1 const char filters_rcs[] = "$Id: filters.c,v 1.45 2002/03/08 16:47:50 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', `domain_split',
10 * `filter_popups', `forward_url', 'redirect_url',
11 * `ij_untrusted_url', `intercept_url', `pcrs_filter_respose',
12 * '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 1.45 2002/03/08 16:47:50 oes
42 * Added choice beween GIF and PNG built-in images
44 * Revision 1.44 2002/03/07 03:49:31 oes
45 * - Fixed compiler warnings etc
46 * - Changed built-in images from GIF to PNG
47 * (with regard to Unisys patent issue)
48 * - Added a 4x4 pattern PNG which is less intrusive
49 * than the logo but also clearly marks the deleted banners
51 * Revision 1.43 2002/01/22 23:51:59 jongfoster
52 * Replacing strsav() with the safer string_append().
54 * Adding missing html_encode() to error message generators. Where encoded
55 * and unencoded versions of a string were provided, removing the unencoded
58 * Revision 1.42 2002/01/17 21:00:32 jongfoster
59 * Moving all our URL and URL pattern parsing code to urlmatch.c.
61 * Using a single, simple url_match(pattern,url) function - rather than
62 * the 3-line match routine which was repeated all over the place.
64 * Renaming free_url to free_url_spec, since it frees a struct url_spec.
66 * Using parse_http_url() to parse URLs without faking a HTTP
67 * request line for parse_http_request().
69 * Revision 1.41 2001/11/13 00:14:07 jongfoster
70 * Fixing stupid bug now I've figured out what || means.
71 * (It always returns 0 or 1, not one of it's paramaters.)
73 * Revision 1.40 2001/10/26 17:37:55 oes
74 * - Re-enabled Netscape 200/404 bug workaround in block_url():
75 * - Removed OS/2 special case
76 * - Made block_url() independant from sed() having been run
77 * - Made trust_url independant from sed() having been run
78 * - Made is_imageurl independant from sed() having been run.
79 * It now checks User-Agent: and Accept: by itself.
82 * Revision 1.39 2001/10/25 03:40:48 david__schmidt
83 * Change in porting tactics: OS/2's EMX porting layer doesn't allow multiple
84 * threads to call select() simultaneously. So, it's time to do a real, live,
85 * native OS/2 port. See defines for __EMX__ (the porting layer) vs. __OS2__
86 * (native). Both versions will work, but using __OS2__ offers multi-threading.
88 * Revision 1.38 2001/10/23 21:32:33 jongfoster
89 * Adding error-checking to selected functions
91 * Revision 1.37 2001/10/22 15:33:56 david__schmidt
92 * Special-cased OS/2 out of the Netscape-abort-on-404-in-js problem in
93 * filters.c. Added a FIXME in front of the offending code. I'll gladly
94 * put in a better/more robust fix for all parties if one is presented...
95 * It seems that just returning 200 instead of 404 would pretty much fix
96 * it for everyone, but I don't know all the history of the problem.
98 * Revision 1.36 2001/10/10 16:44:16 oes
99 * Added match_portlist function
101 * Revision 1.35 2001/10/07 15:41:23 oes
102 * Replaced 6 boolean members of csp with one bitmap (csp->flags)
104 * New function remove_chunked_transfer_coding that strips chunked
105 * transfer coding to plain and is called by pcrs_filter_response
106 * and gif_deanimate_response if neccessary
108 * Improved handling of zero-change re_filter runs
110 * pcrs_filter_response and gif_deanimate_response now remove
111 * chunked transfer codeing before processing the body.
113 * Revision 1.34 2001/09/20 15:49:36 steudten
115 * Fix BUG: Change int size to size_t size in pcrs_filter_response().
116 * See cgi.c fill_template().
118 * Revision 1.33 2001/09/16 17:05:14 jongfoster
119 * Removing unused #include showarg.h
121 * Revision 1.32 2001/09/16 13:21:27 jongfoster
122 * Changes to use new list functions.
124 * Revision 1.31 2001/09/16 11:38:02 jongfoster
125 * Splitting fill_template() into 2 functions:
126 * template_load() loads the file
127 * template_fill() performs the PCRS regexps.
128 * This is because the CGI edit interface has a "table row"
129 * template which is used many times in the page - this
130 * change means it's only loaded from disk once.
132 * Revision 1.30 2001/09/16 11:00:10 jongfoster
133 * New function alloc_http_response, for symmetry with free_http_response
135 * Revision 1.29 2001/09/13 23:32:40 jongfoster
136 * Moving image data to cgi.c rather than cgi.h
137 * Fixing a GPF under Win32 (and any other OS that protects global
138 * constants from being written to).
140 * Revision 1.28 2001/09/10 10:18:51 oes
141 * Silenced compiler warnings
143 * Revision 1.27 2001/08/05 16:06:20 jongfoster
144 * Modifiying "struct map" so that there are now separate header and
145 * "map_entry" structures. This means that functions which modify a
146 * map no longer need to return a pointer to the modified map.
147 * Also, it no longer reverses the order of the entries (which may be
148 * important with some advanced template substitutions).
150 * Revision 1.26 2001/07/30 22:08:36 jongfoster
151 * Tidying up #defines:
152 * - All feature #defines are now of the form FEATURE_xxx
153 * - Permanently turned off WIN_GUI_EDIT
154 * - Permanently turned on WEBDAV and SPLIT_PROXY_ARGS
156 * Revision 1.25 2001/07/26 10:09:46 oes
157 * Made browser detection a little less naive
159 * Revision 1.24 2001/07/25 17:22:51 oes
160 * Added workaround for Netscape bug that prevents display of page when loading a component fails.
162 * Revision 1.23 2001/07/23 13:40:12 oes
163 * Fixed bug that caused document body to be dropped when pcrs joblist was empty.
165 * Revision 1.22 2001/07/18 12:29:34 oes
166 * - Made gif_deanimate_response respect
167 * csp->action->string[ACTION_STRING_DEANIMATE]
168 * - Logging cosmetics
170 * Revision 1.21 2001/07/13 13:59:53 oes
171 * - Introduced gif_deanimate_response which shares the
172 * generic content modification interface of pcrs_filter_response
173 * and acts as a wrapper to deanimate.c:gif_deanimate()
174 * - Renamed re_process_buffer to pcrs_filter_response
175 * - pcrs_filter_response now returns NULL on failiure
176 * - Removed all #ifdef PCRS
178 * Revision 1.20 2001/07/01 17:01:04 oes
179 * Added comments and missing return statement in is_untrusted_url()
181 * Revision 1.19 2001/06/29 21:45:41 oes
182 * Indentation, CRLF->LF, Tab-> Space
184 * Revision 1.18 2001/06/29 13:27:38 oes
185 * - Cleaned up, renamed and reorderd functions
186 * and improved comments
189 * - Ported to CGI platform. Now delivers
190 * http_response or NULL
191 * - Unified HTML and GIF generation (moved image detection
192 * and GIF generation here from jcc.c:chat())
193 * - Fixed HTTP status to:
194 * - 403 (Forbidden) for the "blocked" HTML message
195 * - 200 (OK) for GIF answers
196 * - 302 (Redirect) for redirect to GIF
199 * - Ported to CGI platform. Now delivers
200 * http_response or NULL
201 * - Separated detection of untrusted URL into
202 * (bool)is_untrusted_url
203 * - Added enforcement of untrusted requests
205 * - Moved redirect_url() from cgi.c to here
206 * and ported it to the CGI platform
208 * - Removed logentry from cancelled commit
210 * Revision 1.17 2001/06/09 10:55:28 jongfoster
211 * Changing BUFSIZ ==> BUFFER_SIZE
213 * Revision 1.16 2001/06/07 23:10:26 jongfoster
214 * Allowing unanchored domain patterns to back off and retry
215 * if they partially match. Optimized right-anchored patterns.
216 * Moving ACL and forward files into config file.
217 * Replacing struct gateway with struct forward_spec
219 * Revision 1.15 2001/06/03 19:12:00 oes
220 * extracted-CGI relevant stuff
222 * Revision 1.14 2001/06/01 10:30:55 oes
223 * Added optional left-anchoring to domaincmp
225 * Revision 1.13 2001/05/31 21:21:30 jongfoster
226 * Permissionsfile / actions file changes:
227 * - Changed "permission" to "action" throughout
228 * - changes to file format to allow string parameters
229 * - Moved helper functions to actions.c
231 * Revision 1.12 2001/05/31 17:35:20 oes
233 * - Enhanced domain part globbing with infix and prefix asterisk
234 * matching and optional unanchored operation
236 * Revision 1.11 2001/05/29 11:53:23 oes
237 * "See why" link added to "blocked" page
239 * Revision 1.10 2001/05/29 09:50:24 jongfoster
240 * Unified blocklist/imagelist/permissionslist.
241 * File format is still under discussion, but the internal changes
244 * Also modified interceptor behaviour:
245 * - We now intercept all URLs beginning with one of the following
246 * prefixes (and *only* these prefixes):
248 * * http://ijbswa.sf.net/config/
249 * * http://ijbswa.sourceforge.net/config/
250 * - New interceptors "home page" - go to http://i.j.b/ to see it.
251 * - Internal changes so that intercepted and fast redirect pages
252 * are not replaced with an image.
253 * - Interceptors now have the option to send a binary page direct
254 * to the client. (i.e. ijb-send-banner uses this)
255 * - Implemented show-url-info interceptor. (Which is why I needed
256 * the above interceptors changes - a typical URL is
257 * "http://i.j.b/show-url-info?url=www.somesite.com/banner.gif".
258 * The previous mechanism would not have intercepted that, and
259 * if it had been intercepted then it then it would have replaced
262 * Revision 1.9 2001/05/27 22:17:04 oes
264 * - re_process_buffer no longer writes the modified buffer
265 * to the client, which was very ugly. It now returns the
266 * buffer, which it is then written by chat.
268 * - content_length now adjusts the Content-Length: header
269 * for modified documents rather than crunch()ing it.
270 * (Length info in csp->content_length, which is 0 for
271 * unmodified documents)
273 * - For this to work, sed() is called twice when filtering.
275 * Revision 1.8 2001/05/26 17:13:28 jongfoster
276 * Filled in a function comment.
278 * Revision 1.7 2001/05/26 15:26:15 jongfoster
279 * ACL feature now provides more security by immediately dropping
280 * connections from untrusted hosts.
282 * Revision 1.6 2001/05/26 00:28:36 jongfoster
283 * Automatic reloading of config file.
284 * Removed obsolete SIGHUP support (Unix) and Reload menu option (Win32).
285 * Most of the global variables have been moved to a new
286 * struct configuration_spec, accessed through csp->config->globalname
287 * Most of the globals remaining are used by the Win32 GUI.
289 * Revision 1.5 2001/05/25 22:34:30 jongfoster
292 * Revision 1.4 2001/05/22 18:46:04 oes
294 * - Enabled filtering banners by size rather than URL
295 * by adding patterns that replace all standard banner
296 * sizes with the "Junkbuster" gif to the re_filterfile
298 * - Enabled filtering WebBugs by providing a pattern
299 * which kills all 1x1 images
301 * - Added support for PCRE_UNGREEDY behaviour to pcrs,
302 * which is selected by the (nonstandard and therefore
303 * capital) letter 'U' in the option string.
304 * It causes the quantifiers to be ungreedy by default.
305 * Appending a ? turns back to greedy (!).
307 * - Added a new interceptor ijb-send-banner, which
308 * sends back the "Junkbuster" gif. Without imagelist or
309 * MSIE detection support, or if tinygif = 1, or the
310 * URL isn't recognized as an imageurl, a lame HTML
311 * explanation is sent instead.
313 * - Added new feature, which permits blocking remote
314 * script redirects and firing back a local redirect
316 * The feature is conditionally compiled, i.e. it
317 * can be disabled with --disable-fast-redirects,
318 * plus it must be activated by a "fast-redirects"
319 * line in the config file, has its own log level
320 * and of course wants to be displayed by show-proxy-args
321 * Note: Boy, all the #ifdefs in 1001 locations and
322 * all the fumbling with configure.in and acconfig.h
323 * were *way* more work than the feature itself :-(
325 * - Because a generic redirect template was needed for
326 * this, tinygif = 3 now uses the same.
328 * - Moved GIFs, and other static HTTP response templates
333 * - Removed some >400 CRs again (Jon, you really worked
336 * Revision 1.3 2001/05/20 16:44:47 jongfoster
337 * Removing last hardcoded JunkBusters.com URLs.
339 * Revision 1.2 2001/05/20 01:21:20 jongfoster
340 * Version 2.9.4 checkin.
341 * - Merged popupfile and cookiefile, and added control over PCRS
342 * filtering, in new "permissionsfile".
343 * - Implemented LOG_LEVEL_FATAL, so that if there is a configuration
344 * file error you now get a message box (in the Win32 GUI) rather
345 * than the program exiting with no explanation.
346 * - Made killpopup use the PCRS MIME-type checking and HTTP-header
348 * - Removed tabs from "config"
349 * - Moved duplicated url parsing code in "loaders.c" to a new funcition.
350 * - Bumped up version number.
352 * Revision 1.1.1.1 2001/05/15 13:58:52 oes
353 * Initial import of version 2.9.3 source tree
356 *********************************************************************/
362 #include <sys/types.h>
371 #endif /* ndef __OS2__ */
372 #include <netinet/in.h>
374 #include <winsock2.h>
375 #endif /* ndef _WIN32 */
379 #endif /* def __OS2__ */
387 #include "jbsockets.h"
388 #include "miscutil.h"
392 #include "deanimate.h"
393 #include "urlmatch.h"
399 const char filters_h_rcs[] = FILTERS_H_VERSION;
401 /* Fix a problem with Solaris. There should be no effect on other
403 * Solaris's isspace() is a macro which uses it's argument directly
404 * as an array index. Therefore we need to make sure that high-bit
405 * characters generate +ve values, and ideally we also want to make
406 * the argument match the declared parameter type of "int".
408 #define ijb_isdigit(__X) isdigit((int)(unsigned char)(__X))
412 /*********************************************************************
414 * Function : block_acl
416 * Description : Block this request?
417 * Decide yes or no based on ACL file.
420 * 1 : dst = The proxy or gateway address this is going to.
421 * Or NULL to check all possible targets.
422 * 2 : csp = Current client state (buffers, headers, etc...)
423 * Also includes the client IP address.
425 * Returns : 0 = FALSE (don't block) and 1 = TRUE (do block)
427 *********************************************************************/
428 int block_acl(struct access_control_addr *dst, struct client_state *csp)
430 struct access_control_list *acl = csp->config->acl;
432 /* if not using an access control list, then permit the connection */
438 /* search the list */
441 if ((csp->ip_addr_long & acl->src->mask) == acl->src->addr)
445 /* Just want to check if they have any access */
446 if (acl->action == ACL_PERMIT)
451 else if ( ((dst->addr & acl->dst->mask) == acl->dst->addr)
452 && ((dst->port == acl->dst->port) || (acl->dst->port == 0)))
454 if (acl->action == ACL_PERMIT)
472 /*********************************************************************
474 * Function : acl_addr
476 * Description : Called from `load_config' to parse an ACL address.
479 * 1 : aspec = String specifying ACL address.
480 * 2 : aca = struct access_control_addr to fill in.
482 * Returns : 0 => Ok, everything else is an error.
484 *********************************************************************/
485 int acl_addr(char *aspec, struct access_control_addr *aca)
487 int i, masklength, port;
493 if ((p = strchr(aspec, '/')))
497 if (ijb_isdigit(*p) == 0)
501 masklength = atoi(p);
504 if ((masklength < 0) || (masklength > 32))
509 if ((p = strchr(aspec, ':')))
513 if (ijb_isdigit(*p) == 0)
522 aca->addr = ntohl(resolve_hostname_to_ip(aspec));
524 if (aca->addr == INADDR_NONE)
529 /* build the netmask */
531 for (i=1; i <= masklength ; i++)
533 aca->mask |= (1 << (32 - i));
536 /* now mask off the host portion of the ip address
537 * (i.e. save on the network portion of the address).
539 aca->addr = aca->addr & aca->mask;
544 #endif /* def FEATURE_ACL */
547 /*********************************************************************
549 * Function : match_portlist
551 * Description : Check if a given number is covered by a comma
552 * separated list of numbers and ranges (a,b-c,d,..)
555 * 1 : portlist = String with list
556 * 2 : port = port to check
558 * Returns : 0 => no match
561 *********************************************************************/
562 int match_portlist(const char *portlist, int port)
564 char *min, *max, *next, *portlist_copy;
566 min = next = portlist_copy = strdup(portlist);
569 * Zero-terminate first item and remember offset for next
571 if (NULL != (next = strchr(portlist_copy, (int) ',')))
577 * Loop through all items, checking for match
581 if (NULL == (max = strchr(min, (int) '-')))
584 * No dash, check for equality
586 if (port == atoi(min))
595 * This is a range, so check if between min and max,
596 * or, if max was omitted, between min and 65K
599 if(port >= atoi(min) && port <= (atoi(max) ? atoi(max) : 65535))
613 * Zero-terminate next item and remember offset for n+1
615 if ((NULL != next) && (NULL != (next = strchr(next, (int) ','))))
627 /*********************************************************************
629 * Function : block_url
631 * Description : Called from `chat'. Check to see if we need to block this.
634 * 1 : csp = Current client state (buffers, headers, etc...)
636 * Returns : NULL => unblocked, else HTTP block response
638 *********************************************************************/
639 struct http_response *block_url(struct client_state *csp)
641 #ifdef FEATURE_IMAGE_BLOCKING
643 #endif /* def FEATURE_IMAGE_BLOCKING */
644 struct http_response *rsp;
647 * If it's not blocked, don't block it ;-)
649 if ((csp->action->flags & ACTION_BLOCK) == 0)
655 * Else, prepare a response
657 if (NULL == (rsp = alloc_http_response()))
659 return cgi_error_memory();
663 * If it's an image-url, send back an image or redirect
664 * as specified by the relevant +image action
666 #ifdef FEATURE_IMAGE_BLOCKING
667 if (((csp->action->flags & ACTION_IMAGE_BLOCKER) != 0)
670 /* determine HOW images should be blocked */
671 p = csp->action->string[ACTION_STRING_IMAGE_BLOCKER];
673 /* and handle accordingly: */
674 if ((p == NULL) || (0 == strcmpic(p, "logo")))
676 rsp->body = bindup(image_logo_data, image_logo_length);
677 if (rsp->body == NULL)
679 free_http_response(rsp);
680 return cgi_error_memory();
682 rsp->content_length = image_logo_length;
684 if (enlist_unique_header(rsp->headers, "Content-Type", BUILTIN_IMAGE_MIMETYPE))
686 free_http_response(rsp);
687 return cgi_error_memory();
691 else if (0 == strcmpic(p, "blank"))
693 rsp->body = bindup(image_blank_data, image_blank_length);
694 if (rsp->body == NULL)
696 free_http_response(rsp);
697 return cgi_error_memory();
699 rsp->content_length = image_blank_length;
701 if (enlist_unique_header(rsp->headers, "Content-Type", BUILTIN_IMAGE_MIMETYPE))
703 free_http_response(rsp);
704 return cgi_error_memory();
708 else if (0 == strcmpic(p, "pattern"))
710 rsp->body = bindup(image_pattern_data, image_pattern_length);
711 if (rsp->body == NULL)
713 free_http_response(rsp);
714 return cgi_error_memory();
716 rsp->content_length = image_pattern_length;
718 if (enlist_unique_header(rsp->headers, "Content-Type", BUILTIN_IMAGE_MIMETYPE))
720 free_http_response(rsp);
721 return cgi_error_memory();
727 rsp->status = strdup("302 Local Redirect from Junkbuster");
728 if (rsp->status == NULL)
730 free_http_response(rsp);
731 return cgi_error_memory();
734 if (enlist_unique_header(rsp->headers, "Location", p))
736 free_http_response(rsp);
737 return cgi_error_memory();
742 #endif /* def FEATURE_IMAGE_BLOCKING */
745 * Else, generate an HTML "blocked" message:
749 struct map * exports;
752 * Workaround for stupid Netscape bug which prevents
753 * pages from being displayed if loading a referenced
754 * JavaScript or style sheet fails. So make it appear
755 * as if it succeeded.
757 if ( NULL != (p = get_header_value(csp->headers, "User-Agent:"))
758 && !strncmpic(p, "mozilla", 7) /* Catch Netscape but */
759 && !strstr(p, "Gecko") /* save Mozilla, */
760 && !strstr(p, "compatible") /* MSIE */
761 && !strstr(p, "Opera")) /* and Opera. */
763 rsp->status = strdup("200 Request for blocked URL");
767 rsp->status = strdup("404 Request for blocked URL");
770 if (rsp->status == NULL)
772 free_http_response(rsp);
773 return cgi_error_memory();
776 exports = default_exports(csp, NULL);
779 free_http_response(rsp);
780 return cgi_error_memory();
783 #ifdef FEATURE_FORCE_LOAD
784 err = map(exports, "force-prefix", 1, FORCE_PREFIX, 1);
785 #else /* ifndef FEATURE_FORCE_LOAD */
786 err = map_block_killer(exports, "force-support");
787 #endif /* ndef FEATURE_FORCE_LOAD */
789 if (!err) err = map(exports, "hostport", 1, html_encode(csp->http->hostport), 0);
790 if (!err) err = map(exports, "path", 1, html_encode(csp->http->path), 0);
795 free_http_response(rsp);
796 return cgi_error_memory();
799 err = template_fill_for_cgi(csp, "blocked", exports, rsp);
802 free_http_response(rsp);
803 return cgi_error_memory();
807 return finish_http_response(rsp);
813 /*********************************************************************
815 * Function : trust_url FIXME: I should be called distrust_url
817 * Description : Calls is_untrusted_url to determine if the URL is trusted
818 * and if not, returns a HTTP 304 response with a reject message.
821 * 1 : csp = Current client state (buffers, headers, etc...)
823 * Returns : NULL => trusted, else http_response.
825 *********************************************************************/
826 struct http_response *trust_url(struct client_state *csp)
828 struct http_response *rsp;
829 struct map * exports;
830 char buf[BUFFER_SIZE];
832 struct url_spec **tl;
837 * Don't bother to work on trusted URLs
839 if (!is_untrusted_url(csp))
845 * Else, prepare a response:
847 if (NULL == (rsp = alloc_http_response()))
849 return cgi_error_memory();
852 exports = default_exports(csp, NULL);
855 free_http_response(rsp);
856 return cgi_error_memory();
860 * Export the host, port, and referrer information
862 err = map(exports, "hostport", 1, csp->http->hostport, 1);
863 if (!err) err = map(exports, "path", 1, csp->http->path, 1);
865 if (NULL != (p = get_header_value(csp->headers, "Referer:")))
867 if (!err) err = map(exports, "referrer", 1, html_encode(p), 0);
871 if (!err) err = map(exports, "referrer", 1, "unknown", 1);
877 free_http_response(rsp);
878 return cgi_error_memory();
882 * Export the trust list
885 for (tl = csp->config->trust_list; (t = *tl) ; tl++)
887 sprintf(buf, "<li>%s</li>\n", t->spec);
888 string_append(&p, buf);
890 err = map(exports, "trusted-referrers", 1, p, 0);
895 free_http_response(rsp);
896 return cgi_error_memory();
900 * Export the trust info, if available
902 if (csp->config->trust_info->first)
904 struct list_entry *l;
907 for (l = csp->config->trust_info->first; l ; l = l->next)
909 sprintf(buf, "<li> <a href=%s>%s</a><br>\n",l->str, l->str);
910 string_append(&p, buf);
912 err = map(exports, "trust-info", 1, p, 0);
916 err = map_block_killer(exports, "have-trust-info");
922 free_http_response(rsp);
923 return cgi_error_memory();
927 * Export the force prefix or the force conditional block killer
929 #ifdef FEATURE_FORCE_LOAD
930 err = map(exports, "force-prefix", 1, FORCE_PREFIX, 1);
931 #else /* ifndef FEATURE_FORCE_LOAD */
932 err = map_block_killer(exports, "force-support");
933 #endif /* ndef FEATURE_FORCE_LOAD */
938 free_http_response(rsp);
939 return cgi_error_memory();
945 err = template_fill_for_cgi(csp, "untrusted", exports, rsp);
948 free_http_response(rsp);
949 return cgi_error_memory();
952 return finish_http_response(rsp);
954 #endif /* def FEATURE_TRUST */
957 #ifdef FEATURE_FAST_REDIRECTS
958 /*********************************************************************
960 * Function : redirect_url
962 * Description : Checks for redirection URLs and returns a HTTP redirect
963 * to the destination URL, if necessary
966 * 1 : csp = Current client state (buffers, headers, etc...)
968 * Returns : NULL if URL was clean, HTTP redirect otherwise.
970 *********************************************************************/
971 struct http_response *redirect_url(struct client_state *csp)
974 struct http_response *rsp;
976 p = q = csp->http->path;
977 log_error(LOG_LEVEL_REDIRECTS, "checking path for redirects: %s", p);
980 * find the last URL encoded in the request
982 while ((p = strstr(p, "http://")))
988 * if there was any, generate and return a HTTP redirect
990 if (q != csp->http->path)
992 log_error(LOG_LEVEL_REDIRECTS, "redirecting to: %s", q);
994 if (NULL == (rsp = alloc_http_response()))
996 return cgi_error_memory();
999 if ( enlist_unique_header(rsp->headers, "Location", q)
1000 || (NULL == (rsp->status = strdup("302 Local Redirect from Junkbuster"))) )
1002 free_http_response(rsp);
1003 return cgi_error_memory();
1006 return finish_http_response(rsp);
1014 #endif /* def FEATURE_FAST_REDIRECTS */
1017 #ifdef FEATURE_IMAGE_BLOCKING
1018 /*********************************************************************
1020 * Function : is_imageurl
1022 * Description : Given a URL, decide whether it is an image or not,
1023 * using either the info from a previous +image action
1024 * or, #ifdef FEATURE_IMAGE_DETECT_MSIE, the info from
1025 * the browser's accept header.
1028 * 1 : csp = Current client state (buffers, headers, etc...)
1030 * Returns : True (nonzero) if URL is an image, false (0)
1033 *********************************************************************/
1034 int is_imageurl(struct client_state *csp)
1036 #ifdef FEATURE_IMAGE_DETECT_MSIE
1039 tmp = get_header_value(csp->headers, "User-Agent:");
1040 if (tmp && strstr(tmp, "MSIE"))
1042 tmp = get_header_value(csp->headers, "Accept:");
1043 if (tmp && strstr(tmp, "image/gif"))
1045 /* Client will accept HTML. If this seems counterintuitive,
1055 #endif /* def FEATURE_IMAGE_DETECT_MSIE */
1057 return ((csp->action->flags & ACTION_IMAGE) != 0);
1060 #endif /* def FEATURE_IMAGE_BLOCKING */
1063 #ifdef FEATURE_COOKIE_JAR
1064 /*********************************************************************
1066 * Function : is_untrusted_url
1068 * Description : Should we "distrust" this URL (and block it)?
1070 * Yes if it matches a line in the trustfile, or if the
1071 * referrer matches a line starting with "+" in the
1076 * 1 : csp = Current client state (buffers, headers, etc...)
1078 * Returns : 0 => trusted, 1 => untrusted
1080 *********************************************************************/
1081 int is_untrusted_url(struct client_state *csp)
1083 struct file_list *fl;
1084 struct block_spec *b;
1085 struct url_spec **trusted_url;
1086 struct http_request rhttp[1];
1087 const char * referer;
1091 * If we don't have a trustlist, we trust everybody
1093 if (((fl = csp->tlist) == NULL) || ((b = fl->f) == NULL))
1098 memset(rhttp, '\0', sizeof(*rhttp));
1101 * Do we trust the request URL itself?
1103 for (b = b->next; b ; b = b->next)
1105 if (url_match(b->url, csp->http))
1111 if (NULL == (referer = get_header_value(csp->headers, "Referer:")))
1113 /* no referrer was supplied */
1118 * If not, do we maybe trust its referrer?
1123 * Parse the URL from the referrer
1126 err = parse_http_url(referer, rhttp, csp);
1132 for (trusted_url = csp->config->trust_list; *trusted_url != NULL; trusted_url++)
1134 if (url_match(*trusted_url, rhttp))
1136 /* if the URL's referrer is from a trusted referrer, then
1137 * add the target spec to the trustfile as an unblocked
1138 * domain and return NULL (which means it's OK).
1143 if ((fp = fopen(csp->config->trustfile, "a")))
1147 char * new_entry = strdup("~");
1149 string_append(&new_entry, csp->http->hostport);
1151 path = csp->http->path;
1152 if ( (path[0] == '/')
1154 && ((path_end = strchr(path + 2, '/')) != NULL))
1156 /* since this path points into a user's home space
1157 * be sure to include this spec in the trustfile.
1159 int path_len = path_end - path; /* save offset */
1160 path = strdup(path); /* Copy string */
1163 path_end = path + path_len; /* regenerate ptr to new buffer */
1164 *(path_end + 1) = '\0'; /* Truncate path after '/' */
1166 string_join(&new_entry, path);
1169 if (new_entry != NULL)
1171 fprintf(fp, "%s\n", new_entry);
1176 /* FIXME: No way to handle out-of memory, so mostly ignoring it */
1177 log_error(LOG_LEVEL_ERROR, "Out of memory adding pattern to trust file");
1187 #endif /* def FEATURE_COOKIE_JAR */
1190 /*********************************************************************
1192 * Function : pcrs_filter_response
1194 * Description : Ecexute all text substitutions from all applying
1195 * +filter actions on the text buffer that's been accumulated
1196 * in csp->iob->buf. If this changes the contents, set
1197 * csp->content_length to the modified size and raise the
1198 * CSP_FLAG_MODIFIED flag.
1201 * 1 : csp = Current client state (buffers, headers, etc...)
1203 * Returns : a pointer to the (newly allocated) modified buffer.
1204 * or NULL if there were no hits or something went wrong
1206 *********************************************************************/
1207 char *pcrs_filter_response(struct client_state *csp)
1212 char *old = csp->iob->cur, *new = NULL;
1215 struct file_list *fl;
1216 struct re_filterfile_spec *b;
1217 struct list_entry *filtername;
1222 if (csp->iob->cur >= csp->iob->eod)
1226 size = csp->iob->eod - csp->iob->cur;
1228 if ( ( NULL == (fl = csp->rlist) ) || ( NULL == fl->f) )
1230 log_error(LOG_LEVEL_ERROR, "Unable to get current state of regexp filtering.");
1235 * If the body has a "chunked" transfer-encoding,
1236 * get rid of it first, adjusting size and iob->eod
1238 if (csp->flags & CSP_FLAG_CHUNKED)
1240 log_error(LOG_LEVEL_RE_FILTER, "Need to de-chunk first");
1241 if (0 == (size = remove_chunked_transfer_coding(csp->iob->cur, size)))
1245 csp->iob->eod = csp->iob->cur + size;
1246 csp->flags |= CSP_FLAG_MODIFIED;
1250 * For all applying +filter actions, look if a filter by that
1251 * name exists and if yes, execute it's pcrs_joblist on the
1254 for (b = fl->f; b; b = b->next)
1256 for (filtername = csp->action->multi[ACTION_MULTI_FILTER]->first;
1257 filtername ; filtername = filtername->next)
1259 if (strcmp(b->filtername, filtername->str) == 0)
1261 if ( NULL == b->joblist )
1263 log_error(LOG_LEVEL_RE_FILTER, "Filter %s has empty joblist. Nothing to do.", b->filtername);
1267 log_error(LOG_LEVEL_RE_FILTER, "re_filtering %s%s (size %d) with filter %s...",
1268 csp->http->hostport, csp->http->path, size, b->filtername);
1270 /* Apply all jobs from the joblist */
1271 for (job = b->joblist; NULL != job; job = job->next)
1273 hits += pcrs_execute(job, old, size, &new, &size);
1274 if (old != csp->iob->cur) free(old);
1278 log_error(LOG_LEVEL_RE_FILTER, " ...produced %d hits (new size %d).", hits, size);
1284 * If there were no hits, destroy our copy and let
1285 * chat() use the original in csp->iob
1293 csp->flags |= CSP_FLAG_MODIFIED;
1294 csp->content_length = size;
1302 /*********************************************************************
1304 * Function : gif_deanimate_response
1306 * Description : Deanimate the GIF image that has been accumulated in
1307 * csp->iob->buf, set csp->content_length to the modified
1308 * size and raise the CSP_FLAG_MODIFIED flag.
1311 * 1 : csp = Current client state (buffers, headers, etc...)
1313 * Returns : a pointer to the (newly allocated) modified buffer.
1314 * or NULL in case something went wrong.
1316 *********************************************************************/
1317 char *gif_deanimate_response(struct client_state *csp)
1319 struct binbuffer *in, *out;
1321 size_t size = csp->iob->eod - csp->iob->cur;
1324 * If the body has a "chunked" transfer-encoding,
1325 * get rid of it first, adjusting size and iob->eod
1327 if (csp->flags & CSP_FLAG_CHUNKED)
1329 log_error(LOG_LEVEL_DEANIMATE, "Need to de-chunk first");
1330 if (0 == (size = remove_chunked_transfer_coding(csp->iob->cur, size)))
1334 csp->iob->eod = csp->iob->cur + size;
1335 csp->flags |= CSP_FLAG_MODIFIED;
1338 if ( (NULL == (in = (struct binbuffer *)zalloc(sizeof *in )))
1339 || (NULL == (out = (struct binbuffer *)zalloc(sizeof *out))) )
1341 log_error(LOG_LEVEL_DEANIMATE, "failed! (no mem)");
1345 in->buffer = csp->iob->cur;
1348 if (gif_deanimate(in, out, strncmp("last", csp->action->string[ACTION_STRING_DEANIMATE], 4)))
1350 log_error(LOG_LEVEL_DEANIMATE, "failed! (gif parsing)");
1357 log_error(LOG_LEVEL_DEANIMATE, "Success! GIF shrunk from %d bytes to %d.", size, out->offset);
1358 csp->content_length = out->offset;
1359 csp->flags |= CSP_FLAG_MODIFIED;
1369 /*********************************************************************
1371 * Function : remove_chunked_transfer_coding
1373 * Description : In-situ remove the "chunked" transfer coding as defined
1374 * in rfc2616 from a buffer.
1377 * 1 : buffer = Pointer to the text buffer
1378 * 2 : size = Number of bytes to be processed
1380 * Returns : The new size, i.e. the number of bytes from buffer which
1381 * are occupied by the stripped body, or 0 in case something
1384 *********************************************************************/
1385 int remove_chunked_transfer_coding(char *buffer, const size_t size)
1388 unsigned int chunksize = 0;
1389 char *from_p, *to_p;
1392 from_p = to_p = buffer;
1394 if (sscanf(buffer, "%x", &chunksize) != 1)
1396 log_error(LOG_LEVEL_ERROR, "Invalid first chunksize while stripping \"chunked\" transfer coding");
1400 while (chunksize > 0)
1402 if (NULL == (from_p = strstr(from_p, "\r\n")))
1404 log_error(LOG_LEVEL_ERROR, "Parse error while stripping \"chunked\" transfer coding");
1407 newsize += chunksize;
1410 memmove(to_p, from_p, (size_t) chunksize);
1411 to_p = buffer + newsize;
1412 from_p += chunksize + 2;
1414 if (sscanf(from_p, "%x", &chunksize) != 1)
1416 log_error(LOG_LEVEL_ERROR, "Parse error while stripping \"chunked\" transfer coding");
1421 /* FIXME: Should this get its own loglevel? */
1422 log_error(LOG_LEVEL_RE_FILTER, "De-chunking successful. Shrunk from %d to %d\n", size, newsize);
1428 /*********************************************************************
1430 * Function : url_actions
1432 * Description : Gets the actions for this URL.
1435 * 1 : http = http_request request for blocked URLs
1436 * 2 : csp = Current client state (buffers, headers, etc...)
1440 *********************************************************************/
1441 void url_actions(struct http_request *http,
1442 struct client_state *csp)
1444 struct file_list *fl;
1445 struct url_actions *b;
1447 init_current_action(csp->action);
1449 if (((fl = csp->actions_list) == NULL) || ((b = fl->f) == NULL))
1454 apply_url_actions(csp->action, http, b);
1459 /*********************************************************************
1461 * Function : apply_url_actions
1463 * Description : Applies a list of URL actions.
1466 * 1 : action = Destination.
1467 * 2 : http = Current URL
1468 * 3 : b = list of URL actions to apply
1472 *********************************************************************/
1473 void apply_url_actions(struct current_action_spec *action,
1474 struct http_request *http,
1475 struct url_actions *b)
1479 /* Should never happen */
1483 for (b = b->next; NULL != b; b = b->next)
1485 if (url_match(b->url, http))
1487 merge_current_action(action, b->action);
1493 /*********************************************************************
1495 * Function : forward_url
1497 * Description : Should we forward this to another proxy?
1500 * 1 : http = http_request request for current URL
1501 * 2 : csp = Current client state (buffers, headers, etc...)
1503 * Returns : Pointer to forwarding information.
1505 *********************************************************************/
1506 const struct forward_spec * forward_url(struct http_request *http,
1507 struct client_state *csp)
1509 static const struct forward_spec fwd_default[1] = { FORWARD_SPEC_INITIALIZER };
1510 struct forward_spec *fwd = csp->config->forward;
1519 if (url_match(fwd->url, http))