-const char parsers_rcs[] = "$Id: parsers.c,v 1.1.1.1 2001/05/15 13:59:01 oes Exp $";
+const char parsers_rcs[] = "$Id: parsers.c,v 1.87 2007/01/31 16:21:38 fabiankeil Exp $";
/*********************************************************************
*
* File : $Source: /cvsroot/ijbswa/current/parsers.c,v $
* `client_uagent', `client_x_forwarded',
* `client_x_forwarded_adder', `client_xtra_adder',
* `content_type', `crumble', `destroy_list', `enlist',
- * `flush_socket', `free_http_request', `get_header',
- * `list_to_text', `match', `parse_http_request', `sed',
- * and `server_set_cookie'.
+ * `flush_socket', ``get_header', `sed', `filter_server_header'
+ * `filter_client_header', `filter_header', `crunch_server_header',
+ * `server_content_encoding', `server_content_disposition',
+ * `server_last_modified', `client_accept_language',
+ * `crunch_client_header', `client_if_modified_since',
+ * `client_if_none_match', `get_destination_from_headers',
+ * `parse_header_time' and `server_set_cookie'.
*
- * Copyright : Written by and Copyright (C) 2001 the SourceForge
- * IJBSWA team. http://ijbswa.sourceforge.net
+ * Copyright : Written by and Copyright (C) 2001-2007 the SourceForge
+ * Privoxy team. http://www.privoxy.org/
*
* Based on the Internet Junkbuster originally written
- * by and Copyright (C) 1997 Anonymous Coders and
+ * by and Copyright (C) 1997 Anonymous Coders and
* Junkbusters Corporation. http://www.junkbusters.com
*
- * This program is free software; you can redistribute it
+ * This program is free software; you can redistribute it
* and/or modify it under the terms of the GNU General
* Public License as published by the Free Software
* Foundation; either version 2 of the License, or (at
*
* Revisions :
* $Log: parsers.c,v $
+ * Revision 1.87 2007/01/31 16:21:38 fabiankeil
+ * Search for Max-Forwards headers case-insensitive,
+ * don't generate the "501 unsupported" message for invalid
+ * Max-Forwards values and don't increase negative ones.
+ *
+ * Revision 1.86 2007/01/30 13:05:26 fabiankeil
+ * - Let server_set_cookie() check the expiration date
+ * of cookies and don't touch the ones that are already
+ * expired. Fixes problems with low quality web applications
+ * as described in BR 932612.
+ *
+ * - Adjust comment in client_max_forwards to reality;
+ * remove invalid Max-Forwards headers.
+ *
+ * Revision 1.85 2007/01/26 15:33:46 fabiankeil
+ * Stop filter_header() from unintentionally removing
+ * empty header lines that were enlisted by the continue
+ * hack.
+ *
+ * Revision 1.84 2007/01/24 12:56:52 fabiankeil
+ * - Repeat the request URL before logging any headers.
+ * Makes reading the log easier in case of simultaneous requests.
+ * - If there are more than one Content-Type headers in one request,
+ * use the first one and remove the others.
+ * - Remove "newval" variable in server_content_type().
+ * It's only used once.
+ *
+ * Revision 1.83 2007/01/12 15:03:02 fabiankeil
+ * Correct a cast, check inflateEnd() exit code
+ * to see if we have to, replace sprintf calls
+ * with snprintf.
+ *
+ * Revision 1.82 2007/01/01 19:36:37 fabiankeil
+ * Integrate a modified version of Wil Mahan's
+ * zlib patch (PR #895531).
+ *
+ * Revision 1.81 2006/12/31 22:21:33 fabiankeil
+ * Skip empty filter files in filter_header()
+ * but don't ignore the ones that come afterwards.
+ * Fixes BR 1619208, this time for real.
+ *
+ * Revision 1.80 2006/12/29 19:08:22 fabiankeil
+ * Reverted parts of my last commit
+ * to keep error handling working.
+ *
+ * Revision 1.79 2006/12/29 18:04:40 fabiankeil
+ * Fixed gcc43 conversion warnings.
+ *
+ * Revision 1.78 2006/12/26 17:19:20 fabiankeil
+ * Bringing back the "useless" localtime() call
+ * I removed in revision 1.67. On some platforms
+ * it's necessary to prevent time zone offsets.
+ *
+ * Revision 1.77 2006/12/07 18:44:26 fabiankeil
+ * Rebuild request URL in get_destination_from_headers()
+ * to make sure redirect{pcrs command} works as expected
+ * for intercepted requests.
+ *
+ * Revision 1.76 2006/12/06 19:52:25 fabiankeil
+ * Added get_destination_from_headers().
+ *
+ * Revision 1.75 2006/11/13 19:05:51 fabiankeil
+ * Make pthread mutex locking more generic. Instead of
+ * checking for OSX and OpenBSD, check for FEATURE_PTHREAD
+ * and use mutex locking unless there is an _r function
+ * available. Better safe than sorry.
+ *
+ * Fixes "./configure --disable-pthread" and should result
+ * in less threading-related problems on pthread-using platforms,
+ * but it still doesn't fix BR#1122404.
+ *
+ * Revision 1.74 2006/10/02 16:59:12 fabiankeil
+ * The special header "X-Filter: No" now disables
+ * header filtering as well.
+ *
+ * Revision 1.73 2006/09/23 13:26:38 roro
+ * Replace TABs by spaces in source code.
+ *
+ * Revision 1.72 2006/09/23 12:37:21 fabiankeil
+ * Don't print a log message every time filter_headers is
+ * entered or left. It only creates noise without any real
+ * information.
+ *
+ * Revision 1.71 2006/09/21 19:55:17 fabiankeil
+ * Fix +hide-if-modified-since{-n}.
+ *
+ * Revision 1.70 2006/09/08 12:06:34 fabiankeil
+ * Have hide-if-modified-since interpret the random
+ * range value as minutes instead of hours. Allows
+ * more fine-grained configuration.
+ *
+ * Revision 1.69 2006/09/06 16:25:51 fabiankeil
+ * Always have parse_header_time return a pointer
+ * that actual makes sense, even though we currently
+ * only need it to detect problems.
+ *
+ * Revision 1.68 2006/09/06 10:43:32 fabiankeil
+ * Added config option enable-remote-http-toggle
+ * to specify if Privoxy should recognize special
+ * headers (currently only X-Filter) to change its
+ * behaviour. Disabled by default.
+ *
+ * Revision 1.67 2006/09/04 11:01:26 fabiankeil
+ * After filtering de-chunked instances, remove
+ * "Transfer-Encoding" header entirely instead of changing
+ * it to "Transfer-Encoding: identity", which is invalid.
+ * Thanks Michael Shields <shields@msrl.com>. Fixes PR 1318658.
+ *
+ * Don't use localtime in parse_header_time. An empty time struct
+ * is good enough, it gets overwritten by strptime anyway.
+ *
+ * Revision 1.66 2006/09/03 19:38:28 fabiankeil
+ * Use gmtime_r if available, fallback to gmtime with mutex
+ * protection for MacOSX and use vanilla gmtime for the rest.
+ *
+ * Revision 1.65 2006/08/22 10:55:56 fabiankeil
+ * Changed client_referrer to use the right type (size_t) for
+ * hostlenght and to shorten the temporary referrer string with
+ * '\0' instead of adding a useless line break.
+ *
+ * Revision 1.64 2006/08/17 17:15:10 fabiankeil
+ * - Back to timegm() using GnuPG's replacement if necessary.
+ * Using mktime() and localtime() could add a on hour offset if
+ * the randomize factor was big enough to lead to a summer/wintertime
+ * switch.
+ *
+ * - Removed now-useless Privoxy 3.0.3 compatibility glue.
+ *
+ * - Moved randomization code into pick_from_range().
+ *
+ * - Changed parse_header_time definition.
+ * time_t isn't guaranteed to be signed and
+ * if it isn't, -1 isn't available as error code.
+ * Changed some variable types in client_if_modified_since()
+ * because of the same reason.
+ *
+ * Revision 1.63 2006/08/14 13:18:08 david__schmidt
+ * OS/2 compilation compatibility fixups
+ *
+ * Revision 1.62 2006/08/14 08:58:42 fabiankeil
+ * Changed include from strptime.c to strptime.h
+ *
+ * Revision 1.61 2006/08/14 08:25:19 fabiankeil
+ * Split filter-headers{} into filter-client-headers{}
+ * and filter-server-headers{}.
+ * Added parse_header_time() to share some code.
+ * Replaced timegm() with mktime().
+ *
+ * Revision 1.60 2006/08/12 03:54:37 david__schmidt
+ * Windows service integration
+ *
+ * Revision 1.59 2006/08/03 02:46:41 david__schmidt
+ * Incorporate Fabian Keil's patch work:\rhttp://www.fabiankeil.de/sourcecode/privoxy/
+ *
+ * Revision 1.58 2006/07/18 14:48:47 david__schmidt
+ * Reorganizing the repository: swapping out what was HEAD (the old 3.1 branch)
+ * with what was really the latest development (the v_3_0_branch branch)
+ *
+ * Revision 1.56.2.10 2006/01/21 16:16:08 david__schmidt
+ * Thanks to Edward Carrel for his patch to modernize OSX's\rpthreads support. See bug #1409623.
+ *
+ * Revision 1.56.2.9 2004/10/03 12:53:45 david__schmidt
+ * Add the ability to check jpeg images for invalid
+ * lengths of comment blocks. Defensive strategy
+ * against the exploit:
+ * Microsoft Security Bulletin MS04-028
+ * Buffer Overrun in JPEG Processing (GDI+) Could
+ * Allow Code Execution (833987)
+ * Enabled with +inspect-jpegs in actions files.
+ *
+ * Revision 1.56.2.8 2003/07/11 13:21:25 oes
+ * Excluded text/plain objects from filtering. This fixes a
+ * couple of client-crashing, download corruption and
+ * Privoxy performance issues, whose root cause lies in
+ * web servers labelling content of unknown type as text/plain.
+ *
+ * Revision 1.56.2.7 2003/05/06 12:07:26 oes
+ * Fixed bug #729900: Suspicious HOST: headers are now killed and regenerated if necessary
+ *
+ * Revision 1.56.2.6 2003/04/14 21:28:30 oes
+ * Completing the previous change
+ *
+ * Revision 1.56.2.5 2003/04/14 12:08:16 oes
+ * Added temporary workaround for bug in PHP < 4.2.3
+ *
+ * Revision 1.56.2.4 2003/03/07 03:41:05 david__schmidt
+ * Wrapping all *_r functions (the non-_r versions of them) with mutex semaphores for OSX. Hopefully this will take care of all of those pesky crash reports.
+ *
+ * Revision 1.56.2.3 2002/11/10 04:20:02 hal9
+ * Fix typo: supressed -> suppressed
+ *
+ * Revision 1.56.2.2 2002/09/25 14:59:53 oes
+ * Improved cookie logging
+ *
+ * Revision 1.56.2.1 2002/09/25 14:52:45 oes
+ * Added basic support for OPTIONS and TRACE HTTP methods:
+ * - New parser function client_max_forwards which decrements
+ * the Max-Forwards HTTP header field of OPTIONS and TRACE
+ * requests by one before forwarding
+ * - New parser function client_host which extracts the host
+ * and port information from the HTTP header field if the
+ * request URI was not absolute
+ * - Don't crumble and re-add the Host: header, but only generate
+ * and append if missing
+ *
+ * Revision 1.56 2002/05/12 15:34:22 jongfoster
+ * Fixing typo in a comment
+ *
+ * Revision 1.55 2002/05/08 16:01:07 oes
+ * Optimized add_to_iob:
+ * - Use realloc instead of malloc(), memcpy(), free()
+ * - Expand to powers of two if possible, to get
+ * O(log n) reallocs instead of O(n).
+ * - Moved check for buffer limit here from chat
+ * - Report failure via returncode
+ *
+ * Revision 1.54 2002/04/02 15:03:16 oes
+ * Tiny code cosmetics
+ *
+ * Revision 1.53 2002/03/26 22:29:55 swa
+ * we have a new homepage!
+ *
+ * Revision 1.52 2002/03/24 13:25:43 swa
+ * name change related issues
+ *
+ * Revision 1.51 2002/03/13 00:27:05 jongfoster
+ * Killing warnings
+ *
+ * Revision 1.50 2002/03/12 01:45:35 oes
+ * More verbose logging
+ *
+ * Revision 1.49 2002/03/09 20:03:52 jongfoster
+ * - Making various functions return int rather than size_t.
+ * (Undoing a recent change). Since size_t is unsigned on
+ * Windows, functions like read_socket that return -1 on
+ * error cannot return a size_t.
+ *
+ * THIS WAS A MAJOR BUG - it caused frequent, unpredictable
+ * crashes, and also frequently caused JB to jump to 100%
+ * CPU and stay there. (Because it thought it had just
+ * read ((unsigned)-1) == 4Gb of data...)
+ *
+ * - The signature of write_socket has changed, it now simply
+ * returns success=0/failure=nonzero.
+ *
+ * - Trying to get rid of a few warnings --with-debug on
+ * Windows, I've introduced a new type "jb_socket". This is
+ * used for the socket file descriptors. On Windows, this
+ * is SOCKET (a typedef for unsigned). Everywhere else, it's
+ * an int. The error value can't be -1 any more, so it's
+ * now JB_INVALID_SOCKET (which is -1 on UNIX, and in
+ * Windows it maps to the #define INVALID_SOCKET.)
+ *
+ * - The signature of bind_port has changed.
+ *
+ * Revision 1.48 2002/03/07 03:46:53 oes
+ * Fixed compiler warnings etc
+ *
+ * Revision 1.47 2002/02/20 23:15:13 jongfoster
+ * Parsing functions now handle out-of-memory gracefully by returning
+ * an error code.
+ *
+ * Revision 1.46 2002/01/17 21:03:47 jongfoster
+ * Moving all our URL and URL pattern parsing code to urlmatch.c.
+ *
+ * Revision 1.45 2002/01/09 14:33:03 oes
+ * Added support for localtime_r.
+ *
+ * Revision 1.44 2001/12/14 01:22:54 steudten
+ * Remove 'user:pass@' from 'proto://user:pass@host' for the
+ * new added header 'Host: ..'. (See Req ID 491818)
+ *
+ * Revision 1.43 2001/11/23 00:26:38 jongfoster
+ * Fixing two really stupid errors in my previous commit
+ *
+ * Revision 1.42 2001/11/22 21:59:30 jongfoster
+ * Adding code to handle +no-cookies-keep
+ *
+ * Revision 1.41 2001/11/05 23:43:05 steudten
+ * Add time+date to log files.
+ *
+ * Revision 1.40 2001/10/26 20:13:09 jongfoster
+ * ctype.h is needed in Windows, too.
+ *
+ * Revision 1.39 2001/10/26 17:40:04 oes
+ * Introduced get_header_value()
+ * Removed http->user_agent, csp->referrer and csp->accept_types
+ * Removed client_accept()
+ *
+ * Revision 1.38 2001/10/25 03:40:48 david__schmidt
+ * Change in porting tactics: OS/2's EMX porting layer doesn't allow multiple
+ * threads to call select() simultaneously. So, it's time to do a real, live,
+ * native OS/2 port. See defines for __EMX__ (the porting layer) vs. __OS2__
+ * (native). Both versions will work, but using __OS2__ offers multi-threading.
+ *
+ * Revision 1.37 2001/10/23 21:36:02 jongfoster
+ * Documenting sed()'s error behaviou (doc change only)
+ *
+ * Revision 1.36 2001/10/13 12:51:51 joergs
+ * Removed client_host, (was only required for the old 2.0.2-11 http://noijb.
+ * force-load), instead crumble Host: and add it (again) in client_host_adder
+ * (in case we get a HTTP/1.0 request without Host: header and forward it to
+ * a HTTP/1.1 server/proxy).
+ *
+ * Revision 1.35 2001/10/09 22:39:21 jongfoster
+ * assert.h is also required under Win32, so moving out of #ifndef _WIN32
+ * block.
+ *
+ * Revision 1.34 2001/10/07 18:50:55 oes
+ * Added server_content_encoding, renamed server_transfer_encoding
+ *
+ * Revision 1.33 2001/10/07 18:04:49 oes
+ * Changed server_http11 to server_http and its pattern to "HTTP".
+ * Additional functionality: it now saves the HTTP status into
+ * csp->http->status and sets CT_TABOO for Status 206 (partial range)
+ *
+ * Revision 1.32 2001/10/07 15:43:28 oes
+ * Removed FEATURE_DENY_GZIP and replaced it with client_accept_encoding,
+ * client_te and client_accept_encoding_adder, triggered by the new
+ * +no-compression action. For HTTP/1.1 the Accept-Encoding header is
+ * changed to allow only identity and chunked, and the TE header is
+ * crunched. For HTTP/1.0, Accept-Encoding is crunched.
+ *
+ * parse_http_request no longer does anything than parsing. The rewriting
+ * of http->cmd and version mangling are gone. It now also recognizes
+ * the put and delete methods and saves the url in http->url. Removed
+ * unused variable.
+ *
+ * renamed content_type and content_length to have the server_ prefix
+ *
+ * server_content_type now only works if csp->content_type != CT_TABOO
+ *
+ * added server_transfer_encoding, which
+ * - Sets CT_TABOO to prohibit filtering if encoding compresses
+ * - Raises the CSP_FLAG_CHUNKED flag if Encoding is "chunked"
+ * - Change from "chunked" to "identity" if body was chunked
+ * but has been de-chunked for filtering.
+ *
+ * added server_content_md5 which crunches any Content-MD5 headers
+ * if the body was modified.
+ *
+ * made server_http11 conditional on +downgrade action
+ *
+ * Replaced 6 boolean members of csp with one bitmap (csp->flags)
+ *
+ * Revision 1.31 2001/10/05 14:25:02 oes
+ * Crumble Keep-Alive from Server
+ *
+ * Revision 1.30 2001/09/29 12:56:03 joergs
+ * IJB now changes HTTP/1.1 to HTTP/1.0 in requests and answers.
+ *
+ * Revision 1.29 2001/09/24 21:09:24 jongfoster
+ * Fixing 2 memory leaks that Guy spotted, where the paramater to
+ * enlist() was not being free()d.
+ *
+ * Revision 1.28 2001/09/22 16:32:28 jongfoster
+ * Removing unused #includes.
+ *
+ * Revision 1.27 2001/09/20 15:45:25 steudten
+ *
+ * add casting from size_t to int for printf()
+ * remove local variable shadow s2
+ *
+ * Revision 1.26 2001/09/16 17:05:14 jongfoster
+ * Removing unused #include showarg.h
+ *
+ * Revision 1.25 2001/09/16 13:21:27 jongfoster
+ * Changes to use new list functions.
+ *
+ * Revision 1.24 2001/09/13 23:05:50 jongfoster
+ * Changing the string paramater to the header parsers a "const".
+ *
+ * Revision 1.23 2001/09/12 18:08:19 steudten
+ *
+ * In parse_http_request() header rewriting miss the host value, so
+ * from http://www.mydomain.com the result was just " / " not
+ * http://www.mydomain.com/ in case we forward.
+ *
+ * Revision 1.22 2001/09/10 10:58:53 oes
+ * Silenced compiler warnings
+ *
+ * Revision 1.21 2001/07/31 14:46:00 oes
+ * - Persistant connections now suppressed
+ * - sed() no longer appends empty header to csp->headers
+ *
+ * Revision 1.20 2001/07/30 22:08:36 jongfoster
+ * Tidying up #defines:
+ * - All feature #defines are now of the form FEATURE_xxx
+ * - Permanently turned off WIN_GUI_EDIT
+ * - Permanently turned on WEBDAV and SPLIT_PROXY_ARGS
+ *
+ * Revision 1.19 2001/07/25 17:21:54 oes
+ * client_uagent now saves copy of User-Agent: header value
+ *
+ * Revision 1.18 2001/07/13 14:02:46 oes
+ * - Included fix to repair broken HTTP requests that
+ * don't contain a path, not even '/'.
+ * - Removed all #ifdef PCRS
+ * - content_type now always inspected and classified as
+ * text, gif or other.
+ * - formatting / comments
+ *
+ * Revision 1.17 2001/06/29 21:45:41 oes
+ * Indentation, CRLF->LF, Tab-> Space
+ *
+ * Revision 1.16 2001/06/29 13:32:42 oes
+ * - Fixed a comment
+ * - Adapted free_http_request
+ * - Removed logentry from cancelled commit
+ *
+ * Revision 1.15 2001/06/03 19:12:38 oes
+ * deleted const struct interceptors
+ *
+ * Revision 1.14 2001/06/01 18:49:17 jongfoster
+ * Replaced "list_share" with "list" - the tiny memory gain was not
+ * worth the extra complexity.
+ *
+ * Revision 1.13 2001/05/31 21:30:33 jongfoster
+ * Removed list code - it's now in list.[ch]
+ * Renamed "permission" to "action", and changed many features
+ * to use the actions file rather than the global config.
+ *
+ * Revision 1.12 2001/05/31 17:33:13 oes
+ *
+ * CRLF -> LF
+ *
+ * Revision 1.11 2001/05/29 20:11:19 joergs
+ * '/ * inside comment' warning removed.
+ *
+ * Revision 1.10 2001/05/29 09:50:24 jongfoster
+ * Unified blocklist/imagelist/permissionslist.
+ * File format is still under discussion, but the internal changes
+ * are (mostly) done.
+ *
+ * Also modified interceptor behaviour:
+ * - We now intercept all URLs beginning with one of the following
+ * prefixes (and *only* these prefixes):
+ * * http://i.j.b/
+ * * http://ijbswa.sf.net/config/
+ * * http://ijbswa.sourceforge.net/config/
+ * - New interceptors "home page" - go to http://i.j.b/ to see it.
+ * - Internal changes so that intercepted and fast redirect pages
+ * are not replaced with an image.
+ * - Interceptors now have the option to send a binary page direct
+ * to the client. (i.e. ijb-send-banner uses this)
+ * - Implemented show-url-info interceptor. (Which is why I needed
+ * the above interceptors changes - a typical URL is
+ * "http://i.j.b/show-url-info?url=www.somesite.com/banner.gif".
+ * The previous mechanism would not have intercepted that, and
+ * if it had been intercepted then it then it would have replaced
+ * it with an image.)
+ *
+ * Revision 1.9 2001/05/28 17:26:33 jongfoster
+ * Fixing segfault if last header was crunched.
+ * Fixing Windows build (snprintf() is _snprintf() under Win32, but we
+ * can use the cross-platform sprintf() instead.)
+ *
+ * Revision 1.8 2001/05/27 22:17:04 oes
+ *
+ * - re_process_buffer no longer writes the modified buffer
+ * to the client, which was very ugly. It now returns the
+ * buffer, which it is then written by chat.
+ *
+ * - content_length now adjusts the Content-Length: header
+ * for modified documents rather than crunch()ing it.
+ * (Length info in csp->content_length, which is 0 for
+ * unmodified documents)
+ *
+ * - For this to work, sed() is called twice when filtering.
+ *
+ * Revision 1.7 2001/05/27 13:19:06 oes
+ * Patched Joergs solution for the content-length in.
+ *
+ * Revision 1.6 2001/05/26 13:39:32 jongfoster
+ * Only crunches Content-Length header if applying RE filtering.
+ * Without this fix, Microsoft Windows Update wouldn't work.
+ *
+ * Revision 1.5 2001/05/26 00:28:36 jongfoster
+ * Automatic reloading of config file.
+ * Removed obsolete SIGHUP support (Unix) and Reload menu option (Win32).
+ * Most of the global variables have been moved to a new
+ * struct configuration_spec, accessed through csp->config->globalname
+ * Most of the globals remaining are used by the Win32 GUI.
+ *
+ * Revision 1.4 2001/05/22 18:46:04 oes
+ *
+ * - Enabled filtering banners by size rather than URL
+ * by adding patterns that replace all standard banner
+ * sizes with the "Junkbuster" gif to the re_filterfile
+ *
+ * - Enabled filtering WebBugs by providing a pattern
+ * which kills all 1x1 images
+ *
+ * - Added support for PCRE_UNGREEDY behaviour to pcrs,
+ * which is selected by the (nonstandard and therefore
+ * capital) letter 'U' in the option string.
+ * It causes the quantifiers to be ungreedy by default.
+ * Appending a ? turns back to greedy (!).
+ *
+ * - Added a new interceptor ijb-send-banner, which
+ * sends back the "Junkbuster" gif. Without imagelist or
+ * MSIE detection support, or if tinygif = 1, or the
+ * URL isn't recognized as an imageurl, a lame HTML
+ * explanation is sent instead.
+ *
+ * - Added new feature, which permits blocking remote
+ * script redirects and firing back a local redirect
+ * to the browser.
+ * The feature is conditionally compiled, i.e. it
+ * can be disabled with --disable-fast-redirects,
+ * plus it must be activated by a "fast-redirects"
+ * line in the config file, has its own log level
+ * and of course wants to be displayed by show-proxy-args
+ * Note: Boy, all the #ifdefs in 1001 locations and
+ * all the fumbling with configure.in and acconfig.h
+ * were *way* more work than the feature itself :-(
+ *
+ * - Because a generic redirect template was needed for
+ * this, tinygif = 3 now uses the same.
+ *
+ * - Moved GIFs, and other static HTTP response templates
+ * to project.h
+ *
+ * - Some minor fixes
+ *
+ * - Removed some >400 CRs again (Jon, you really worked
+ * a lot! ;-)
+ *
+ * Revision 1.3 2001/05/20 01:21:20 jongfoster
+ * Version 2.9.4 checkin.
+ * - Merged popupfile and cookiefile, and added control over PCRS
+ * filtering, in new "permissionsfile".
+ * - Implemented LOG_LEVEL_FATAL, so that if there is a configuration
+ * file error you now get a message box (in the Win32 GUI) rather
+ * than the program exiting with no explanation.
+ * - Made killpopup use the PCRS MIME-type checking and HTTP-header
+ * skipping.
+ * - Removed tabs from "config"
+ * - Moved duplicated url parsing code in "loaders.c" to a new funcition.
+ * - Bumped up version number.
+ *
+ * Revision 1.2 2001/05/17 23:02:36 oes
+ * - Made referrer option accept 'L' as a substitute for '§'
+ *
* Revision 1.1.1.1 2001/05/15 13:59:01 oes
* Initial import of version 2.9.3 source tree
*
#include "config.h"
+#ifndef _WIN32
#include <stdio.h>
#include <sys/types.h>
+#endif
+
#include <stdlib.h>
#include <ctype.h>
+#include <assert.h>
#include <string.h>
+#include <time.h>
-#ifndef _WIN32
+#ifdef FEATURE_ZLIB
+#include <zlib.h>
+#endif
+
+#if !defined(_WIN32) && !defined(__OS2__)
#include <unistd.h>
#endif
#include "project.h"
+
+#ifdef FEATURE_PTHREAD
+#include "jcc.h"
+/* jcc.h is for mutex semapores only */
+#endif /* def FEATURE_PTHREAD */
+#include "list.h"
#include "parsers.h"
#include "encode.h"
-#include "filters.h"
-#include "loaders.h"
-#include "showargs.h"
-#include "jcc.h"
#include "ssplit.h"
#include "errlog.h"
#include "jbsockets.h"
#include "miscutil.h"
+#include "list.h"
+
+#ifndef HAVE_STRPTIME
+#include "strptime.h"
+#endif
const char parsers_h_rcs[] = PARSERS_H_VERSION;
* as an array index. Therefore we need to make sure that high-bit
* characters generate +ve values, and ideally we also want to make
* the argument match the declared parameter type of "int".
- *
- * Why did they write a character function that can't take a simple
+ *
+ * Why did they write a character function that can't take a simple
* "char" argument? Doh!
*/
#define ijb_isupper(__X) isupper((int)(unsigned char)(__X))
const struct parsers client_patterns[] = {
- { "referer:", 8, client_referrer },
+ { "referer:", 8, client_referrer },
{ "user-agent:", 11, client_uagent },
- { "ua-", 3, client_ua },
- { "from:", 5, client_from },
- { "cookie:", 7, client_send_cookie },
+ { "ua-", 3, client_ua },
+ { "from:", 5, client_from },
+ { "cookie:", 7, client_send_cookie },
{ "x-forwarded-for:", 16, client_x_forwarded },
- { "proxy-connection:", 17, crumble },
-#ifdef DENY_GZIP
- { "Accept-Encoding: gzip", 21, crumble },
-#endif /* def DENY_GZIP */
-#if defined(DETECT_MSIE_IMAGES)
- { "Accept:", 7, client_accept },
-#endif /* defined(DETECT_MSIE_IMAGES) */
-#ifdef FORCE_LOAD
+ { "Accept-Encoding:", 16, client_accept_encoding },
+ { "TE:", 3, client_te },
{ "Host:", 5, client_host },
-#endif /* def FORCE_LOAD */
-/* { "if-modified-since:", 18, crumble }, */
- { NULL, 0, NULL }
+ { "if-modified-since:", 18, client_if_modified_since },
+ { "Keep-Alive:", 11, crumble },
+ { "connection:", 11, crumble },
+ { "proxy-connection:", 17, crumble },
+ { "max-forwards:", 13, client_max_forwards },
+ { "Accept-Language:", 16, client_accept_language },
+ { "if-none-match:", 14, client_if_none_match },
+ { "X-Filter:", 9, client_x_filter },
+ { "*", 0, crunch_client_header },
+ { "*", 0, filter_client_header },
+ { NULL, 0, NULL }
};
-const struct interceptors intercept_patterns[] = {
- { "show-proxy-args", 14, show_proxy_args },
-#ifdef TRUST_FILES
- { "ij-untrusted-url", 14, ij_untrusted_url },
-#endif /* def TRUST_FILES */
+const struct parsers server_patterns[] = {
+ { "HTTP", 4, server_http },
+ { "set-cookie:", 11, server_set_cookie },
+ { "connection:", 11, crumble },
+ { "Content-Type:", 13, server_content_type },
+ { "Content-Length:", 15, server_content_length },
+ { "Content-MD5:", 12, server_content_md5 },
+ { "Content-Encoding:", 17, server_content_encoding },
+ { "Transfer-Encoding:", 18, server_transfer_coding },
+ { "Keep-Alive:", 11, crumble },
+ { "content-disposition:", 20, server_content_disposition },
+ { "Last-Modified:", 14, server_last_modified },
+ { "*", 0, crunch_server_header },
+ { "*", 0, filter_server_header },
{ NULL, 0, NULL }
};
-const struct parsers server_patterns[] = {
- { "set-cookie:", 11, server_set_cookie },
- { "connection:", 11, crumble },
-#ifdef PCRS
- { "Content-Type:", 13, content_type },
- { "Content-Length:", 15, crumble },
-#endif /* def PCRS */
+const struct parsers server_patterns_light[] = {
+ { "Content-Length:", 15, server_content_length },
+ { "Transfer-Encoding:", 18, server_transfer_coding },
+#ifdef FEATURE_ZLIB
+ { "Content-Encoding:", 17, server_content_encoding },
+#endif /* def FEATURE_ZLIB */
{ NULL, 0, NULL }
};
-
-void (* const add_client_headers[])(struct client_state *) = {
+const add_header_func_ptr add_client_headers[] = {
+ client_host_adder,
client_cookie_adder,
client_x_forwarded_adder,
client_xtra_adder,
+ /* Temporarily disabled: client_accept_encoding_adder, */
+ connection_close_adder,
NULL
};
-void (* const add_server_headers[])(struct client_state *) = {
+const add_header_func_ptr add_server_headers[] = {
+ connection_close_adder,
NULL
};
-
-/*********************************************************************
- *
- * Function : match
- *
- * Description : Do a `strncmpic' on every pattern in pats.
- *
- * Parameters :
- * 1 : buf = a string to match to a list of patterns
- * 2 : pats = list of strings to compare against buf.
- *
- * Returns : Return the matching "struct parsers *",
- * or NULL if no pattern matches.
- *
- *********************************************************************/
-static const struct parsers *match(char *buf, const struct parsers *pats)
-{
- const struct parsers *v;
-
- if (buf == NULL)
- {
- /* hit me */
- log_error(LOG_LEVEL_ERROR, "NULL parameter to match()");
- return(NULL);
- }
-
- for (v = pats; v->str ; v++)
- {
- if (strncmpic(buf, v->str, v->len) == 0)
- {
- return(v);
- }
- }
- return(NULL);
-
-}
-
-
/*********************************************************************
*
* Function : flush_socket
* file, the results are not portable.
*
*********************************************************************/
-int flush_socket(int fd, struct client_state *csp)
+int flush_socket(jb_socket fd, struct client_state *csp)
{
struct iob *iob = csp->iob;
- int n = iob->eod - iob->cur;
+ int len = iob->eod - iob->cur;
- if (n <= 0)
+ if (len <= 0)
{
return(0);
}
- n = write_socket(fd, iob->cur, n);
+ if (write_socket(fd, iob->cur, (size_t)len))
+ {
+ return(-1);
+ }
iob->eod = iob->cur = iob->buf;
- return(n);
+ return(len);
}
*
* Function : add_to_iob
*
- * Description : Add content to the buffered page.
+ * Description : Add content to the buffered page, expanding the
+ * buffer if necessary.
*
* Parameters :
* 1 : csp = Current client state (buffers, headers, etc...)
* 2 : buf = holds the content to be added to the page
* 3 : n = number of bytes to be added
*
- * Returns : Number of bytes in the content buffer.
+ * Returns : JB_ERR_OK on success, JB_ERR_MEMORY if out-of-memory
+ * or buffer limit reached.
*
*********************************************************************/
-int add_to_iob(struct client_state *csp, char *buf, int n)
+jb_err add_to_iob(struct client_state *csp, char *buf, int n)
{
struct iob *iob = csp->iob;
- int have, need;
+ size_t used, offset, need, want;
char *p;
- have = iob->eod - iob->cur;
+ if (n <= 0) return JB_ERR_OK;
+
+ used = (size_t)(iob->eod - iob->buf);
+ offset = (size_t)(iob->cur - iob->buf);
+ need = used + (size_t)n + 1;
+
+ /*
+ * If the buffer can't hold the new data, extend it first.
+ * Use the next power of two if possible, else use the actual need.
+ */
+ if (need > csp->config->buffer_limit)
+ {
+ log_error(LOG_LEVEL_ERROR, "Buffer limit reached while extending the buffer (iob)");
+ return JB_ERR_MEMORY;
+ }
- if (n <= 0)
+ if (need > iob->size)
{
- return(have);
+ for (want = csp->iob->size ? csp->iob->size : 512; want <= need;) want *= 2;
+
+ if (want <= csp->config->buffer_limit && NULL != (p = (char *)realloc(iob->buf, want)))
+ {
+ iob->size = want;
+ }
+ else if (NULL != (p = (char *)realloc(iob->buf, need)))
+ {
+ iob->size = need;
+ }
+ else
+ {
+ log_error(LOG_LEVEL_ERROR, "Extending the buffer (iob) failed: %E");
+ return JB_ERR_MEMORY;
+ }
+
+ /* Update the iob pointers */
+ iob->cur = p + offset;
+ iob->eod = p + used;
+ iob->buf = p;
}
- need = have + n;
+ /* copy the new data into the iob buffer */
+ memcpy(iob->eod, buf, (size_t)n);
+
+ /* point to the end of the data */
+ iob->eod += n;
+
+ /* null terminate == cheap insurance */
+ *iob->eod = '\0';
+
+ return JB_ERR_OK;
+
+}
+
+
+#ifdef FEATURE_ZLIB
+/*********************************************************************
+ *
+ * Function : decompress_iob
+ *
+ * Description : Decompress buffered page, expanding the
+ * buffer as necessary. csp->iob->cur
+ * should point to the the beginning of the
+ * compressed data block.
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ *
+ * Returns : JB_ERR_OK on success,
+ * JB_ERR_MEMORY if out-of-memory limit reached, and
+ * JB_ERR_COMPRESS if error decompressing buffer.
+ *
+ *********************************************************************/
+jb_err decompress_iob(struct client_state *csp)
+{
+ char *buf; /* new, uncompressed buffer */
+ char *cur; /* Current iob position (to keep the original
+ * iob->cur unmodified if we return early) */
+ size_t bufsize; /* allocated size of the new buffer */
+ size_t skip_size; /* Number of bytes at the beginning of the iob
+ that we should NOT decompress. */
+ int status; /* return status of the inflate() call */
+ z_stream zstr; /* used by calls to zlib */
+
+ bufsize = csp->iob->size;
+ skip_size = (size_t)(csp->iob->cur - csp->iob->buf);
+
+ cur = csp->iob->cur;
- if ((p = (char *)malloc(need + 1)) == NULL)
+ if (bufsize < 10)
{
- log_error(LOG_LEVEL_ERROR, "malloc() iob failed: %E");
- return(-1);
+ /*
+ * This is to protect the parsing of gzipped data,
+ * but it should(?) be valid for deflated data also.
+ */
+ log_error (LOG_LEVEL_ERROR, "Buffer too small decompressing iob");
+ return JB_ERR_COMPRESS;
}
- if (have)
+ if (csp->content_type & CT_GZIP)
{
- /* there is something in the buffer - save it */
- memcpy(p, iob->cur, have);
+ /*
+ * Our task is slightly complicated by the facts that data
+ * compressed by gzip does not include a zlib header, and
+ * that there is no easily accessible interface in zlib to
+ * handle a gzip header. We strip off the gzip header by
+ * hand, and later inform zlib not to expect a header.
+ */
- /* replace the buffer with the new space */
- freez(iob->buf);
- iob->buf = p;
+ /*
+ * Strip off the gzip header. Please see RFC 1952 for more
+ * explanation of the appropriate fields.
+ */
+ if ((*cur++ != (char)0x1f)
+ || (*cur++ != (char)0x8b)
+ || (*cur++ != Z_DEFLATED))
+ {
+ log_error (LOG_LEVEL_ERROR, "Invalid gzip header when decompressing");
+ return JB_ERR_COMPRESS;
+ }
+ else
+ {
+ int flags = *cur++;
+ /*
+ * XXX: These magic numbers should be replaced
+ * with macros to give a better idea what they do.
+ */
+ if (flags & 0xe0)
+ {
+ /* The gzip header has reserved bits set; bail out. */
+ log_error (LOG_LEVEL_ERROR, "Invalid gzip header flags when decompressing");
+ return JB_ERR_COMPRESS;
+ }
+ cur += 6;
+
+ /* Skip extra fields if necessary. */
+ if (flags & 0x04)
+ {
+ /*
+ * Skip a given number of bytes, specified
+ * as a 16-bit little-endian value.
+ */
+ /*
+ * XXX: This code used to be:
+ *
+ * csp->iob->cur += *csp->iob->cur++ + (*csp->iob->cur++ << 8);
+ *
+ * which I had to change into:
+ *
+ * cur += *cur++ + (*cur++ << 8);
+ *
+ * at which point gcc43 finally noticed that the value
+ * of cur is undefined (it depends on which of the
+ * summands is evaluated first).
+ *
+ * I haven't come across a site where this
+ * code is actually executed yet, but I hope
+ * it works anyway.
+ */
+ int skip_bytes;
+ skip_bytes = *cur++;
+ skip_bytes = *cur++ << 8;
+
+ assert(skip_bytes == *csp->iob->cur-2 + ((*csp->iob->cur-1) << 8));
+
+ /*
+ * The number of bytes to skip should be positive
+ * and we'd like to stay in the buffer.
+ */
+ if((skip_bytes < 0) || (skip_bytes >= (csp->iob->eod - cur)))
+ {
+ log_error (LOG_LEVEL_ERROR,
+ "Unreasonable amount of bytes to skip (%d). Stopping decompression",
+ skip_bytes);
+ return JB_ERR_COMPRESS;
+ }
+ log_error (LOG_LEVEL_INFO,
+ "Skipping %d bytes for gzip compression. Does this sound right?",
+ skip_bytes);
+ cur += skip_bytes;
+ }
+
+ /* Skip the filename if necessary. */
+ if (flags & 0x08)
+ {
+ /* A null-terminated string follows. */
+ while (*cur++);
+ }
+
+ /* Skip the comment if necessary. */
+ if (flags & 0x10)
+ {
+ while (*cur++);
+ }
- /* point to the end of the data */
- p += have;
+ /* Skip the CRC if necessary. */
+ if (flags & 0x02)
+ {
+ cur += 2;
+ }
+ }
+ }
+ else if (csp->content_type & CT_DEFLATE)
+ {
+ /*
+ * XXX: The debug level should be lowered
+ * before the next stable release.
+ */
+ log_error (LOG_LEVEL_INFO, "Decompressing deflated iob: %d", *cur);
+ /*
+ * In theory (that is, according to RFC 1950), deflate-compressed
+ * data should begin with a two-byte zlib header and have an
+ * adler32 checksum at the end. It seems that in practice only
+ * the raw compressed data is sent. Note that this means that
+ * we are not RFC 1950-compliant here, but the advantage is that
+ * this actually works. :)
+ *
+ * We add a dummy null byte to tell zlib where the data ends,
+ * and later inform it not to expect a header.
+ *
+ * Fortunately, add_to_iob() has thoughtfully null-terminated
+ * the buffer; we can just increment the end pointer to include
+ * the dummy byte.
+ */
+ csp->iob->eod++;
}
else
{
- /* the buffer is empty, free it and reinitialize */
- freez(iob->buf);
- iob->buf = p;
+ log_error (LOG_LEVEL_ERROR,
+ "Unable to determine compression format for decompression");
+ return JB_ERR_COMPRESS;
}
- /* copy the new data into the iob buffer */
- memcpy(p, buf, n);
+ /* Set up the fields required by zlib. */
+ zstr.next_in = (Bytef *)cur;
+ zstr.avail_in = (unsigned int)(csp->iob->eod - cur);
+ zstr.zalloc = Z_NULL;
+ zstr.zfree = Z_NULL;
+ zstr.opaque = Z_NULL;
- /* point to the end of the data */
- p += n;
+ /*
+ * Passing -MAX_WBITS to inflateInit2 tells the library
+ * that there is no zlib header.
+ */
+ if (inflateInit2 (&zstr, -MAX_WBITS) != Z_OK)
+ {
+ log_error (LOG_LEVEL_ERROR, "Error initializing decompression");
+ return JB_ERR_COMPRESS;
+ }
- /* null terminate == cheap insurance */
- *p = '\0';
+ /*
+ * Next, we allocate new storage for the inflated data.
+ * We don't modify the existing iob yet, so in case there
+ * is error in decompression we can recover gracefully.
+ */
+ buf = zalloc (bufsize);
+ if (NULL == buf)
+ {
+ log_error (LOG_LEVEL_ERROR, "Out of memory decompressing iob");
+ return JB_ERR_MEMORY;
+ }
+
+ assert(bufsize >= skip_size);
+ memcpy(buf, csp->iob->buf, skip_size);
+ zstr.avail_out = bufsize - skip_size;
+ zstr.next_out = (Bytef *)buf + skip_size;
+
+ /* Try to decompress the whole stream in one shot. */
+ while (Z_BUF_ERROR == (status = inflate(&zstr, Z_FINISH)))
+ {
+ /* We need to allocate more memory for the output buffer. */
+
+ char *tmpbuf; /* used for realloc'ing the buffer */
+ size_t oldbufsize = bufsize; /* keep track of the old bufsize */
+
+ /*
+ * If zlib wants more data then there's a problem, because
+ * the complete compressed file should have been buffered.
+ */
+ if (0 == zstr.avail_in)
+ {
+ log_error(LOG_LEVEL_ERROR, "Unexpected end of compressed iob");
+ return JB_ERR_COMPRESS;
+ }
+
+ /*
+ * If we tried the limit and still didn't have enough
+ * memory, just give up.
+ */
+ if (bufsize == csp->config->buffer_limit)
+ {
+ log_error(LOG_LEVEL_ERROR, "Buffer limit reached while decompressing iob");
+ return JB_ERR_MEMORY;
+ }
+
+ /* Try doubling the buffer size each time. */
+ bufsize *= 2;
+
+ /* Don't exceed the buffer limit. */
+ if (bufsize > csp->config->buffer_limit)
+ {
+ bufsize = csp->config->buffer_limit;
+ }
+
+ /* Try to allocate the new buffer. */
+ tmpbuf = realloc(buf, bufsize);
+ if (NULL == tmpbuf)
+ {
+ log_error(LOG_LEVEL_ERROR, "Out of memory decompressing iob");
+ freez(buf);
+ return JB_ERR_MEMORY;
+ }
+ else
+ {
+ char *oldnext_out = (char *)zstr.next_out;
+
+ /*
+ * Update the fields for inflate() to use the new
+ * buffer, which may be in a location different from
+ * the old one.
+ */
+ zstr.avail_out += bufsize - oldbufsize;
+ zstr.next_out = (Bytef *)tmpbuf + bufsize - zstr.avail_out;
+
+ /*
+ * Compare with an uglier method of calculating these values
+ * that doesn't require the extra oldbufsize variable.
+ */
+ assert(zstr.avail_out == tmpbuf + bufsize - (char *)zstr.next_out);
+ assert((char *)zstr.next_out == tmpbuf + ((char *)oldnext_out - buf));
+ assert(zstr.avail_out > 0);
+
+ buf = tmpbuf;
+ }
+ }
+
+ if (Z_STREAM_ERROR == inflateEnd(&zstr))
+ {
+ log_error(LOG_LEVEL_ERROR,
+ "Inconsistent stream state after decompression: %s", zstr.msg);
+ /*
+ * XXX: Intentionally no return.
+ *
+ * According to zlib.h, Z_STREAM_ERROR is returned
+ * "if the stream state was inconsistent".
+ *
+ * I assume in this case inflate()'s status
+ * would also be something different than Z_STREAM_END
+ * so this check should be redundant, but lets see.
+ */
+ }
- /* set the pointers to the new values */
- iob->cur = iob->buf;
- iob->eod = p;
+ if (status != Z_STREAM_END)
+ {
+ /* We failed to decompress the stream. */
+ log_error(LOG_LEVEL_ERROR,
+ "Error in decompressing to the buffer (iob): %s", zstr.msg);
+ return JB_ERR_COMPRESS;
+ }
- return(need);
+ /*
+ * Finally, we can actually update the iob, since the
+ * decompression was successful. First, free the old
+ * buffer.
+ */
+ freez(csp->iob->buf);
+
+ /* Now, update the iob to use the new buffer. */
+ csp->iob->buf = buf;
+ csp->iob->cur = csp->iob->buf + skip_size;
+ csp->iob->eod = (char *)zstr.next_out;
+ csp->iob->size = bufsize;
+
+ /*
+ * Make sure the new uncompressed iob obeys some minimal
+ * consistency conditions.
+ */
+ if ((csp->iob->buf < csp->iob->cur)
+ && (csp->iob->cur <= csp->iob->eod)
+ && (csp->iob->eod <= csp->iob->buf + csp->iob->size))
+ {
+ char t = csp->iob->cur[100];
+ csp->iob->cur[100] = '\0';
+ /*
+ * XXX: The debug level should be lowered
+ * before the next stable release.
+ */
+ log_error(LOG_LEVEL_INFO, "Sucessfully decompressed: %s", csp->iob->cur);
+ csp->iob->cur[100] = t;
+ return JB_ERR_OK;
+ }
+ else
+ {
+ /* It seems that zlib did something weird. */
+ log_error(LOG_LEVEL_ERROR,
+ "Unexpected error decompressing the buffer (iob): %d==%d, %d>%d, %d<%d",
+ csp->iob->cur, csp->iob->buf + skip_size, csp->iob->eod, csp->iob->buf,
+ csp->iob->eod, csp->iob->buf + csp->iob->size);
+ return JB_ERR_COMPRESS;
+ }
}
+#endif /* defined(FEATURE_ZLIB) */
/*********************************************************************
*p = '\0';
ret = strdup(iob->cur);
+ if (ret == NULL)
+ {
+ /* FIXME No way to handle error properly */
+ log_error(LOG_LEVEL_FATAL, "Out of memory in get_header()");
+ }
iob->cur = p+1;
- if ((q = strchr(ret, '\r'))) *q = '\0';
+ if ((q = strchr(ret, '\r')) != NULL) *q = '\0';
- /* is this a blank linke (i.e. the end of the header) ? */
+ /* is this a blank line (i.e. the end of the header) ? */
if (*ret == '\0')
{
freez(ret);
/*********************************************************************
*
- * Function : enlist
+ * Function : get_header_value
*
- * Description : Append a string into a specified string list.
+ * Description : Get the value of a given header from a chained list
+ * of header lines or return NULL if no such header is
+ * present in the list.
*
* Parameters :
- * 1 : h = pointer to list 'dummy' header
- * 2 : s = string to add to the list
+ * 1 : header_list = pointer to list
+ * 2 : header_name = string with name of header to look for.
+ * Trailing colon required, capitalization
+ * doesn't matter.
*
- * Returns : N/A
+ * Returns : NULL if not found, else value of header
*
*********************************************************************/
-void enlist(struct list *h, const char *s)
+char *get_header_value(const struct list *header_list, const char *header_name)
{
- struct list *n = (struct list *)malloc(sizeof(*n));
- struct list *l;
+ struct list_entry *cur_entry;
+ char *ret = NULL;
+ size_t length = 0;
- if (n)
- {
- n->str = strdup(s);
- n->next = NULL;
+ assert(header_list);
+ assert(header_name);
+ length = strlen(header_name);
- if ((l = h->last))
- {
- l->next = n;
- }
- else
+ for (cur_entry = header_list->first; cur_entry ; cur_entry = cur_entry->next)
+ {
+ if (cur_entry->str)
{
- h->next = n;
+ if (!strncmpic(cur_entry->str, header_name, length))
+ {
+ /*
+ * Found: return pointer to start of value
+ */
+ ret = (char *) (cur_entry->str + length);
+ while (*ret && ijb_isspace(*ret)) ret++;
+ return(ret);
+ }
}
-
- h->last = n;
}
-}
+ /*
+ * Not found
+ */
+ return NULL;
+}
/*********************************************************************
*
- * Function : destroy_list
+ * Function : sed
+ *
+ * Description : add, delete or modify lines in the HTTP header streams.
+ * On entry, it receives a linked list of headers space
+ * that was allocated dynamically (both the list nodes
+ * and the header contents).
*
- * Description : Destroy a string list (opposite of enlist)
+ * As a side effect it frees the space used by the original
+ * header lines.
*
* Parameters :
- * 1 : h = pointer to list 'dummy' header
+ * 1 : pats = list of patterns to match against headers
+ * 2 : more_headers = list of functions to add more
+ * headers (client or server)
+ * 3 : csp = Current client state (buffers, headers, etc...)
*
- * Returns : N/A
+ * Returns : Single pointer to a fully formed header, or NULL
+ * on out-of-memory error.
*
*********************************************************************/
-void destroy_list(struct list *h)
+char *sed(const struct parsers pats[],
+ const add_header_func_ptr more_headers[],
+ struct client_state *csp)
{
- struct list *p, *n;
+ struct list_entry *p;
+ const struct parsers *v;
+ const add_header_func_ptr *f;
+ jb_err err = JB_ERR_OK;
+ int first_run;
+
+ /*
+ * If filtering is enabled, sed is run twice,
+ * but most of the work needs to be done only once.
+ */
+ first_run = (more_headers != NULL ) ? 1 : 0;
- for (p = h->next; p ; p = n)
+ if (first_run) /* Parse and print */
{
- n = p->next;
- freez(p->str);
- freez(p);
+ log_error(LOG_LEVEL_HEADER, "scanning headers for: %s", csp->http->url);
+ for (v = pats; (err == JB_ERR_OK) && (v->str != NULL) ; v++)
+ {
+ for (p = csp->headers->first; (err == JB_ERR_OK) && (p != NULL) ; p = p->next)
+ {
+ /* Header crunch()ed in previous run? -> ignore */
+ if (p->str == NULL) continue;
+
+ if (v == pats) log_error(LOG_LEVEL_HEADER, "scan: %s", p->str);
+
+ /* Does the current parser handle this header? */
+ if ((strncmpic(p->str, v->str, v->len) == 0) || (v->len == CHECK_EVERY_HEADER_REMAINING))
+ {
+ err = v->parser(csp, (char **)&(p->str));
+ }
+ }
+ }
+ /* place any additional headers on the csp->headers list */
+ for (f = more_headers; (err == JB_ERR_OK) && (*f) ; f++)
+ {
+ err = (*f)(csp);
+ }
+ }
+ else /* Parse only */
+ {
+ /*
+ * The second run is only needed if the body was modified
+ * and the content-lenght has changed.
+ */
+ if (strncmpic(csp->http->cmd, "HEAD", 4))
+ {
+ /*XXX: Code duplication */
+ for (v = pats; (err == JB_ERR_OK) && (v->str != NULL) ; v++)
+ {
+ for (p = csp->headers->first; (err == JB_ERR_OK) && (p != NULL) ; p = p->next)
+ {
+ /* Header crunch()ed in previous run? -> ignore */
+ if (p->str == NULL) continue;
+
+ /* Does the current parser handle this header? */
+ if (strncmpic(p->str, v->str, v->len) == 0)
+ {
+ err = v->parser(csp, (char **)&(p->str));
+ }
+ }
+ }
+ }
}
- memset(h, '\0', sizeof(*h));
+ if (err != JB_ERR_OK)
+ {
+ return NULL;
+ }
+ return list_to_text(csp->headers);
}
+/* here begins the family of parser functions that reformat header lines */
+
/*********************************************************************
*
- * Function : list_to_text
+ * Function : filter_server_header
*
- * Description : "Flaten" a string list into 1 long \r\n delimited string.
+ * Description : Checks if server header filtering is enabled.
+ * If it is, filter_header is called to do the work.
*
* Parameters :
- * 1 : h = pointer to list 'dummy' header
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
*
- * Returns : NULL on malloc error, else new long string.
+ * Returns : JB_ERR_OK on success and always succeeds
*
*********************************************************************/
-static char *list_to_text(struct list *h)
+jb_err filter_server_header(struct client_state *csp, char **header)
{
- struct list *p;
- char *ret = NULL;
- char *s;
- int size;
+ if (csp->action->flags & ACTION_FILTER_SERVER_HEADERS)
+ {
+ filter_header(csp, header);
+ }
+ return(JB_ERR_OK);
+}
+
+/*********************************************************************
+ *
+ * Function : filter_client_header
+ *
+ * Description : Checks if client header filtering is enabled.
+ * If it is, filter_header is called to do the work.
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
+ *
+ * Returns : JB_ERR_OK on success and always succeeds
+ *
+ *********************************************************************/
+jb_err filter_client_header(struct client_state *csp, char **header)
+{
+ if (csp->action->flags & ACTION_FILTER_CLIENT_HEADERS)
+ {
+ filter_header(csp, header);
+ }
+ return(JB_ERR_OK);
+}
+
+/*********************************************************************
+ *
+ * Function : filter_header
+ *
+ * Description : Executes all text substitutions from all applying
+ * +filter actions on the header.
+ * Most of the code was copied from pcrs_filter_response,
+ * including the rather short variable names
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
+ *
+ * Returns : JB_ERR_OK on success and always succeeds
+ *
+ *********************************************************************/
+jb_err filter_header(struct client_state *csp, char **header)
+{
+ int hits=0;
+ int matches;
+ size_t size = strlen(*header);
+
+ char *newheader = NULL;
+ pcrs_job *job;
- size = 0;
+ struct file_list *fl;
+ struct re_filterfile_spec *b;
+ struct list_entry *filtername;
- for (p = h->next; p ; p = p->next)
+ int i, found_filters = 0;
+
+ /*
+ * Need to check the set of re_filterfiles...
+ */
+ for (i = 0; i < MAX_AF_FILES; i++)
{
- if (p->str)
+ fl = csp->rlist[i];
+ if (NULL != fl)
{
- size += strlen(p->str) + 2;
+ if (NULL != fl->f)
+ {
+ found_filters = 1;
+ break;
+ }
}
}
- if ((ret = (char *)malloc(size + 1)) == NULL)
+ if (0 == found_filters)
{
- return(NULL);
+ log_error(LOG_LEVEL_ERROR, "Unable to get current state of regexp filtering.");
+ return(JB_ERR_OK);
}
- ret[size] = '\0';
-
- s = ret;
-
- for (p = h->next; p ; p = p->next)
+ for (i = 0; i < MAX_AF_FILES; i++)
{
- if (p->str)
+ fl = csp->rlist[i];
+ if ((NULL == fl) || (NULL == fl->f))
{
- strcpy(s, p->str);
- s += strlen(s);
- *s++ = '\r'; *s++ = '\n';
+ /*
+ * Either there are no filter files
+ * left, or this filter file just
+ * contains no valid filters.
+ *
+ * Continue to be sure we don't miss
+ * valid filter files that are chained
+ * after empty or invalid ones.
+ */
+ continue;
+ }
+ /*
+ * For all applying +filter actions, look if a filter by that
+ * name exists and if yes, execute its pcrs_joblist on the
+ * buffer.
+ */
+ for (b = fl->f; b; b = b->next)
+ {
+ for (filtername = csp->action->multi[ACTION_MULTI_FILTER]->first;
+ filtername ; filtername = filtername->next)
+ {
+ if (strcmp(b->name, filtername->str) == 0)
+ {
+ int current_hits = 0;
+
+ if ( NULL == b->joblist )
+ {
+ log_error(LOG_LEVEL_RE_FILTER, "Filter %s has empty joblist. Nothing to do.", b->name);
+ continue;
+ }
+
+ log_error(LOG_LEVEL_RE_FILTER, "re_filtering %s (size %d) with filter %s...",
+ *header, size, b->name);
+
+ /* Apply all jobs from the joblist */
+ for (job = b->joblist; NULL != job; job = job->next)
+ {
+ matches = pcrs_execute(job, *header, size, &newheader, &size);
+ if ( 0 < matches )
+ {
+ current_hits += matches;
+ log_error(LOG_LEVEL_HEADER, "Transforming \"%s\" to \"%s\"", *header, newheader);
+ freez(*header);
+ *header = newheader;
+ }
+ else if ( 0 == matches )
+ {
+ /* Filter doesn't change header */
+ freez(newheader);
+ }
+ else
+ {
+ /* RegEx failure */
+ log_error(LOG_LEVEL_ERROR, "Filtering \'%s\' with \'%s\' didn't work out: %s",
+ *header, b->name, pcrs_strerror(matches));
+ if( newheader != NULL)
+ {
+ log_error(LOG_LEVEL_ERROR, "Freeing what's left: %s", newheader);
+ freez(newheader);
+ }
+ }
+ }
+ log_error(LOG_LEVEL_RE_FILTER, " ...produced %d hits (new size %d).", current_hits, size);
+ hits += current_hits;
+ }
+ }
}
}
- return(ret);
+ /*
+ * Additionally checking for hits is important because if
+ * the continue hack is triggered, server headers can
+ * arrive empty to separate multiple heads from each other.
+ */
+ if ((0 == size) && hits)
+ {
+ log_error(LOG_LEVEL_HEADER, "Removing empty header %s", *header);
+ freez(*header);
+ }
+ return(JB_ERR_OK);
}
/*********************************************************************
*
- * Function : sed
+ * Function : crumble
*
- * Description : add, delete or modify lines in the HTTP header streams.
- * On entry, it receives a linked list of headers space
- * that was allocated dynamically (both the list nodes
- * and the header contents).
+ * Description : This is called if a header matches a pattern to "crunch"
*
- * As a side effect it frees the space used by the original
- * header lines.
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
+ *
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
+ *
+ *********************************************************************/
+jb_err crumble(struct client_state *csp, char **header)
+{
+ log_error(LOG_LEVEL_HEADER, "crumble crunched: %s!", *header);
+ freez(*header);
+ return JB_ERR_OK;
+}
+
+/*********************************************************************
+ *
+ * Function : crunch_server_header
+ *
+ * Description : Crunch server header if it matches a string supplied by the
+ * user. Called from `sed'.
*
* Parameters :
- * 1 : pats = list of patterns to match against headers
- * 2 : more_headers = list of functions to add more
- * headers (client or server)
- * 3 : csp = Current client state (buffers, headers, etc...)
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
*
- * Returns : Single pointer to a fully formed header.
+ * Returns : JB_ERR_OK on success and always succeeds
*
*********************************************************************/
-char *sed(const struct parsers pats[], void (* const more_headers[])(struct client_state *), struct client_state *csp)
+jb_err crunch_server_header(struct client_state *csp, char **header)
{
- struct list *p;
- const struct parsers *v;
- char *hdr;
- void (* const *f)();
+ const char *crunch_pattern;
+ /*Is there a header to crunch*/
- for (p = csp->headers->next; p ; p = p->next)
+ if ((csp->action->flags & ACTION_CRUNCH_SERVER_HEADER))
{
- log_error(LOG_LEVEL_HEADER, "scan: %s", p->str);
+ crunch_pattern = csp->action->string[ACTION_STRING_SERVER_HEADER];
- if ((v = match(p->str, pats)))
+ /*Is the current header the lucky one?*/
+ if (strstr(*header, crunch_pattern))
{
- hdr = v->parser(v, p->str, csp);
- freez(p->str);
- p->str = hdr;
+ log_error(LOG_LEVEL_HEADER, "Crunching server header: %s (contains: %s)", *header, crunch_pattern);
+ freez(*header);
}
-
}
- /* place any additional headers on the csp->headers list */
- for (f = more_headers; *f ; f++)
- {
- (*f)(csp);
- }
-
- /* add the blank line at the end of the header */
- enlist(csp->headers, "");
-
- hdr = list_to_text(csp->headers);
-
- return(hdr);
-
+ return JB_ERR_OK;
}
/*********************************************************************
*
- * Function : free_http_request
+ * Function : server_content_type
*
- * Description : Freez a http_request structure
+ * Description : Set the content-type for filterable types (text/.*,
+ * .*xml.*, javascript and image/gif) unless filtering has been
+ * forbidden (CT_TABOO) while parsing earlier headers.
+ * NOTE: Since text/plain is commonly used by web servers
+ * for files whose correct type is unknown, we don't
+ * set CT_TEXT for it.
*
* Parameters :
- * 1 : http = points to a http_request structure to free
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
*
- * Returns : N/A
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
*
*********************************************************************/
-void free_http_request(struct http_request *http)
+jb_err server_content_type(struct client_state *csp, char **header)
{
- freez(http->cmd);
- freez(http->gpc);
- freez(http->host);
- freez(http->hostport);
- freez(http->path);
- freez(http->ver);
+ /* Remove header if it isn't the first Content-Type header */
+ if(csp->content_type && (csp->content_type != CT_TABOO))
+ {
+ /*
+ * Another, slightly slower, way to see if
+ * we already parsed another Content-Type header.
+ */
+ assert(NULL != get_header_value(csp->headers, "Content-Type:"));
+
+ log_error(LOG_LEVEL_ERROR,
+ "Multiple Content-Type headers. Removing and ignoring: \'%s\'",
+ *header);
+ freez(*header);
+
+ return JB_ERR_OK;
+ }
+ if (!(csp->content_type & CT_TABOO))
+ {
+ if ((strstr(*header, " text/") && !strstr(*header, "plain"))
+ || strstr(*header, "xml")
+ || strstr(*header, "application/x-javascript"))
+ {
+ csp->content_type |= CT_TEXT;
+ }
+ else if (strstr(*header, " image/gif"))
+ {
+ csp->content_type |= CT_GIF;
+ }
+ else if (strstr(*header, " image/jpeg"))
+ {
+ csp->content_type |= CT_JPEG;
+ }
+ else
+ {
+ csp->content_type = 0;
+ }
+ }
+ /*
+ * Are we enabling text mode by force?
+ */
+ if (csp->action->flags & ACTION_FORCE_TEXT_MODE)
+ {
+ /*
+ * Do we really have to?
+ */
+ if (csp->content_type & CT_TEXT)
+ {
+ log_error(LOG_LEVEL_HEADER, "Text mode is already enabled.");
+ }
+ else
+ {
+ csp->content_type |= CT_TEXT;
+ log_error(LOG_LEVEL_HEADER, "Text mode enabled by force. Take cover!");
+ }
+ }
+ /*
+ * Are we messing with the content type?
+ */
+ if (csp->action->flags & ACTION_CONTENT_TYPE_OVERWRITE)
+ {
+ /*
+ * Make sure the user doesn't accidently
+ * change the content type of binary documents.
+ */
+ if (csp->content_type & CT_TEXT)
+ {
+ freez(*header);
+ *header = strdup("Content-Type: ");
+ string_append(header, csp->action->string[ACTION_STRING_CONTENT_TYPE]);
+
+ if (header == NULL)
+ {
+ log_error(LOG_LEVEL_HEADER, "Insufficient memory to replace Content-Type!");
+ return JB_ERR_MEMORY;
+ }
+ log_error(LOG_LEVEL_HEADER, "Modified: %s!", *header);
+ }
+ else
+ {
+ log_error(LOG_LEVEL_HEADER, "%s not replaced. It doesn't look like text. "
+ "Enable force-text-mode if you know what you're doing.", *header);
+ }
+ }
+ return JB_ERR_OK;
}
/*********************************************************************
*
- * Function : parse_http_request
+ * Function : server_transfer_coding
*
- * Description : Parse out the host and port from the URL. Find the
- * hostname & path, port (if ':'), and/or password (if '@')
+ * Description : - Prohibit filtering (CT_TABOO) if transfer coding compresses
+ * - Raise the CSP_FLAG_CHUNKED flag if coding is "chunked"
+ * - Remove header if body was chunked but has been
+ * de-chunked for filtering.
*
* Parameters :
- * 1 : req = URL (or is it URI?) to break down
- * 2 : http = pointer to the http structure to hold elements
- * 3 : csp = Current client state (buffers, headers, etc...)
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
*
- * Returns : N/A
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
*
*********************************************************************/
-void parse_http_request(char *req, struct http_request *http, struct client_state *csp)
+jb_err server_transfer_coding(struct client_state *csp, char **header)
{
- char *buf, *v[10], *url, *p;
- int n;
-
- memset(http, '\0', sizeof(*http));
-
- http->cmd = strdup(req);
-
- buf = strdup(req);
-
- n = ssplit(buf, " \r\n", v, SZ(v), 1, 1);
+ /*
+ * Turn off pcrs and gif filtering if body compressed
+ */
+ if (strstr(*header, "gzip") || strstr(*header, "compress") || strstr(*header, "deflate"))
+ {
+#ifdef FEATURE_ZLIB
+ /*
+ * XXX: Added to test if we could use CT_GZIP and CT_DEFLATE here.
+ */
+ log_error(LOG_LEVEL_INFO, "Marking content type for %s as CT_TABOO because of %s.",
+ csp->http->cmd, *header);
+#endif /* def FEATURE_ZLIB */
+ csp->content_type = CT_TABOO;
+ }
- if (n == 3)
+ /*
+ * Raise flag if body chunked
+ */
+ if (strstr(*header, "chunked"))
{
- /* this could be a CONNECT request */
- if (strcmpic(v[0], "connect") == 0)
+ csp->flags |= CSP_FLAG_CHUNKED;
+
+ /*
+ * If the body was modified, it has been de-chunked first
+ * and the header must be removed.
+ *
+ * FIXME: If there is more than one transfer encoding,
+ * only the "chunked" part should be removed here.
+ */
+ if (csp->flags & CSP_FLAG_MODIFIED)
{
- http->ssl = 1;
- http->gpc = strdup(v[0]);
- http->hostport = strdup(v[1]);
- http->ver = strdup(v[2]);
+ log_error(LOG_LEVEL_HEADER, "Removing: %s", *header);
+ freez(*header);
}
+ }
-#ifdef WEBDAV
+ return JB_ERR_OK;
+}
-/* This next line is a little ugly, but it simplifies the if statement below. */
-/* Basically if using webDAV, we want the OR condition to use these too. */
-/*
- * by haroon
- * These are the headers as defined in RFC2518 to add webDAV support
- */
-
-#define OR_WEBDAV || \
- (0 == strcmpic(v[0], "propfind")) || \
- (0 == strcmpic(v[0], "proppatch")) || \
- (0 == strcmpic(v[0], "move")) || \
- (0 == strcmpic(v[0], "copy")) || \
- (0 == strcmpic(v[0], "mkcol")) || \
- (0 == strcmpic(v[0], "lock")) || \
- (0 == strcmpic(v[0], "unlock"))
+/*********************************************************************
+ *
+ * Function : server_content_encoding
+ *
+ * Description : This function is run twice for each request,
+ * unless FEATURE_ZLIB and filtering are disabled.
+ *
+ * The first run is used to check if the content
+ * is compressed, if FEATURE_ZLIB is disabled
+ * filtering is then disabled as well, if FEATURE_ZLIB
+ * is enabled the content is marked for decompression.
+ *
+ * The second run is used to remove the Content-Encoding
+ * header if the decompression was successful.
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
+ *
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
+ *
+ *********************************************************************/
+jb_err server_content_encoding(struct client_state *csp, char **header)
+{
+#ifdef FEATURE_ZLIB
+ if ((csp->flags & CSP_FLAG_MODIFIED)
+ && (csp->content_type & (CT_GZIP | CT_DEFLATE)))
+ {
+ /*
+ * We successfully decompressed the content,
+ * and have to clean the header now, so the
+ * client no longer expects compressed data..
+ *
+ * XXX: There is a difference between cleaning
+ * and removing it completely.
+ */
+ log_error(LOG_LEVEL_HEADER, "Crunching: %s", *header);
+ freez(*header);
+ }
+ else if (strstr(*header, "gzip"))
+ {
+ /* Mark for gzip decompression */
+ csp->content_type |= CT_GZIP;
+ }
+ else if (strstr(*header, "deflate"))
+ {
+ /* Mark for zlib decompression */
+ csp->content_type |= CT_DEFLATE;
+ }
+ else if (strstr(*header, "compress"))
+ {
+ /*
+ * We can't decompress this; therefore we can't filter
+ * it either.
+ */
+ csp->content_type |= CT_TABOO;
+ }
+#else /* !defined(FEATURE_ZLIB) */
+ if (strstr(*header, "gzip") || strstr(*header, "compress") || strstr(*header, "deflate"))
+ {
+ /*
+ * Body is compressed, turn off pcrs and gif filtering.
+ */
+ csp->content_type |= CT_TABOO;
+ }
+#endif /* !defined(FEATURE_ZLIB) */
-#else /* No webDAV support is enabled. Provide an empty OR_WEBDAV macro. */
+ return JB_ERR_OK;
-#define OR_WEBDAV
+}
-#endif
- /* or it could be a GET or a POST (possibly webDAV too) */
- if ((strcmpic(v[0], "get") == 0) ||
- (strcmpic(v[0], "head") == 0) OR_WEBDAV ||
- (strcmpic(v[0], "post") == 0))
+/*********************************************************************
+ *
+ * Function : server_content_length
+ *
+ * Description : Adjust Content-Length header if we modified
+ * the body.
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
+ *
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
+ *
+ *********************************************************************/
+jb_err server_content_length(struct client_state *csp, char **header)
+{
+ const size_t max_header_length = 80;
+ if (csp->content_length != 0) /* Content length could have been modified */
+ {
+ /*
+ * XXX: Shouldn't we check if csp->content_length
+ * is different than the original value?
+ */
+ freez(*header);
+ *header = (char *) zalloc(max_header_length);
+ if (*header == NULL)
{
- http->ssl = 0;
- http->gpc = strdup(v[0]);
- url = v[1];
- http->ver = strdup(v[2]);
-
- if (strncmpic(url, "http://", 7) == 0)
- {
- url += 7;
- }
- else if (strncmpic(url, "https://", 8) == 0)
- {
- url += 8;
- }
- else
- {
- url = NULL;
- }
-
- if (url && (p = strchr(url, '/')))
- {
- http->path = strdup(p);
- *p = '\0';
- http->hostport = strdup(url);
- }
+ return JB_ERR_MEMORY;
}
+
+ snprintf(*header, max_header_length, "Content-Length: %d",
+ (int)csp->content_length);
+ log_error(LOG_LEVEL_HEADER, "Adjusted Content-Length to %d",
+ (int)csp->content_length);
}
- freez(buf);
+ return JB_ERR_OK;
+}
- if (http->hostport == NULL)
+/*********************************************************************
+ *
+ * Function : server_content_md5
+ *
+ * Description : Crumble any Content-MD5 headers if the document was
+ * modified. FIXME: Should we re-compute instead?
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
+ *
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
+ *
+ *********************************************************************/
+jb_err server_content_md5(struct client_state *csp, char **header)
+{
+ if (csp->flags & CSP_FLAG_MODIFIED)
{
- free_http_request(http);
- return;
+ log_error(LOG_LEVEL_HEADER, "Crunching Content-MD5");
+ freez(*header);
}
- buf = strdup(http->hostport);
+ return JB_ERR_OK;
+}
+/*********************************************************************
+ *
+ * Function : server_content_disposition
+ *
+ * Description : If enabled, blocks or modifies the "content-disposition" header.
+ * Called from `sed'.
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
+ *
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
+ *
+ *********************************************************************/
+jb_err server_content_disposition(struct client_state *csp, char **header)
+{
+ const char *newval;
- /* check if url contains password */
- n = ssplit(buf, "@", v, SZ(v), 1, 1);
- if (n == 2)
+ /*
+ * Are we messing with the content-disposition header?
+ */
+ if ((csp->action->flags & ACTION_HIDE_CONTENT_DISPOSITION) == 0)
{
- char * newbuf = NULL;
- newbuf = strdup(v[1]);
- freez(buf);
- buf = newbuf;
+ /*Me tinks not*/
+ return JB_ERR_OK;
}
- n = ssplit(buf, ":", v, SZ(v), 1, 1);
+ newval = csp->action->string[ACTION_STRING_CONTENT_DISPOSITION];
- if (n == 1)
+ if ((newval == NULL) || (0 == strcmpic(newval, "block")) )
{
- http->host = strdup(v[0]);
- http->port = 80;
+ /*
+ * Blocking content-disposition header
+ */
+ log_error(LOG_LEVEL_HEADER, "Crunching %s!", *header);
+ freez(*header);
+ return JB_ERR_OK;
+ }
+ else
+ {
+ /*
+ * Replacing content-disposition header
+ */
+ freez(*header);
+ *header = strdup("content-disposition: ");
+ string_append(header, newval);
+
+ if (*header == NULL)
+ {
+ log_error(LOG_LEVEL_HEADER, "Insufficent memory. content-disposition header not fully replaced.");
+ }
+ else
+ {
+ log_error(LOG_LEVEL_HEADER, "content-disposition header crunched and replaced with: %s", *header);
+ }
}
+ return (*header == NULL) ? JB_ERR_MEMORY : JB_ERR_OK;
+}
+
+/*********************************************************************
+ *
+ * Function : server_last_modified
+ *
+ * Description : Changes Last-Modified header to the actual date
+ * to help hide-if-modified-since.
+ * Called from `sed'.
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
+ *
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
+ *
+ *********************************************************************/
+jb_err server_last_modified(struct client_state *csp, char **header)
+{
+ const char *newval;
+ char buf[BUFFER_SIZE];
- if (n == 2)
+ char newheader[50];
+#ifdef HAVE_GMTIME_R
+ struct tm gmt;
+#endif
+ struct tm *timeptr = NULL;
+ time_t now, last_modified;
+ long int rtime;
+ long int days, hours, minutes, seconds;
+
+ /*
+ * Are we messing with the Last-Modified header?
+ */
+ if ((csp->action->flags & ACTION_OVERWRITE_LAST_MODIFIED) == 0)
{
- http->host = strdup(v[0]);
- http->port = atoi(v[1]);
+ /*Nope*/
+ return JB_ERR_OK;
}
- freez(buf);
+ newval = csp->action->string[ACTION_STRING_LAST_MODIFIED];
- if (http->host == NULL)
+ if (0 == strcmpic(newval, "block") )
{
- free_http_request(http);
+ /*
+ * Blocking Last-Modified header. Useless but why not.
+ */
+ log_error(LOG_LEVEL_HEADER, "Crunching %s!", *header);
+ freez(*header);
+ return JB_ERR_OK;
}
+ else if (0 == strcmpic(newval, "reset-to-request-time"))
+ {
+ /*
+ * Setting Last-Modified Header to now.
+ */
+ get_http_time(0, buf);
+ freez(*header);
+ *header = strdup("Last-Modified: ");
+ string_append(header, buf);
- if (http->path == NULL)
+ if (*header == NULL)
+ {
+ log_error(LOG_LEVEL_HEADER, "Insufficent memory. Last-Modified header got lost, boohoo.");
+ }
+ else
+ {
+ log_error(LOG_LEVEL_HEADER, "Reset to present time: %s", *header);
+ }
+ }
+ else if (0 == strcmpic(newval, "randomize"))
{
- http->path = strdup("");
+ log_error(LOG_LEVEL_HEADER, "Randomizing: %s", *header);
+ now = time(NULL);
+#ifdef HAVE_GMTIME_R
+ timeptr = gmtime_r(&now, &gmt);
+#elif FEATURE_PTHREAD
+ pthread_mutex_lock(&gmtime_mutex);
+ timeptr = gmtime(&now);
+ pthread_mutex_unlock(&gmtime_mutex);
+#else
+ timeptr = gmtime(&now);
+#endif
+ if ((timeptr = parse_header_time(*header, &last_modified)) == NULL)
+ {
+ log_error(LOG_LEVEL_HEADER, "Couldn't parse: %s (crunching!)", *header);
+ freez(*header);
+ }
+ else
+ {
+ rtime = (long int)difftime(now, last_modified);
+ if (rtime)
+ {
+ rtime = pick_from_range(rtime);
+ last_modified += rtime;
+#ifdef HAVE_GMTIME_R
+ timeptr = gmtime_r(&last_modified, &gmt);
+#elif FEATURE_PTHREAD
+ pthread_mutex_lock(&gmtime_mutex);
+ timeptr = gmtime(&last_modified);
+ pthread_mutex_unlock(&gmtime_mutex);
+#else
+ timeptr = gmtime(&last_modified);
+#endif
+ strftime(newheader, sizeof(newheader), "%a, %d %b %Y %H:%M:%S GMT", timeptr);
+ freez(*header);
+ *header = strdup("Last-Modified: ");
+ string_append(header, newheader);
+
+ if (*header == NULL)
+ {
+ log_error(LOG_LEVEL_ERROR, "Insufficent memory, header crunched without replacement.");
+ return JB_ERR_MEMORY;
+ }
+
+ if(LOG_LEVEL_HEADER & debug) /* Save cycles if the user isn't interested. */
+ {
+ days = rtime / (3600 * 24);
+ hours = rtime / 3600 % 24;
+ minutes = rtime / 60 % 60;
+ seconds = rtime % 60;
+
+ log_error(LOG_LEVEL_HEADER, "Randomized: %s (added %d da%s %d hou%s %d minut%s %d second%s",
+ *header, days, (days == 1) ? "y" : "ys", hours, (hours == 1) ? "r" : "rs",
+ minutes, (minutes == 1) ? "e" : "es", seconds, (seconds == 1) ? ")" : "s)");
+ }
+ }
+ else
+ {
+ log_error(LOG_LEVEL_HEADER, "Randomized ... or not. No time difference to work with.");
+ }
+ }
}
+ return JB_ERR_OK;
}
-
-/* here begins the family of parser functions that reformat header lines */
-
-
/*********************************************************************
*
- * Function : crumble
+ * Function : client_accept_encoding
*
- * Description : This is called if a header matches a pattern to "crunch"
+ * Description : Rewrite the client's Accept-Encoding header so that
+ * if doesn't allow compression, if the action applies.
+ * Note: For HTTP/1.0 the absence of the header is enough.
*
* Parameters :
- * 1 : v = Pointer to parsers structure, which basically holds
- * headers (client or server) that we want to "crunch"
- * 2 : s = header (from sed) to "crunch"
- * 3 : csp = Current client state (buffers, headers, etc...)
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
*
- * Returns : Always NULL.
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
*
*********************************************************************/
-char *crumble(const struct parsers *v, char *s, struct client_state *csp)
+jb_err client_accept_encoding(struct client_state *csp, char **header)
{
- log_error(LOG_LEVEL_HEADER, "crunch!");
- return(NULL);
+ if ((csp->action->flags & ACTION_NO_COMPRESSION) != 0)
+ {
+ log_error(LOG_LEVEL_HEADER, "Suppressed offer to compress content");
+
+ freez(*header);
+
+ /* Temporarily disable the correct behaviour to
+ * work around a PHP bug.
+ *
+ * if (!strcmpic(csp->http->ver, "HTTP/1.1"))
+ * {
+ * *header = strdup("Accept-Encoding: identity;q=1.0, *;q=0");
+ * if (*header == NULL)
+ * {
+ * return JB_ERR_MEMORY;
+ * }
+ * }
+ *
+ */
+ }
+ return JB_ERR_OK;
}
-#ifdef PCRS
-
/*********************************************************************
*
- * Function : content_type
+ * Function : client_te
*
- * Description : Is this a text/* or javascript MIME Type?
+ * Description : Rewrite the client's TE header so that
+ * if doesn't allow compression, if the action applies.
*
* Parameters :
- * 1 : v = ignored
- * 2 : s = header string we are "considering"
- * 3 : csp = Current client state (buffers, headers, etc...)
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
*
- * Returns : A duplicate string pointer to this header (ie. pass thru)
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
*
*********************************************************************/
-char *content_type(const struct parsers *v, char *s, struct client_state *csp)
+jb_err client_te(struct client_state *csp, char **header)
{
- if (strstr (s, " text/") || strstr (s, "application/x-javascript"))
- csp->is_text = 1;
- else
- csp->is_text = 0;
-
- return(strdup(s));
+ if ((csp->action->flags & ACTION_NO_COMPRESSION) != 0)
+ {
+ freez(*header);
+ log_error(LOG_LEVEL_HEADER, "Suppressed offer to compress transfer");
+ }
+ return JB_ERR_OK;
}
-#endif /* def PCRS */
-
-
/*********************************************************************
*
* Function : client_referrer
* Called from `sed'.
*
* Parameters :
- * 1 : v = ignored
- * 2 : s = header (from sed) to "crunch"
- * 3 : csp = Current client state (buffers, headers, etc...)
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
*
- * Returns : NULL if crunched, or a malloc'ed string with the original
- * or modified header
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
*
*********************************************************************/
-char *client_referrer(const struct parsers *v, char *s, struct client_state *csp)
+jb_err client_referrer(struct client_state *csp, char **header)
{
-#ifdef FORCE_LOAD
+ const char *newval;
+ const char *host;
+ char *referer;
+ size_t hostlenght;
+
+#ifdef FEATURE_FORCE_LOAD
/* Since the referrer can include the prefix even
- * even if the request itself is non-forced, we must
- * clean it unconditionally
+ * if the request itself is non-forced, we must
+ * clean it unconditionally
*/
- strclean(s, FORCE_PREFIX);
-#endif /* def FORCE_LOAD */
+ strclean(*header, FORCE_PREFIX);
+#endif /* def FEATURE_FORCE_LOAD */
- csp->referrer = strdup(s);
-
- if (referrer == NULL)
+ /*
+ * Are we sending referer?
+ */
+ if ((csp->action->flags & ACTION_HIDE_REFERER) == 0)
{
- log_error(LOG_LEVEL_HEADER, "crunch!");
- return(NULL);
+ return JB_ERR_OK;
}
- if (*referrer == '.')
+ newval = csp->action->string[ACTION_STRING_REFERER];
+
+ if ((0 != strcmpic(newval, "conditional-block")))
+ {
+ freez(*header);
+ }
+ if ((newval == NULL) || (0 == strcmpic(newval, "block")) )
{
- return(strdup(s));
+ /*
+ * Blocking referer
+ */
+ log_error(LOG_LEVEL_HEADER, "Referer crunched!");
+ return JB_ERR_OK;
}
-
- if (*referrer == '@')
+ else if (0 == strcmpic(newval, "conditional-block"))
{
- if (csp->send_user_cookie)
+ /*
+ * Block referer if host has changed.
+ */
+
+ if (NULL == (host = strdup(csp->http->hostport)))
+ {
+ freez(*header);
+ log_error(LOG_LEVEL_HEADER, "Referer crunched! Couldn't allocate memory for temporary host copy.");
+ return JB_ERR_MEMORY;
+ }
+ if (NULL == (referer = strdup(*header)))
{
- return(strdup(s));
+ freez(*header);
+ freez(host);
+ log_error(LOG_LEVEL_HEADER, "Referer crunched! Couldn't allocate memory for temporary referer copy.");
+ return JB_ERR_MEMORY;
+ }
+ hostlenght = strlen(host);
+ if ( hostlenght < (strlen(referer)-17) ) /*referer begins with 'Referer: http[s]://'*/
+ {
+ /*Shorten referer to make sure the referer is blocked
+ *if www.example.org/www.example.com-shall-see-the-referer/
+ *links to www.example.com/
+ */
+ referer[hostlenght+17] = '\0';
+ }
+ if ( 0 == strstr(referer, host)) /*Host has changed*/
+ {
+ log_error(LOG_LEVEL_HEADER, "New host is: %s. Crunching %s!", host, *header);
+ freez(*header);
}
else
{
- log_error(LOG_LEVEL_HEADER, "crunch!");
- return(NULL);
+ log_error(LOG_LEVEL_HEADER, "%s (not modified, still on %s)", *header, host);
}
+ freez(referer);
+ freez(host);
+ return JB_ERR_OK;
}
+ else if (0 != strcmpic(newval, "forge"))
+ {
+ /*
+ * We have a specific (fixed) referer we want to send.
+ */
+ if ((0 != strncmpic(newval, "http://", 7)) && (0 != strncmpic(newval, "https://", 8)))
+ {
+ log_error(LOG_LEVEL_HEADER, "Parameter: +referrer{%s} is a bad idea, but I don't care.", newval);
+ }
+ *header = strdup("Referer: ");
+ string_append(header, newval);
+ log_error(LOG_LEVEL_HEADER, "Referer overwritten with: %s", *header);
+
+ return (*header == NULL) ? JB_ERR_MEMORY : JB_ERR_OK;
+ }
+ else
+ {
+ /*
+ * Forge a referer as http://[hostname:port of REQUEST]/
+ * to fool stupid checks for in-site links
+ */
+
+ *header = strdup("Referer: http://");
+ string_append(header, csp->http->hostport);
+ string_append(header, "/");
+ log_error(LOG_LEVEL_HEADER, "Referer forged to: %s", *header);
+
+ return (*header == NULL) ? JB_ERR_MEMORY : JB_ERR_OK;
+ }
+}
+
+/*********************************************************************
+ *
+ * Function : client_accept_language
+ *
+ * Description : Handle the "Accept-Language" config setting properly.
+ * Called from `sed'.
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
+ *
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
+ *
+ *********************************************************************/
+jb_err client_accept_language(struct client_state *csp, char **header)
+{
+ const char *newval;
/*
- * New option § or L: Forge a referer as http://[hostname:port of REQUEST]/
- * to fool stupid checks for in-site links
+ * Are we messing with the Accept-Language?
*/
+ if ((csp->action->flags & ACTION_HIDE_ACCEPT_LANGUAGE) == 0)
+ {
+ /*I don't think so*/
+ return JB_ERR_OK;
+ }
+
+ newval = csp->action->string[ACTION_STRING_LANGUAGE];
- if (*referrer == '§' || *referrer == 'L')
+ if ((newval == NULL) || (0 == strcmpic(newval, "block")) )
{
- if (csp->send_user_cookie)
+ /*
+ * Blocking Accept-Language header
+ */
+ log_error(LOG_LEVEL_HEADER, "Crunching Accept-Language!");
+ freez(*header);
+ return JB_ERR_OK;
+ }
+ else
+ {
+ /*
+ * Replacing Accept-Language header
+ */
+ freez(*header);
+ *header = strdup("Accept-Language: ");
+ string_append(header, newval);
+
+ if (*header == NULL)
{
- return(strdup(s));
+ log_error(LOG_LEVEL_ERROR,
+ "Insufficent memory. Accept-Language header crunched without replacement.");
}
else
{
- log_error(LOG_LEVEL_HEADER, "crunch+forge!");
- s = strsav(NULL, "Referer: ");
- s = strsav(s, "http://");
- s = strsav(s, csp->http->hostport);
- s = strsav(s, "/");
- return(s);
+ log_error(LOG_LEVEL_HEADER,
+ "Accept-Language header crunched and replaced with: %s", *header);
}
}
+ return (*header == NULL) ? JB_ERR_MEMORY : JB_ERR_OK;
+}
- log_error(LOG_LEVEL_HEADER, "modified");
-
- s = strsav( NULL, "Referer: " );
- s = strsav( s, referrer );
- return(s);
+/*********************************************************************
+ *
+ * Function : crunch_client_header
+ *
+ * Description : Crunch client header if it matches a string supplied by the
+ * user. Called from `sed'.
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
+ *
+ * Returns : JB_ERR_OK on success and always succeeds
+ *
+ *********************************************************************/
+jb_err crunch_client_header(struct client_state *csp, char **header)
+{
+ const char *crunch_pattern;
+ /*Is there a header to crunch*/
+
+ if ((csp->action->flags & ACTION_CRUNCH_CLIENT_HEADER))
+ {
+ crunch_pattern = csp->action->string[ACTION_STRING_CLIENT_HEADER];
+ /*Is the current header the lucky one?*/
+ if (strstr(*header, crunch_pattern))
+ {
+ log_error(LOG_LEVEL_HEADER, "Crunching client header: %s (contains: %s)", *header, crunch_pattern);
+ freez(*header);
+ }
+ }
+ return JB_ERR_OK;
}
*
* Function : client_uagent
*
- * Description : Handle the "user-agent" config setting properly.
+ * Description : Handle the "user-agent" config setting properly
+ * and remember its original value to enable browser
+ * bug workarounds. Called from `sed'.
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
+ *
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
+ *
+ *********************************************************************/
+jb_err client_uagent(struct client_state *csp, char **header)
+{
+ const char *newval;
+
+ if ((csp->action->flags & ACTION_HIDE_USER_AGENT) == 0)
+ {
+ return JB_ERR_OK;
+ }
+
+ newval = csp->action->string[ACTION_STRING_USER_AGENT];
+ if (newval == NULL)
+ {
+ return JB_ERR_OK;
+ }
+
+ freez(*header);
+ *header = strdup("User-Agent: ");
+ string_append(header, newval);
+
+ log_error(LOG_LEVEL_HEADER, "Modified: %s", *header);
+
+ return (*header == NULL) ? JB_ERR_MEMORY : JB_ERR_OK;
+}
+
+/*********************************************************************
+ *
+ * Function : client_ua
+ *
+ * Description : Handle "ua-" headers properly. Called from `sed'.
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
+ *
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
+ *
+ *********************************************************************/
+jb_err client_ua(struct client_state *csp, char **header)
+{
+ if ((csp->action->flags & ACTION_HIDE_USER_AGENT) != 0)
+ {
+ log_error(LOG_LEVEL_HEADER, "crunched User-Agent!");
+ freez(*header);
+ }
+
+ return JB_ERR_OK;
+}
+
+
+/*********************************************************************
+ *
+ * Function : client_from
+ *
+ * Description : Handle the "from" config setting properly.
* Called from `sed'.
*
* Parameters :
- * 1 : v = ignored
- * 2 : s = header (from sed) to "crunch"
- * 3 : csp = Current client state (buffers, headers, etc...)
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
*
- * Returns : A malloc'ed pointer to the default agent, or
- * a malloc'ed string pointer to this header (ie. pass thru).
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
*
*********************************************************************/
-char *client_uagent(const struct parsers *v, char *s, struct client_state *csp)
+jb_err client_from(struct client_state *csp, char **header)
{
-#ifdef DETECT_MSIE_IMAGES
- if (strstr (s, "MSIE "))
+ const char *newval;
+
+ if ((csp->action->flags & ACTION_HIDE_FROM) == 0)
{
- /* This is Microsoft Internet Explorer.
- * Enable auto-detect.
- */
- csp->accept_types |= ACCEPT_TYPE_IS_MSIE;
+ return JB_ERR_OK;
}
-#endif /* def DETECT_MSIE_IMAGES */
- if (uagent == NULL)
+ freez(*header);
+
+ newval = csp->action->string[ACTION_STRING_FROM];
+
+ /*
+ * Are we blocking the e-mail address?
+ */
+ if ((newval == NULL) || (0 == strcmpic(newval, "block")) )
{
- log_error(LOG_LEVEL_HEADER, "default");
- return(strdup(DEFAULT_USER_AGENT));
+ log_error(LOG_LEVEL_HEADER, "crunched From!");
+ return JB_ERR_OK;
}
- if (*uagent == '.')
+ log_error(LOG_LEVEL_HEADER, " modified");
+
+ *header = strdup("From: ");
+ string_append(header, newval);
+
+ return (*header == NULL) ? JB_ERR_MEMORY : JB_ERR_OK;
+}
+
+
+/*********************************************************************
+ *
+ * Function : client_send_cookie
+ *
+ * Description : Handle the "cookie" header properly. Called from `sed'.
+ * If cookie is accepted, add it to the cookie_list,
+ * else we crunch it. Mmmmmmmmmmm ... cookie ......
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
+ *
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
+ *
+ *********************************************************************/
+jb_err client_send_cookie(struct client_state *csp, char **header)
+{
+ jb_err result = JB_ERR_OK;
+
+ if ((csp->action->flags & ACTION_NO_COOKIE_READ) == 0)
{
- return(strdup(s));
+ /* strlen("cookie: ") == 8 */
+ result = enlist(csp->cookie_list, *header + 8);
+ }
+ else
+ {
+ log_error(LOG_LEVEL_HEADER, "Crunched outgoing cookie -- yum!");
}
- if (*uagent == '@')
+ /*
+ * Always remove the cookie here. The cookie header
+ * will be sent at the end of the header.
+ */
+ freez(*header);
+
+ return result;
+}
+
+
+/*********************************************************************
+ *
+ * Function : client_x_forwarded
+ *
+ * Description : Handle the "x-forwarded-for" config setting properly,
+ * also used in the add_client_headers list. Called from `sed'.
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
+ *
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
+ *
+ *********************************************************************/
+jb_err client_x_forwarded(struct client_state *csp, char **header)
+{
+ if ((csp->action->flags & ACTION_HIDE_FORWARDED) == 0)
+ {
+ /* Save it so we can re-add it later */
+ freez(csp->x_forwarded);
+ csp->x_forwarded = *header;
+
+ /*
+ * Always set *header = NULL, since this information
+ * will be sent at the end of the header.
+ */
+ *header = NULL;
+ }
+ else
+ {
+ freez(*header);
+ log_error(LOG_LEVEL_HEADER, "crunched x-forwarded-for!");
+ }
+
+ return JB_ERR_OK;
+}
+
+
+/*********************************************************************
+ *
+ * Function : client_max_forwards
+ *
+ * Description : If the HTTP method is OPTIONS or TRACE, subtract one
+ * from the value of the Max-Forwards header field.
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
+ *
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
+ *
+ *********************************************************************/
+jb_err client_max_forwards(struct client_state *csp, char **header)
+{
+ int max_forwards;
+
+ if ((0 == strcmpic(csp->http->gpc, "trace")) ||
+ (0 == strcmpic(csp->http->gpc, "options")))
{
- if (csp->send_user_cookie)
+ assert(*(*header+12) == ':');
+ if (1 == sscanf(*header+12, ": %u", &max_forwards))
{
- return(strdup(s));
+ if (max_forwards > 0)
+ {
+ snprintf(*header, strlen(*header)+1, "Max-Forwards: %u", --max_forwards);
+ log_error(LOG_LEVEL_HEADER, "Max-Forwards value for %s request reduced to %u.",
+ csp->http->gpc, max_forwards);
+ }
+ else if (max_forwards < 0)
+ {
+ log_error(LOG_LEVEL_ERROR, "Crunching invalid header: %s", *header);
+ freez(*header);
+ }
+ else
+ {
+ /*
+ * Not supposed to be reached. direct_response() which
+ * was already called earlier in chat() should have
+ * intercepted the request.
+ */
+ log_error(LOG_LEVEL_ERROR,
+ "Non-intercepted %s request with Max-Forwards zero!", csp->http->gpc);
+ assert(max_forwards != 0);
+ }
}
else
{
- log_error(LOG_LEVEL_HEADER, "default");
- return(strdup(DEFAULT_USER_AGENT));
+ log_error(LOG_LEVEL_ERROR, "Crunching invalid header: %s", *header);
+ freez(*header);
}
}
- log_error(LOG_LEVEL_HEADER, "modified");
-
- s = strsav( NULL, "User-Agent: " );
- s = strsav( s, uagent );
- return(s);
-
+ return JB_ERR_OK;
}
/*********************************************************************
*
- * Function : client_ua
+ * Function : client_host
*
- * Description : Handle "ua-" headers properly. Called from `sed'.
+ * Description : If the request URI did not contain host and
+ * port information, parse and evaluate the Host
+ * header field.
+ *
+ * Also, kill ill-formed HOST: headers as sent by
+ * Apple's iTunes software when used with a proxy.
*
* Parameters :
- * 1 : v = ignored
- * 2 : s = header (from sed) to "crunch"
- * 3 : csp = Current client state (buffers, headers, etc...)
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
*
- * Returns : NULL if crunched, or a malloc'ed string to original header
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
*
*********************************************************************/
-char *client_ua(const struct parsers *v, char *s, struct client_state *csp)
+jb_err client_host(struct client_state *csp, char **header)
{
- if (uagent == NULL)
- {
- log_error(LOG_LEVEL_HEADER, "crunch!");
- return(NULL);
- }
+ char *p, *q;
- if (*uagent == '.')
+ /*
+ * If the header field name is all upper-case, chances are that it's
+ * an ill-formed one from iTunes. BTW, killing innocent headers here is
+ * not a problem -- they are regenerated later.
+ */
+ if ((*header)[1] == 'O')
{
- return(strdup(s));
+ log_error(LOG_LEVEL_HEADER, "Killed all-caps Host header line: %s", *header);
+ freez(*header);
+ return JB_ERR_OK;
}
- if (*uagent == '@')
+ if (!csp->http->hostport || (*csp->http->hostport == '*') ||
+ *csp->http->hostport == ' ' || *csp->http->hostport == '\0')
{
- if (csp->send_user_cookie)
+
+ if (NULL == (p = strdup((*header)+6)))
+ {
+ return JB_ERR_MEMORY;
+ }
+ chomp(p);
+ if (NULL == (q = strdup(p)))
+ {
+ freez(p);
+ return JB_ERR_MEMORY;
+ }
+
+ freez(csp->http->hostport);
+ csp->http->hostport = p;
+ freez(csp->http->host);
+ csp->http->host = q;
+ q = strchr(csp->http->host, ':');
+ if (q != NULL)
{
- return(strdup(s));
+ /* Terminate hostname and evaluate port string */
+ *q++ = '\0';
+ csp->http->port = atoi(q);
}
else
{
- log_error(LOG_LEVEL_HEADER, "crunch!");
- return(NULL);
+ csp->http->port = csp->http->ssl ? 443 : 80;
}
- }
- log_error(LOG_LEVEL_HEADER, "crunch!");
- return(NULL);
+ log_error(LOG_LEVEL_HEADER, "New host and port from Host field: %s = %s:%d",
+ csp->http->hostport, csp->http->host, csp->http->port);
+ }
+ return JB_ERR_OK;
}
-
/*********************************************************************
*
- * Function : client_from
+ * Function : client_if_modified_since
*
- * Description : Handle the "from" config setting properly.
- * Called from `sed'.
+ * Description : Remove or modify the If-Modified-Since header.
*
* Parameters :
- * 1 : v = ignored
- * 2 : s = header (from sed) to "crunch"
- * 3 : csp = Current client state (buffers, headers, etc...)
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
*
- * Returns : NULL if crunched, or a malloc'ed string to
- * modified/original header.
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
*
*********************************************************************/
-char *client_from(const struct parsers *v, char *s, struct client_state *csp)
+jb_err client_if_modified_since(struct client_state *csp, char **header)
{
- /* if not set, zap it */
- if (from == NULL)
+ char newheader[50];
+#ifdef HAVE_GMTIME_R
+ struct tm gmt;
+#endif
+ struct tm *timeptr = NULL;
+ time_t tm = 0;
+ const char *newval;
+ long int rtime;
+ long int hours, minutes, seconds;
+ int negative = 0;
+ char * endptr;
+
+ if ( 0 == strcmpic(*header, "If-Modified-Since: Wed, 08 Jun 1955 12:00:00 GMT"))
{
- log_error(LOG_LEVEL_HEADER, "crunch!");
- return(NULL);
+ /*
+ * The client got an error message because of a temporary problem,
+ * the problem is gone and the client now tries to revalidate our
+ * error message on the real server. The revalidation would always
+ * end with the transmission of the whole document and there is
+ * no need to expose the bogus If-Modified-Since header.
+ */
+ log_error(LOG_LEVEL_HEADER, "Crunching useless If-Modified-Since header.");
+ freez(*header);
}
-
- if (*from == '.')
+ else if (csp->action->flags & ACTION_HIDE_IF_MODIFIED_SINCE)
{
- return(strdup(s));
- }
-
- log_error(LOG_LEVEL_HEADER, " modified");
+ newval = csp->action->string[ACTION_STRING_IF_MODIFIED_SINCE];
- s = strsav( NULL, "From: " );
- s = strsav( s, from );
- return(s);
+ if ((0 == strcmpic(newval, "block")))
+ {
+ log_error(LOG_LEVEL_HEADER, "Crunching %s", *header);
+ freez(*header);
+ }
+ else /* add random value */
+ {
+ if ((timeptr = parse_header_time(*header, &tm)) == NULL)
+ {
+ log_error(LOG_LEVEL_HEADER, "Couldn't parse: %s (crunching!)", *header);
+ freez(*header);
+ }
+ else
+ {
+ rtime = strtol(newval, &endptr, 0);
+ if(rtime)
+ {
+ log_error(LOG_LEVEL_HEADER, "Randomizing: %s (random range: %d minut%s)",
+ *header, rtime, (rtime == 1 || rtime == -1) ? "e": "es");
+ if(rtime < 0)
+ {
+ rtime *= -1;
+ negative = 1;
+ }
+ rtime *= 60;
+ rtime = pick_from_range(rtime);
+ }
+ else
+ {
+ log_error(LOG_LEVEL_ERROR, "Random range is 0. Assuming time transformation test.",
+ *header);
+ }
+ tm += rtime * (negative ? -1 : 1);
+#ifdef HAVE_GMTIME_R
+ timeptr = gmtime_r(&tm, &gmt);
+#elif FEATURE_PTHREAD
+ pthread_mutex_lock(&gmtime_mutex);
+ timeptr = gmtime(&tm);
+ pthread_mutex_unlock(&gmtime_mutex);
+#else
+ timeptr = gmtime(&tm);
+#endif
+ strftime(newheader, sizeof(newheader), "%a, %d %b %Y %H:%M:%S GMT", timeptr);
+
+ freez(*header);
+ *header = strdup("If-Modified-Since: ");
+ string_append(header, newheader);
+
+ if (*header == NULL)
+ {
+ log_error(LOG_LEVEL_HEADER, "Insufficent memory, header crunched without replacement.");
+ return JB_ERR_MEMORY;
+ }
+
+ if(LOG_LEVEL_HEADER & debug) /* Save cycles if the user isn't interested. */
+ {
+ hours = rtime / 3600;
+ minutes = rtime / 60 % 60;
+ seconds = rtime % 60;
+
+ log_error(LOG_LEVEL_HEADER, "Randomized: %s (%s %d hou%s %d minut%s %d second%s",
+ *header, (negative) ? "subtracted" : "added", hours, (hours == 1) ? "r" : "rs",
+ minutes, (minutes == 1) ? "e" : "es", seconds, (seconds == 1) ? ")" : "s)");
+ }
+ }
+ }
+ }
+ return JB_ERR_OK;
}
-
/*********************************************************************
*
- * Function : client_send_cookie
+ * Function : client_if_none_match
*
- * Description : Handle the "cookie" header properly. Called from `sed'.
- * If cookie is accepted, add it to the cookie_list,
- * else we crunch it. Mmmmmmmmmmm ... cookie ......
+ * Description : Remove the If-None-Match header.
*
* Parameters :
- * 1 : v = pattern of cookie `sed' found matching
- * 2 : s = header (from sed) to "crunch"
- * 3 : csp = Current client state (buffers, headers, etc...)
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
*
- * Returns : Always NULL.
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
*
*********************************************************************/
-char *client_send_cookie(const struct parsers *v, char *s, struct client_state *csp)
+jb_err client_if_none_match(struct client_state *csp, char **header)
{
- if (csp->send_user_cookie)
- {
- enlist(csp->cookie_list, s + v->len + 1);
- }
- else
- {
- log_error(LOG_LEVEL_HEADER, " crunch!");
+ if (csp->action->flags & ACTION_CRUNCH_IF_NONE_MATCH)
+ {
+ log_error(LOG_LEVEL_HEADER, "Crunching %s", *header);
+ freez(*header);
}
- /*
- * Always return NULL here. The cookie header
- * will be sent at the end of the header.
- */
- return(NULL);
-
+ return JB_ERR_OK;
}
-
/*********************************************************************
*
- * Function : client_x_forwarded
+ * Function : client_x_filter
*
- * Description : Handle the "x-forwarded-for" config setting properly,
- * also used in the add_client_headers list. Called from `sed'.
+ * Description : Disables filtering if the client set "X-Filter: No".
+ * Called from `sed'.
*
* Parameters :
- * 1 : v = ignored
- * 2 : s = header (from sed) to "crunch"
- * 3 : csp = Current client state (buffers, headers, etc...)
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
*
- * Returns : Always NULL.
+ * Returns : JB_ERR_OK on success
*
*********************************************************************/
-char *client_x_forwarded(const struct parsers *v, char *s, struct client_state *csp)
+jb_err client_x_filter(struct client_state *csp, char **header)
{
- if (add_forwarded)
+ if ( 0 == strcmpic(*header, "X-Filter: No"))
{
- csp->x_forwarded = strdup(s);
+ if (!(csp->config->feature_flags & RUNTIME_FEATURE_HTTP_TOGGLE))
+ {
+ log_error(LOG_LEVEL_INFO, "Ignored the client's request to fetch without filtering.");
+ }
+ else
+ {
+ if (csp->action->flags & ACTION_FORCE_TEXT_MODE)
+ {
+ log_error(LOG_LEVEL_HEADER,
+ "force-text-mode overruled the client's request to fetch without filtering!");
+ }
+ else
+ {
+ csp->content_type = CT_TABOO;
+ csp->action->flags &= ~ACTION_FILTER_SERVER_HEADERS;
+ csp->action->flags &= ~ACTION_FILTER_CLIENT_HEADERS;
+ log_error(LOG_LEVEL_HEADER, "Accepted the client's request to fetch without filtering.");
+ }
+ log_error(LOG_LEVEL_HEADER, "Crunching %s", *header);
+ freez(*header);
+ }
}
-
- /*
- * Always return NULL, since this information
- * will be sent at the end of the header.
- */
-
- return(NULL);
-
+ return JB_ERR_OK;
}
-#if defined(DETECT_MSIE_IMAGES)
+/* the following functions add headers directly to the header list */
+
/*********************************************************************
*
- * Function : client_accept
+ * Function : client_host_adder
*
- * Description : Detect whether the client wants HTML or an image.
- * Clients do not always make this information available
- * in a sane way. Always passes the header through
- * the proxy unchanged.
+ * Description : Adds the Host: header field if it is missing.
+ * Called from `sed'.
*
* Parameters :
- * 1 : v = Ignored.
- * 2 : s = Header string. Null terminated.
- * 3 : csp = Current client state (buffers, headers, etc...)
+ * 1 : csp = Current client state (buffers, headers, etc...)
*
- * Returns : Duplicate of argument s.
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
*
*********************************************************************/
-char *client_accept(const struct parsers *v, char *s, struct client_state *csp)
+jb_err client_host_adder(struct client_state *csp)
{
-#ifdef DETECT_MSIE_IMAGES
- if (strstr (s, "image/gif"))
+ char *p;
+ jb_err err;
+
+ if ( !csp->http->hostport || !*(csp->http->hostport))
{
- /* Client will accept HTML. If this seems counterintuitive,
- * blame Microsoft.
- */
- csp->accept_types |= ACCEPT_TYPE_MSIE_HTML;
+ return JB_ERR_OK;
+ }
+
+ /*
+ * remove 'user:pass@' from 'proto://user:pass@host'
+ */
+ if ( (p = strchr( csp->http->hostport, '@')) != NULL )
+ {
+ p++;
}
else
{
- csp->accept_types |= ACCEPT_TYPE_MSIE_IMAGE;
+ p = csp->http->hostport;
}
-#endif /* def DETECT_MSIE_IMAGES */
- return(strdup(s));
+ log_error(LOG_LEVEL_HEADER, "addh-unique: Host: %s", p);
+ err = enlist_unique_header(csp->headers, "Host", p);
+ return err;
}
-#endif /* defined(DETECT_MSIE_IMAGES) */
-
-
-
-/* the following functions add headers directly to the header list */
/*********************************************************************
* Parameters :
* 1 : csp = Current client state (buffers, headers, etc...)
*
- * Returns : N/A
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
*
*********************************************************************/
-void client_cookie_adder(struct client_state *csp)
+jb_err client_cookie_adder(struct client_state *csp)
{
- struct list *l;
- char *tmp = NULL;
- char *e;
+ struct list_entry *lst;
+ char *tmp;
+ struct list_entry *list1 = csp->cookie_list->first;
+ struct list_entry *list2 = csp->action->multi[ACTION_MULTI_WAFER]->first;
+ int first_cookie = 1;
+ jb_err err;
+
+ if ((list1 == NULL) && (list2 == NULL))
+ {
+ /* Nothing to do */
+ return JB_ERR_OK;
+ }
- for (l = csp->cookie_list->next; l ; l = l->next)
+ tmp = strdup("Cookie: ");
+
+ for (lst = list1; lst ; lst = lst->next)
{
- if (tmp)
+ if (first_cookie)
+ {
+ first_cookie = 0;
+ }
+ else
{
- tmp = strsav(tmp, "; ");
+ string_append(&tmp, "; ");
}
- tmp = strsav(tmp, l->str);
+ string_append(&tmp, lst->str);
}
- for (l = wafer_list->next; l ; l = l->next)
+ for (lst = list2; lst ; lst = lst->next)
{
- if (tmp)
+ if (first_cookie)
{
- tmp = strsav(tmp, "; ");
+ first_cookie = 0;
}
-
- if ((e = cookie_encode(l->str)))
+ else
{
- tmp = strsav(tmp, e);
- freez(e);
+ string_append(&tmp, "; ");
}
+ string_join(&tmp, cookie_encode(lst->str));
}
- if (tmp)
+ if (tmp == NULL)
{
- char *ret;
+ return JB_ERR_MEMORY;
+ }
- ret = strdup("Cookie: ");
- ret = strsav(ret, tmp);
- log_error(LOG_LEVEL_HEADER, "addh: %s", ret);
- enlist(csp->headers, ret);
- freez(tmp);
- freez(ret);
+ log_error(LOG_LEVEL_HEADER, "addh: %s", tmp);
+ err = enlist(csp->headers, tmp);
+ free(tmp);
+ return err;
+}
+
+
+/*********************************************************************
+ *
+ * Function : client_accept_encoding_adder
+ *
+ * Description : Add an Accept-Encoding header to the client's request
+ * that disables compression if the action applies, and
+ * the header is not already there. Called from `sed'.
+ * Note: For HTTP/1.0, the absence of the header is enough.
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ *
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
+ *
+ *********************************************************************/
+jb_err client_accept_encoding_adder(struct client_state *csp)
+{
+ if ( ((csp->action->flags & ACTION_NO_COMPRESSION) != 0)
+ && (!strcmpic(csp->http->ver, "HTTP/1.1")) )
+ {
+ return enlist_unique(csp->headers, "Accept-Encoding: identity;q=1.0, *;q=0", 16);
}
+ return JB_ERR_OK;
}
* Parameters :
* 1 : csp = Current client state (buffers, headers, etc...)
*
- * Returns : N/A
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
*
*********************************************************************/
-void client_xtra_adder(struct client_state *csp)
+jb_err client_xtra_adder(struct client_state *csp)
{
- struct list *l;
+ struct list_entry *lst;
+ jb_err err;
- for (l = xtra_list->next; l ; l = l->next)
+ for (lst = csp->action->multi[ACTION_MULTI_ADD_HEADER]->first;
+ lst ; lst = lst->next)
{
- log_error(LOG_LEVEL_HEADER, "addh: %s", l->str);
- enlist(csp->headers, l->str);
+ log_error(LOG_LEVEL_HEADER, "addh: %s", lst->str);
+ err = enlist(csp->headers, lst->str);
+ if (err)
+ {
+ return err;
+ }
+
}
+ return JB_ERR_OK;
}
* Parameters :
* 1 : csp = Current client state (buffers, headers, etc...)
*
- * Returns : N/A
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
*
*********************************************************************/
-void client_x_forwarded_adder(struct client_state *csp)
+jb_err client_x_forwarded_adder(struct client_state *csp)
{
char *p = NULL;
+ jb_err err;
- if (add_forwarded == 0) return;
+ if ((csp->action->flags & ACTION_HIDE_FORWARDED) != 0)
+ {
+ return JB_ERR_OK;
+ }
if (csp->x_forwarded)
{
- p = strsav(p, csp->x_forwarded);
- p = strsav(p, ", ");
- p = strsav(p, csp->ip_addr_str);
+ p = strdup(csp->x_forwarded);
+ string_append(&p, ", ");
}
else
{
- p = strsav(p, "X-Forwarded-For: ");
- p = strsav(p, csp->ip_addr_str);
+ p = strdup("X-Forwarded-For: ");
+ }
+ string_append(&p, csp->ip_addr_str);
+
+ if (p == NULL)
+ {
+ return JB_ERR_MEMORY;
}
log_error(LOG_LEVEL_HEADER, "addh: %s", p);
- enlist(csp->headers, p);
+ err = enlist(csp->headers, p);
+ free(p);
+ return err;
}
/*********************************************************************
*
- * Function : server_set_cookie
+ * Function : connection_close_adder
*
- * Description : Handle the server "cookie" header properly.
- * Log cookie to the jar file. Then "crunch" it,
- * or accept it. Called from `sed'.
+ * Description : Adds a "Connection: close" header to csp->headers
+ * as a temporary fix for the needed but missing HTTP/1.1
+ * support. Called from `sed'.
+ * FIXME: This whole function shouldn't be neccessary!
*
* Parameters :
- * 1 : v = parser pattern that matched this header
- * 2 : s = header that matched this pattern
- * 3 : csp = Current client state (buffers, headers, etc...)
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ *
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
+ *
+ *********************************************************************/
+jb_err connection_close_adder(struct client_state *csp)
+{
+ log_error(LOG_LEVEL_HEADER, "Adding: Connection: close");
+ return enlist(csp->headers, "Connection: close");
+}
+
+
+/*********************************************************************
*
- * Returns : `crumble' or a newly malloc'ed string.
+ * Function : server_http
+ *
+ * Description : - Save the HTTP Status into csp->http->status
+ * - Set CT_TABOO to prevent filtering if the answer
+ * is a partial range (HTTP status 206)
+ * - Rewrite HTTP/1.1 answers to HTTP/1.0 if +downgrade
+ * action applies.
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
+ *
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
*
*********************************************************************/
-char *server_set_cookie(const struct parsers *v, char *s, struct client_state *csp)
+jb_err server_http(struct client_state *csp, char **header)
{
-#ifdef JAR_FILES
- if (jar)
+ sscanf(*header, "HTTP/%*d.%*d %d", &(csp->http->status));
+ if (csp->http->status == 206)
{
- fprintf(jar, "%s\t%s\n", csp->http->host, (s + v->len + 1));
+ csp->content_type = CT_TABOO;
}
-#endif /* def JAR_FILES */
- if (csp->accept_server_cookie == 0)
+ if ((csp->action->flags & ACTION_DOWNGRADE) != 0)
{
- return(crumble(v, s, csp));
+ (*header)[7] = '0';
+ log_error(LOG_LEVEL_HEADER, "Downgraded answer to HTTP/1.0");
}
- return(strdup(s));
-
+ return JB_ERR_OK;
}
-#ifdef FORCE_LOAD
/*********************************************************************
*
- * Function : client_host
+ * Function : server_set_cookie
*
- * Description : Clean the FORCE_PREFIX out of the 'host' http
- * header, if we use force
+ * Description : Handle the server "cookie" header properly.
+ * Log cookie to the jar file. Then "crunch",
+ * accept or rewrite it to a session cookie.
+ * Called from `sed'.
+ *
+ * TODO: Allow the user to specify a new expiration
+ * time to cause the cookie to expire even before the
+ * browser is closed.
*
* Parameters :
- * 1 : v = ignored
- * 2 : s = header (from sed) to clean
- * 3 : csp = Current client state (buffers, headers, etc...)
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
*
- * Returns : A malloc'ed pointer to the cleaned host header
+ * Returns : JB_ERR_OK on success, or
+ * JB_ERR_MEMORY on out-of-memory error.
*
*********************************************************************/
-char *client_host(const struct parsers *v, char *s, struct client_state *csp)
+jb_err server_set_cookie(struct client_state *csp, char **header)
{
- char *cleanhost = strdup(s);
-
- if(csp->force)
- strclean(cleanhost, FORCE_PREFIX);
+ time_t now;
+ time_t cookie_time;
+ struct tm tm_now;
+ struct tm tm_cookie;
+ time(&now);
+
+#ifdef FEATURE_COOKIE_JAR
+ if (csp->config->jar)
+ {
+ /*
+ * Write timestamp into outbuf.
+ *
+ * Complex because not all OSs have tm_gmtoff or
+ * the %z field in strftime()
+ */
+ char tempbuf[ BUFFER_SIZE ];
- return(cleanhost);
+#ifdef HAVE_LOCALTIME_R
+ tm_now = *localtime_r(&now, &tm_now);
+#elif FEATURE_PTHREAD
+ pthread_mutex_lock(&localtime_mutex);
+ tm_now = *localtime (&now);
+ pthread_mutex_unlock(&localtime_mutex);
+#else
+ tm_now = *localtime (&now);
+#endif
+ strftime(tempbuf, BUFFER_SIZE-6, "%b %d %H:%M:%S ", &tm_now);
+
+ /* strlen("set-cookie: ") = 12 */
+ fprintf(csp->config->jar, "%s %s\t%s\n", tempbuf, csp->http->host, *header + 12);
+ }
+#endif /* def FEATURE_COOKIE_JAR */
+
+ if ((csp->action->flags & ACTION_NO_COOKIE_SET) != 0)
+ {
+ log_error(LOG_LEVEL_HEADER, "Crunched incoming cookie -- yum!");
+ return crumble(csp, header);
+ }
+ else if ((csp->action->flags & ACTION_NO_COOKIE_KEEP) != 0)
+ {
+ /* Flag whether or not to log a message */
+ int changed = 0;
+
+ /* A variable to store the tag we're working on */
+ char *cur_tag;
+
+ /* Skip "Set-Cookie:" (11 characters) in header */
+ cur_tag = *header + 11;
+
+ /* skip whitespace between "Set-Cookie:" and value */
+ while (*cur_tag && ijb_isspace(*cur_tag))
+ {
+ cur_tag++;
+ }
+
+ /* Loop through each tag in the cookie */
+ while (*cur_tag)
+ {
+ /* Find next tag */
+ char *next_tag = strchr(cur_tag, ';');
+ if (next_tag != NULL)
+ {
+ /* Skip the ';' character itself */
+ next_tag++;
+
+ /* skip whitespace ";" and start of tag */
+ while (*next_tag && ijb_isspace(*next_tag))
+ {
+ next_tag++;
+ }
+ }
+ else
+ {
+ /* "Next tag" is the end of the string */
+ next_tag = cur_tag + strlen(cur_tag);
+ }
+
+ /*
+ * Check the expiration date to see
+ * if the cookie is still valid, if yes,
+ * rewrite it to a session cookie.
+ */
+ if (strncmpic(cur_tag, "expires=", 8) == 0)
+ {
+ char *match;
+ /*
+ * Try the valid time formats we know about.
+ *
+ * XXX: Maybe the log messages should be removed
+ * for the next stable release. They just exist to
+ * see which time format gets the most hits and
+ * should be checked for first.
+ */
+ if (NULL != (match = strptime(cur_tag, "expires=%a, %e-%b-%y %H:%M:%S ", &tm_cookie)))
+ {
+ log_error(LOG_LEVEL_HEADER,
+ "cookie \'%s\' send by %s appears to be using time format 1.",
+ csp->http->url, *header);
+ }
+ else if (NULL != (match = strptime(cur_tag, "expires=%A, %e-%b-%Y %H:%M:%S ", &tm_cookie)))
+ {
+ log_error(LOG_LEVEL_HEADER,
+ "cookie \'%s\' send by %s appears to be using time format 2.",
+ csp->http->url, *header);
+
+ }
+ else if (NULL != (match = strptime(cur_tag, "expires=%a, %e-%b-%Y %H:%M:%S ", &tm_cookie)))
+ {
+ log_error(LOG_LEVEL_HEADER,
+ "cookie \'%s\' send by %s appears to be using time format 3.",
+ csp->http->url, *header);
+ }
+
+ /* Did any of them match? */
+ if (NULL == match)
+ {
+ /*
+ * Nope, treat it as if it was still valid.
+ *
+ * XXX: Should we remove the whole cookie instead?
+ */
+ log_error(LOG_LEVEL_ERROR,
+ "Can't parse %s. Unsupported time format?", cur_tag);
+ memmove(cur_tag, next_tag, strlen(next_tag) + 1);
+ changed = 1;
+ }
+ else
+ {
+ /*
+ * Yes. Check if the cookie is still valid.
+ *
+ * If the cookie is already expired it's probably
+ * a delete cookie and even if it isn't, the browser
+ * will discard it anyway.
+ */
+
+ /*
+ * XXX: timegm() isn't available on some AmigaOS
+ * versions and our replacement doesn't work.
+ *
+ * Our options are to either:
+ *
+ * - disable session-cookies-only completely if timegm
+ * is missing,
+ *
+ * - to simply remove all expired tags, like it has
+ * been done until Privoxy 3.0.6 and to live with
+ * the consequence that it can cause login/logout
+ * problems on servers that don't validate their
+ * input properly, or
+ *
+ * - to replace it with mktime in which
+ * case there is a slight chance of valid cookies
+ * passing as already expired.
+ *
+ * This is the way it's currently done and it's not
+ * as bad as it sounds. If the missing GMT offset is
+ * enough to change the result of the expiration check
+ * the cookie will be only valid for a few hours
+ * anyway, which in many cases will be shorter
+ * than a browser session.
+ */
+ cookie_time = timegm(&tm_cookie);
+ if (cookie_time - now < 0)
+ {
+ log_error(LOG_LEVEL_HEADER,
+ "Cookie \'%s\' is already expired and can pass unmodified.", *header);
+ /* Just in case some clown sets more then one expiration date */
+ cur_tag = next_tag;
+ }
+ else
+ {
+ log_error(LOG_LEVEL_HEADER,
+ "Cookie \'%s\' is still valid and has to be rewritten.", *header);
+
+ /*
+ * Delete the tag by copying the rest of the string over it.
+ * (Note that we cannot just use "strcpy(cur_tag, next_tag)",
+ * since the behaviour of strcpy is undefined for overlapping
+ * strings.)
+ */
+ memmove(cur_tag, next_tag, strlen(next_tag) + 1);
+
+ /* That changed the header, need to issue a log message */
+ changed = 1;
+
+ /*
+ * Note that the next tag has now been moved to *cur_tag,
+ * so we do not need to update the cur_tag pointer.
+ */
+ }
+ }
+
+ }
+ else
+ {
+ /* Move on to next cookie tag */
+ cur_tag = next_tag;
+ }
+ }
+
+ if (changed)
+ {
+ assert(NULL != *header);
+ log_error(LOG_LEVEL_HEADER, "Cookie rewritten to a temporary one: %s",
+ *header);
+ }
+ }
+
+ return JB_ERR_OK;
}
-#endif /* def FORCE_LOAD */
-
-
-#ifdef FORCE_LOAD
+
+
+#ifdef FEATURE_FORCE_LOAD
/*********************************************************************
*
* Function : strclean
*
- * Description : In-Situ-Eliminate all occurances of substring in
+ * Description : In-Situ-Eliminate all occurances of substring in
* string
*
* Parameters :
*********************************************************************/
int strclean(const char *string, const char *substring)
{
- int hits = 0, len = strlen(substring);
+ int hits = 0;
+ size_t len;
char *pos, *p;
- while((pos = strstr(string, substring)))
+ len = strlen(substring);
+
+ while((pos = strstr(string, substring)) != NULL)
{
p = pos + len;
do
{
- *(p - len) = *p;
+ *(p - len) = *p;
}
while (*p++ != '\0');
return(hits);
}
-#endif /* def FORCE_LOAD */
+#endif /* def FEATURE_FORCE_LOAD */
+
+/*********************************************************************
+ *
+ * Function : parse_header_time
+ *
+ * Description : Transforms time inside a HTTP header into
+ * the usual time format.
+ *
+ * Parameters :
+ * 1 : header = header to parse
+ * 2 : tm = storage for the resulting time in seconds
+ *
+ * Returns : Time struct containing the header time, or
+ * NULL in case of a parsing problems.
+ *
+ *********************************************************************/
+struct tm *parse_header_time(char *header, time_t *tm) {
+
+ char * timestring;
+ struct tm gmt;
+ struct tm * timeptr;
+
+ /*
+ * Initializing gmt to prevent time zone offsets.
+ *
+ * While this is only necessary on some platforms
+ * (mingw32 for example), I don't know how to
+ * detect these automatically and doing it everywhere
+ * shouldn't hurt.
+ */
+ time(tm);
+#ifdef HAVE_LOCALTIME_R
+ gmt = *localtime_r(tm, &gmt);
+#elif FEATURE_PTHREAD
+ pthread_mutex_lock(&localtime_mutex);
+ gmt = *localtime(tm);
+ pthread_mutex_unlock(&localtime_mutex);
+#else
+ gmt = *localtime(tm);
+#endif
+
+ /* Skipping header name */
+ timestring = strstr(header, ": ");
+ if (strptime(timestring, ": %a, %d %b %Y %H:%M:%S", &gmt) == NULL)
+ {
+ timeptr = NULL;
+ }
+ else
+ {
+ *tm = timegm(&gmt);
+ timeptr = &gmt;
+ }
+ return(timeptr);
+
+}
+
+/*********************************************************************
+ *
+ * Function : get_destination_from_headers
+ *
+ * Description : Parse the "Host:" header to get the request's destination.
+ * Only needed if the client's request was forcefully
+ * redirected into Privoxy.
+ *
+ * Code mainly copied from client_host() which is currently
+ * run too late for this purpose.
+ *
+ * Parameters :
+ * 1 : headers = List of headers (one of them hopefully being
+ * the "Host:" header)
+ * 2 : http = storage for the result (host, port and hostport).
+ *
+ * Returns : JB_ERR_MEMORY in case of memory problems,
+ * JB_ERR_PARSE if the host header couldn't be found,
+ * JB_ERR_OK otherwise.
+ *
+ *********************************************************************/
+jb_err get_destination_from_headers(const struct list *headers, struct http_request *http)
+{
+ char *q;
+ char *p;
+ char *host;
+
+ host = get_header_value(headers, "Host:");
+
+ if (NULL == host)
+ {
+ log_error(LOG_LEVEL_ERROR, "No \"Host:\" header found.");
+ return JB_ERR_PARSE;
+ }
+
+ if (NULL == (p = strdup((host))))
+ {
+ log_error(LOG_LEVEL_ERROR, "Out of memory while parsing \"Host:\" header");
+ return JB_ERR_MEMORY;
+ }
+ chomp(p);
+ if (NULL == (q = strdup(p)))
+ {
+ freez(p);
+ log_error(LOG_LEVEL_ERROR, "Out of memory while parsing \"Host:\" header");
+ return JB_ERR_MEMORY;
+ }
+
+ freez(http->hostport);
+ http->hostport = p;
+ freez(http->host);
+ http->host = q;
+ q = strchr(http->host, ':');
+ if (q != NULL)
+ {
+ /* Terminate hostname and evaluate port string */
+ *q++ = '\0';
+ http->port = atoi(q);
+ }
+ else
+ {
+ http->port = http->ssl ? 443 : 80;
+ }
+
+ /* Rebuild request URL */
+ freez(http->url);
+ http->url = strdup(http->ssl ? "https://" : "http://");
+ string_append(&http->url, http->hostport);
+ string_append(&http->url, http->path);
+ if (http->url == NULL)
+ {
+ return JB_ERR_MEMORY;
+ }
+
+ log_error(LOG_LEVEL_HEADER, "Destination extracted from \"Host:\" header. New request URL: %s",
+ http->url);
+
+ return JB_ERR_OK;
+
+}
/*