Privoxy"> ]>
Privoxy &p-version; User Manual Copyright &my-copy; 2001-2011 by Privoxy Developers $Id: user-manual.sgml,v 2.143 2011/11/20 17:16:36 fabiankeil Exp $ This is here to keep vim syntax file from breaking :/ If I knew enough to fix it, I would. PLEASE DO NOT REMOVE! HB: hal@foobox.net ]]> The Privoxy User Manual gives users information on how to install, configure and use Privoxy. &p-intro; You can find the latest version of the Privoxy User Manual at http://www.privoxy.org/user-manual/. Please see the Contact section on how to contact the developers. Introduction This documentation is included with the current &p-status; version of Privoxy, v.&p-version;. Since this is a &p-status; version, not all new features are well tested. This documentation may be slightly out of sync as a result (especially with CVS sources). And there may be bugs, though hopefully not many! ]]> Features In addition to the core features of ad blocking and cookie management, Privoxy provides many supplemental features, that give the end-user more control, more privacy and more freedom: &newfeatures; Installation Privoxy is available both in convenient pre-compiled packages for a wide range of operating systems, and as raw source code. For most users, we recommend using the packages, which can be downloaded from our Privoxy Project Page. Note: On some platforms, the installer may remove previously installed versions, if found. (See below for your platform). In any case be sure to backup your old configuration if it is valuable to you. See the note to upgraders section below. Binary Packages How to install the binary packages depends on your operating system: Red Hat and Fedora RPMs RPMs can be installed with rpm -Uvh privoxy-&p-version;-1.rpm, and will use /etc/privoxy for the location of configuration files. Note that on Red Hat, Privoxy will not be automatically started on system boot. You will need to enable that using chkconfig, ntsysv, or similar methods. If you have problems with failed dependencies, try rebuilding the SRC RPM: rpm --rebuild privoxy-&p-version;-1.src.rpm. This will use your locally installed libraries and RPM version. Also note that if you have a Junkbuster RPM installed on your system, you need to remove it first, because the packages conflict. Otherwise, RPM will try to remove Junkbuster automatically if found, before installing Privoxy. Debian and Ubuntu DEBs can be installed with apt-get install privoxy, and will use /etc/privoxy for the location of configuration files. Windows Just double-click the installer, which will guide you through the installation process. You will find the configuration files in the same directory as you installed Privoxy in. Version 3.0.5 beta introduced full Windows service functionality. On Windows only, the Privoxy program has two new command line arguments to install and uninstall Privoxy as a service. Arguments: --install[:service_name] --uninstall[:service_name] After invoking Privoxy with --install, you will need to bring up the Windows service console to assign the user you want Privoxy to run under, and whether or not you want it to run whenever the system starts. You can start the Windows services console with the following command: services.msc. If you do not take the manual step of modifying Privoxy's service settings, it will not start. Note too that you will need to give Privoxy a user account that actually exists, or it will not be permitted to write to its log and configuration files. Solaris <!--, NetBSD, HP-UX--> Create a new directory, cd to it, then unzip and untar the archive. For the most part, you'll have to figure out where things go. OS/2 First, make sure that no previous installations of Junkbuster and / or Privoxy are left on your system. Check that no Junkbuster or Privoxy objects are in your startup folder. Then, just double-click the WarpIN self-installing archive, which will guide you through the installation process. A shadow of the Privoxy executable will be placed in your startup folder so it will start automatically whenever OS/2 starts. The directory you choose to install Privoxy into will contain all of the configuration files. Mac OS X Unzip the downloaded file (you can either double-click on the zip file icon from the Finder, or from the desktop if you downloaded it there). Then, double-click on the package installer icon and follow the installation process. The privoxy service will automatically start after a successful installation (in addition to every time your computer starts up). To prevent the privoxy service from automatically starting when your computer starts up, remove or rename the folder named /Library/StartupItems/Privoxy. To manually start or stop the privoxy service, use the Privoxy Utility for Mac OS X. This application controls the privoxy service (e.g. starting and stopping the service as well as uninstalling the software). AmigaOS Copy and then unpack the lha archive to a suitable location. All necessary files will be installed into Privoxy directory, including all configuration and log files. To uninstall, just remove this directory. FreeBSD Privoxy is part of FreeBSD's Ports Collection, you can build and install it with cd /usr/ports/www/privoxy; make install clean. If you don't use the ports, you can fetch and install the package with pkg_add -r privoxy. The port skeleton and the package can also be downloaded from the File Release Page, but there's no reason to use them unless you're interested in the beta releases which are only available there. Gentoo Gentoo source packages (Ebuilds) for Privoxy are contained in the Gentoo Portage Tree (they are not on the download page, but there is a Gentoo section, where you can see when a new Privoxy Version is added to the Portage Tree). Before installing Privoxy under Gentoo just do first emerge --sync to get the latest changes from the Portage tree. With emerge privoxy you install the latest version. Configuration files are in /etc/privoxy, the documentation is in /usr/share/doc/privoxy-&p-version; and the Log directory is in /var/log/privoxy. Building from Source The most convenient way to obtain the Privoxy sources is to download the source tarball from our project download page. If you like to live on the bleeding edge and are not afraid of using possibly unstable development versions, you can check out the up-to-the-minute version directly from the CVS repository. &buildsource; Keeping your Installation Up-to-Date As user feedback comes in and development continues, we will make updated versions of both the main actions file (as a separate package) and the software itself (including the actions file) available for download. If you wish to receive an email notification whenever we release updates of Privoxy or the actions file, subscribe to our announce mailing list, ijbswa-announce@lists.sourceforge.net. In order not to lose your personal changes and adjustments when updating to the latest default.action file we strongly recommend that you use user.action and user.filter for your local customizations of Privoxy. See the Chapter on actions files for details. What's New in this Release Privoxy 3.0.18 is a stable release. The changes since 3.0.17 stable are: Bug fixes: If a generated redirect URL contains characters RFC 3986 doesn't permit, they are (re)encoded. Not doing this makes Privoxy versions from 3.0.5 to 3.0.17 susceptible to HTTP response splitting (CWE-113) attacks if the +fast-redirects{check-decoded-url} action is used. Fix a logic bug that could cause Privoxy to reuse a server socket after it got tainted by a server-header-tagger-induced block that was triggered before the whole server response had been read. If keep-alive was enabled and the request following the blocked one was to the same host and using the same forwarding settings, Privoxy would send it on the tainted server socket. While the server would simply treat it as a pipelined request, Privoxy would later on fail to properly parse the server's response as it would try to parse the unread data from the first response as server headers for the second one. Regression introduced in 3.0.17. When implying keep-alive in client_connection(), remember that the client didn't. Fixes a regression introduced in 3.0.13 that would cause Privoxy to wait for additional client requests after receiving a HTTP/1.1 request with "Connection: close" set and connection sharing enabled. With clients which terminates the client connection after detecting that the whole body has been received it doesn't really matter, but with clients that don't the connection would be kept open until it timed out. Fix a subtle race condition between prepare_csp_for_next_request() and sweep(). A thread preparing itself for the next client request could briefly appear to be inactive. If all other threads were already using more recent files, the thread could get its files swept away under its feet. So far this has only been reproduced while stress testing in valgrind while touching action files in a loop. It's unlikely to have caused any actual problems in the real world. Disable filters if SDCH compression is used unless filtering is forced. If SDCH was combined with a supported compression algorithm, Privoxy previously could try to decompress it and ditch the Content-Encoding header even though the SDCH compression wasn't dealt with. Reported by zebul666 in #3225863. Make a copy of the --user value and only mess with that when splitting user and group. On some operating systems modifying the value directly is reflected in the output of ps and friends and can be misleading. Reported by zepard in #3292710. If forwarded-connect-retries is set, only retry if Privoxy is actually forwarding the request. Previously direct connections would be retried as well. Fixed a small memory leak when retrying connections with IPv6 support enabled. Remove an incorrect assertion in compile_dynamic_pcrs_job_list() It could be triggered by a pcrs job with an invalid pcre pattern (for example one that contains a lone quantifier). If the --user argument user[.group] contains a dot, always bail out if no group has been specified. Previously the intended, but undocumented (and apparently untested), behaviour was to try interpreting the whole argument as user name, but the detection was flawed and checked for '0' instead of '\0', thus merely preventing group names beginning with a zero. In html_code_map[], use a numeric character reference instead of ' which wasn't standardized before XHTML 1.0. Fix an invalid free when compiled with FEATURE_GRACEFUL_TERMINATION and shut down through http://config.privoxy.org/die In get_actions(), fix the "temporary" backwards compatibility hack to accept block actions without reason. It also covered other actions that should be rejected as invalid. Reported by Billy Crook. General improvements: Privoxy can (re)compress buffered content before delivering it to the client. Disabled by default as most users wouldn't benefit from it. The +fast-redirects{check-decoded-url} action checks URL segments separately. If there are other parameters behind the redirect URL, this makes it unnecessary to cut them off by additionally using a +redirect{} pcrs command. Initial patch submitted by Jamie Zawinski in #3429848. When loading action sections, verify that the referenced filters exist. Currently missing filters only result in an error message, but eventually the severity will be upgraded to fatal. Allow to bind to multiple separate addresses. Patch set submitted by Petr Pisar in #3354485. Set socket_error to errno if connecting fails in rfc2553_connect_to(). Previously rejected direct connections could be incorrectly reported as DNS issues if Privoxy was compiled with IPv6 support. Adjust url_code_map[] so spaces are replaced with %20 instead of '+' While '+' can be used by client's submitting form data, this is not actually what Privoxy is using the lookups for. This is more of a cosmetic issue and doesn't fix any known problems. When compiled without FEATURE_FAST_REDIRECTS, do not silently ignore +fast-redirect{} directives Added a workaround for GNU libc's strptime() reporting negative year values when the parsed year is only specified with two digits. On affected systems cookies with such a date would not be turned into session cookies by the +session-cookies-only action. Reported by Vaeinoe in #3403560 Fixed bind failures with certain GNU libc versions if no non-loopback IP address has been configured on the system. This is mainly an issue if the system is using DHCP and Privoxy is started before the network is completely configured. Reported by Raphael Marichez in #3349356. Additional insight from Petr Pisar. Privoxy log messages now use the ISO 8601 date format %Y-%m-%d. It's only slightly longer than the old format, but contains the full date including the year and allows sorting by date (when grepping in multiple log files) without hassle. In get_last_url(), do not bother trying to decode URLs that do not contain at least one '%' sign. It reduces the log noise and a number of unnecessary memory allocations. In case of SOCKS5 failures, dump the socks response in the log message. Simplify the signal setup in main(). Streamline socks5_connect() slightly. In socks5_connect(), require a complete socks response from the server. Previously Privoxy didn't care how much data the server response contained as long as the first two bytes contained the expected values. While at it, shrink the buffer size so Privoxy can't read more than a whole socks response. In chat(), do not bother to generate a client request in case of direct CONNECT requests. It will not be used anyway. Reduce server_last_modified()'s stack size. Shorten get_http_time() by using strftime(). Constify the known_http_methods pointers in unknown_method(). Constify the time_formats pointers in parse_header_time(). Constify the formerly_valid_actions pointers in action_used_to_be_valid(). Introduce a GNUMakefile MAN_PAGE variable that defaults to privoxy.1. The Debian package uses section 8 for the man page and this should simplify the patch. Deduplicate the INADDR_NONE definition for Solaris by moving it to jbsockets.h In block_url(), ditch the obsolete workaround for ancient Netscape versions that supposedly couldn't properly deal with status code 403. Remove a useless NULL pointer check in load_trustfile(). Remove two useless NULL pointer checks in load_one_re_filterfile(). Change url_code_map[] from an array of pointers to an array of arrays It removes an unnecessary layer of indirection and on 64bit system reduces the size of the binary a bit. Fix various typos. Fixes taken from Debian's 29_typos.dpatch by Roland Rosenfeld. Add a dok-tidy GNUMakefile target to clean up the messy HTML generated by the other dok targets. GNUisms in the GNUMakefile have been removed. Change the HTTP version in static responses to 1.1 Synced config.sub and config.guess with upstream 2011-11-11/386c7218162c145f5f9e1ff7f558a3fbb66c37c5. Add a dedicated function to parse the values of toggles. Reduces duplicated code in load_config() and provides better error handling. Invalid or missing toggle values are now a fatal error instead of being silently ignored. Terminate HTML lines in static error messages with \n instead of \r\n. Simplify cgi_error_unknown() a bit. In LogPutString(), don't bother looking at pszText when not actually logging anything. Change ssplit()'s fourth parameter from int to size_t. Fixes a clang complaint. Add a warning that the statistics currently can't be trusted. Mention Privoxy-Log-Parser's --statistics option as an alternative for the time being. In rfc2553_connect_to(), start setting cgi->error_message on error. Change the expected status code returned for http://p.p/die depending on whether or not FEATURE_GRACEFUL_TERMINATION is available. In cgi_die(), mark the client connection for closing. If the client will fetch the style sheet through another connection it gets the main thread out of the accept() state and should thus trigger the actual shutdown. Add a proper CGI message for cgi_die(). Don't enforce a logical line length limit in read_config_line(). Slightly refactor server_last_modified() to remove useless gmtime*() calls. In get_content_type(), also recognize '.jpeg' as JPEG extension. Add '.png' to the list of recognized file extensions in get_content_type(). In block_url(), consistently use the block reason "Request blocked by Privoxy" In two places the reason was "Request for blocked URL" which hides the fact that the request got blocked by Privoxy and isn't necessarily correct as the block may be due to tags. In listen_loop(), reload the configuration files after accepting a new connection instead of before. Previously the first connection that arrived after a configuration change would still be handled with the old configuration. In chat()'s receive-data loop, skip a client socket check if the socket will be written to right away anyway. This can increase the transfer speed for unfiltered content on fast network connections. The socket timeout is used for SOCKS negotiations as well which previously couldn't timeout. Don't keep the client connection alive if any configuration file changed since the time the connection came in. This is closer to Privoxy's behaviour before keep-alive support for client connection has been added and also less confusing in general. Treat all Content-Type header values containing the pattern 'script' as a sign of text. Reported by pribog in #3134970. Action file improvements: Moved the site-specific block pattern section below the one for the generic patterns so for requests that are matched in both, the block reason for the domain is shown which is usually more useful than showing the one for the generic pattern. Remove -prevent-compression from the fragile alias. It's no longer used anywhere by default and isn't known to break stuff anyway. Add a (disabled) section to block various Facebook tracking URLs. Reported by Dan Stahlke in #3421764. Add a (disabled) section to rewrite and redirect click-tracking URLs used on news.google.com. Reported by Dan Stahlke in #3421755. Unblock linuxcounter.net/. Reported by Dan Stahlke in #3422612. Block 'www91.intel.com/' which is used by Omniture. Reported by Adam Piggott in #3167370. Disable the handle-as-empty-doc-returns-ok option and mark it as deprecated. Reminded by tceverling in #2790091. Add ".ivwbox.de/" to the "Cross-site user tracking" section. Reported by Nettozahler in #3172525. Unblock and fast-redirect ".awin1.com/.*=http://". Reported by Adam Piggott in #3170921. Block "b.collective-media.net/". Widen the Debian popcon exception to "qa.debian.org/popcon". Seen in Debian's 05_default_action.dpatch by Roland Rosenfeld. Block ".gemius.pl/" which only seems to be used for user tracking. Reported by johnd16 in #3002731. Additional input from Lee and movax. Disable banners-by-size filters for '.thinkgeek.com/'. The filter only seems to catch pictures of the inventory. Block requests for 'go.idmnet.bbelements.com/please/showit/'. Reported by kacperdominik in #3372959. Unblock adainitiative.org/. Add a fast-redirects exception for '.googleusercontent.com/.*=cache'. Add a fast-redirects exception for webcache.googleusercontent.com/. Unblock http://adassier.wordpress.com/ and http://adassier.files.wordpress.com/. Filter file improvements: Let the yahoo filter hide '.ads'. Let the msn filter hide overlay ads for Facebook 'likes' in search results and elements with the id 's_notf_div'. They only seem to be used to advertise site 'enhancements'. Let the js-events filter additionally disarm setInterval(). Suggested by dg1727 in #3423775. Documentation improvements: Clarify the effect of compiling Privoxy with zlib support. Suggested by dg1727 in #3423782. Point out that the SourceForge messaging system works like a black hole and should thus not be used to contact individual developers. Mention some of the problems one can experience when not explicitly configuring an IP addresses as listen address. Explicitly mention that hostnames can be used instead of IP addresses for the listen-address, that only the first address returned will be used and what happens if the address is invalid. Requested by Calestyo in #3302213. Log message improvements: If only the server connection is kept alive, do not pretend to wait for a new client request. Remove a superfluous log message in forget_connection(). In chat(), properly report missing server responses as such instead of calling them empty. In forwarded_connect(), fix a log message nobody should ever see. Fix a log message in socks5_connect(), a failed write operation was logged as failed read operation. Let load_one_actions_file() properly complain about a missing '{' at the beginning of the file. Simply stating that a line is invalid isn't particularly helpful. Do not claim to listen on a socket until Privoxy actually does. Patch submitted by Petr Pisar #3354485 Prevent a duplicated LOG_LEVEL_CLF message when sending out the "no-server-data" response. Also log the client socket when dropping a connection. Include the destination host in the 'Request ... marked for blocking. limit-connect{...} doesn't allow CONNECT ...' message Patch submitted by Saperski in #3296250. Prevent a duplicated log message if none of the resolved IP addresses were reachable. In connect_to(), do not pretend to retry if forwarded-connect-retries is zero or unset. When a specified user or group can't be found, put the name in single-quotes when logging it. In rfc2553_connect_to(), explain getnameinfo() errors better. Remove a useless log message in chat(). When retrying to connect, also log the maximum number of connection attempts. Rephrase a log message in compile_dynamic_pcrs_job_list(). Divide the error code and its meaning with a colon. Call the pcrs job dynamic and not the filter. Filters may contain dynamic and non-dynamic pcrs jobs at the same time. Only mention the name of the filter or tagger, but don't claim it's a filter when it could be a tagger. In a fatal error message in load_one_actions_file(), cover both URL and TAG patterns. In pcrs_strerror(), properly report unknown positive error code values as such. Previously they were handled like 0 (no error). In compile_dynamic_pcrs_job_list(), also log the actual error code as pcrs_strerror() doesn't handle all errors reported by pcre. Don't bother trying to continue chatting if the client didn't ask for it. Reduces log noise a bit. Make two fatal error message in load_one_actions_file() more descriptive. In cgi_send_user_manual(), log when rejecting a file name due to '/' or '..'. In load_file(), log a message if opening a file failed. The CGI error message alone isn't too helpful. In connection_destination_matches(), improve two log messages to help understand why the destinations don't match. Rephrase a log message in serve(). Client request arrival should be differentiated from closed client connections now. In serve(), log if a client connection isn't reused due to a configuration file change. Let mark_server_socket_tainted() always mark the server socket tainted, just don't talk about it in cases where it has no effect. It doesn't change Privoxy's behaviour, but makes understanding the log file easier. configure: Added a --disable-ipv6-support switch for platforms where support is detected but doesn't actually work. Do not check for the existence of strerror() and memmove() twice Remove a useless test for setpgrp(2). Privoxy doesn't need it and it can cause problems when cross-compiling. Rename the --disable-acl-files switch to --disable-acl-support. Since about 2001, ACL directives are specified in the standard config file. Update the URL of the 'Removing outdated PCRE version after the next stable release' posting. The old URL stopped working after one of SF's recent site "optimizations". Reported by Han Liu. Privoxy-Regression-Test: Added --shuffle-tests option to increase the chances of detection race conditions. Added a --local-test-file option that allows to use Privoxy-Regression-Test without Privoxy. Added tests for missing socks4 and socks4a forwarders. The --privoxy-address option now works with IPv6 addresses containing brackets, too. Perform limited sanity checks for parameters that are supposed to have numerical values. Added a --sleep-time option to specify a number of seconds to sleep between tests, defaults to 0. Disable the range-requests tagger for tests that break if it's enabled. Log messages use the ISO 8601 date format %Y-%m-%d. Fix spelling in two error messages. In the --help output, include a list of supported tests and their default levels. Adjust the tests to properly deal with FEATURE_TOGGLE being disabled. Privoxy-Log-Parser: Perform limited sanity checks for command line parameters that are supposed to have numerical values. Implement a --unbreak-lines-only option to try to revert MUA breakage. Accept and highlight: Added header: Content-Encoding: deflate Accept and highlight: Compressed content from 29258 to 8630 bytes. Accept and highlight: Client request arrived in time on socket 21. Highlight: Didn't receive data in time: a.fsdn.com:443 Accept log messages with ISO 8601 time stamps, too. uagen: Bump generated Firefox version to 8.0. Only randomize the release date if the new --randomize-release-date option is enabled. Firefox versions after 4 use a fixed date string without meaning. Note to Upgraders A quick list of things to be aware of before upgrading from earlier versions of Privoxy: The recommended way to upgrade &my-app; is to backup your old configuration files, install the new ones, verify that &my-app; is working correctly and finally merge back your changes using diff and maybe patch. There are a number of new features in each &my-app; release and most of them have to be explicitly enabled in the configuration files. Old configuration files obviously don't do that and due to syntax changes using old configuration files with a new &my-app; isn't always possible anyway. Note that some installers remove earlier versions completely, including configuration files, therefore you should really save any important configuration files! On the other hand, other installers don't overwrite existing configuration files, thinking you will want to do that yourself. standard.action has been merged into the default.action file. In the default configuration only fatal errors are logged now. You can change that in the debug section of the configuration file. You may also want to enable more verbose logging until you verified that the new &my-app; version is working as expected. Three other config file settings are now off by default: enable-remote-toggle, enable-remote-http-toggle, and enable-edit-actions. If you use or want these, you will need to explicitly enable them, and be aware of the security issues involved. Quickstart to Using Privoxy Install Privoxy. See the Installation Section below for platform specific information. Advanced users and those who want to offer Privoxy service to more than just their local machine should check the main config file, especially the security-relevant options. These are off by default. Start Privoxy, if the installation program has not done this already (may vary according to platform). See the section Starting Privoxy. Set your browser to use Privoxy as HTTP and HTTPS (SSL) proxy by setting the proxy configuration for address of 127.0.0.1 and port 8118. DO NOT activate proxying for FTP or any protocols besides HTTP and HTTPS (SSL) unless you intend to prevent your browser from using these protocols. Flush your browser's disk and memory caches, to remove any cached ad images. If using Privoxy to manage cookies, you should remove any currently stored cookies too. A default installation should provide a reasonable starting point for most. There will undoubtedly be occasions where you will want to adjust the configuration, but that can be dealt with as the need arises. Little to no initial configuration is required in most cases, you may want to enable the web-based action editor though. Be sure to read the warnings first. See the Configuration section for more configuration options, and how to customize your installation. You might also want to look at the next section for a quick introduction to how Privoxy blocks ads and banners. If you experience ads that slip through, innocent images that are blocked, or otherwise feel the need to fine-tune Privoxy's behavior, take a look at the actions files. As a quick start, you might find the richly commented examples helpful. You can also view and edit the actions files through the web-based user interface. The Appendix Troubleshooting: Anatomy of an Action has hints on how to understand and debug actions that misbehave. Please see the section Contacting the Developers on how to report bugs, problems with websites or to get help. Now enjoy surfing with enhanced control, comfort and privacy! Quickstart to Ad Blocking Ad blocking is but one of Privoxy's array of features. Many of these features are for the technically minded advanced user. But, ad and banner blocking is surely common ground for everybody. This section will provide a quick summary of ad blocking so you can get up to speed quickly without having to read the more extensive information provided below, though this is highly recommended. First a bit of a warning ... blocking ads is much like blocking SPAM: the more aggressive you are about it, the more likely you are to block things that were not intended. And the more likely that some things may not work as intended. So there is a trade off here. If you want extreme ad free browsing, be prepared to deal with more problem sites, and to spend more time adjusting the configuration to solve these unintended consequences. In short, there is not an easy way to eliminate all ads. Either take the easy way and settle for most ads blocked with the default configuration, or jump in and tweak it for your personal surfing habits and preferences. Secondly, a brief explanation of Privoxy's actions. Actions in this context, are the directives we use to tell Privoxy to perform some task relating to HTTP transactions (i.e. web browsing). We tell Privoxy to take some action. Each action has a unique name and function. While there are many potential actions in Privoxy's arsenal, only a few are used for ad blocking. Actions, and action configuration files, are explained in depth below. Actions are specified in Privoxy's configuration, followed by one or more URLs to which the action should apply. URLs can actually be URL type patterns that use wildcards so they can apply potentially to a range of similar URLs. The actions, together with the URL patterns are called a section. When you connect to a website, the full URL will either match one or more of the sections as defined in Privoxy's configuration, or not. If so, then Privoxy will perform the respective actions. If not, then nothing special happens. Furthermore, web pages may contain embedded, secondary URLs that your web browser will use to load additional components of the page, as it parses the original page's HTML content. An ad image for instance, is just an URL embedded in the page somewhere. The image itself may be on the same server, or a server somewhere else on the Internet. Complex web pages will have many such embedded URLs. &my-app; can deal with each URL individually, so, for instance, the main page text is not touched, but images from such-and-such server are blocked. The most important actions for basic ad blocking are: block, handle-as-image, handle-as-empty-document,and set-image-blocker: block - this is perhaps the single most used action, and is particularly important for ad blocking. This action stops any contact between your browser and any URL patterns that match this action's configuration. It can be used for blocking ads, but also anything that is determined to be unwanted. By itself, it simply stops any communication with the remote server and sends Privoxy's own built-in BLOCKED page instead to let you now what has happened (with some exceptions, see below). handle-as-image - tells Privoxy to treat this URL as an image. Privoxy's default configuration already does this for all common image types (e.g. GIF), but there are many situations where this is not so easy to determine. So we'll force it in these cases. This is particularly important for ad blocking, since only if we know that it's an image of some kind, can we replace it with an image of our choosing, instead of the Privoxy BLOCKED page (which would only result in a broken image icon). There are some limitations to this though. For instance, you can't just brute-force an image substitution for an entire HTML page in most situations. handle-as-empty-document - sends an empty document instead of Privoxy's normal BLOCKED HTML page. This is useful for file types that are neither HTML nor images, such as blocking JavaScript files. set-image-blocker - tells Privoxy what to display in place of an ad image that has hit a block rule. For this to come into play, the URL must match a block action somewhere in the configuration, and, it must also match an handle-as-image action. The configuration options on what to display instead of the ad are:    pattern - a checkerboard pattern, so that an ad replacement is obvious. This is the default.    blank - A very small empty GIF image is displayed. This is the so-called invisible configuration option.    http://<URL> - A redirect to any image anywhere of the user's choosing (advanced usage). Advanced users will eventually want to explore &my-app; filters as well. Filters are very different from blocks. A block blocks a site, page, or unwanted contented. Filters are a way of filtering or modifying what is actually on the page. An example filter usage: a text replacement of no-no for nasty-word. That is a very simple example. This process can be used for ad blocking, but it is more in the realm of advanced usage and has some pitfalls to be wary off. The quickest way to adjust any of these settings is with your browser through the special Privoxy editor at http://config.privoxy.org/show-status (shortcut: http://p.p/show-status). This is an internal page, and does not require Internet access. Note that as of Privoxy 3.0.7 beta the action editor is disabled by default. Check the enable-edit-actions section in the configuration file to learn why and in which cases it's safe to enable again. If you decided to enable the action editor, select the appropriate actions file, and click Edit. It is best to put personal or local preferences in user.action since this is not meant to be overwritten during upgrades, and will over-ride the settings in other files. Here you can insert new actions, and URLs for ad blocking or other purposes, and make other adjustments to the configuration. Privoxy will detect these changes automatically. A quick and simple step by step example: Right click on the ad image to be blocked, then select Copy Link Location from the pop-up menu. Set your browser to http://config.privoxy.org/show-status Find user.action in the top section, and click on Edit:
Actions Files in Use [ Screenshot of Actions Files in Use ]
You should have a section with only block listed under Actions:. If not, click a Insert new section below button, and in the new section that just appeared, click the Edit button right under the word Actions:. This will bring up a list of all actions. Find block near the top, and click in the Enabled column, then Submit just below the list. Now, in the block actions section, click the Add button, and paste the URL the browser got from Copy Link Location. Remove the http:// at the beginning of the URL. Then, click Submit (or OK if in a pop-up window). Now go back to the original page, and press SHIFT-Reload (or flush all browser caches). The image should be gone now.
This is a very crude and simple example. There might be good reasons to use a wildcard pattern match to include potentially similar images from the same site. For a more extensive explanation of patterns, and the entire actions concept, see the Actions section. For advanced users who want to hand edit their config files, you might want to now go to the Actions Files Tutorial. The ideas explained therein also apply to the web-based editor. There are also various filters that can be used for ad blocking (filters are a special subset of actions). These fall into the advanced usage category, and are explained in depth in later sections.
Starting Privoxy Before launching Privoxy for the first time, you will want to configure your browser(s) to use Privoxy as a HTTP and HTTPS (SSL) proxy. The default is 127.0.0.1 (or localhost) for the proxy address, and port 8118 (earlier versions used port 8000). This is the one configuration step that must be done ! Please note that Privoxy can only proxy HTTP and HTTPS traffic. It will not work with FTP or other protocols.
Proxy Configuration Showing Mozilla/Netscape HTTP and HTTPS (SSL) Settings [ Screenshot of Mozilla Proxy Configuration ]
With Firefox, this is typically set under: Tools -> Options -> Advanced -> Network ->Connection -> Settings Or optionally on some platforms: Edit -> Preferences -> General -> Connection Settings -> Manual Proxy Configuration With Netscape (and Mozilla), this can be set under: Edit -> Preferences -> Advanced -> Proxies -> HTTP Proxy For Internet Explorer v.5-7: Tools -> Internet Options -> Connections -> LAN Settings Then, check Use Proxy and fill in the appropriate info (Address: 127.0.0.1, Port: 8118). Include HTTPS (SSL), if you want HTTPS proxy support too (sometimes labeled Secure). Make sure any checkboxes like Use the same proxy server for all protocols is UNCHECKED. You want only HTTP and HTTPS (SSL)!
Proxy Configuration Showing Internet Explorer HTTP and HTTPS (Secure) Settings [ Screenshot of IE Proxy Configuration ]
After doing this, flush your browser's disk and memory caches to force a re-reading of all pages and to get rid of any ads that may be cached. Remove any cookies, if you want Privoxy to manage that. You are now ready to start enjoying the benefits of using Privoxy! Privoxy itself is typically started by specifying the main configuration file to be used on the command line. If no configuration file is specified on the command line, Privoxy will look for a file named config in the current directory. Except on Win32 where it will try config.txt. Red Hat and Fedora A default Red Hat installation may not start &my-app; upon boot. It will use the file /etc/privoxy/config as its main configuration file. # /etc/rc.d/init.d/privoxy start Or ... # service privoxy start Debian We use a script. Note that Debian typically starts &my-app; upon booting per default. It will use the file /etc/privoxy/config as its main configuration file. # /etc/init.d/privoxy start Windows Click on the &my-app; Icon to start Privoxy. If no configuration file is specified on the command line, Privoxy will look for a file named config.txt. Note that Windows will automatically start &my-app; when the system starts if you chose that option when installing. Privoxy can run with full Windows service functionality. On Windows only, the &my-app; program has two new command line arguments to install and uninstall &my-app; as a service. See the Windows Installation instructions for details. Solaris, NetBSD, FreeBSD, HP-UX and others Example Unix startup command: # /usr/sbin/privoxy /etc/privoxy/config OS/2 During installation, Privoxy is configured to start automatically when the system restarts. You can start it manually by double-clicking on the Privoxy icon in the Privoxy folder. Mac OS X After downloading the privoxy software, unzip the downloaded file by double-clicking on the zip file icon. Then, double-click on the installer package icon and follow the installation process. The privoxy service will automatically start after a successful installation. In addition, the privoxy service will automatically start every time your computer starts up. To prevent the privoxy service from automatically starting when your computer starts up, remove or rename the folder named /Library/StartupItems/Privoxy. A simple application named Privoxy Utility has been created which enables administrators to easily start and stop the privoxy service. In addition, the Privoxy Utility presents a simple way for administrators to edit the various privoxy config files. A method to uninstall the software is also available. An administrator username and password must be supplied in order for the Privoxy Utility to perform any of the tasks. AmigaOS Start Privoxy (with RUN <>NIL:) in your startnet script (AmiTCP), in s:user-startup (RoadShow), as startup program in your startup script (Genesis), or as startup action (Miami and MiamiDx). Privoxy will automatically quit when you quit your TCP/IP stack (just ignore the harmless warning your TCP/IP stack may display that Privoxy is still running). Gentoo A script is again used. It will use the file /etc/privoxy/config as its main configuration file. /etc/init.d/privoxy start Note that Privoxy is not automatically started at boot time by default. You can change this with the rc-update command. rc-update add privoxy default Command Line Options Privoxy may be invoked with the following command-line options: --version Print version info and exit. Unix only. --help Print short usage info and exit. Unix only. --no-daemon Don't become a daemon, i.e. don't fork and become process group leader, and don't detach from controlling tty. Unix only. --pidfile FILE On startup, write the process ID to FILE. Delete the FILE on exit. Failure to create or delete the FILE is non-fatal. If no FILE option is given, no PID file will be used. Unix only. --user USER[.GROUP] After (optionally) writing the PID file, assume the user ID of USER, and if included the GID of GROUP. Exit if the privileges are not sufficient to do so. Unix only. --chroot Before changing to the user ID given in the --user option, chroot to that user's home directory, i.e. make the kernel pretend to the &my-app; process that the directory tree starts there. If set up carefully, this can limit the impact of possible vulnerabilities in &my-app; to the files contained in that hierarchy. Unix only. --pre-chroot-nslookup hostname Specifies a hostname to look up before doing a chroot. On some systems, initializing the resolver library involves reading config files from /etc and/or loading additional shared libraries from /lib. On these systems, doing a hostname lookup before the chroot reduces the number of files that must be copied into the chroot tree. For fastest startup speed, a good value is a hostname that is not in /etc/hosts but that your local name server (listed in /etc/resolv.conf) can resolve without recursion (that is, without having to ask any other name servers). The hostname need not exist, but if it doesn't, an error message (which can be ignored) will be output. configfile If no configfile is included on the command line, Privoxy will look for a file named config in the current directory (except on Win32 where it will look for config.txt instead). Specify full path to avoid confusion. If no config file is found, Privoxy will fail to start. On MS Windows only there are two additional command-line options to allow Privoxy to install and run as a service. See the Window Installation section for details.
Privoxy Configuration All Privoxy configuration is stored in text files. These files can be edited with a text editor. Many important aspects of Privoxy can also be controlled easily with a web browser. Controlling Privoxy with Your Web Browser Privoxy's user interface can be reached through the special URL http://config.privoxy.org/ (shortcut: http://p.p/), which is a built-in page and works without Internet access. You will see the following section:     Privoxy Menu         ▪  View & change the current configuration         ▪  View the source code version numbers         ▪  View the request headers.         ▪  Look up which actions apply to a URL and why         ▪  Toggle Privoxy on or off         ▪  Documentation This should be self-explanatory. Note the first item leads to an editor for the actions files, which is where the ad, banner, cookie, and URL blocking magic is configured as well as other advanced features of Privoxy. This is an easy way to adjust various aspects of Privoxy configuration. The actions file, and other configuration files, are explained in detail below. Toggle Privoxy On or Off is handy for sites that might have problems with your current actions and filters. You can in fact use it as a test to see whether it is Privoxy causing the problem or not. Privoxy continues to run as a proxy in this case, but all manipulation is disabled, i.e. Privoxy acts like a normal forwarding proxy. There is even a toggle Bookmarklet offered, so that you can toggle Privoxy with one click from your browser. Note that several of the features described above are disabled by default in Privoxy 3.0.7 beta and later. Check the configuration file to learn why and in which cases it's safe to enable them again. Configuration Files Overview For Unix, *BSD and Linux, all configuration files are located in /etc/privoxy/ by default. For MS Windows, OS/2, and AmigaOS these are all in the same directory as the Privoxy executable. The installed defaults provide a reasonable starting point, though some settings may be aggressive by some standards. For the time being, the principle configuration files are: The main configuration file is named config on Linux, Unix, BSD, OS/2, and AmigaOS and config.txt on Windows. This is a required file. match-all.action is used to define which actions relating to banner-blocking, images, pop-ups, content modification, cookie handling etc should be applied by default. It should be the first actions file loaded. default.action defines many exceptions (both positive and negative) from the default set of actions that's configured in match-all.action. It should be the second actions file loaded and shouldn't be edited by the user. Multiple actions files may be defined in config. These are processed in the order they are defined. Local customizations and locally preferred exceptions to the default policies as defined in match-all.action (which you will most probably want to define sooner or later) are best applied in user.action, where you can preserve them across upgrades. The file isn't installed by all installers, but you can easily create it yourself with a text editor. There is also a web based editor that can be accessed from http://config.privoxy.org/show-status (Shortcut: http://p.p/show-status) for the various actions files. Filter files (the filter file) can be used to re-write the raw page content, including viewable text as well as embedded HTML and JavaScript, and whatever else lurks on any given web page. The filtering jobs are only pre-defined here; whether to apply them or not is up to the actions files. default.filter includes various filters made available for use by the developers. Some are much more intrusive than others, and all should be used with caution. You may define additional filter files in config as you can with actions files. We suggest user.filter for any locally defined filters or customizations. The syntax of the configuration and filter files may change between different Privoxy versions, unfortunately some enhancements cost backwards compatibility. All files use the # character to denote a comment (the rest of the line will be ignored) and understand line continuation through placing a backslash ("\") as the very last character in a line. If the # is preceded by a backslash, it looses its special function. Placing a # in front of an otherwise valid configuration line to prevent it from being interpreted is called "commenting out" that line. Blank lines are ignored. The actions files and filter files can use Perl style regular expressions for maximum flexibility. After making any changes, there is no need to restart Privoxy in order for the changes to take effect. Privoxy detects such changes automatically. Note, however, that it may take one or two additional requests for the change to take effect. When changing the listening address of Privoxy, these wake up requests must obviously be sent to the old listening address. While under development, the configuration content is subject to change. The below documentation may not be accurate by the time you read this. Also, what constitutes a default setting, may change, so please check all your configuration files on important issues. ]]> &config; Actions Files The actions files are used to define what actions Privoxy takes for which URLs, and thus determines how ad images, cookies and various other aspects of HTTP content and transactions are handled, and on which sites (or even parts thereof). There are a number of such actions, with a wide range of functionality. Each action does something a little different. These actions give us a veritable arsenal of tools with which to exert our control, preferences and independence. Actions can be combined so that their effects are aggregated when applied against a given set of URLs. There are three action files included with Privoxy with differing purposes: match-all.action - is used to define which actions relating to banner-blocking, images, pop-ups, content modification, cookie handling etc should be applied by default. It should be the first actions file loaded default.action - defines many exceptions (both positive and negative) from the default set of actions that's configured in match-all.action. It is a set of rules that should work reasonably well as-is for most users. This file is only supposed to be edited by the developers. It should be the second actions file loaded. user.action - is intended to be for local site preferences and exceptions. As an example, if your ISP or your bank has specific requirements, and need special handling, this kind of thing should go here. This file will not be upgraded. Edit Set to Cautious Set to Medium Set to Advanced These have increasing levels of aggressiveness and have no influence on your browsing unless you select them explicitly in the editor. A default installation should be pre-set to Cautious. New users should try this for a while before adjusting the settings to more aggressive levels. The more aggressive the settings, then the more likelihood there is of problems such as sites not working as they should. The Edit button allows you to turn each action on/off individually for fine-tuning. The Cautious button changes the actions list to low/safe settings which will activate ad blocking and a minimal set of &my-app;'s features, and subsequently there will be less of a chance for accidental problems. The Medium button sets the list to a medium level of other features and a low level set of privacy features. The Advanced button sets the list to a high level of ad blocking and medium level of privacy. See the chart below. The latter three buttons over-ride any changes via with the Edit button. More fine-tuning can be done in the lower sections of this internal page. While the actions file editor allows to enable these settings in all actions files, they are only supposed to be enabled in the first one to make sure you don't unintentionally overrule earlier rules. The default profiles, and their associated actions, as pre-defined in default.action are: Default Configurations Feature Cautious Medium Advanced Ad-blocking Aggressiveness medium high high Ad-filtering by size no yes yes Ad-filtering by link no no yes Pop-up killing blocks only blocks only blocks only Privacy Features low medium medium/high Cookie handling none session-only kill Referer forging no yes yes GIF de-animation no yes yes Fast redirects no no yes HTML taming no no yes JavaScript taming no no yes Web-bug killing no yes yes Image tag reordering no yes yes
The list of actions files to be used are defined in the main configuration file, and are processed in the order they are defined (e.g. default.action is typically processed before user.action). The content of these can all be viewed and edited from http://config.privoxy.org/show-status. The over-riding principle when applying actions, is that the last action that matches a given URL wins. The broadest, most general rules go first (defined in default.action), followed by any exceptions (typically also in default.action), which are then followed lastly by any local preferences (typically in user.action). Generally, user.action has the last word. An actions file typically has multiple sections. If you want to use aliases in an actions file, you have to place the (optional) alias section at the top of that file. Then comes the default set of rules which will apply universally to all sites and pages (be very careful with using such a universal set in user.action or any other actions file after default.action, because it will override the result from consulting any previous file). And then below that, exceptions to the defined universal policies. You can regard user.action as an appendix to default.action, with the advantage that it is a separate file, which makes preserving your personal settings across Privoxy upgrades easier. Actions can be used to block anything you want, including ads, banners, or just some obnoxious URL whose content you would rather not see. Cookies can be accepted or rejected, or accepted only during the current browser session (i.e. not written to disk), content can be modified, some JavaScripts tamed, user-tracking fooled, and much more. See below for a complete list of actions. Finding the Right Mix Note that some actions, like cookie suppression or script disabling, may render some sites unusable that rely on these techniques to work properly. Finding the right mix of actions is not always easy and certainly a matter of personal taste. And, things can always change, requiring refinements in the configuration. In general, it can be said that the more aggressive your default settings (in the top section of the actions file) are, the more exceptions for trusted sites you will have to make later. If, for example, you want to crunch all cookies per default, you'll have to make exceptions from that rule for sites that you regularly use and that require cookies for actually useful purposes, like maybe your bank, favorite shop, or newspaper. We have tried to provide you with reasonable rules to start from in the distribution actions files. But there is no general rule of thumb on these things. There just are too many variables, and sites are constantly changing. Sooner or later you will want to change the rules (and read this chapter again :). How to Edit The easiest way to edit the actions files is with a browser by using our browser-based editor, which can be reached from http://config.privoxy.org/show-status. Note: the config file option enable-edit-actions must be enabled for this to work. The editor allows both fine-grained control over every single feature on a per-URL basis, and easy choosing from wholesale sets of defaults like Cautious, Medium or Advanced. Warning: the Advanced setting is more aggressive, and will be more likely to cause problems for some sites. Experienced users only! If you prefer plain text editing to GUIs, you can of course also directly edit the the actions files with your favorite text editor. Look at default.action which is richly commented with many good examples. How Actions are Applied to Requests Actions files are divided into sections. There are special sections, like the alias sections which will be discussed later. For now let's concentrate on regular sections: They have a heading line (often split up to multiple lines for readability) which consist of a list of actions, separated by whitespace and enclosed in curly braces. Below that, there is a list of URL and tag patterns, each on a separate line. To determine which actions apply to a request, the URL of the request is compared to all URL patterns in each action file. Every time it matches, the list of applicable actions for the request is incrementally updated, using the heading of the section in which the pattern is located. The same is done again for tags and tag patterns later on. If multiple applying sections set the same action differently, the last match wins. If not, the effects are aggregated. E.g. a URL might match a regular section with a heading line of { +handle-as-image }, then later another one with just { +block }, resulting in both actions to apply. And there may well be cases where you will want to combine actions together. Such a section then might look like: { +handle-as-image +block{Banner ads.} } # Block these as if they were images. Send no block page. banners.example.com media.example.com/.*banners .example.com/images/ads/ You can trace this process for URL patterns and any given URL by visiting http://config.privoxy.org/show-url-info. Examples and more detail on this is provided in the Appendix, Troubleshooting: Anatomy of an Action section. Patterns As mentioned, Privoxy uses patterns to determine what actions might apply to which sites and pages your browser attempts to access. These patterns use wild card type pattern matching to achieve a high degree of flexibility. This allows one expression to be expanded and potentially match against many similar patterns. Generally, an URL pattern has the form <domain><port>/<path>, where the <domain>, the <port> and the <path> are optional. (This is why the special / pattern matches all URLs). Note that the protocol portion of the URL pattern (e.g. http://) should not be included in the pattern. This is assumed already! The pattern matching syntax is different for the domain and path parts of the URL. The domain part uses a simple globbing type matching technique, while the path part uses more flexible Regular Expressions (POSIX 1003.2). The port part of a pattern is a decimal port number preceded by a colon (:). If the domain part contains a numerical IPv6 address, it has to be put into angle brackets (<, >). www.example.com/ is a domain-only pattern and will match any request to www.example.com, regardless of which document on that server is requested. So ALL pages in this domain would be covered by the scope of this action. Note that a simple example.com is different and would NOT match. www.example.com means exactly the same. For domain-only patterns, the trailing / may be omitted. www.example.com/index.html matches all the documents on www.example.com whose name starts with /index.html. www.example.com/index.html$ matches only the single document /index.html on www.example.com. /index.html$ matches the document /index.html, regardless of the domain, i.e. on any web server anywhere. / Matches any URL because there's no requirement for either the domain or the path to match anything. :8000/ Matches any URL pointing to TCP port 8000. <2001:db8::1>/ Matches any URL with the host address 2001:db8::1. (Note that the real URL uses plain brackets, not angle brackets.) index.html matches nothing, since it would be interpreted as a domain name and there is no top-level domain called .html. So its a mistake. The Domain Pattern The matching of the domain part offers some flexible options: if the domain starts or ends with a dot, it becomes unanchored at that end. For example: .example.com matches any domain with first-level domain com and second-level domain example. For example www.example.com, example.com and foo.bar.baz.example.com. Note that it wouldn't match if the second-level domain was another-example. www. matches any domain that STARTS with www. (It also matches the domain www but most of the time that doesn't matter.) .example. matches any domain that CONTAINS .example.. And, by the way, also included would be any files or documents that exist within that domain since no path limitations are specified. (Correctly speaking: It matches any FQDN that contains example as a domain.) This might be www.example.com, news.example.de, or www.example.net/cgi/testing.pl for instance. All these cases are matched. Additionally, there are wild-cards that you can use in the domain names themselves. These work similarly to shell globbing type wild-cards: * represents zero or more arbitrary characters (this is equivalent to the Regular Expression based syntax of .*), ? represents any single character (this is equivalent to the regular expression syntax of a simple .), and you can define character classes in square brackets which is similar to the same regular expression technique. All of this can be freely mixed: ad*.example.com matches adserver.example.com, ads.example.com, etc but not sfads.example.com *ad*.example.com matches all of the above, and then some. .?pix.com matches www.ipix.com, pictures.epix.com, a.b.c.d.e.upix.com etc. www[1-9a-ez].example.c* matches www1.example.com, www4.example.cc, wwwd.example.cy, wwwz.example.com etc., but not wwww.example.com. While flexible, this is not the sophistication of full regular expression based syntax. The Path Pattern Privoxy uses modern POSIX 1003.2 Regular Expressions for matching the path portion (after the slash), and is thus more flexible. There is an Appendix with a brief quick-start into regular expressions, you also might want to have a look at your operating system's documentation on regular expressions (try man re_format). Note that the path pattern is automatically left-anchored at the /, i.e. it matches as if it would start with a ^ (regular expression speak for the beginning of a line). Please also note that matching in the path is CASE INSENSITIVE by default, but you can switch to case sensitive at any point in the pattern by using the (?-i) switch: www.example.com/(?-i)PaTtErN.* will match only documents whose path starts with PaTtErN in exactly this capitalization. .example.com/.* Is equivalent to just .example.com, since any documents within that domain are matched with or without the .* regular expression. This is redundant .example.com/.*/index.html$ Will match any page in the domain of example.com that is named index.html, and that is part of some path. For example, it matches www.example.com/testing/index.html but NOT www.example.com/index.html because the regular expression called for at least two /'s, thus the path requirement. It also would match www.example.com/testing/index_html, because of the special meta-character .. .example.com/(.*/)?index\.html$ This regular expression is conditional so it will match any page named index.html regardless of path which in this case can have one or more /'s. And this one must contain exactly .html (but does not have to end with that!). .example.com/(.*/)(ads|banners?|junk) This regular expression will match any path of example.com that contains any of the words ads, banner, banners (because of the ?) or junk. The path does not have to end in these words, just contain them. .example.com/(.*/)(ads|banners?|junk)/.*\.(jpe?g|gif|png)$ This is very much the same as above, except now it must end in either .jpg, .jpeg, .gif or .png. So this one is limited to common image formats. There are many, many good examples to be found in default.action, and more tutorials below in Appendix on regular expressions. The Tag Pattern Tag patterns are used to change the applying actions based on the request's tags. Tags can be created with either the client-header-tagger or the server-header-tagger action. Tag patterns have to start with TAG:, so &my-app; can tell them apart from URL patterns. Everything after the colon including white space, is interpreted as a regular expression with path pattern syntax, except that tag patterns aren't left-anchored automatically (&my-app; doesn't silently add a ^, you have to do it yourself if you need it). To match all requests that are tagged with foo your pattern line should be TAG:^foo$, TAG:foo would work as well, but it would also match requests whose tags contain foo somewhere. TAG: foo wouldn't work as it requires white space. Sections can contain URL and tag patterns at the same time, but tag patterns are checked after the URL patterns and thus always overrule them, even if they are located before the URL patterns. Once a new tag is added, Privoxy checks right away if it's matched by one of the tag patterns and updates the action settings accordingly. As a result tags can be used to activate other tagger actions, as long as these other taggers look for headers that haven't already be parsed. For example you could tag client requests which use the POST method, then use this tag to activate another tagger that adds a tag if cookies are sent, and then use a block action based on the cookie tag. This allows the outcome of one action, to be input into a subsequent action. However if you'd reverse the position of the described taggers, and activated the method tagger based on the cookie tagger, no method tags would be created. The method tagger would look for the request line, but at the time the cookie tag is created, the request line has already been parsed. While this is a limitation you should be aware of, this kind of indirection is seldom needed anyway and even the example doesn't make too much sense. Actions All actions are disabled by default, until they are explicitly enabled somewhere in an actions file. Actions are turned on if preceded with a +, and turned off if preceded with a -. So a +action means do that action, e.g. +block means please block URLs that match the following patterns, and -block means don't block URLs that match the following patterns, even if +block previously applied. Again, actions are invoked by placing them on a line, enclosed in curly braces and separated by whitespace, like in {+some-action -some-other-action{some-parameter}}, followed by a list of URL patterns, one per line, to which they apply. Together, the actions line and the following pattern lines make up a section of the actions file. Actions fall into three categories: Boolean, i.e the action can only be enabled or disabled. Syntax: +name # enable action name -name # disable action name Example: +handle-as-image Parameterized, where some value is required in order to enable this type of action. Syntax: +name{param} # enable action and set parameter to param, # overwriting parameter from previous match if necessary -name # disable action. The parameter can be omitted Note that if the URL matches multiple positive forms of a parameterized action, the last match wins, i.e. the params from earlier matches are simply ignored. Example: +hide-user-agent{Mozilla/5.0 (X11; U; FreeBSD i386; en-US; rv:1.8.1.4) Gecko/20070602 Firefox/2.0.0.4} Multi-value. These look exactly like parameterized actions, but they behave differently: If the action applies multiple times to the same URL, but with different parameters, all the parameters from all matches are remembered. This is used for actions that can be executed for the same request repeatedly, like adding multiple headers, or filtering through multiple filters. Syntax: +name{param} # enable action and add param to the list of parameters -name{param} # remove the parameter param from the list of parameters # If it was the last one left, disable the action. -name # disable this action completely and remove all parameters from the list Examples: +add-header{X-Fun-Header: Some text} and +filter{html-annoyances} If nothing is specified in any actions file, no actions are taken. So in this case Privoxy would just be a normal, non-blocking, non-filtering proxy. You must specifically enable the privacy and blocking features you need (although the provided default actions files will give a good starting point). Later defined action sections always over-ride earlier ones of the same type. So exceptions to any rules you make, should come in the latter part of the file (or in a file that is processed later when using multiple actions files such as user.action). For multi-valued actions, the actions are applied in the order they are specified. Actions files are processed in the order they are defined in config (the default installation has three actions files). It also quite possible for any given URL to match more than one pattern (because of wildcards and regular expressions), and thus to trigger more than one set of actions! Last match wins. The list of valid Privoxy actions are: add-header Typical use: Confuse log analysis, custom applications Effect: Sends a user defined HTTP header to the web server. Type: Multi-value. Parameter: Any string value is possible. Validity of the defined HTTP headers is not checked. It is recommended that you use the X- prefix for custom headers. Notes: This action may be specified multiple times, in order to define multiple headers. This is rarely needed for the typical user. If you don't know what HTTP headers are, you definitely don't need to worry about this one. Headers added by this action are not modified by other actions. Example usage: +add-header{X-User-Tracking: sucks} block Typical use: Block ads or other unwanted content Effect: Requests for URLs to which this action applies are blocked, i.e. the requests are trapped by &my-app; and the requested URL is never retrieved, but is answered locally with a substitute page or image, as determined by the handle-as-image, set-image-blocker, and handle-as-empty-document actions. Type: Parameterized. Parameter: A block reason that should be given to the user. Notes: Privoxy sends a special BLOCKED page for requests to blocked pages. This page contains the block reason given as parameter, a link to find out why the block action applies, and a click-through to the blocked content (the latter only if the force feature is available and enabled). A very important exception occurs if both block and handle-as-image, apply to the same request: it will then be replaced by an image. If set-image-blocker (see below) also applies, the type of image will be determined by its parameter, if not, the standard checkerboard pattern is sent. It is important to understand this process, in order to understand how Privoxy deals with ads and other unwanted content. Blocking is a core feature, and one upon which various other features depend. The filter action can perform a very similar task, by blocking banner images and other content through rewriting the relevant URLs in the document's HTML source, so they don't get requested in the first place. Note that this is a totally different technique, and it's easy to confuse the two. Example usage (section): {+block{No nasty stuff for you.}} # Block and replace with "blocked" page .nasty-stuff.example.com {+block{Doubleclick banners.} +handle-as-image} # Block and replace with image .ad.doubleclick.net .ads.r.us/banners/ {+block{Layered ads.} +handle-as-empty-document} # Block and then ignore adserver.example.net/.*\.js$ change-x-forwarded-for Typical use: Improve privacy by not forwarding the source of the request in the HTTP headers. Effect: Deletes the X-Forwarded-For: HTTP header from the client request, or adds a new one. Type: Parameterized. Parameter: block to delete the header. add to create the header (or append the client's IP address to an already existing one). Notes: It is safe and recommended to use block. Forwarding the source address of the request may make sense in some multi-user setups but is also a privacy risk. Example usage: +change-x-forwarded-for{block} client-header-filter Typical use: Rewrite or remove single client headers. Effect: All client headers to which this action applies are filtered on-the-fly through the specified regular expression based substitutions. Type: Parameterized. Parameter: The name of a client-header filter, as defined in one of the filter files. Notes: Client-header filters are applied to each header on its own, not to all at once. This makes it easier to diagnose problems, but on the downside you can't write filters that only change header x if header y's value is z. You can do that by using tags though. Client-header filters are executed after the other header actions have finished and use their output as input. If the request URL gets changed, &my-app; will detect that and use the new one. This can be used to rewrite the request destination behind the client's back, for example to specify a Tor exit relay for certain requests. Please refer to the filter file chapter to learn which client-header filters are available by default, and how to create your own. Example usage (section): # Hide Tor exit notation in Host and Referer Headers {+client-header-filter{hide-tor-exit-notation}} / client-header-tagger Typical use: Block requests based on their headers. Effect: Client headers to which this action applies are filtered on-the-fly through the specified regular expression based substitutions, the result is used as tag. Type: Parameterized. Parameter: The name of a client-header tagger, as defined in one of the filter files. Notes: Client-header taggers are applied to each header on its own, and as the header isn't modified, each tagger sees the original. Client-header taggers are the first actions that are executed and their tags can be used to control every other action. Example usage (section): # Tag every request with the User-Agent header {+client-header-tagger{user-agent}} / # Tagging itself doesn't change the action # settings, sections with TAG patterns do: # # If it's a download agent, use a different forwarding proxy, # show the real User-Agent and make sure resume works. {+forward-override{forward-socks5 10.0.0.2:2222 .} \ -hide-if-modified-since \ -overwrite-last-modified \ -hide-user-agent \ -filter \ -deanimate-gifs \ } TAG:^User-Agent: NetBSD-ftp/ TAG:^User-Agent: Novell ZYPP Installer TAG:^User-Agent: RPM APT-HTTP/ TAG:^User-Agent: fetch libfetch/ TAG:^User-Agent: Ubuntu APT-HTTP/ TAG:^User-Agent: MPlayer/ content-type-overwrite Typical use: Stop useless download menus from popping up, or change the browser's rendering mode Effect: Replaces the Content-Type: HTTP server header. Type: Parameterized. Parameter: Any string. Notes: The Content-Type: HTTP server header is used by the browser to decide what to do with the document. The value of this header can cause the browser to open a download menu instead of displaying the document by itself, even if the document's format is supported by the browser. The declared content type can also affect which rendering mode the browser chooses. If XHTML is delivered as text/html, many browsers treat it as yet another broken HTML document. If it is send as application/xml, browsers with XHTML support will only display it, if the syntax is correct. If you see a web site that proudly uses XHTML buttons, but sets Content-Type: text/html, you can use &my-app; to overwrite it with application/xml and validate the web master's claim inside your XHTML-supporting browser. If the syntax is incorrect, the browser will complain loudly. You can also go the opposite direction: if your browser prints error messages instead of rendering a document falsely declared as XHTML, you can overwrite the content type with text/html and have it rendered as broken HTML document. By default content-type-overwrite only replaces Content-Type: headers that look like some kind of text. If you want to overwrite it unconditionally, you have to combine it with force-text-mode. This limitation exists for a reason, think twice before circumventing it. Most of the time it's easier to replace this action with a custom server-header filter. It allows you to activate it for every document of a certain site and it will still only replace the content types you aimed at. Of course you can apply content-type-overwrite to a whole site and then make URL based exceptions, but it's a lot more work to get the same precision. Example usage (sections): # Check if www.example.net/ really uses valid XHTML { +content-type-overwrite{application/xml} } www.example.net/ # but leave the content type unmodified if the URL looks like a style sheet {-content-type-overwrite} www.example.net/.*\.css$ www.example.net/.*style crunch-client-header Typical use: Remove a client header Privoxy has no dedicated action for. Effect: Deletes every header sent by the client that contains the string the user supplied as parameter. Type: Parameterized. Parameter: Any string. Notes: This action allows you to block client headers for which no dedicated Privoxy action exists. Privoxy will remove every client header that contains the string you supplied as parameter. Regular expressions are not supported and you can't use this action to block different headers in the same request, unless they contain the same string. crunch-client-header is only meant for quick tests. If you have to block several different headers, or only want to modify parts of them, you should use a client-header filter. Don't block any header without understanding the consequences. Example usage (section): # Block the non-existent "Privacy-Violation:" client header { +crunch-client-header{Privacy-Violation:} } / crunch-if-none-match Typical use: Prevent yet another way to track the user's steps between sessions. Effect: Deletes the If-None-Match: HTTP client header. Type: Boolean. Parameter: N/A Notes: Removing the If-None-Match: HTTP client header is useful for filter testing, where you want to force a real reload instead of getting status code 304 which would cause the browser to use a cached copy of the page. It is also useful to make sure the header isn't used as a cookie replacement (unlikely but possible). Blocking the If-None-Match: header shouldn't cause any caching problems, as long as the If-Modified-Since: header isn't blocked or missing as well. It is recommended to use this action together with hide-if-modified-since and overwrite-last-modified. Example usage (section): # Let the browser revalidate cached documents but don't # allow the server to use the revalidation headers for user tracking. {+hide-if-modified-since{-60} \ +overwrite-last-modified{randomize} \ +crunch-if-none-match} / crunch-incoming-cookies Typical use: Prevent the web server from setting HTTP cookies on your system Effect: Deletes any Set-Cookie: HTTP headers from server replies. Type: Boolean. Parameter: N/A Notes: This action is only concerned with incoming HTTP cookies. For outgoing HTTP cookies, use crunch-outgoing-cookies. Use both to disable HTTP cookies completely. It makes no sense at all to use this action in conjunction with the session-cookies-only action, since it would prevent the session cookies from being set. See also filter-content-cookies. Example usage: +crunch-incoming-cookies crunch-server-header Typical use: Remove a server header Privoxy has no dedicated action for. Effect: Deletes every header sent by the server that contains the string the user supplied as parameter. Type: Parameterized. Parameter: Any string. Notes: This action allows you to block server headers for which no dedicated Privoxy action exists. Privoxy will remove every server header that contains the string you supplied as parameter. Regular expressions are not supported and you can't use this action to block different headers in the same request, unless they contain the same string. crunch-server-header is only meant for quick tests. If you have to block several different headers, or only want to modify parts of them, you should use a custom server-header filter. Don't block any header without understanding the consequences. Example usage (section): # Crunch server headers that try to prevent caching { +crunch-server-header{no-cache} } / crunch-outgoing-cookies Typical use: Prevent the web server from reading any HTTP cookies from your system Effect: Deletes any Cookie: HTTP headers from client requests. Type: Boolean. Parameter: N/A Notes: This action is only concerned with outgoing HTTP cookies. For incoming HTTP cookies, use crunch-incoming-cookies. Use both to disable HTTP cookies completely. It makes no sense at all to use this action in conjunction with the session-cookies-only action, since it would prevent the session cookies from being read. Example usage: +crunch-outgoing-cookies deanimate-gifs Typical use: Stop those annoying, distracting animated GIF images. Effect: De-animate GIF animations, i.e. reduce them to their first or last image. Type: Parameterized. Parameter: last or first Notes: This will also shrink the images considerably (in bytes, not pixels!). If the option first is given, the first frame of the animation is used as the replacement. If last is given, the last frame of the animation is used instead, which probably makes more sense for most banner animations, but also has the risk of not showing the entire last frame (if it is only a delta to an earlier frame). You can safely use this action with patterns that will also match non-GIF objects, because no attempt will be made at anything that doesn't look like a GIF. Example usage: +deanimate-gifs{last} downgrade-http-version Typical use: Work around (very rare) problems with HTTP/1.1 Effect: Downgrades HTTP/1.1 client requests and server replies to HTTP/1.0. Type: Boolean. Parameter: N/A Notes: This is a left-over from the time when Privoxy didn't support important HTTP/1.1 features well. It is left here for the unlikely case that you experience HTTP/1.1-related problems with some server out there. Note that enabling this action is only a workaround. It should not be enabled for sites that work without it. While it shouldn't break any pages, it has an (usually negative) performance impact. If you come across a site where enabling this action helps, please report it, so the cause of the problem can be analyzed. If the problem turns out to be caused by a bug in Privoxy it should be fixed so the following release works without the work around. Example usage (section): {+downgrade-http-version} problem-host.example.com fast-redirects Typical use: Fool some click-tracking scripts and speed up indirect links. Effect: Detects redirection URLs and redirects the browser without contacting the redirection server first. Type: Parameterized. Parameter: simple-check to just search for the string http:// to detect redirection URLs. check-decoded-url to decode URLs (if necessary) before searching for redirection URLs. Notes: Many sites, like yahoo.com, don't just link to other sites. Instead, they will link to some script on their own servers, giving the destination as a parameter, which will then redirect you to the final target. URLs resulting from this scheme typically look like: http://www.example.org/click-tracker.cgi?target=http%3a//www.example.net/. Sometimes, there are even multiple consecutive redirects encoded in the URL. These redirections via scripts make your web browsing more traceable, since the server from which you follow such a link can see where you go to. Apart from that, valuable bandwidth and time is wasted, while your browser asks the server for one redirect after the other. Plus, it feeds the advertisers. This feature is currently not very smart and is scheduled for improvement. If it is enabled by default, you will have to create some exceptions to this action. It can lead to failures in several ways: Not every URLs with other URLs as parameters is evil. Some sites offer a real service that requires this information to work. For example a validation service needs to know, which document to validate. fast-redirects assumes that every URL parameter that looks like another URL is a redirection target, and will always redirect to the last one. Most of the time the assumption is correct, but if it isn't, the user gets redirected anyway. Another failure occurs if the URL contains other parameters after the URL parameter. The URL: http://www.example.org/?redirect=http%3a//www.example.net/&foo=bar. contains the redirection URL http://www.example.net/, followed by another parameter. fast-redirects doesn't know that and will cause a redirect to http://www.example.net/&foo=bar. Depending on the target server configuration, the parameter will be silently ignored or lead to a page not found error. You can prevent this problem by first using the redirect action to remove the last part of the URL, but it requires a little effort. To detect a redirection URL, fast-redirects only looks for the string http://, either in plain text (invalid but often used) or encoded as http%3a//. Some sites use their own URL encoding scheme, encrypt the address of the target server or replace it with a database id. In theses cases fast-redirects is fooled and the request reaches the redirection server where it probably gets logged. Example usage: { +fast-redirects{simple-check} } one.example.com { +fast-redirects{check-decoded-url} } another.example.com/testing filter Typical use: Get rid of HTML and JavaScript annoyances, banner advertisements (by size), do fun text replacements, add personalized effects, etc. Effect: All instances of text-based type, most notably HTML and JavaScript, to which this action applies, can be filtered on-the-fly through the specified regular expression based substitutions. (Note: as of version 3.0.3 plain text documents are exempted from filtering, because web servers often use the text/plain MIME type for all files whose type they don't know.) Type: Parameterized. Parameter: The name of a content filter, as defined in the filter file. Filters can be defined in one or more files as defined by the filterfile option in the config file. default.filter is the collection of filters supplied by the developers. Locally defined filters should go in their own file, such as user.filter. When used in its negative form, and without parameters, all filtering is completely disabled. Notes: For your convenience, there are a number of pre-defined filters available in the distribution filter file that you can use. See the examples below for a list. Filtering requires buffering the page content, which may appear to slow down page rendering since nothing is displayed until all content has passed the filters. (The total time until the page is completely rendered doesn't change much, but it may be perceived as slower since the page is not incrementally displayed.) This effect will be more noticeable on slower connections. Rolling your own filters requires a knowledge of Regular Expressions and HTML. This is very powerful feature, and potentially very intrusive. Filters should be used with caution, and where an equivalent action is not available. The amount of data that can be filtered is limited to the buffer-limit option in the main config file. The default is 4096 KB (4 Megs). Once this limit is exceeded, the buffered data, and all pending data, is passed through unfiltered. Inappropriate MIME types, such as zipped files, are not filtered at all. (Again, only text-based types except plain text). Encrypted SSL data (from HTTPS servers) cannot be filtered either, since this would violate the integrity of the secure transaction. In some situations it might be necessary to protect certain text, like source code, from filtering by defining appropriate -filter exceptions. Compressed content can't be filtered either, but if &my-app; is compiled with zlib support and a supported compression algorithm is used (gzip or deflate), &my-app; can first decompress the content and then filter it. If you use a &my-app; version without zlib support, but want filtering to work on as much documents as possible, even those that would normally be sent compressed, you must use the prevent-compression action in conjunction with filter. Content filtering can achieve some of the same effects as the block action, i.e. it can be used to block ads and banners. But the mechanism works quite differently. One effective use, is to block ad banners based on their size (see below), since many of these seem to be somewhat standardized. Feedback with suggestions for new or improved filters is particularly welcome! The below list has only the names and a one-line description of each predefined filter. There are more verbose explanations of what these filters do in the filter file chapter. Example usage (with filters from the distribution default.filter file). See the Predefined Filters section for more explanation on each: +filter{js-annoyances} # Get rid of particularly annoying JavaScript abuse. +filter{js-events} # Kill all JS event bindings and timers (Radically destructive! Only for extra nasty sites). +filter{html-annoyances} # Get rid of particularly annoying HTML abuse. +filter{content-cookies} # Kill cookies that come in the HTML or JS content. +filter{refresh-tags} # Kill automatic refresh tags (for dial-on-demand setups). +filter{unsolicited-popups} # Disable only unsolicited pop-up windows. Useful if your browser lacks this ability. +filter{all-popups} # Kill all popups in JavaScript and HTML. Useful if your browser lacks this ability. +filter{img-reorder} # Reorder attributes in <img> tags to make the banners-by-* filters more effective. +filter{banners-by-size} # Kill banners by size. +filter{banners-by-link} # Kill banners by their links to known clicktrackers. +filter{webbugs} # Squish WebBugs (1x1 invisible GIFs used for user tracking). +filter{tiny-textforms} # Extend those tiny textareas up to 40x80 and kill the hard wrap. +filter{jumping-windows} # Prevent windows from resizing and moving themselves. +filter{frameset-borders} # Give frames a border and make them resizable. +filter{demoronizer} # Fix MS's non-standard use of standard charsets. +filter{shockwave-flash} # Kill embedded Shockwave Flash objects. +filter{quicktime-kioskmode} # Make Quicktime movies saveable. +filter{fun} # Text replacements for subversive browsing fun! +filter{crude-parental} # Crude parental filtering. Note that this filter doesn't work reliably. +filter{ie-exploits} # Disable some known Internet Explorer bug exploits. +filter{site-specifics} # Cure for site-specific problems. Don't apply generally! +filter{no-ping} # Removes non-standard ping attributes in <a> and <area> tags. +filter{google} # CSS-based block for Google text ads. Also removes a width limitation and the toolbar advertisement. +filter{yahoo} # CSS-based block for Yahoo text ads. Also removes a width limitation. +filter{msn} # CSS-based block for MSN text ads. Also removes tracking URLs and a width limitation. +filter{blogspot} # Cleans up some Blogspot blogs. Read the fine print before using this. force-text-mode Typical use: Force Privoxy to treat a document as if it was in some kind of text format. Effect: Declares a document as text, even if the Content-Type: isn't detected as such. Type: Boolean. Parameter: N/A Notes: As explained above, Privoxy tries to only filter files that are in some kind of text format. The same restrictions apply to content-type-overwrite. force-text-mode declares a document as text, without looking at the Content-Type: first. Think twice before activating this action. Filtering binary data with regular expressions can cause file damage. Example usage: +force-text-mode forward-override Typical use: Change the forwarding settings based on User-Agent or request origin Effect: Overrules the forward directives in the configuration file. Type: Multi-value. Parameter: forward . to use a direct connection without any additional proxies. forward 127.0.0.1:8123 to use the HTTP proxy listening at 127.0.0.1 port 8123. forward-socks4a 127.0.0.1:9050 . to use the socks4a proxy listening at 127.0.0.1 port 9050. Replace forward-socks4a with forward-socks4 to use a socks4 connection (with local DNS resolution) instead, use forward-socks5 for socks5 connections (with remote DNS resolution). forward-socks4a 127.0.0.1:9050 proxy.example.org:8000 to use the socks4a proxy listening at 127.0.0.1 port 9050 to reach the HTTP proxy listening at proxy.example.org port 8000. Replace forward-socks4a with forward-socks4 to use a socks4 connection (with local DNS resolution) instead, use forward-socks5 for socks5 connections (with remote DNS resolution). Notes: This action takes parameters similar to the forward directives in the configuration file, but without the URL pattern. It can be used as replacement, but normally it's only used in cases where matching based on the request URL isn't sufficient. Please read the description for the forward directives before using this action. Forwarding to the wrong people will reduce your privacy and increase the chances of man-in-the-middle attacks. If the ports are missing or invalid, default values will be used. This might change in the future and you shouldn't rely on it. Otherwise incorrect syntax causes Privoxy to exit. Use the show-url-info CGI page to verify that your forward settings do what you thought the do. Example usage: # Always use direct connections for requests previously tagged as # User-Agent: fetch libfetch/2.0 and make sure # resuming downloads continues to work. # This way you can continue to use Tor for your normal browsing, # without overloading the Tor network with your FreeBSD ports updates # or downloads of bigger files like ISOs. # Note that HTTP headers are easy to fake and therefore their # values are as (un)trustworthy as your clients and users. {+forward-override{forward .} \ -hide-if-modified-since \ -overwrite-last-modified \ } TAG:^User-Agent: fetch libfetch/2\.0$ handle-as-empty-document Typical use: Mark URLs that should be replaced by empty documents if they get blocked Effect: This action alone doesn't do anything noticeable. It just marks URLs. If the block action also applies, the presence or absence of this mark decides whether an HTML BLOCKED page, or an empty document will be sent to the client as a substitute for the blocked content. The empty document isn't literally empty, but actually contains a single space. Type: Boolean. Parameter: N/A Notes: Some browsers complain about syntax errors if JavaScript documents are blocked with Privoxy's default HTML page; this option can be used to silence them. And of course this action can also be used to eliminate the &my-app; BLOCKED message in frames. The content type for the empty document can be specified with content-type-overwrite{}, but usually this isn't necessary. Example usage: # Block all documents on example.org that end with ".js", # but send an empty document instead of the usual HTML message. {+block{Blocked JavaScript} +handle-as-empty-document} example.org/.*\.js$ handle-as-image Typical use: Mark URLs as belonging to images (so they'll be replaced by images if they do get blocked, rather than HTML pages) Effect: This action alone doesn't do anything noticeable. It just marks URLs as images. If the block action also applies, the presence or absence of this mark decides whether an HTML blocked page, or a replacement image (as determined by the set-image-blocker action) will be sent to the client as a substitute for the blocked content. Type: Boolean. Parameter: N/A Notes: The below generic example section is actually part of default.action. It marks all URLs with well-known image file name extensions as images and should be left intact. Users will probably only want to use the handle-as-image action in conjunction with block, to block sources of banners, whose URLs don't reflect the file type, like in the second example section. Note that you cannot treat HTML pages as images in most cases. For instance, (in-line) ad frames require an HTML page to be sent, or they won't display properly. Forcing handle-as-image in this situation will not replace the ad frame with an image, but lead to error messages. Example usage (sections): # Generic image extensions: # {+handle-as-image} /.*\.(gif|jpg|jpeg|png|bmp|ico)$ # These don't look like images, but they're banners and should be # blocked as images: # {+block{Nasty banners.} +handle-as-image} nasty-banner-server.example.com/junk.cgi\?output=trash hide-accept-language Typical use: Pretend to use different language settings. Effect: Deletes or replaces the Accept-Language: HTTP header in client requests. Type: Parameterized. Parameter: Keyword: block, or any user defined value. Notes: Faking the browser's language settings can be useful to make a foreign User-Agent set with hide-user-agent more believable. However some sites with content in different languages check the Accept-Language: to decide which one to take by default. Sometimes it isn't possible to later switch to another language without changing the Accept-Language: header first. Therefore it's a good idea to either only change the Accept-Language: header to languages you understand, or to languages that aren't wide spread. Before setting the Accept-Language: header to a rare language, you should consider that it helps to make your requests unique and thus easier to trace. If you don't plan to change this header frequently, you should stick to a common language. Example usage (section): # Pretend to use Canadian language settings. {+hide-accept-language{en-ca} \ +hide-user-agent{Mozilla/5.0 (X11; U; OpenBSD i386; en-CA; rv:1.8.0.4) Gecko/20060628 Firefox/1.5.0.4} \ } / hide-content-disposition Typical use: Prevent download menus for content you prefer to view inside the browser. Effect: Deletes or replaces the Content-Disposition: HTTP header set by some servers. Type: Parameterized. Parameter: Keyword: block, or any user defined value. Notes: Some servers set the Content-Disposition: HTTP header for documents they assume you want to save locally before viewing them. The Content-Disposition: header contains the file name the browser is supposed to use by default. In most browsers that understand this header, it makes it impossible to just view the document, without downloading it first, even if it's just a simple text file or an image. Removing the Content-Disposition: header helps to prevent this annoyance, but some browsers additionally check the Content-Type: header, before they decide if they can display a document without saving it first. In these cases, you have to change this header as well, before the browser stops displaying download menus. It is also possible to change the server's file name suggestion to another one, but in most cases it isn't worth the time to set it up. This action will probably be removed in the future, use server-header filters instead. Example usage: # Disarm the download link in Sourceforge's patch tracker { -filter \ +content-type-overwrite{text/plain}\ +hide-content-disposition{block} } .sourceforge.net/tracker/download\.php hide-if-modified-since Typical use: Prevent yet another way to track the user's steps between sessions. Effect: Deletes the If-Modified-Since: HTTP client header or modifies its value. Type: Parameterized. Parameter: Keyword: block, or a user defined value that specifies a range of hours. Notes: Removing this header is useful for filter testing, where you want to force a real reload instead of getting status code 304, which would cause the browser to use a cached copy of the page. Instead of removing the header, hide-if-modified-since can also add or subtract a random amount of time to/from the header's value. You specify a range of minutes where the random factor should be chosen from and Privoxy does the rest. A negative value means subtracting, a positive value adding. Randomizing the value of the If-Modified-Since: makes it less likely that the server can use the time as a cookie replacement, but you will run into caching problems if the random range is too high. It is a good idea to only use a small negative value and let overwrite-last-modified handle the greater changes. It is also recommended to use this action together with crunch-if-none-match, otherwise it's more or less pointless. Example usage (section): # Let the browser revalidate but make tracking based on the time less likely. {+hide-if-modified-since{-60} \ +overwrite-last-modified{randomize} \ +crunch-if-none-match} / hide-from-header Typical use: Keep your (old and ill) browser from telling web servers your email address Effect: Deletes any existing From: HTTP header, or replaces it with the specified string. Type: Parameterized. Parameter: Keyword: block, or any user defined value. Notes: The keyword block will completely remove the header (not to be confused with the block action). Alternately, you can specify any value you prefer to be sent to the web server. If you do, it is a matter of fairness not to use any address that is actually used by a real person. This action is rarely needed, as modern web browsers don't send From: headers anymore. Example usage: +hide-from-header{block} or +hide-from-header{spam-me-senseless@sittingduck.example.com} hide-referrer Typical use: Conceal which link you followed to get to a particular site Effect: Deletes the Referer: (sic) HTTP header from the client request, or replaces it with a forged one. Type: Parameterized. Parameter: conditional-block to delete the header completely if the host has changed. conditional-forge to forge the header if the host has changed. block to delete the header unconditionally. forge to pretend to be coming from the homepage of the server we are talking to. Any other string to set a user defined referrer. Notes: conditional-block is the only parameter, that isn't easily detected in the server's log file. If it blocks the referrer, the request will look like the visitor used a bookmark or typed in the address directly. Leaving the referrer unmodified for requests on the same host allows the server owner to see the visitor's click path, but in most cases she could also get that information by comparing other parts of the log file: for example the User-Agent if it isn't a very common one, or the user's IP address if it doesn't change between different requests. Always blocking the referrer, or using a custom one, can lead to failures on servers that check the referrer before they answer any requests, in an attempt to prevent their content from being embedded or linked to elsewhere. Both conditional-block and forge will work with referrer checks, as long as content and valid referring page are on the same host. Most of the time that's the case. hide-referer is an alternate spelling of hide-referrer and the two can be can be freely substituted with each other. (referrer is the correct English spelling, however the HTTP specification has a bug - it requires it to be spelled as referer.) Example usage: +hide-referrer{forge} or +hide-referrer{http://www.yahoo.com/} hide-user-agent Typical use: Try to conceal your type of browser and client operating system Effect: Replaces the value of the User-Agent: HTTP header in client requests with the specified value. Type: Parameterized. Parameter: Any user-defined string. Notes: This can lead to problems on web sites that depend on looking at this header in order to customize their content for different browsers (which, by the way, is NOT the right thing to do: good web sites work browser-independently). Using this action in multi-user setups or wherever different types of browsers will access the same Privoxy is not recommended. In single-user, single-browser setups, you might use it to delete your OS version information from the headers, because it is an invitation to exploit known bugs for your OS. It is also occasionally useful to forge this in order to access sites that won't let you in otherwise (though there may be a good reason in some cases). More information on known user-agent strings can be found at http://www.user-agents.org/ and http://en.wikipedia.org/wiki/User_agent. Example usage: +hide-user-agent{Netscape 6.1 (X11; I; Linux 2.4.18 i686)} limit-connect Typical use: Prevent abuse of Privoxy as a TCP proxy relay or disable SSL for untrusted sites Effect: Specifies to which ports HTTP CONNECT requests are allowable. Type: Parameterized. Parameter: A comma-separated list of ports or port ranges (the latter using dashes, with the minimum defaulting to 0 and the maximum to 65K). Notes: By default, i.e. if no limit-connect action applies, Privoxy allows HTTP CONNECT requests to all ports. Use limit-connect if fine-grained control is desired for some or all destinations. The CONNECT methods exists in HTTP to allow access to secure websites (https:// URLs) through proxies. It works very simply: the proxy connects to the server on the specified port, and then short-circuits its connections to the client and to the remote server. This means CONNECT-enabled proxies can be used as TCP relays very easily. Privoxy relays HTTPS traffic without seeing the decoded content. Websites can leverage this limitation to circumvent &my-app;'s filters. By specifying an invalid port range you can disable HTTPS entirely. Example usages: +limit-connect{443} # Port 443 is OK. +limit-connect{80,443} # Ports 80 and 443 are OK. +limit-connect{-3, 7, 20-100, 500-} # Ports less than 3, 7, 20 to 100 and above 500 are OK. +limit-connect{-} # All ports are OK +limit-connect{,} # No HTTPS/SSL traffic is allowed prevent-compression Typical use: Ensure that servers send the content uncompressed, so it can be passed through filters. Effect: Removes the Accept-Encoding header which can be used to ask for compressed transfer. Type: Boolean. Parameter: N/A Notes: More and more websites send their content compressed by default, which is generally a good idea and saves bandwidth. But the filter and deanimate-gifs actions need access to the uncompressed data. When compiled with zlib support (available since &my-app; 3.0.7), content that should be filtered is decompressed on-the-fly and you don't have to worry about this action. If you are using an older &my-app; version, or one that hasn't been compiled with zlib support, this action can be used to convince the server to send the content uncompressed. Most text-based instances compress very well, the size is seldom decreased by less than 50%, for markup-heavy instances like news feeds saving more than 90% of the original size isn't unusual. Not using compression will therefore slow down the transfer, and you should only enable this action if you really need it. As of &my-app; 3.0.7 it's disabled in all predefined action settings. Note that some (rare) ill-configured sites don't handle requests for uncompressed documents correctly. Broken PHP applications tend to send an empty document body, some IIS versions only send the beginning of the content. If you enable prevent-compression per default, you might want to add exceptions for those sites. See the example for how to do that. Example usage (sections): # Selectively turn off compression, and enable a filter # { +filter{tiny-textforms} +prevent-compression } # Match only these sites .google. sourceforge.net sf.net # Or instead, we could set a universal default: # { +prevent-compression } / # Match all sites # Then maybe make exceptions for broken sites: # { -prevent-compression } .compusa.com/ overwrite-last-modified Typical use: Prevent yet another way to track the user's steps between sessions. Effect: Deletes the Last-Modified: HTTP server header or modifies its value. Type: Parameterized. Parameter: One of the keywords: block, reset-to-request-time and randomize Notes: Removing the Last-Modified: header is useful for filter testing, where you want to force a real reload instead of getting status code 304, which would cause the browser to reuse the old version of the page. The randomize option overwrites the value of the Last-Modified: header with a randomly chosen time between the original value and the current time. In theory the server could send each document with a different Last-Modified: header to track visits without using cookies. Randomize makes it impossible and the browser can still revalidate cached documents. reset-to-request-time overwrites the value of the Last-Modified: header with the current time. You could use this option together with hide-if-modified-since to further customize your random range. The preferred parameter here is randomize. It is safe to use, as long as the time settings are more or less correct. If the server sets the Last-Modified: header to the time of the request, the random range becomes zero and the value stays the same. Therefore you should later randomize it a second time with hided-if-modified-since, just to be sure. It is also recommended to use this action together with crunch-if-none-match. Example usage: # Let the browser revalidate without being tracked across sessions { +hide-if-modified-since{-60} \ +overwrite-last-modified{randomize} \ +crunch-if-none-match} / redirect Typical use: Redirect requests to other sites. Effect: Convinces the browser that the requested document has been moved to another location and the browser should get it from there. Type: Parameterized Parameter: An absolute URL or a single pcrs command. Notes: Requests to which this action applies are answered with a HTTP redirect to URLs of your choosing. The new URL is either provided as parameter, or derived by applying a single pcrs command to the original URL. This action will be ignored if you use it together with block. It can be combined with fast-redirects{check-decoded-url} to redirect to a decoded version of a rewritten URL. Use this action carefully, make sure not to create redirection loops and be aware that using your own redirects might make it possible to fingerprint your requests. In case of problems with your redirects, or simply to watch them working, enable debug 128. Example usages: # Replace example.com's style sheet with another one { +redirect{http://localhost/css-replacements/example.com.css} } example.com/stylesheet\.css # Create a short, easy to remember nickname for a favorite site # (relies on the browser accept and forward invalid URLs to &my-app;) { +redirect{http://www.privoxy.org/user-manual/actions-file.html} } a # Always use the expanded view for Undeadly.org articles # (Note the $ at the end of the URL pattern to make sure # the request for the rewritten URL isn't redirected as well) {+redirect{s@$@&mode=expanded@}} undeadly.org/cgi\?action=article&sid=\d*$ # Redirect Google search requests to MSN {+redirect{s@^http://[^/]*/search\?q=([^&]*).*@http://search.msn.com/results.aspx?q=$1@}} .google.com/search # Redirect MSN search requests to Yahoo {+redirect{s@^http://[^/]*/results\.aspx\?q=([^&]*).*@http://search.yahoo.com/search?p=$1@}} search.msn.com//results\.aspx\?q= # Redirect remote requests for this manual # to the local version delivered by Privoxy {+redirect{s@^http://www@http://config@}} www.privoxy.org/user-manual/ server-header-filter Typical use: Rewrite or remove single server headers. Effect: All server headers to which this action applies are filtered on-the-fly through the specified regular expression based substitutions. Type: Parameterized. Parameter: The name of a server-header filter, as defined in one of the filter files. Notes: Server-header filters are applied to each header on its own, not to all at once. This makes it easier to diagnose problems, but on the downside you can't write filters that only change header x if header y's value is z. You can do that by using tags though. Server-header filters are executed after the other header actions have finished and use their output as input. Please refer to the filter file chapter to learn which server-header filters are available by default, and how to create your own. Example usage (section): {+server-header-filter{html-to-xml}} example.org/xml-instance-that-is-delivered-as-html {+server-header-filter{xml-to-html}} example.org/instance-that-is-delivered-as-xml-but-is-not server-header-tagger Typical use: Enable or disable filters based on the Content-Type header. Effect: Server headers to which this action applies are filtered on-the-fly through the specified regular expression based substitutions, the result is used as tag. Type: Parameterized. Parameter: The name of a server-header tagger, as defined in one of the filter files. Notes: Server-header taggers are applied to each header on its own, and as the header isn't modified, each tagger sees the original. Server-header taggers are executed before all other header actions that modify server headers. Their tags can be used to control all of the other server-header actions, the content filters and the crunch actions (redirect and block). Obviously crunching based on tags created by server-header taggers doesn't prevent the request from showing up in the server's log file. Example usage (section): # Tag every request with the content type declared by the server {+server-header-tagger{content-type}} / session-cookies-only Typical use: Allow only temporary session cookies (for the current browser session only). Effect: Deletes the expires field from Set-Cookie: server headers. Most browsers will not store such cookies permanently and forget them in between sessions. Type: Boolean. Parameter: N/A Notes: This is less strict than crunch-incoming-cookies / crunch-outgoing-cookies and allows you to browse websites that insist or rely on setting cookies, without compromising your privacy too badly. Most browsers will not permanently store cookies that have been processed by session-cookies-only and will forget about them between sessions. This makes profiling cookies useless, but won't break sites which require cookies so that you can log in for transactions. This is generally turned on for all sites, and is the recommended setting. It makes no sense at all to use session-cookies-only together with crunch-incoming-cookies or crunch-outgoing-cookies. If you do, cookies will be plainly killed. Note that it is up to the browser how it handles such cookies without an expires field. If you use an exotic browser, you might want to try it out to be sure. This setting also has no effect on cookies that may have been stored previously by the browser before starting Privoxy. These would have to be removed manually. Privoxy also uses the content-cookies filter to block some types of cookies. Content cookies are not effected by session-cookies-only. Example usage: +session-cookies-only set-image-blocker Typical use: Choose the replacement for blocked images Effect: This action alone doesn't do anything noticeable. If both block and handle-as-image also apply, i.e. if the request is to be blocked as an image, then the parameter of this action decides what will be sent as a replacement. Type: Parameterized. Parameter: pattern to send a built-in checkerboard pattern image. The image is visually decent, scales very well, and makes it obvious where banners were busted. blank to send a built-in transparent image. This makes banners disappear completely, but makes it hard to detect where Privoxy has blocked images on a given page and complicates troubleshooting if Privoxy has blocked innocent images, like navigation icons. target-url to send a redirect to target-url. You can redirect to any image anywhere, even in your local filesystem via file:/// URL. (But note that not all browsers support redirecting to a local file system). A good application of redirects is to use special Privoxy-built-in URLs, which send the built-in images, as target-url. This has the same visual effect as specifying blank or pattern in the first place, but enables your browser to cache the replacement image, instead of requesting it over and over again. Notes: The URLs for the built-in images are http://config.privoxy.org/send-banner?type=type, where type is either blank or pattern. There is a third (advanced) type, called auto. It is NOT to be used in set-image-blocker, but meant for use from filters. Auto will select the type of image that would have applied to the referring page, had it been an image. Example usage: Built-in pattern: +set-image-blocker{pattern} Redirect to the BSD daemon: +set-image-blocker{http://www.freebsd.org/gifs/dae_up3.gif} Redirect to the built-in pattern for better caching: +set-image-blocker{http://config.privoxy.org/send-banner?type=pattern} Summary Note that many of these actions have the potential to cause a page to misbehave, possibly even not to display at all. There are many ways a site designer may choose to design his site, and what HTTP header content, and other criteria, he may depend on. There is no way to have hard and fast rules for all sites. See the Appendix for a brief example on troubleshooting actions. Aliases Custom actions, known to Privoxy as aliases, can be defined by combining other actions. These can in turn be invoked just like the built-in actions. Currently, an alias name can contain any character except space, tab, =, { and }, but we strongly recommend that you only use a to z, 0 to 9, +, and -. Alias names are not case sensitive, and are not required to start with a + or - sign, since they are merely textually expanded. Aliases can be used throughout the actions file, but they must be defined in a special section at the top of the file! And there can only be one such section per actions file. Each actions file may have its own alias section, and the aliases defined in it are only visible within that file. There are two main reasons to use aliases: One is to save typing for frequently used combinations of actions, the other one is a gain in flexibility: If you decide once how you want to handle shops by defining an alias called shop, you can later change your policy on shops in one place, and your changes will take effect everywhere in the actions file where the shop alias is used. Calling aliases by their purpose also makes your actions files more readable. Currently, there is one big drawback to using aliases, though: Privoxy's built-in web-based action file editor honors aliases when reading the actions files, but it expands them before writing. So the effects of your aliases are of course preserved, but the aliases themselves are lost when you edit sections that use aliases with it. Now let's define some aliases... # Useful custom aliases we can use later. # # Note the (required!) section header line and that this section # must be at the top of the actions file! # {{alias}} # These aliases just save typing later: # (Note that some already use other aliases!) # +crunch-all-cookies = +crunch-incoming-cookies +crunch-outgoing-cookies -crunch-all-cookies = -crunch-incoming-cookies -crunch-outgoing-cookies +block-as-image = +block{Blocked image.} +handle-as-image allow-all-cookies = -crunch-all-cookies -session-cookies-only -filter{content-cookies} # These aliases define combinations of actions # that are useful for certain types of sites: # fragile = -block -filter -crunch-all-cookies -fast-redirects -hide-referrer -prevent-compression shop = -crunch-all-cookies -filter{all-popups} # Short names for other aliases, for really lazy people ;-) # c0 = +crunch-all-cookies c1 = -crunch-all-cookies ...and put them to use. These sections would appear in the lower part of an actions file and define exceptions to the default actions (as specified further up for the / pattern): # These sites are either very complex or very keen on # user data and require minimal interference to work: # {fragile} .office.microsoft.com .windowsupdate.microsoft.com # Gmail is really mail.google.com, not gmail.com mail.google.com # Shopping sites: # Allow cookies (for setting and retrieving your customer data) # {shop} .quietpc.com .worldpay.com # for quietpc.com mybank.example.com # These shops require pop-ups: # {-filter{all-popups} -filter{unsolicited-popups}} .dabs.com .overclockers.co.uk Aliases like shop and fragile are typically used for problem sites that require more than one action to be disabled in order to function properly. Actions Files Tutorial The above chapters have shown which actions files there are and how they are organized, how actions are specified and applied to URLs, how patterns work, and how to define and use aliases. Now, let's look at an example match-all.action, default.action and user.action file and see how all these pieces come together: match-all.action Remember all actions are disabled when matching starts, so we have to explicitly enable the ones we want. While the match-all.action file only contains a single section, it is probably the most important one. It has only one pattern, /, but this pattern matches all URLs. Therefore, the set of actions used in this default section will be applied to all requests as a start. It can be partly or wholly overridden by other actions files like default.action and user.action, but it will still be largely responsible for your overall browsing experience. Again, at the start of matching, all actions are disabled, so there is no need to disable any actions here. (Remember: a + preceding the action name enables the action, a - disables!). Also note how this long line has been made more readable by splitting it into multiple lines with line continuation. { \ +change-x-forwarded-for{block} \ +hide-from-header{block} \ +set-image-blocker{pattern} \ } / # Match all URLs The default behavior is now set. default.action If you aren't a developer, there's no need for you to edit the default.action file. It is maintained by the &my-app; developers and if you disagree with some of the sections, you should overrule them in your user.action. Understanding the default.action file can help you with your user.action, though. The first section in this file is a special section for internal use that prevents older &my-app; versions from reading the file: ########################################################################## # Settings -- Don't change! For internal Privoxy use ONLY. ########################################################################## {{settings}} for-privoxy-version=3.0.11 After that comes the (optional) alias section. We'll use the example section from the above chapter on aliases, that also explains why and how aliases are used: ########################################################################## # Aliases ########################################################################## {{alias}} # These aliases just save typing later: # (Note that some already use other aliases!) # +crunch-all-cookies = +crunch-incoming-cookies +crunch-outgoing-cookies -crunch-all-cookies = -crunch-incoming-cookies -crunch-outgoing-cookies +block-as-image = +block{Blocked image.} +handle-as-image mercy-for-cookies = -crunch-all-cookies -session-cookies-only -filter{content-cookies} # These aliases define combinations of actions # that are useful for certain types of sites: # fragile = -block -filter -crunch-all-cookies -fast-redirects -hide-referrer shop = -crunch-all-cookies -filter{all-popups} The first of our specialized sections is concerned with fragile sites, i.e. sites that require minimum interference, because they are either very complex or very keen on tracking you (and have mechanisms in place that make them unusable for people who avoid being tracked). We will simply use our pre-defined fragile alias instead of stating the list of actions explicitly: ########################################################################## # Exceptions for sites that'll break under the default action set: ########################################################################## # "Fragile" Use a minimum set of actions for these sites (see alias above): # { fragile } .office.microsoft.com # surprise, surprise! .windowsupdate.microsoft.com mail.google.com Shopping sites are not as fragile, but they typically require cookies to log in, and pop-up windows for shopping carts or item details. Again, we'll use a pre-defined alias: # Shopping sites: # { shop } .quietpc.com .worldpay.com # for quietpc.com .jungle.com .scan.co.uk The fast-redirects action, which may have been enabled in match-all.action, breaks some sites. So disable it for popular sites where we know it misbehaves: { -fast-redirects } login.yahoo.com edit.*.yahoo.com .google.com .altavista.com/.*(like|url|link):http .altavista.com/trans.*urltext=http .nytimes.com It is important that Privoxy knows which URLs belong to images, so that if they are to be blocked, a substitute image can be sent, rather than an HTML page. Contacting the remote site to find out is not an option, since it would destroy the loading time advantage of banner blocking, and it would feed the advertisers information about you. We can mark any URL as an image with the handle-as-image action, and marking all URLs that end in a known image file extension is a good start: ########################################################################## # Images: ########################################################################## # Define which file types will be treated as images, in case they get # blocked further down this file: # { +handle-as-image } /.*\.(gif|jpe?g|png|bmp|ico)$ And then there are known banner sources. They often use scripts to generate the banners, so it won't be visible from the URL that the request is for an image. Hence we block them and mark them as images in one go, with the help of our +block-as-image alias defined above. (We could of course just as well use +block +handle-as-image here.) Remember that the type of the replacement image is chosen by the set-image-blocker action. Since all URLs have matched the default section with its +set-image-blocker{pattern} action before, it still applies and needn't be repeated: # Known ad generators: # { +block-as-image } ar.atwola.com .ad.doubleclick.net .ad.*.doubleclick.net .a.yimg.com/(?:(?!/i/).)*$ .a[0-9].yimg.com/(?:(?!/i/).)*$ bs*.gsanet.com .qkimg.net One of the most important jobs of Privoxy is to block banners. Many of these can be blocked by the filter{banners-by-size} action, which we enabled above, and which deletes the references to banner images from the pages while they are loaded, so the browser doesn't request them anymore, and hence they don't need to be blocked here. But this naturally doesn't catch all banners, and some people choose not to use filters, so we need a comprehensive list of patterns for banner URLs here, and apply the block action to them. First comes many generic patterns, which do most of the work, by matching typical domain and path name components of banners. Then comes a list of individual patterns for specific sites, which is omitted here to keep the example short: ########################################################################## # Block these fine banners: ########################################################################## { +block{Banner ads.} } # Generic patterns: # ad*. .*ads. banner?. count*. /.*count(er)?\.(pl|cgi|exe|dll|asp|php[34]?) /(?:.*/)?(publicite|werbung|rekla(ma|me|am)|annonse|maino(kset|nta|s)?)/ # Site-specific patterns (abbreviated): # .hitbox.com It's quite remarkable how many advertisers actually call their banner servers ads.company.com, or call the directory in which the banners are stored simply banners. So the above generic patterns are surprisingly effective. But being very generic, they necessarily also catch URLs that we don't want to block. The pattern .*ads. e.g. catches nasty-ads.nasty-corp.com as intended, but also downloads.sourcefroge.net or adsl.some-provider.net. So here come some well-known exceptions to the +block section above. Note that these are exceptions to exceptions from the default! Consider the URL downloads.sourcefroge.net: Initially, all actions are deactivated, so it wouldn't get blocked. Then comes the defaults section, which matches the URL, but just deactivates the block action once again. Then it matches .*ads., an exception to the general non-blocking policy, and suddenly +block applies. And now, it'll match .*loads., where -block applies, so (unless it matches again further down) it ends up with no block action applying. ########################################################################## # Save some innocent victims of the above generic block patterns: ########################################################################## # By domain: # { -block } adv[io]*. # (for advogato.org and advice.*) adsl. # (has nothing to do with ads) adobe. # (has nothing to do with ads either) ad[ud]*. # (adult.* and add.*) .edu # (universities don't host banners (yet!)) .*loads. # (downloads, uploads etc) # By path: # /.*loads/ # Site-specific: # www.globalintersec.com/adv # (adv = advanced) www.ugu.com/sui/ugu/adv Filtering source code can have nasty side effects, so make an exception for our friends at sourceforge.net, and all paths with cvs in them. Note that -filter disables all filters in one fell swoop! # Don't filter code! # { -filter } /(.*/)?cvs bugzilla. developer. wiki. .sourceforge.net The actual default.action is of course much more comprehensive, but we hope this example made clear how it works. user.action So far we are painting with a broad brush by setting general policies, which would be a reasonable starting point for many people. Now, you might want to be more specific and have customized rules that are more suitable to your personal habits and preferences. These would be for narrowly defined situations like your ISP or your bank, and should be placed in user.action, which is parsed after all other actions files and hence has the last word, over-riding any previously defined actions. user.action is also a safe place for your personal settings, since default.action is actively maintained by the Privoxy developers and you'll probably want to install updated versions from time to time. So let's look at a few examples of things that one might typically do in user.action: # My user.action file. <fred@example.com> As aliases are local to the actions file that they are defined in, you can't use the ones from default.action, unless you repeat them here: # Aliases are local to the file they are defined in. # (Re-)define aliases for this file: # {{alias}} # # These aliases just save typing later, and the alias names should # be self explanatory. # +crunch-all-cookies = +crunch-incoming-cookies +crunch-outgoing-cookies -crunch-all-cookies = -crunch-incoming-cookies -crunch-outgoing-cookies allow-all-cookies = -crunch-all-cookies -session-cookies-only allow-popups = -filter{all-popups} +block-as-image = +block{Blocked as image.} +handle-as-image -block-as-image = -block # These aliases define combinations of actions that are useful for # certain types of sites: # fragile = -block -crunch-all-cookies -filter -fast-redirects -hide-referrer shop = -crunch-all-cookies allow-popups # Allow ads for selected useful free sites: # allow-ads = -block -filter{banners-by-size} -filter{banners-by-link} # Alias for specific file types that are text, but might have conflicting # MIME types. We want the browser to force these to be text documents. handle-as-text = -filter +-content-type-overwrite{text/plain} +-force-text-mode -hide-content-disposition Say you have accounts on some sites that you visit regularly, and you don't want to have to log in manually each time. So you'd like to allow persistent cookies for these sites. The allow-all-cookies alias defined above does exactly that, i.e. it disables crunching of cookies in any direction, and the processing of cookies to make them only temporary. { allow-all-cookies } sourceforge.net .yahoo.com .msdn.microsoft.com .redhat.com Your bank is allergic to some filter, but you don't know which, so you disable them all: { -filter } .your-home-banking-site.com Some file types you may not want to filter for various reasons: # Technical documentation is likely to contain strings that might # erroneously get altered by the JavaScript-oriented filters: # .tldp.org /(.*/)?selfhtml/ # And this stupid host sends streaming video with a wrong MIME type, # so that Privoxy thinks it is getting HTML and starts filtering: # stupid-server.example.com/ Example of a simple block action. Say you've seen an ad on your favourite page on example.com that you want to get rid of. You have right-clicked the image, selected copy image location and pasted the URL below while removing the leading http://, into a { +block{} } section. Note that { +handle-as-image } need not be specified, since all URLs ending in .gif will be tagged as images by the general rules as set in default.action anyway: { +block{Nasty ads.} } www.example.com/nasty-ads/sponsor\.gif another.example.net/more/junk/here/ The URLs of dynamically generated banners, especially from large banner farms, often don't use the well-known image file name extensions, which makes it impossible for Privoxy to guess the file type just by looking at the URL. You can use the +block-as-image alias defined above for these cases. Note that objects which match this rule but then turn out NOT to be an image are typically rendered as a broken image icon by the browser. Use cautiously. { +block-as-image } .doubleclick.net .fastclick.net /Realmedia/ads/ ar.atwola.com/ Now you noticed that the default configuration breaks Forbes Magazine, but you were too lazy to find out which action is the culprit, and you were again too lazy to give feedback, so you just used the fragile alias on the site, and -- whoa! -- it worked. The fragile aliases disables those actions that are most likely to break a site. Also, good for testing purposes to see if it is Privoxy that is causing the problem or not. We later find other regular sites that misbehave, and add those to our personalized list of troublemakers: { fragile } .forbes.com webmail.example.com .mybank.com You like the fun text replacements in default.filter, but it is disabled in the distributed actions file. So you'd like to turn it on in your private, update-safe config, once and for all: { +filter{fun} } / # For ALL sites! Note that the above is not really a good idea: There are exceptions to the filters in default.action for things that really shouldn't be filtered, like code on CVS->Web interfaces. Since user.action has the last word, these exceptions won't be valid for the fun filtering specified here. You might also worry about how your favourite free websites are funded, and find that they rely on displaying banner advertisements to survive. So you might want to specifically allow banners for those sites that you feel provide value to you: { allow-ads } .sourceforge.net .slashdot.org .osdn.net Note that allow-ads has been aliased to -block, -filter{banners-by-size}, and -filter{banners-by-link} above. Invoke another alias here to force an over-ride of the MIME type application/x-sh which typically would open a download type dialog. In my case, I want to look at the shell script, and then I can save it should I choose to. { handle-as-text } /.*\.sh$ user.action is generally the best place to define exceptions and additions to the default policies of default.action. Some actions are safe to have their default policies set here though. So let's set a default policy to have a blank image as opposed to the checkerboard pattern for ALL sites. / of course matches all URL paths and patterns: { +set-image-blocker{blank} } / # ALL sites
Filter Files On-the-fly text substitutions need to be defined in a filter file. Once defined, they can then be invoked as an action. &my-app; supports three different filter actions: filter to rewrite the content that is send to the client, client-header-filter to rewrite headers that are send by the client, and server-header-filter to rewrite headers that are send by the server. &my-app; also supports two tagger actions: client-header-tagger and server-header-tagger. Taggers and filters use the same syntax in the filter files, the difference is that taggers don't modify the text they are filtering, but use a rewritten version of the filtered text as tag. The tags can then be used to change the applying actions through sections with tag-patterns. Multiple filter files can be defined through the filterfile config directive. The filters as supplied by the developers are located in default.filter. It is recommended that any locally defined or modified filters go in a separately defined file such as user.filter. Common tasks for content filters are to eliminate common annoyances in HTML and JavaScript, such as pop-up windows, exit consoles, crippled windows without navigation tools, the infamous <BLINK> tag etc, to suppress images with certain width and height attributes (standard banner sizes or web-bugs), or just to have fun. Enabled content filters are applied to any content whose Content Type header is recognised as a sign of text-based content, with the exception of text/plain. Use the force-text-mode action to also filter other content. Substitutions are made at the source level, so if you want to roll your own filters, you should first be familiar with HTML syntax, and, of course, regular expressions. Just like the actions files, the filter file is organized in sections, which are called filters here. Each filter consists of a heading line, that starts with one of the keywords FILTER:, CLIENT-HEADER-FILTER: or SERVER-HEADER-FILTER: followed by the filter's name, and a short (one line) description of what it does. Below that line come the jobs, i.e. lines that define the actual text substitutions. By convention, the name of a filter should describe what the filter eliminates. The comment is used in the web-based user interface. Once a filter called name has been defined in the filter file, it can be invoked by using an action of the form +filter{name} in any actions file. Filter definitions start with a header line that contains the filter type, the filter name and the filter description. A content filter header line for a filter called foo could look like this: FILTER: foo Replace all "foo" with "bar" Below that line, and up to the next header line, come the jobs that define what text replacements the filter executes. They are specified in a syntax that imitates Perl's s/// operator. If you are familiar with Perl, you will find this to be quite intuitive, and may want to look at the PCRS documentation for the subtle differences to Perl behaviour. Most notably, the non-standard option letter U is supported, which turns the default to ungreedy matching. If you are new to Regular Expressions, you might want to take a look at the Appendix on regular expressions, and see the Perl manual for the s/// operator's syntax and Perl-style regular expressions in general. The below examples might also help to get you started. Filter File Tutorial Now, let's complete our foo content filter. We have already defined the heading, but the jobs are still missing. Since all it does is to replace foo with bar, there is only one (trivial) job needed: s/foo/bar/ But wait! Didn't the comment say that all occurrences of foo should be replaced? Our current job will only take care of the first foo on each page. For global substitution, we'll need to add the g option: s/foo/bar/g Our complete filter now looks like this: FILTER: foo Replace all "foo" with "bar" s/foo/bar/g Let's look at some real filters for more interesting examples. Here you see a filter that protects against some common annoyances that arise from JavaScript abuse. Let's look at its jobs one after the other: FILTER: js-annoyances Get rid of particularly annoying JavaScript abuse # Get rid of JavaScript referrer tracking. Test page: http://www.randomoddness.com/untitled.htm # s|(<script.*)document\.referrer(.*</script>)|$1"Not Your Business!"$2|Usg Following the header line and a comment, you see the job. Note that it uses | as the delimiter instead of /, because the pattern contains a forward slash, which would otherwise have to be escaped by a backslash (\). Now, let's examine the pattern: it starts with the text <script.* enclosed in parentheses. Since the dot matches any character, and * means: Match an arbitrary number of the element left of myself, this matches <script, followed by any text, i.e. it matches the whole page, from the start of the first <script> tag. That's more than we want, but the pattern continues: document\.referrer matches only the exact string document.referrer. The dot needed to be escaped, i.e. preceded by a backslash, to take away its special meaning as a joker, and make it just a regular dot. So far, the meaning is: Match from the start of the first <script> tag in a the page, up to, and including, the text document.referrer, if both are present in the page (and appear in that order). But there's still more pattern to go. The next element, again enclosed in parentheses, is .*</script>. You already know what .* means, so the whole pattern translates to: Match from the start of the first <script> tag in a page to the end of the last <script> tag, provided that the text document.referrer appears somewhere in between. This is still not the whole story, since we have ignored the options and the parentheses: The portions of the page matched by sub-patterns that are enclosed in parentheses, will be remembered and be available through the variables $1, $2, ... in the substitute. The U option switches to ungreedy matching, which means that the first .* in the pattern will only eat up all text in between <script and the first occurrence of document.referrer, and that the second .* will only span the text up to the first </script> tag. Furthermore, the s option says that the match may span multiple lines in the page, and the g option again means that the substitution is global. So, to summarize, the pattern means: Match all scripts that contain the text document.referrer. Remember the parts of the script from (and including) the start tag up to (and excluding) the string document.referrer as $1, and the part following that string, up to and including the closing tag, as $2. Now the pattern is deciphered, but wasn't this about substituting things? So lets look at the substitute: $1"Not Your Business!"$2 is easy to read: The text remembered as $1, followed by "Not Your Business!" (including the quotation marks!), followed by the text remembered as $2. This produces an exact copy of the original string, with the middle part (the document.referrer) replaced by "Not Your Business!". The whole job now reads: Replace document.referrer by "Not Your Business!" wherever it appears inside a <script> tag. Note that this job won't break JavaScript syntax, since both the original and the replacement are syntactically valid string objects. The script just won't have access to the referrer information anymore. We'll show you two other jobs from the JavaScript taming department, but this time only point out the constructs of special interest: # The status bar is for displaying link targets, not pointless blahblah # s/window\.status\s*=\s*(['"]).*?\1/dUmMy=1/ig \s stands for whitespace characters (space, tab, newline, carriage return, form feed), so that \s* means: zero or more whitespace. The ? in .*? makes this matching of arbitrary text ungreedy. (Note that the U option is not set). The ['"] construct means: a single or a double quote. Finally, \1 is a back-reference to the first parenthesis just like $1 above, with the difference that in the pattern, a backslash indicates a back-reference, whereas in the substitute, it's the dollar. So what does this job do? It replaces assignments of single- or double-quoted strings to the window.status object with a dummy assignment (using a variable name that is hopefully odd enough not to conflict with real variables in scripts). Thus, it catches many cases where e.g. pointless descriptions are displayed in the status bar instead of the link target when you move your mouse over links. # Kill OnUnload popups. Yummy. Test: http://www.zdnet.com/zdsubs/yahoo/tree/yfs.html # s/(<body [^>]*)onunload(.*>)/$1never$2/iU Including the OnUnload event binding in the HTML DOM was a CRIME. When I close a browser window, I want it to close and die. Basta. This job replaces the onunload attribute in <body> tags with the dummy word never. Note that the i option makes the pattern matching case-insensitive. Also note that ungreedy matching alone doesn't always guarantee a minimal match: In the first parenthesis, we had to use [^>]* instead of .* to prevent the match from exceeding the <body> tag if it doesn't contain OnUnload, but the page's content does. The last example is from the fun department: FILTER: fun Fun text replacements # Spice the daily news: # s/microsoft(?!\.com)/MicroSuck/ig Note the (?!\.com) part (a so-called negative lookahead) in the job's pattern, which means: Don't match, if the string .com appears directly following microsoft in the page. This prevents links to microsoft.com from being trashed, while still replacing the word everywhere else. # Buzzword Bingo (example for extended regex syntax) # s* industry[ -]leading \ | cutting[ -]edge \ | customer[ -]focused \ | market[ -]driven \ | award[ -]winning # Comments are OK, too! \ | high[ -]performance \ | solutions[ -]based \ | unmatched \ | unparalleled \ | unrivalled \ *<font color="red"><b>BINGO!</b></font> \ *igx The x option in this job turns on extended syntax, and allows for e.g. the liberal use of (non-interpreted!) whitespace for nicer formatting. You get the idea? The Pre-defined Filters The distribution default.filter file contains a selection of pre-defined filters for your convenience: js-annoyances The purpose of this filter is to get rid of particularly annoying JavaScript abuse. To that end, it replaces JavaScript references to the browser's referrer information with the string "Not Your Business!". This compliments the hide-referrer action on the content level. removes the bindings to the DOM's unload event which we feel has no right to exist and is responsible for most exit consoles, i.e. nasty windows that pop up when you close another one. removes code that causes new windows to be opened with undesired properties, such as being full-screen, non-resizeable, without location, status or menu bar etc. Use with caution. This is an aggressive filter, and can break sites that rely heavily on JavaScript. js-events This is a very radical measure. It removes virtually all JavaScript event bindings, which means that scripts can not react to user actions such as mouse movements or clicks, window resizing etc, anymore. Use with caution! We strongly discourage using this filter as a default since it breaks many legitimate scripts. It is meant for use only on extra-nasty sites (should you really need to go there). html-annoyances This filter will undo many common instances of HTML based abuse. The BLINK and MARQUEE tags are neutralized (yeah baby!), and browser windows will be created as resizeable (as of course they should be!), and will have location, scroll and menu bars -- even if specified otherwise. content-cookies Most cookies are set in the HTTP dialog, where they can be intercepted by the crunch-incoming-cookies and crunch-outgoing-cookies actions. But web sites increasingly make use of HTML meta tags and JavaScript to sneak cookies to the browser on the content level. This filter disables most HTML and JavaScript code that reads or sets cookies. It cannot detect all clever uses of these types of code, so it should not be relied on as an absolute fix. Use it wherever you would also use the cookie crunch actions. refresh tags Disable any refresh tags if the interval is greater than nine seconds (so that redirections done via refresh tags are not destroyed). This is useful for dial-on-demand setups, or for those who find this HTML feature annoying. unsolicited-popups This filter attempts to prevent only unsolicited pop-up windows from opening, yet still allow pop-up windows that the user has explicitly chosen to open. It was added in version 3.0.1, as an improvement over earlier such filters. Technical note: The filter works by redefining the window.open JavaScript function to a dummy function, PrivoxyWindowOpen(), during the loading and rendering phase of each HTML page access, and restoring the function afterward. This is recommended only for browsers that cannot perform this function reliably themselves. And be aware that some sites require such windows in order to function normally. Use with caution. all-popups Attempt to prevent all pop-up windows from opening. Note this should be used with even more discretion than the above, since it is more likely to break some sites that require pop-ups for normal usage. Use with caution. img-reorder This is a helper filter that has no value if used alone. It makes the banners-by-size and banners-by-link (see below) filters more effective and should be enabled together with them. banners-by-size This filter removes image tags purely based on what size they are. Fortunately for us, many ads and banner images tend to conform to certain standardized sizes, which makes this filter quite effective for ad stripping purposes. Occasionally this filter will cause false positives on images that are not ads, but just happen to be of one of the standard banner sizes. Recommended only for those who require extreme ad blocking. The default block rules should catch 95+% of all ads without this filter enabled. banners-by-link This is an experimental filter that attempts to kill any banners if their URLs seem to point to known or suspected click trackers. It is currently not of much value and is not recommended for use by default. webbugs Webbugs are small, invisible images (technically 1X1 GIF images), that are used to track users across websites, and collect information on them. As an HTML page is loaded by the browser, an embedded image tag causes the browser to contact a third-party site, disclosing the tracking information through the requested URL and/or cookies for that third-party domain, without the user ever becoming aware of the interaction with the third-party site. HTML-ized spam also uses a similar technique to verify email addresses. This filter removes the HTML code that loads such webbugs. tiny-textforms A rather special-purpose filter that can be used to enlarge textareas (those multi-line text boxes in web forms) and turn off hard word wrap in them. It was written for the sourceforge.net tracker system where such boxes are a nuisance, but it can be handy on other sites, too. It is not recommended to use this filter as a default. jumping-windows Many consider windows that move, or resize themselves to be abusive. This filter neutralizes the related JavaScript code. Note that some sites might not display or behave as intended when using this filter. Use with caution. frameset-borders Some web designers seem to assume that everyone in the world will view their web sites using the same browser brand and version, screen resolution etc, because only that assumption could explain why they'd use static frame sizes, yet prevent their frames from being resized by the user, should they be too small to show their whole content. This filter removes the related HTML code. It should only be applied to sites which need it. demoronizer Many Microsoft products that generate HTML use non-standard extensions (read: violations) of the ISO 8859-1 aka Latin-1 character set. This can cause those HTML documents to display with errors on standard-compliant platforms. This filter translates the MS-only characters into Latin-1 equivalents. It is not necessary when using MS products, and will cause corruption of all documents that use 8-bit character sets other than Latin-1. It's mostly worthwhile for Europeans on non-MS platforms, if weird garbage characters sometimes appear on some pages, or user agents that don't correct for this on the fly. shockwave-flash A filter for shockwave haters. As the name suggests, this filter strips code out of web pages that is used to embed shockwave flash objects. quicktime-kioskmode Change HTML code that embeds Quicktime objects so that kioskmode, which prevents saving, is disabled. fun Text replacements for subversive browsing fun. Make fun of your favorite Monopolist or play buzzword bingo. crude-parental A demonstration-only filter that shows how Privoxy can be used to delete web content on a keyword basis. ie-exploits An experimental collection of text replacements to disable malicious HTML and JavaScript code that exploits known security holes in Internet Explorer. Presently, it only protects against Nimda and a cross-site scripting bug, and would need active maintenance to provide more substantial protection. site-specifics Some web sites have very specific problems, the cure for which doesn't apply anywhere else, or could even cause damage on other sites. This is a collection of such site-specific cures which should only be applied to the sites they were intended for, which is what the supplied default.action file does. Users shouldn't need to change anything regarding this filter. google A CSS based block for Google text ads. Also removes a width limitation and the toolbar advertisement. yahoo Another CSS based block, this time for Yahoo text ads. And removes a width limitation as well. msn Another CSS based block, this time for MSN text ads. And removes tracking URLs, as well as a width limitation. blogspot Cleans up some Blogspot blogs. Read the fine print before using this one! This filter also intentionally removes some navigation stuff and sets the page width to 100%. As a result, some rounded corners would appear to early or not at all and as fixing this would require a browser that understands background-size (CSS3), they are removed instead. xml-to-html Server-header filter to change the Content-Type from xml to html. html-to-xml Server-header filter to change the Content-Type from html to xml. no-ping Removes the non-standard ping attribute from anchor and area HTML tags. hide-tor-exit-notation Client-header filter to remove the Tor exit node notation found in Host and Referer headers. If &my-app; and Tor are chained and &my-app; is configured to use socks4a, one can use http://www.example.org.foobar.exit/ to access the host www.example.org through the Tor exit node foobar. As the HTTP client isn't aware of this notation, it treats the whole string www.example.org.foobar.exit as host and uses it for the Host and Referer headers. From the server's point of view the resulting headers are invalid and can cause problems. An invalid Referer header can trigger hot-linking protections, an invalid Host header will make it impossible for the server to find the right vhost (several domains hosted on the same IP address). This client-header filter removes the foo.exit part in those headers to prevent the mentioned problems. Note that it only modifies the HTTP headers, it doesn't make it impossible for the server to detect your Tor exit node based on the IP address the request is coming from. Privoxy's Template Files All Privoxy built-in pages, i.e. error pages such as the 404 - No Such Domain error page, the BLOCKED page and all pages of its web-based user interface, are generated from templates. (Privoxy must be running for the above links to work as intended.) These templates are stored in a subdirectory of the configuration directory called templates. On Unixish platforms, this is typically /etc/privoxy/templates/. The templates are basically normal HTML files, but with place-holders (called symbols or exports), which Privoxy fills at run time. It is possible to edit the templates with a normal text editor, should you want to customize them. (Not recommended for the casual user). Should you create your own custom templates, you should use the config setting templdir to specify an alternate location, so your templates do not get overwritten during upgrades. Note that just like in configuration files, lines starting with # are ignored when the templates are filled in. The place-holders are of the form @name@, and you will find a list of available symbols, which vary from template to template, in the comments at the start of each file. Note that these comments are not always accurate, and that it's probably best to look at the existing HTML code to find out which symbols are supported and what they are filled in with. A special application of this substitution mechanism is to make whole blocks of HTML code disappear when a specific symbol is set. We use this for many purposes, one of them being to include the beta warning in all our user interface (CGI) pages when Privoxy is in an alpha or beta development stage: <!-- @if-unstable-start --> ... beta warning HTML code goes here ... <!-- if-unstable-end@ --> If the "unstable" symbol is set, everything in between and including @if-unstable-start and if-unstable-end@ will disappear, leaving nothing but an empty comment: <!-- --> There's also an if-then-else construct and an #include mechanism, but you'll sure find out if you are inclined to edit the templates ;-) All templates refer to a style located at http://config.privoxy.org/send-stylesheet. This is, of course, locally served by Privoxy and the source for it can be found and edited in the cgi-style.css template. Contacting the Developers, Bug Reporting and Feature Requests &contacting; Privoxy Copyright, License and History ©right; License &license; History &history; Authors &p-authors; See Also &seealso; Appendix Regular Expressions Privoxy uses Perl-style regular expressions in its actions files and filter file, through the PCRE and PCRS libraries. If you are reading this, you probably don't understand what regular expressions are, or what they can do. So this will be a very brief introduction only. A full explanation would require a book ;-) Regular expressions provide a language to describe patterns that can be run against strings of characters (letter, numbers, etc), to see if they match the string or not. The patterns are themselves (sometimes complex) strings of literal characters, combined with wild-cards, and other special characters, called meta-characters. The meta-characters have special meanings and are used to build complex patterns to be matched against. Perl Compatible Regular Expressions are an especially convenient dialect of the regular expression language. To make a simple analogy, we do something similar when we use wild-card characters when listing files with the dir command in DOS. *.* matches all filenames. The special character here is the asterisk which matches any and all characters. We can be more specific and use ? to match just individual characters. So dir file?.text would match file1.txt, file2.txt, etc. We are pattern matching, using a similar technique to regular expressions! Regular expressions do essentially the same thing, but are much, much more powerful. There are many more special characters and ways of building complex patterns however. Let's look at a few of the common ones, and then some examples: . - Matches any single character, e.g. a, A, 4, :, or @. ? - The preceding character or expression is matched ZERO or ONE times. Either/or. + - The preceding character or expression is matched ONE or MORE times. * - The preceding character or expression is matched ZERO or MORE times. \ - The escape character denotes that the following character should be taken literally. This is used where one of the special characters (e.g. .) needs to be taken literally and not as a special meta-character. Example: example\.com, makes sure the period is recognized only as a period (and not expanded to its meta-character meaning of any single character). [ ] - Characters enclosed in brackets will be matched if any of the enclosed characters are encountered. For instance, [0-9] matches any numeric digit (zero through nine). As an example, we can combine this with + to match any digit one of more times: [0-9]+. ( ) - parentheses are used to group a sub-expression, or multiple sub-expressions. | - The bar character works like an or conditional statement. A match is successful if the sub-expression on either side of | matches. As an example: /(this|that) example/ uses grouping and the bar character and would match either this example or that example, and nothing else. These are just some of the ones you are likely to use when matching URLs with Privoxy, and is a long way from a definitive list. This is enough to get us started with a few simple examples which may be more illuminating: /.*/banners/.* - A simple example that uses the common combination of . and * to denote any character, zero or more times. In other words, any string at all. So we start with a literal forward slash, then our regular expression pattern (.*) another literal forward slash, the string banners, another forward slash, and lastly another .*. We are building a directory path here. This will match any file with the path that has a directory named banners in it. The .* matches any characters, and this could conceivably be more forward slashes, so it might expand into a much longer looking path. For example, this could match: /eye/hate/spammers/banners/annoy_me_please.gif, or just /banners/annoying.html, or almost an infinite number of other possible combinations, just so it has banners in the path somewhere. And now something a little more complex: /.*/adv((er)?ts?|ertis(ing|ements?))?/ - We have several literal forward slashes again (/), so we are building another expression that is a file path statement. We have another .*, so we are matching against any conceivable sub-path, just so it matches our expression. The only true literal that must match our pattern is adv, together with the forward slashes. What comes after the adv string is the interesting part. Remember the ? means the preceding expression (either a literal character or anything grouped with (...) in this case) can exist or not, since this means either zero or one match. So ((er)?ts?|ertis(ing|ements?)) is optional, as are the individual sub-expressions: (er), (ing|ements?), and the s. The | means or. We have two of those. For instance, (ing|ements?), can expand to match either ing OR ements?. What is being done here, is an attempt at matching as many variations of advertisement, and similar, as possible. So this would expand to match just adv, or advert, or adverts, or advertising, or advertisement, or advertisements. You get the idea. But it would not match advertizements (with a z). We could fix that by changing our regular expression to: /.*/adv((er)?ts?|erti(s|z)(ing|ements?))?/, which would then match either spelling. /.*/advert[0-9]+\.(gif|jpe?g) - Again another path statement with forward slashes. Anything in the square brackets [ ] can be matched. This is using 0-9 as a shorthand expression to mean any digit one through nine. It is the same as saying 0123456789. So any digit matches. The + means one or more of the preceding expression must be included. The preceding expression here is what is in the square brackets -- in this case, any digit one through nine. Then, at the end, we have a grouping: (gif|jpe?g). This includes a |, so this needs to match the expression on either side of that bar character also. A simple gif on one side, and the other side will in turn match either jpeg or jpg, since the ? means the letter e is optional and can be matched once or not at all. So we are building an expression here to match image GIF or JPEG type image file. It must include the literal string advert, then one or more digits, and a . (which is now a literal, and not a special character, since it is escaped with \), and lastly either gif, or jpeg, or jpg. Some possible matches would include: //advert1.jpg, /nasty/ads/advert1234.gif, /banners/from/hell/advert99.jpg. It would not match advert1.gif (no leading slash), or /adverts232.jpg (the expression does not include an s), or /advert1.jsp (jsp is not in the expression anywhere). We are barely scratching the surface of regular expressions here so that you can understand the default Privoxy configuration files, and maybe use this knowledge to customize your own installation. There is much, much more that can be done with regular expressions. Now that you know enough to get started, you can learn more on your own :/ More reading on Perl Compatible Regular expressions: http://perldoc.perl.org/perlre.html For information on regular expression based substitutions and their applications in filters, please see the filter file tutorial in this manual. Privoxy's Internal Pages Since Privoxy proxies each requested web page, it is easy for Privoxy to trap certain special URLs. In this way, we can talk directly to Privoxy, and see how it is configured, see how our rules are being applied, change these rules and other configuration options, and even turn Privoxy's filtering off, all with a web browser. The URLs listed below are the special ones that allow direct access to Privoxy. Of course, Privoxy must be running to access these. If not, you will get a friendly error message. Internet access is not necessary either. Privoxy main page:
http://config.privoxy.org/
There is a shortcut: http://p.p/ (But it doesn't provide a fall-back to a real page, in case the request is not sent through Privoxy)
Show information about the current configuration, including viewing and editing of actions files:
http://config.privoxy.org/show-status
Show the source code version numbers:
http://config.privoxy.org/show-version
Show the browser's request headers:
http://config.privoxy.org/show-request
Show which actions apply to a URL and why:
http://config.privoxy.org/show-url-info
Toggle Privoxy on or off. This feature can be turned off/on in the main config file. When toggled off, Privoxy continues to run, but only as a pass-through proxy, with no actions taking place:
http://config.privoxy.org/toggle
Short cuts. Turn off, then on:
http://config.privoxy.org/toggle?set=disable
http://config.privoxy.org/toggle?set=enable
These may be bookmarked for quick reference. See next. Bookmarklets Below are some bookmarklets to allow you to easily access a mini version of some of Privoxy's special pages. They are designed for MS Internet Explorer, but should work equally well in Netscape, Mozilla, and other browsers which support JavaScript. They are designed to run directly from your bookmarks - not by clicking the links below (although that should work for testing). To save them, right-click the link and choose Add to Favorites (IE) or Add Bookmark (Netscape). You will get a warning that the bookmark may not be safe - just click OK. Then you can run the Bookmarklet directly from your favorites/bookmarks. For even faster access, you can put them on the Links bar (IE) or the Personal Toolbar (Netscape), and run them with a single click. Privoxy - Enable Privoxy - Disable Privoxy - Toggle Privoxy (Toggles between enabled and disabled) Privoxy- View Status Privoxy - Why? Credit: The site which gave us the general idea for these bookmarklets is www.bookmarklets.com. They have more information about bookmarklets.
Chain of Events Let's take a quick look at how some of Privoxy's core features are triggered, and the ensuing sequence of events when a web page is requested by your browser: First, your web browser requests a web page. The browser knows to send the request to Privoxy, which will in turn, relay the request to the remote web server after passing the following tests: Privoxy traps any request for its own internal CGI pages (e.g http://p.p/) and sends the CGI page back to the browser. Next, Privoxy checks to see if the URL matches any +block patterns. If so, the URL is then blocked, and the remote web server will not be contacted. +handle-as-image and +handle-as-empty-document are then checked, and if there is no match, an HTML BLOCKED page is sent back to the browser. Otherwise, if it does match, an image is returned for the former, and an empty text document for the latter. The type of image would depend on the setting of +set-image-blocker (blank, checkerboard pattern, or an HTTP redirect to an image elsewhere). Untrusted URLs are blocked. If URLs are being added to the trust file, then that is done. If the URL pattern matches the +fast-redirects action, it is then processed. Unwanted parts of the requested URL are stripped. Now the rest of the client browser's request headers are processed. If any of these match any of the relevant actions (e.g. +hide-user-agent, etc.), headers are suppressed or forged as determined by these actions and their parameters. Now the web server starts sending its response back (i.e. typically a web page). First, the server headers are read and processed to determine, among other things, the MIME type (document type) and encoding. The headers are then filtered as determined by the +crunch-incoming-cookies, +session-cookies-only, and +downgrade-http-version actions. If any +filter action or +deanimate-gifs action applies (and the document type fits the action), the rest of the page is read into memory (up to a configurable limit). Then the filter rules (from default.filter and any other filter files) are processed against the buffered content. Filters are applied in the order they are specified in one of the filter files. Animated GIFs, if present, are reduced to either the first or last frame, depending on the action setting.The entire page, which is now filtered, is then sent by Privoxy back to your browser. If neither a +filter action or +deanimate-gifs matches, then Privoxy passes the raw data through to the client browser as it becomes available. As the browser receives the now (possibly filtered) page content, it reads and then requests any URLs that may be embedded within the page source, e.g. ad images, stylesheets, JavaScript, other HTML documents (e.g. frames), sounds, etc. For each of these objects, the browser issues a separate request (this is easily viewable in Privoxy's logs). And each such request is in turn processed just as above. Note that a complex web page will have many, many such embedded URLs. If these secondary requests are to a different server, then quite possibly a very differing set of actions is triggered. NOTE: This is somewhat of a simplistic overview of what happens with each URL request. For the sake of brevity and simplicity, we have focused on Privoxy's core features only. Troubleshooting: Anatomy of an Action The way Privoxy applies actions and filters to any given URL can be complex, and not always so easy to understand what is happening. And sometimes we need to be able to see just what Privoxy is doing. Especially, if something Privoxy is doing is causing us a problem inadvertently. It can be a little daunting to look at the actions and filters files themselves, since they tend to be filled with regular expressions whose consequences are not always so obvious. One quick test to see if Privoxy is causing a problem or not, is to disable it temporarily. This should be the first troubleshooting step. See the Bookmarklets section on a quick and easy way to do this (be sure to flush caches afterward!). Looking at the logs is a good idea too. (Note that both the toggle feature and logging are enabled via config file settings, and may need to be turned on.) Another easy troubleshooting step to try is if you have done any customization of your installation, revert back to the installed defaults and see if that helps. There are times the developers get complaints about one thing or another, and the problem is more related to a customized configuration issue. Privoxy also provides the http://config.privoxy.org/show-url-info page that can show us very specifically how actions are being applied to any given URL. This is a big help for troubleshooting. First, enter one URL (or partial URL) at the prompt, and then Privoxy will tell us how the current configuration will handle it. This will not help with filtering effects (i.e. the +filter action) from one of the filter files since this is handled very differently and not so easy to trap! It also will not tell you about any other URLs that may be embedded within the URL you are testing. For instance, images such as ads are expressed as URLs within the raw page source of HTML pages. So you will only get info for the actual URL that is pasted into the prompt area -- not any sub-URLs. If you want to know about embedded URLs like ads, you will have to dig those out of the HTML source. Use your browser's View Page Source option for this. Or right click on the ad, and grab the URL. Let's try an example, google.com, and look at it one section at a time in a sample configuration (your real configuration may vary): Matches for http://www.google.com: In file: default.action [ View ] [ Edit ] {+change-x-forwarded-for{block} +deanimate-gifs {last} +fast-redirects {check-decoded-url} +filter {refresh-tags} +filter {img-reorder} +filter {banners-by-size} +filter {webbugs} +filter {jumping-windows} +filter {ie-exploits} +hide-from-header {block} +hide-referrer {forge} +session-cookies-only +set-image-blocker {pattern} / { -session-cookies-only } .google.com { -fast-redirects } .google.com In file: user.action [ View ] [ Edit ] (no matches in this file) This is telling us how we have defined our actions, and which ones match for our test case, google.com. Displayed is all the actions that are available to us. Remember, the + sign denotes on. - denotes off. So some are on here, but many are off. Each example we try may provide a slightly different end result, depending on our configuration directives. The first listing is for our default.action file. The large, multi-line listing, is how the actions are set to match for all URLs, i.e. our default settings. If you look at your actions file, this would be the section just below the aliases section near the top. This will apply to all URLs as signified by the single forward slash at the end of the listing -- / . But we have defined additional actions that would be exceptions to these general rules, and then we list specific URLs (or patterns) that these exceptions would apply to. Last match wins. Just below this then are two explicit matches for .google.com. The first is negating our previous cookie setting, which was for +session-cookies-only (i.e. not persistent). So we will allow persistent cookies for google, at least that is how it is in this example. The second turns off any +fast-redirects action, allowing this to take place unmolested. Note that there is a leading dot here -- .google.com. This will match any hosts and sub-domains, in the google.com domain also, such as www.google.com or mail.google.com. But it would not match www.google.de! So, apparently, we have these two actions defined as exceptions to the general rules at the top somewhere in the lower part of our default.action file, and google.com is referenced somewhere in these latter sections. Then, for our user.action file, we again have no hits. So there is nothing google-specific that we might have added to our own, local configuration. If there was, those actions would over-rule any actions from previously processed files, such as default.action. user.action typically has the last word. This is the best place to put hard and fast exceptions, And finally we pull it all together in the bottom section and summarize how Privoxy is applying all its actions to google.com: Final results: -add-header -block +change-x-forwarded-for{block} -client-header-filter{hide-tor-exit-notation} -content-type-overwrite -crunch-client-header -crunch-if-none-match -crunch-incoming-cookies -crunch-outgoing-cookies -crunch-server-header +deanimate-gifs {last} -downgrade-http-version -fast-redirects -filter {js-events} -filter {content-cookies} -filter {all-popups} -filter {banners-by-link} -filter {tiny-textforms} -filter {frameset-borders} -filter {demoronizer} -filter {shockwave-flash} -filter {quicktime-kioskmode} -filter {fun} -filter {crude-parental} -filter {site-specifics} -filter {js-annoyances} -filter {html-annoyances} +filter {refresh-tags} -filter {unsolicited-popups} +filter {img-reorder} +filter {banners-by-size} +filter {webbugs} +filter {jumping-windows} +filter {ie-exploits} -filter {google} -filter {yahoo} -filter {msn} -filter {blogspot} -filter {no-ping} -force-text-mode -handle-as-empty-document -handle-as-image -hide-accept-language -hide-content-disposition +hide-from-header {block} -hide-if-modified-since +hide-referrer {forge} -hide-user-agent -limit-connect -overwrite-last-modified -prevent-compression -redirect -server-header-filter{xml-to-html} -server-header-filter{html-to-xml} -session-cookies-only +set-image-blocker {pattern} Notice the only difference here to the previous listing, is to fast-redirects and session-cookies-only, which are activated specifically for this site in our configuration, and thus show in the Final Results. Now another example, ad.doubleclick.net: { +block{Domains starts with "ad"} } ad*. { +block{Domain contains "ad"} } .ad. { +block{Doubleclick banner server} +handle-as-image } .[a-vx-z]*.doubleclick.net We'll just show the interesting part here - the explicit matches. It is matched three different times. Two +block{} sections, and a +block{} +handle-as-image, which is the expanded form of one of our aliases that had been defined as: +block-as-image. (Aliases are defined in the first section of the actions file and typically used to combine more than one action.) Any one of these would have done the trick and blocked this as an unwanted image. This is unnecessarily redundant since the last case effectively would also cover the first. No point in taking chances with these guys though ;-) Note that if you want an ad or obnoxious URL to be invisible, it should be defined as ad.doubleclick.net is done here -- as both a +block{} and an +handle-as-image. The custom alias +block-as-image just simplifies the process and make it more readable. One last example. Let's try http://www.example.net/adsl/HOWTO/. This one is giving us problems. We are getting a blank page. Hmmm ... Matches for http://www.example.net/adsl/HOWTO/: In file: default.action [ View ] [ Edit ] {-add-header -block +change-x-forwarded-for{block} -client-header-filter{hide-tor-exit-notation} -content-type-overwrite -crunch-client-header -crunch-if-none-match -crunch-incoming-cookies -crunch-outgoing-cookies -crunch-server-header +deanimate-gifs -downgrade-http-version +fast-redirects {check-decoded-url} -filter {js-events} -filter {content-cookies} -filter {all-popups} -filter {banners-by-link} -filter {tiny-textforms} -filter {frameset-borders} -filter {demoronizer} -filter {shockwave-flash} -filter {quicktime-kioskmode} -filter {fun} -filter {crude-parental} -filter {site-specifics} -filter {js-annoyances} -filter {html-annoyances} +filter {refresh-tags} -filter {unsolicited-popups} +filter {img-reorder} +filter {banners-by-size} +filter {webbugs} +filter {jumping-windows} +filter {ie-exploits} -filter {google} -filter {yahoo} -filter {msn} -filter {blogspot} -filter {no-ping} -force-text-mode -handle-as-empty-document -handle-as-image -hide-accept-language -hide-content-disposition +hide-from-header{block} +hide-referer{forge} -hide-user-agent -overwrite-last-modified +prevent-compression -redirect -server-header-filter{xml-to-html} -server-header-filter{html-to-xml} +session-cookies-only +set-image-blocker{blank} } / { +block{Path contains "ads".} +handle-as-image } /ads Ooops, the /adsl/ is matching /ads in our configuration! But we did not want this at all! Now we see why we get the blank page. It is actually triggering two different actions here, and the effects are aggregated so that the URL is blocked, and &my-app; is told to treat the block as if it were an image. But this is, of course, all wrong. We could now add a new action below this (or better in our own user.action file) that explicitly un blocks ( {-block}) paths with adsl in them (remember, last match in the configuration wins). There are various ways to handle such exceptions. Example: { -block } /adsl Now the page displays ;-) Remember to flush your browser's caches when making these kinds of changes to your configuration to insure that you get a freshly delivered page! Or, try using Shift+Reload. But now what about a situation where we get no explicit matches like we did with: { +block{Path starts with "ads".} +handle-as-image } /ads That actually was very helpful and pointed us quickly to where the problem was. If you don't get this kind of match, then it means one of the default rules in the first section of default.action is causing the problem. This would require some guesswork, and maybe a little trial and error to isolate the offending rule. One likely cause would be one of the +filter actions. These tend to be harder to troubleshoot. Try adding the URL for the site to one of aliases that turn off +filter: { shop } .quietpc.com .worldpay.com # for quietpc.com .jungle.com .scan.co.uk .forbes.com { shop } is an alias that expands to { -filter -session-cookies-only }. Or you could do your own exception to negate filtering: { -filter } # Disable ALL filter actions for sites in this section .forbes.com developer.ibm.com localhost This would turn off all filtering for these sites. This is best put in user.action, for local site exceptions. Note that when a simple domain pattern is used by itself (without the subsequent path portion), all sub-pages within that domain are included automatically in the scope of the action. Images that are inexplicably being blocked, may well be hitting the +filter{banners-by-size} rule, which assumes that images of certain sizes are ad banners (works well most of the time since these tend to be standardized). { fragile } is an alias that disables most actions that are the most likely to cause trouble. This can be used as a last resort for problem sites. { fragile } # Handle with care: easy to break mail.google. mybank.example.com Remember to flush caches! Note that the mail.google reference lacks the TLD portion (e.g. .com). This will effectively match any TLD with google in it, such as mail.google.de., just as an example. If this still does not work, you will have to go through the remaining actions one by one to find which one(s) is causing the problem.