));
1161 if (domain[strlen(domain) - 1] == '.')
1163 ret->unanchored |= ANCHOR_RIGHT;
1166 if (domain[0] == '.')
1168 ret->unanchored |= ANCHOR_LEFT;
1171 ret->dbuf = strdup(domain);
1173 /* map to lower case */
1174 for (p = ret->dbuf; *p ; p++) *p = tolower(*p);
1176 /* split the domain name into components */
1177 ret->dcnt = ssplit(ret->dbuf, ".", v, SZ(v), 1, 1);
1181 memset(ret, '\0', sizeof(ret));
1185 /* save a copy of the pointers in dvec */
1186 size = ret->dcnt * sizeof(*ret->dvec);
1188 if ((ret->dvec = (char **)malloc(size)))
1190 memcpy(ret->dvec, v, size);
1198 /*********************************************************************
1200 * Function : simple_domaincmp
1202 * Description : Domain-wise Compare fqdn's. The comparison is
1203 * both left- and right-anchored. The individual
1204 * domain names are compared with simplematch().
1205 * This is only used by domaincmp.
1208 * 1 : pv = array of patterns to compare
1209 * 2 : fv = array of domain components to compare
1210 * 3 : len = length of the arrays (both arrays are the
1211 * same length - if they weren't, it couldn't
1212 * possibly be a match).
1214 * Returns : 0 => domains are equivalent, else no match.
1216 *********************************************************************/
1217 static int simple_domaincmp(char **pv, char **fv, int len)
1221 for (n = 0; n < len; n++)
1223 if (simplematch(pv[n], fv[n]))
1234 /*********************************************************************
1236 * Function : domaincmp
1238 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
1239 * pattern->unachored, the comparison is un-, left-,
1240 * right-anchored, or both.
1241 * The individual domain names are compared with
1245 * 1 : pattern = a domain that may contain a '*' as a wildcard.
1246 * 2 : fqdn = domain name against which the patterns are compared.
1248 * Returns : 0 => domains are equivalent, else no match.
1250 *********************************************************************/
1251 int domaincmp(struct url_spec *pattern, struct url_spec *fqdn)
1253 char **pv, **fv; /* vectors */
1255 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
1257 plen = pattern->dcnt;
1262 /* fqdn is too short to match this pattern */
1269 if (unanchored == ANCHOR_LEFT)
1274 * Convert this into a fully anchored pattern with
1275 * the fqdn and pattern the same length
1277 fv += (flen - plen); /* flen - plen >= 0 due to check above */
1278 return simple_domaincmp(pv, fv, plen);
1280 else if (unanchored == 0)
1282 /* Fully anchored, check length */
1287 return simple_domaincmp(pv, fv, plen);
1289 else if (unanchored == ANCHOR_RIGHT)
1291 /* Left anchored, ignore all extra in fqdn */
1292 return simple_domaincmp(pv, fv, plen);
1298 int maxn = flen - plen;
1299 for (n = 0; n <= maxn; n++)
1301 if (!simple_domaincmp(pv, fv, plen))
1306 * Doesn't match from start of fqdn
1307 * Try skipping first part of fqdn