1 const char filters_rcs[] = "$Id: filters.c,v 1.23 2001/07/23 13:40:12 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 1.23 2001/07/23 13:40:12 oes
42 * Fixed bug that caused document body to be dropped when pcrs joblist was empty.
44 * Revision 1.22 2001/07/18 12:29:34 oes
45 * - Made gif_deanimate_response respect
46 * csp->action->string[ACTION_STRING_DEANIMATE]
49 * Revision 1.21 2001/07/13 13:59:53 oes
50 * - Introduced gif_deanimate_response which shares the
51 * generic content modification interface of pcrs_filter_response
52 * and acts as a wrapper to deanimate.c:gif_deanimate()
53 * - Renamed re_process_buffer to pcrs_filter_response
54 * - pcrs_filter_response now returns NULL on failiure
55 * - Removed all #ifdef PCRS
57 * Revision 1.20 2001/07/01 17:01:04 oes
58 * Added comments and missing return statement in is_untrusted_url()
60 * Revision 1.19 2001/06/29 21:45:41 oes
61 * Indentation, CRLF->LF, Tab-> Space
63 * Revision 1.18 2001/06/29 13:27:38 oes
64 * - Cleaned up, renamed and reorderd functions
65 * and improved comments
68 * - Ported to CGI platform. Now delivers
69 * http_response or NULL
70 * - Unified HTML and GIF generation (moved image detection
71 * and GIF generation here from jcc.c:chat())
72 * - Fixed HTTP status to:
73 * - 403 (Forbidden) for the "blocked" HTML message
74 * - 200 (OK) for GIF answers
75 * - 302 (Redirect) for redirect to GIF
78 * - Ported to CGI platform. Now delivers
79 * http_response or NULL
80 * - Separated detection of untrusted URL into
81 * (bool)is_untrusted_url
82 * - Added enforcement of untrusted requests
84 * - Moved redirect_url() from cgi.c to here
85 * and ported it to the CGI platform
87 * - Removed logentry from cancelled commit
89 * Revision 1.17 2001/06/09 10:55:28 jongfoster
90 * Changing BUFSIZ ==> BUFFER_SIZE
92 * Revision 1.16 2001/06/07 23:10:26 jongfoster
93 * Allowing unanchored domain patterns to back off and retry
94 * if they partially match. Optimized right-anchored patterns.
95 * Moving ACL and forward files into config file.
96 * Replacing struct gateway with struct forward_spec
98 * Revision 1.15 2001/06/03 19:12:00 oes
99 * extracted-CGI relevant stuff
101 * Revision 1.14 2001/06/01 10:30:55 oes
102 * Added optional left-anchoring to domaincmp
104 * Revision 1.13 2001/05/31 21:21:30 jongfoster
105 * Permissionsfile / actions file changes:
106 * - Changed "permission" to "action" throughout
107 * - changes to file format to allow string parameters
108 * - Moved helper functions to actions.c
110 * Revision 1.12 2001/05/31 17:35:20 oes
112 * - Enhanced domain part globbing with infix and prefix asterisk
113 * matching and optional unanchored operation
115 * Revision 1.11 2001/05/29 11:53:23 oes
116 * "See why" link added to "blocked" page
118 * Revision 1.10 2001/05/29 09:50:24 jongfoster
119 * Unified blocklist/imagelist/permissionslist.
120 * File format is still under discussion, but the internal changes
123 * Also modified interceptor behaviour:
124 * - We now intercept all URLs beginning with one of the following
125 * prefixes (and *only* these prefixes):
127 * * http://ijbswa.sf.net/config/
128 * * http://ijbswa.sourceforge.net/config/
129 * - New interceptors "home page" - go to http://i.j.b/ to see it.
130 * - Internal changes so that intercepted and fast redirect pages
131 * are not replaced with an image.
132 * - Interceptors now have the option to send a binary page direct
133 * to the client. (i.e. ijb-send-banner uses this)
134 * - Implemented show-url-info interceptor. (Which is why I needed
135 * the above interceptors changes - a typical URL is
136 * "http://i.j.b/show-url-info?url=www.somesite.com/banner.gif".
137 * The previous mechanism would not have intercepted that, and
138 * if it had been intercepted then it then it would have replaced
141 * Revision 1.9 2001/05/27 22:17:04 oes
143 * - re_process_buffer no longer writes the modified buffer
144 * to the client, which was very ugly. It now returns the
145 * buffer, which it is then written by chat.
147 * - content_length now adjusts the Content-Length: header
148 * for modified documents rather than crunch()ing it.
149 * (Length info in csp->content_length, which is 0 for
150 * unmodified documents)
152 * - For this to work, sed() is called twice when filtering.
154 * Revision 1.8 2001/05/26 17:13:28 jongfoster
155 * Filled in a function comment.
157 * Revision 1.7 2001/05/26 15:26:15 jongfoster
158 * ACL feature now provides more security by immediately dropping
159 * connections from untrusted hosts.
161 * Revision 1.6 2001/05/26 00:28:36 jongfoster
162 * Automatic reloading of config file.
163 * Removed obsolete SIGHUP support (Unix) and Reload menu option (Win32).
164 * Most of the global variables have been moved to a new
165 * struct configuration_spec, accessed through csp->config->globalname
166 * Most of the globals remaining are used by the Win32 GUI.
168 * Revision 1.5 2001/05/25 22:34:30 jongfoster
171 * Revision 1.4 2001/05/22 18:46:04 oes
173 * - Enabled filtering banners by size rather than URL
174 * by adding patterns that replace all standard banner
175 * sizes with the "Junkbuster" gif to the re_filterfile
177 * - Enabled filtering WebBugs by providing a pattern
178 * which kills all 1x1 images
180 * - Added support for PCRE_UNGREEDY behaviour to pcrs,
181 * which is selected by the (nonstandard and therefore
182 * capital) letter 'U' in the option string.
183 * It causes the quantifiers to be ungreedy by default.
184 * Appending a ? turns back to greedy (!).
186 * - Added a new interceptor ijb-send-banner, which
187 * sends back the "Junkbuster" gif. Without imagelist or
188 * MSIE detection support, or if tinygif = 1, or the
189 * URL isn't recognized as an imageurl, a lame HTML
190 * explanation is sent instead.
192 * - Added new feature, which permits blocking remote
193 * script redirects and firing back a local redirect
195 * The feature is conditionally compiled, i.e. it
196 * can be disabled with --disable-fast-redirects,
197 * plus it must be activated by a "fast-redirects"
198 * line in the config file, has its own log level
199 * and of course wants to be displayed by show-proxy-args
200 * Note: Boy, all the #ifdefs in 1001 locations and
201 * all the fumbling with configure.in and acconfig.h
202 * were *way* more work than the feature itself :-(
204 * - Because a generic redirect template was needed for
205 * this, tinygif = 3 now uses the same.
207 * - Moved GIFs, and other static HTTP response templates
212 * - Removed some >400 CRs again (Jon, you really worked
215 * Revision 1.3 2001/05/20 16:44:47 jongfoster
216 * Removing last hardcoded JunkBusters.com URLs.
218 * Revision 1.2 2001/05/20 01:21:20 jongfoster
219 * Version 2.9.4 checkin.
220 * - Merged popupfile and cookiefile, and added control over PCRS
221 * filtering, in new "permissionsfile".
222 * - Implemented LOG_LEVEL_FATAL, so that if there is a configuration
223 * file error you now get a message box (in the Win32 GUI) rather
224 * than the program exiting with no explanation.
225 * - Made killpopup use the PCRS MIME-type checking and HTTP-header
227 * - Removed tabs from "config"
228 * - Moved duplicated url parsing code in "loaders.c" to a new funcition.
229 * - Bumped up version number.
231 * Revision 1.1.1.1 2001/05/15 13:58:52 oes
232 * Initial import of version 2.9.3 source tree
235 *********************************************************************/
241 #include <sys/types.h>
248 #include <netinet/in.h>
250 #include <winsock2.h>
257 #include "showargs.h"
261 #include "jbsockets.h"
263 #include "jbsockets.h"
264 #include "miscutil.h"
268 #include "deanimate.h"
274 const char filters_h_rcs[] = FILTERS_H_VERSION;
276 /* Fix a problem with Solaris. There should be no effect on other
278 * Solaris's isspace() is a macro which uses it's argument directly
279 * as an array index. Therefore we need to make sure that high-bit
280 * characters generate +ve values, and ideally we also want to make
281 * the argument match the declared parameter type of "int".
283 #define ijb_isdigit(__X) isdigit((int)(unsigned char)(__X))
287 /*********************************************************************
289 * Function : block_acl
291 * Description : Block this request?
292 * Decide yes or no based on ACL file.
295 * 1 : dst = The proxy or gateway address this is going to.
296 * Or NULL to check all possible targets.
297 * 2 : csp = Current client state (buffers, headers, etc...)
298 * Also includes the client IP address.
300 * Returns : 0 = FALSE (don't block) and 1 = TRUE (do block)
302 *********************************************************************/
303 int block_acl(struct access_control_addr *dst, struct client_state *csp)
305 struct access_control_list *acl = csp->config->acl;
307 /* if not using an access control list, then permit the connection */
313 /* search the list */
316 if ((csp->ip_addr_long & acl->src->mask) == acl->src->addr)
320 /* Just want to check if they have any access */
321 if (acl->action == ACL_PERMIT)
326 else if ( ((dst->addr & acl->dst->mask) == acl->dst->addr)
327 && ((dst->port == acl->dst->port) || (acl->dst->port == 0)))
329 if (acl->action == ACL_PERMIT)
347 /*********************************************************************
349 * Function : acl_addr
351 * Description : Called from `load_aclfile' to parse an ACL address.
354 * 1 : aspec = String specifying ACL address.
355 * 2 : aca = struct access_control_addr to fill in.
357 * Returns : 0 => Ok, everything else is an error.
359 *********************************************************************/
360 int acl_addr(char *aspec, struct access_control_addr *aca)
362 int i, masklength, port;
368 if ((p = strchr(aspec, '/')))
372 if (ijb_isdigit(*p) == 0)
376 masklength = atoi(p);
379 if ((masklength < 0) || (masklength > 32))
384 if ((p = strchr(aspec, ':')))
388 if (ijb_isdigit(*p) == 0)
397 aca->addr = ntohl(resolve_hostname_to_ip(aspec));
401 log_error(LOG_LEVEL_ERROR, "can't resolve address for %s", aspec);
405 /* build the netmask */
407 for (i=1; i <= masklength ; i++)
409 aca->mask |= (1 << (32 - i));
412 /* now mask off the host portion of the ip address
413 * (i.e. save on the network portion of the address).
415 aca->addr = aca->addr & aca->mask;
420 #endif /* def ACL_FILES */
423 /*********************************************************************
425 * Function : block_url
427 * Description : Called from `chat'. Check to see if we need to block this.
430 * 1 : csp = Current client state (buffers, headers, etc...)
432 * Returns : NULL => unblocked, else HTTP block response
434 *********************************************************************/
435 struct http_response *block_url(struct client_state *csp)
438 struct http_response *rsp;
439 struct map *exports = NULL;
442 * If it's not blocked, don't block it ;-)
444 if ((csp->action->flags & ACTION_BLOCK) == 0)
450 * Else, prepare a response
452 if (NULL == ( rsp = (struct http_response *)zalloc(sizeof(*rsp))))
458 * If it's an image-url, send back an image or redirect
459 * as specified by the relevant +image action
461 #ifdef IMAGE_BLOCKING
462 if (((csp->action->flags & ACTION_IMAGE_BLOCKER) != 0)
465 /* determine HOW images should be blocked */
466 p = csp->action->string[ACTION_STRING_IMAGE_BLOCKER];
468 /* and handle accordingly: */
469 if ((p == NULL) || (0 == strcmpic(p, "logo")))
471 rsp->body = bindup(JBGIF, sizeof(JBGIF));
472 rsp->content_length = sizeof(JBGIF);
473 enlist_unique_header(rsp->headers, "Content-Type", "image/gif");
476 else if (0 == strcmpic(p, "blank"))
478 rsp->body = bindup(BLANKGIF, sizeof(BLANKGIF));
479 rsp->content_length = sizeof(BLANKGIF);
480 enlist_unique_header(rsp->headers, "Content-Type", "image/gif");
485 rsp->status = strdup("302 Local Redirect from Junkbuster");
486 enlist_unique_header(rsp->headers, "Location", p);
490 #endif /* def IMAGE_BLOCKING */
493 * Else, generate an HTML "blocked" message:
497 exports = default_exports(csp, NULL);
499 exports = map(exports, "force-prefix", 1, FORCE_PREFIX, 1);
501 exports = map_block_killer(exports, "force-support");
502 #endif /* ndef FORCE_LOAD */
504 exports = map(exports, "hostport", 1, csp->http->hostport, 1);
505 exports = map(exports, "hostport-html", 1, html_encode(csp->http->hostport), 0);
506 exports = map(exports, "path", 1, csp->http->path, 1);
507 exports = map(exports, "path-html", 1, html_encode(csp->http->path), 0);
509 rsp->body = fill_template(csp, "blocked", exports);
513 * Workaround for stupid Netscape bug which prevents
514 * pages from being displayed if loading a referenced
515 * JavaScript or style sheet fails. So make it appear
516 * as if it succeeded.
518 if (csp->http->user_agent && !strncmpic(csp->http->user_agent, "mozilla", 7))
520 rsp->status = strdup("200 Request for blocked URL");
524 rsp->status = strdup("404 Request for blocked URL");
529 return(finish_http_response(rsp));
535 /*********************************************************************
537 * Function : trust_url FIXME: I should be called distrust_url
539 * Description : Calls is_untrusted_url to determine if the URL is trusted
540 * and if not, returns a HTTP 304 response with a reject message.
543 * 1 : csp = Current client state (buffers, headers, etc...)
545 * Returns : NULL => trusted, else http_response.
547 *********************************************************************/
548 struct http_response *trust_url(struct client_state *csp)
550 struct http_response *rsp;
551 struct map *exports = NULL;
552 char buf[BUFFER_SIZE], *p = NULL;
553 struct url_spec **tl, *t;
556 * Don't bother to work on trusted URLs
558 if (!is_untrusted_url(csp))
564 * Else, prepare a response:
566 if (NULL == ( rsp = (struct http_response *)zalloc(sizeof(*rsp))))
570 exports = default_exports(csp, NULL);
573 * Export the host, port, and referrer information
575 exports = map(exports, "hostport", 1, csp->http->hostport, 1);
576 exports = map(exports, "path", 1, csp->http->path, 1);
577 exports = map(exports, "hostport-html", 1, html_encode(csp->http->hostport), 0);
578 exports = map(exports, "path-html", 1, html_encode(csp->http->path), 0);
580 if (csp->referrer && strlen(csp->referrer) > 9)
582 exports = map(exports, "referrer", 1, csp->referrer + 9, 1);
583 exports = map(exports, "referrer-html", 1, html_encode(csp->referrer + 9), 0);
587 exports = map(exports, "referrer", 1, "unknown", 1);
588 exports = map(exports, "referrer-html", 1, "unknown", 1);
592 * Export the trust list
594 for (tl = csp->config->trust_list; (t = *tl) ; tl++)
596 sprintf(buf, "<li>%s</li>\n", t->spec);
599 exports = map(exports, "trusted-referrers", 1, p, 0);
603 * Export the trust info, if available
605 if (csp->config->trust_info->next)
609 for (l = csp->config->trust_info->next; l ; l = l->next)
611 sprintf(buf, "<li> <a href=%s>%s</a><br>\n",l->str, l->str);
614 exports = map(exports, "trust-info", 1, p, 0);
618 exports = map_block_killer(exports, "have-trust-info");
622 * Export the force prefix or the force conditional block killer
625 exports = map(exports, "force-prefix", 1, FORCE_PREFIX, 1);
627 exports = map_block_killer(exports, "force-support");
628 #endif /* ndef FORCE_LOAD */
633 rsp->body = fill_template(csp, "untrusted", exports);
636 return(finish_http_response(rsp));
639 #endif /* def TRUST_FILES */
642 #ifdef FAST_REDIRECTS
643 /*********************************************************************
645 * Function : redirect_url
647 * Description : Checks for redirection URLs and returns a HTTP redirect
648 * to the destination URL, if necessary
651 * 1 : csp = Current client state (buffers, headers, etc...)
653 * Returns : NULL if URL was clean, HTTP redirect otherwise.
655 *********************************************************************/
656 struct http_response *redirect_url(struct client_state *csp)
659 struct http_response *rsp;
661 p = q = csp->http->path;
662 log_error(LOG_LEVEL_REDIRECTS, "checking path for redirects: %s", p);
665 * find the last URL encoded in the request
667 while (p = strstr(p, "http://"))
673 * if there was any, generate and return a HTTP redirect
675 if (q != csp->http->path)
677 log_error(LOG_LEVEL_REDIRECTS, "redirecting to: %s", q);
679 if (NULL == ( rsp = zalloc(sizeof(*rsp))))
684 rsp->status = strdup("302 Local Redirect from Junkbuster");
685 enlist_unique_header(rsp->headers, "Location", q);
687 return(finish_http_response(rsp));
695 #endif /* def FAST_REDIRECTS */
698 #ifdef IMAGE_BLOCKING
699 /*********************************************************************
701 * Function : is_imageurl
703 * Description : Given a URL, decide whether it is an image or not,
704 * using either the info from a previous +image action
705 * or, #ifdef DETECT_MSIE_IMAGES, the info from the
706 * browser's accept header.
709 * 1 : csp = Current client state (buffers, headers, etc...)
711 * Returns : True (nonzero) if URL is an image, false (0)
714 *********************************************************************/
715 int is_imageurl(struct client_state *csp)
717 #ifdef DETECT_MSIE_IMAGES
718 if ((csp->accept_types
719 & (ACCEPT_TYPE_IS_MSIE|ACCEPT_TYPE_MSIE_IMAGE|ACCEPT_TYPE_MSIE_HTML))
720 == (ACCEPT_TYPE_IS_MSIE|ACCEPT_TYPE_MSIE_IMAGE))
724 else if ((csp->accept_types
725 & (ACCEPT_TYPE_IS_MSIE|ACCEPT_TYPE_MSIE_IMAGE|ACCEPT_TYPE_MSIE_HTML))
726 == (ACCEPT_TYPE_IS_MSIE|ACCEPT_TYPE_MSIE_HTML))
732 return ((csp->action->flags & ACTION_IMAGE) != 0);
735 #endif /* def IMAGE_BLOCKING */
739 /*********************************************************************
741 * Function : is_untrusted_url
743 * Description : Should we "distrust" this URL (and block it)?
745 * Yes if it matches a line in the trustfile, or if the
746 * referrer matches a line starting with "+" in the
751 * 1 : csp = Current client state (buffers, headers, etc...)
753 * Returns : 0 => trusted, 1 => untrusted
755 *********************************************************************/
756 int is_untrusted_url(struct client_state *csp)
758 struct file_list *fl;
759 struct block_spec *b;
760 struct url_spec url[1], **tl, *t;
761 struct http_request rhttp[1];
765 * If we don't have a trustlist, we trust everybody
767 if (((fl = csp->tlist) == NULL) || ((b = fl->f) == NULL))
774 * Do we trust the request URL itself?
776 *url = dsplit(csp->http->host);
778 /* if splitting the domain fails, punt */
779 if (url->dbuf == NULL) return(0);
781 memset(rhttp, '\0', sizeof(*rhttp));
783 for (b = b->next; b ; b = b->next)
785 if ((b->url->port == 0) || (b->url->port == csp->http->port))
787 if ((b->url->domain[0] == '\0') || (domaincmp(b->url, url) == 0))
789 if ((b->url->path == NULL) ||
791 (regexec(b->url->preg, csp->http->path, 0, NULL, 0) == 0)
793 (strncmp(b->url->path, csp->http->path, b->url->pathlen) == 0)
800 if (b->reject == 0) return(0);
811 if ((csp->referrer == NULL)|| (strlen(csp->referrer) <= 9))
813 /* no referrer was supplied */
817 /* forge a URL from the referrer so we can use
818 * convert_url() to parse it into its components.
822 p = strsav(p, "GET ");
823 p = strsav(p, csp->referrer + 9); /* skip over "Referer: " */
824 p = strsav(p, " HTTP/1.0");
826 parse_http_request(p, rhttp, csp);
829 if (rhttp->cmd == NULL)
836 * If not, do we maybe trust its referrer?
838 *url = dsplit(rhttp->host);
840 /* if splitting the domain fails, punt */
841 if (url->dbuf == NULL) return(1);
843 for (tl = csp->config->trust_list; (t = *tl) ; tl++)
845 if ((t->port == 0) || (t->port == rhttp->port))
847 if ((t->domain[0] == '\0') || domaincmp(t, url) == 0)
849 if ((t->path == NULL) ||
851 (regexec(t->preg, rhttp->path, 0, NULL, 0) == 0)
853 (strncmp(t->path, rhttp->path, t->pathlen) == 0)
857 /* if the URL's referrer is from a trusted referrer, then
858 * add the target spec to the trustfile as an unblocked
859 * domain and return NULL (which means it's OK).
867 if ((fp = fopen(csp->config->trustfile, "a")))
872 h = strsav(h, csp->http->hostport);
878 /* since this path points into a user's home space
879 * be sure to include this spec in the trustfile.
881 if ((p = strchr(p, '/')))
884 h = strsav(h, csp->http->path); /* FIXME: p?! */
889 fprintf(fp, "%s\n", h);
900 #endif /* def TRUST_FILES */
903 /*********************************************************************
905 * Function : pcrs_filter_response
907 * Description : Apply all the pcrs jobs from the joblist (re_filterfile)
908 * to the text buffer that's been accumulated in
909 * csp->iob->buf and set csp->content_length to the modified
913 * 1 : csp = Current client state (buffers, headers, etc...)
915 * Returns : a pointer to the (newly allocated) modified buffer.
916 * or NULL in case something went wrong
918 *********************************************************************/
919 char *pcrs_filter_response(struct client_state *csp)
922 int size = csp->iob->eod - csp->iob->cur;
924 char *old = csp->iob->cur, *new = NULL;
927 struct file_list *fl;
928 struct re_filterfile_spec *b;
930 /* Sanity first ;-) */
936 if ( ( NULL == (fl = csp->rlist) ) || ( NULL == (b = fl->f) ) )
938 log_error(LOG_LEVEL_ERROR, "Unable to get current state of regexp filtering.");
942 if ( NULL == b->joblist )
944 log_error(LOG_LEVEL_RE_FILTER, "Empty joblist. Nothing to do.");
948 log_error(LOG_LEVEL_RE_FILTER, "re_filtering %s%s (size %d) ...",
949 csp->http->hostport, csp->http->path, size);
951 /* Apply all jobs from the joblist */
952 for (job = b->joblist; NULL != job; job = job->next)
954 hits += pcrs_execute(job, old, size, &new, &size);
955 if (old != csp->iob->cur) free(old);
959 log_error(LOG_LEVEL_RE_FILTER, " produced %d hits (new size %d).", hits, size);
961 csp->content_length = size;
963 /* fwiw, reset the iob */
970 /*********************************************************************
972 * Function : gif_deanimate_response
974 * Description : Deanimate the GIF image that has been accumulated in
975 * csp->iob->buf and set csp->content_length to the modified
979 * 1 : csp = Current client state (buffers, headers, etc...)
981 * Returns : a pointer to the (newly allocated) modified buffer.
982 * or NULL in case something went wrong.
984 *********************************************************************/
985 char *gif_deanimate_response(struct client_state *csp)
987 struct binbuffer *in, *out;
989 int size = csp->iob->eod - csp->iob->cur;
991 if ( (NULL == (in = (struct binbuffer *)zalloc(sizeof *in )))
992 || (NULL == (out = (struct binbuffer *)zalloc(sizeof *out))) )
994 log_error(LOG_LEVEL_DEANIMATE, "failed! (no mem)");
998 in->buffer = csp->iob->cur;
1001 if (gif_deanimate(in, out, strncmp("last", csp->action->string[ACTION_STRING_DEANIMATE], 4)))
1003 log_error(LOG_LEVEL_DEANIMATE, "failed! (gif parsing)");
1010 log_error(LOG_LEVEL_DEANIMATE, "Success! GIF shrunk from %d bytes to %d.", size, out->offset);
1011 csp->content_length = out->offset;
1021 /*********************************************************************
1023 * Function : url_actions
1025 * Description : Gets the actions for this URL.
1028 * 1 : http = http_request request for blocked URLs
1029 * 2 : csp = Current client state (buffers, headers, etc...)
1033 *********************************************************************/
1034 void url_actions(struct http_request *http,
1035 struct client_state *csp)
1037 struct file_list *fl;
1038 struct url_actions *b;
1040 init_current_action(csp->action);
1042 if (((fl = csp->actions_list) == NULL) || ((b = fl->f) == NULL))
1047 apply_url_actions(csp->action, http, b);
1052 /*********************************************************************
1054 * Function : apply_url_actions
1056 * Description : Applies a list of URL actions.
1059 * 1 : action = Destination.
1060 * 2 : http = Current URL
1061 * 3 : b = list of URL actions to apply
1065 *********************************************************************/
1066 void apply_url_actions(struct current_action_spec *action,
1067 struct http_request *http,
1068 struct url_actions *b)
1070 struct url_spec url[1];
1074 /* Should never happen */
1078 *url = dsplit(http->host);
1080 /* if splitting the domain fails, punt */
1081 if (url->dbuf == NULL)
1086 for (b = b->next; NULL != b; b = b->next)
1088 if ((b->url->port == 0) || (b->url->port == http->port))
1090 if ((b->url->domain[0] == '\0') || (domaincmp(b->url, url) == 0))
1092 if ((b->url->path == NULL) ||
1094 (regexec(b->url->preg, http->path, 0, NULL, 0) == 0)
1096 (strncmp(b->url->path, http->path, b->url->pathlen) == 0)
1100 merge_current_action(action, b->action);
1111 /*********************************************************************
1113 * Function : forward_url
1115 * Description : Should we forward this to another proxy?
1118 * 1 : http = http_request request for current URL
1119 * 2 : csp = Current client state (buffers, headers, etc...)
1121 * Returns : Pointer to forwarding information.
1123 *********************************************************************/
1124 const struct forward_spec * forward_url(struct http_request *http,
1125 struct client_state *csp)
1127 static const struct forward_spec fwd_default[1] = { 0 }; /* All zeroes */
1128 struct forward_spec *fwd = csp->config->forward;
1129 struct url_spec url[1];
1133 return(fwd_default);
1136 *url = dsplit(http->host);
1138 /* if splitting the domain fails, punt */
1139 if (url->dbuf == NULL)
1141 return(fwd_default);
1146 if ((fwd->url->port == 0) || (fwd->url->port == http->port))
1148 if ((fwd->url->domain[0] == '\0') || (domaincmp(fwd->url, url) == 0))
1150 if ((fwd->url->path == NULL) ||
1152 (regexec(fwd->url->preg, http->path, 0, NULL, 0) == 0)
1154 (strncmp(fwd->url->path, http->path, fwd->url->pathlen) == 0)
1169 return(fwd_default);
1174 /*********************************************************************
1178 * Description : Takes a domain and returns a pointer to a url_spec
1179 * structure populated with dbuf, dcnt and dvec. The
1180 * other fields in the structure that is returned are zero.
1183 * 1 : domain = a URL address
1185 * Returns : url_spec structure populated with dbuf, dcnt and dvec.
1187 *********************************************************************/
1188 struct url_spec dsplit(char *domain)
1190 struct url_spec ret[1];
1191 char *v[BUFFER_SIZE];
1195 memset(ret, '\0', sizeof(*ret));
1197 if (domain[strlen(domain) - 1] == '.')
1199 ret->unanchored |= ANCHOR_RIGHT;
1202 if (domain[0] == '.')
1204 ret->unanchored |= ANCHOR_LEFT;
1207 ret->dbuf = strdup(domain);
1209 /* map to lower case */
1210 for (p = ret->dbuf; *p ; p++) *p = tolower(*p);
1212 /* split the domain name into components */
1213 ret->dcnt = ssplit(ret->dbuf, ".", v, SZ(v), 1, 1);
1217 memset(ret, '\0', sizeof(ret));
1221 /* save a copy of the pointers in dvec */
1222 size = ret->dcnt * sizeof(*ret->dvec);
1224 if ((ret->dvec = (char **)malloc(size)))
1226 memcpy(ret->dvec, v, size);
1234 /*********************************************************************
1236 * Function : simple_domaincmp
1238 * Description : Domain-wise Compare fqdn's. The comparison is
1239 * both left- and right-anchored. The individual
1240 * domain names are compared with simplematch().
1241 * This is only used by domaincmp.
1244 * 1 : pv = array of patterns to compare
1245 * 2 : fv = array of domain components to compare
1246 * 3 : len = length of the arrays (both arrays are the
1247 * same length - if they weren't, it couldn't
1248 * possibly be a match).
1250 * Returns : 0 => domains are equivalent, else no match.
1252 *********************************************************************/
1253 static int simple_domaincmp(char **pv, char **fv, int len)
1257 for (n = 0; n < len; n++)
1259 if (simplematch(pv[n], fv[n]))
1270 /*********************************************************************
1272 * Function : domaincmp
1274 * Description : Domain-wise Compare fqdn's. Governed by the bimap in
1275 * pattern->unachored, the comparison is un-, left-,
1276 * right-anchored, or both.
1277 * The individual domain names are compared with
1281 * 1 : pattern = a domain that may contain a '*' as a wildcard.
1282 * 2 : fqdn = domain name against which the patterns are compared.
1284 * Returns : 0 => domains are equivalent, else no match.
1286 *********************************************************************/
1287 int domaincmp(struct url_spec *pattern, struct url_spec *fqdn)
1289 char **pv, **fv; /* vectors */
1291 int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
1293 plen = pattern->dcnt;
1298 /* fqdn is too short to match this pattern */
1305 if (unanchored == ANCHOR_LEFT)
1310 * Convert this into a fully anchored pattern with
1311 * the fqdn and pattern the same length
1313 fv += (flen - plen); /* flen - plen >= 0 due to check above */
1314 return simple_domaincmp(pv, fv, plen);
1316 else if (unanchored == 0)
1318 /* Fully anchored, check length */
1323 return simple_domaincmp(pv, fv, plen);
1325 else if (unanchored == ANCHOR_RIGHT)
1327 /* Left anchored, ignore all extra in fqdn */
1328 return simple_domaincmp(pv, fv, plen);
1334 int maxn = flen - plen;
1335 for (n = 0; n <= maxn; n++)
1337 if (!simple_domaincmp(pv, fv, plen))
1342 * Doesn't match from start of fqdn
1343 * Try skipping first part of fqdn