1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN""http://www.w3.org/TR/html4/loose.dtd">
5 >The Main Configuration File</TITLE
8 CONTENT="Modular DocBook HTML Stylesheet Version 1.79"><LINK
10 TITLE="Privoxy 3.0.16 User Manual"
11 HREF="index.html"><LINK
13 TITLE="Privoxy Configuration"
14 HREF="configuration.html"><LINK
17 HREF="actions-file.html"><LINK
20 HREF="../p_doc.css"><META
21 HTTP-EQUIV="Content-Type"
24 <LINK REL="STYLESHEET" TYPE="text/css" HREF="p_doc.css">
36 SUMMARY="Header navigation table"
45 >Privoxy 3.0.16 User Manual</TH
53 HREF="configuration.html"
67 HREF="actions-file.html"
82 >7. The Main Configuration File</A
85 > Again, the main configuration file is named <TT
89 Linux/Unix/BSD and OS/2, and <TT
93 Configuration lines consist of an initial keyword followed by a list of
94 values, all separated by whitespace (any number of spaces or tabs). For
100 CLASS="LITERALLAYOUT"
105 >confdir /etc/privoxy</I
112 > Assigns the value <TT
119 > and thus indicates that the configuration
120 directory is named <SPAN
122 >"/etc/privoxy/"</SPAN
125 > All options in the config file except for <TT
132 > are optional. Watch out in the below description
133 for what happens if you leave them unset.</P
135 > The main config file controls all aspects of <SPAN
139 operation that are not location dependent (i.e. they apply universally, no matter
140 where you may be surfing). Like the filter and action files, the config file is
141 a plain text file and can be modified with a text editor like emacs, vim or
149 >7.1. Local Set-up Documentation</A
152 > If you intend to operate <SPAN
156 than just yourself, it might be a good idea to let them know how to reach
157 you, what you block and why you do that, your policies, etc.
165 >7.1.1. user-manual</A
176 > Location of the <SPAN
186 >A fully qualified URI</P
201 >Effect if unset:</DT
205 HREF="http://www.privoxy.org/user-manual/"
207 >http://www.privoxy.org/<TT
214 will be used, where <TT
229 > The User Manual URI is the single best source of information on
233 >, and is used for help links from some
234 of the internal CGI pages. The manual itself is normally packaged with the
235 binary distributions, so you probably want to set this to a locally
242 > The best all purpose solution is simply to put the full local
261 >  user-manual  /usr/share/doc/privoxy/user-manual</PRE
268 > The User Manual is then available to anyone with access to
272 >, by following the built-in URL:
275 >http://config.privoxy.org/user-manual/</TT
277 (or the shortcut: <TT
279 >http://p.p/user-manual/</TT
283 > If the documentation is not on the local system, it can be accessed
284 from a remote server, as:
295 >  user-manual  http://example.com/privoxy/user-manual/</PRE
320 > If set, this option should be <SPAN
324 >the first option in the config
327 >, because it is used while the config file is being read
343 NAME="TRUST-INFO-URL"
344 >7.1.2. trust-info-url</A
355 > A URL to be displayed in the error page that users will see if access to an untrusted page is denied.
377 >Effect if unset:</DT
380 > No links are displayed on the "untrusted" error page.
387 > The value of this option only matters if the experimental trust mechanism has been
389 HREF="config.html#TRUSTFILE"
400 > If you use the trust mechanism, it is a good idea to write up some on-line
401 documentation about your trust policy and to specify the URL(s) here.
402 Use multiple times for multiple URLs.
405 > The URL(s) should be added to the trustfile as well, so users don't end up
406 locked out from the information on why they were locked out in the first place!
418 >7.1.3. admin-address</A
429 > An email address to reach the <SPAN
454 >Effect if unset:</DT
457 > No email address is displayed on error pages and the CGI user interface.
471 are unset, the whole "Local Privoxy Support" box on all generated pages will
483 NAME="PROXY-INFO-URL"
484 >7.1.4. proxy-info-url</A
495 > A URL to documentation about the local <SPAN
499 configuration or policies.
521 >Effect if unset:</DT
524 > No link to local documentation is displayed on error pages and the CGI user interface.
538 are unset, the whole "Local Privoxy Support" box on all generated pages will
542 > This URL shouldn't be blocked ;-)
555 >7.2. Configuration and Log File Locations</A
561 > can (and normally does) use a number of
562 other files for additional configuration, help and logging.
563 This section of the configuration file tells <SPAN
567 where to find those other files. </P
569 > The user running <SPAN
573 permission for all configuration files, and write permission to any files
574 that would be modified, such as log files and actions files.</P
592 >The directory where the other configuration files are located.</P
604 >/etc/privoxy (Unix) <SPAN
613 > installation dir (Windows) </P
616 >Effect if unset:</DT
660 >An alternative directory where the templates are loaded from.</P
675 >Effect if unset:</DT
678 >The templates are assumed to be located in confdir/template.</P
687 > original templates are usually
688 overwritten with each update. Use this option to relocate customized
689 templates that should be kept. As template variables might change
690 between updates, you shouldn't expect templates to work with
694 > releases other than the one
695 they were part of, though.
718 > The directory where all logging takes place
735 >/var/log/privoxy (Unix) <SPAN
744 > installation dir (Windows) </P
747 >Effect if unset:</DT
780 >7.2.4. actionsfile</A
783 NAME="DEFAULT.ACTION"
786 NAME="STANDARD.ACTION"
801 HREF="actions-file.html"
810 >Complete file name, relative to <TT
826 CLASS="LITERALLAYOUT"
827 > match-all.action # Actions that are applied to all sites and maybe overruled later on.</P
834 CLASS="LITERALLAYOUT"
835 > default.action # Main actions file</P
842 CLASS="LITERALLAYOUT"
843 > user.action # User customizations</P
853 >Effect if unset:</DT
856 > No actions are taken at all. More or less neutral proxying.
866 > lines are permitted, and are in fact recommended!
870 The default values are <TT
877 > actions file maintained by the developers, and
881 >, where you can make your personal additions.
885 Actions files contain all the per site and per URL configuration for
886 ad blocking, cookie management, privacy considerations, etc.
887 There is no point in using <SPAN
891 least one actions file.
894 > Note that since Privoxy 3.0.7, the complete filename, including the <SPAN
898 extension has to be specified. The syntax change was necessary to be consistent
899 with the other file options and to allow previously forbidden characters.
911 >7.2.5. filterfile</A
914 NAME="DEFAULT.FILTER"
926 HREF="filter-file.html"
935 >File name, relative to <TT
944 >default.filter (Unix) <SPAN
950 > default.filter.txt (Windows)</P
953 >Effect if unset:</DT
956 > No textual content filtering takes place, i.e. all
960 HREF="actions-file.html#FILTER"
969 actions in the actions files are turned neutral.
979 > lines are permitted.
983 HREF="filter-file.html"
985 > contain content modification
987 HREF="appendix.html#REGEX"
988 >regular expressions</A
989 >. These rules permit
990 powerful changes on the content of Web pages, and optionally the headers
991 as well, e.g., you could try to disable your favorite JavaScript annoyances,
992 re-write the actual displayed text, or just have some fun
993 playing buzzword bingo with web pages.
1000 HREF="actions-file.html#FILTER"
1009 actions rely on the relevant filter (<TT
1015 to be defined in a filter file!
1018 > A pre-defined filter file called <TT
1022 a number of useful filters for common problems is included in the distribution.
1023 See the section on the <TT
1026 HREF="actions-file.html#FILTER"
1033 > It is recommended to place any locally adapted filters into a separate
1054 CLASS="VARIABLELIST"
1060 > The log file to use
1067 >File name, relative to <TT
1080 >Unset (commented out)</I
1082 >. When activated: logfile (Unix) <SPAN
1088 > privoxy.log (Windows).</P
1091 >Effect if unset:</DT
1094 > No logfile is written.
1101 > The logfile is where all logging and error messages are written. The level
1102 of detail and number of messages are set with the <TT
1106 option (see below). The logfile can be useful for tracking down a problem with
1110 > (e.g., it's not blocking an ad you
1111 think it should block) and it can help you to monitor what your browser
1115 > Depending on the debug options below, the logfile may be a privacy risk
1116 if third parties can get access to it. As most users will never look
1120 > 3.0.7 and later only log fatal
1124 > For most troubleshooting purposes, you will have to change that,
1125 please refer to the debugging section for details.
1128 > Your logfile will grow indefinitely, and you will probably want to
1129 periodically remove it. On Unix systems, you can do this with a cron job
1133 >). For Red Hat based Linux distributions, a
1137 > script has been included.
1140 > Any log files must be writable by whatever user <SPAN
1144 is being run as (on Unix, default user id is <SPAN
1159 >7.2.7. trustfile</A
1164 CLASS="VARIABLELIST"
1170 > The name of the trust file to use
1177 >File name, relative to <TT
1190 >Unset (commented out)</I
1192 >. When activated: trust (Unix) <SPAN
1198 > trust.txt (Windows)</P
1201 >Effect if unset:</DT
1204 > The entire trust mechanism is disabled.
1211 > The trust mechanism is an experimental feature for building white-lists and should
1212 be used with care. It is <SPAN
1218 > recommended for the casual user.
1221 > If you specify a trust file, <SPAN
1225 access to sites that are specified in the trustfile. Sites can be listed
1232 > character limits access to this site
1233 only (and any sub-paths within this site), e.g.
1236 >~www.example.com</TT
1240 >~www.example.com/features/news.html</TT
1244 > Or, you can designate sites as <SPAN
1248 >trusted referrers</I
1251 prepending the name with a <TT
1254 > character. The effect is that
1255 access to untrusted sites will be granted -- but only if a link from this
1256 trusted referrer was used to get there. The link target will then be added
1260 > so that future, direct accesses will be
1261 granted. Sites added via this mechanism do not become trusted referrers
1262 themselves (i.e. they are added with a <TT
1266 There is a limit of 512 such entries, after which new entries will not be
1270 > If you use the <TT
1273 > operator in the trust file, it may grow
1274 considerably over time.
1277 > It is recommended that <SPAN
1283 >--disable-force</TT
1286 >--disable-toggle</TT
1290 > --disable-editor</TT
1291 > options, if this feature is to be
1295 > Possible applications include limiting Internet access for children.
1311 > These options are mainly useful when tracing a problem.
1312 Note that you might also want to invoke
1320 command line option when debugging.
1333 CLASS="VARIABLELIST"
1339 > Key values that determine what information gets logged.
1352 >0 (i.e.: only fatal errors (that cause Privoxy to exit) are logged)</P
1355 >Effect if unset:</DT
1358 > Default value is used (see above).
1365 > The available debug levels are:
1375 CLASS="PROGRAMLISTING"
1376 > debug 1 # Log the destination for each request <SPAN
1379 > let through. See also debug 1024.
1380 debug 2 # show each connection status
1381 debug 4 # show I/O status
1382 debug 8 # show header parsing
1383 debug 16 # log all data written to the network into the logfile
1384 debug 32 # debug force feature
1385 debug 64 # debug regular expression filters
1386 debug 128 # debug redirects
1387 debug 256 # debug GIF de-animation
1388 debug 512 # Common Log Format
1389 debug 1024 # Log the destination for requests <SPAN
1392 > didn't let through, and the reason why.
1393 debug 2048 # CGI user interface
1394 debug 4096 # Startup banner and warnings.
1395 debug 8192 # Non-fatal errors</PRE
1402 > To select multiple debug levels, you can either add them or use
1409 > A debug level of 1 is informative because it will show you each request
1410 as it happens. <SPAN
1414 >1, 1024, 4096 and 8192 are recommended</I
1417 so that you will notice when things go wrong. The other levels are
1418 probably only of interest if you are hunting down a specific problem.
1419 They can produce a hell of an output (especially 16).
1426 > used to ship with the debug levels recommended above enabled by
1427 default, but due to privacy concerns 3.0.7 and later are configured to
1428 only log fatal errors.
1431 > If you are used to the more verbose settings, simply enable the debug lines
1435 > If you want to use pure CLF (Common Log Format), you should set <SPAN
1445 > and not enable anything else.
1451 > has a hard-coded limit for the
1452 length of log messages. If it's reached, messages are logged truncated
1453 and marked with <SPAN
1455 >"... [too long, truncated]"</SPAN
1459 > Please don't file any support requests without trying to reproduce
1460 the problem with increased debug level first. Once you read the log
1461 messages, you may even be able to solve the problem on your own.
1472 NAME="SINGLE-THREADED"
1473 >7.3.2. single-threaded</A
1478 CLASS="VARIABLELIST"
1484 > Whether to run only one server thread.
1512 >Effect if unset:</DT
1515 > Multi-threaded (or, where unavailable: forked) operation, i.e. the ability to
1516 serve multiple requests simultaneously.
1523 > This option is only there for debugging purposes.
1528 >It will drastically reduce performance.</I
1547 CLASS="VARIABLELIST"
1553 > The hostname shown on the CGI pages.
1575 >Effect if unset:</DT
1578 > The hostname provided by the operating system is used.
1585 > On some misconfigured systems resolving the hostname fails or
1586 takes too much time and slows Privoxy down. Setting a fixed hostname
1587 works around the problem.
1590 > In other circumstances it might be desirable to show a hostname
1591 other than the one returned by the operating system. For example
1592 if the system has several different hostnames and you don't want
1593 to use the first one.
1596 > Note that Privoxy does not validate the specified hostname value.
1608 NAME="ACCESS-CONTROL"
1609 >7.4. Access Control and Security</A
1612 > This section of the config file controls the security-relevant aspects
1623 NAME="LISTEN-ADDRESS"
1624 >7.4.1. listen-address</A
1629 CLASS="VARIABLELIST"
1635 > The IP address and TCP port on which <SPAN
1639 listen for client requests.
1665 >Effect if unset:</DT
1668 > Bind to 127.0.0.1 (IPv4 localhost), port 8118. This is suitable and
1669 recommended for home users who run <SPAN
1673 the same machine as their browser.
1680 > You will need to configure your browser(s) to this proxy address and port.
1683 > If you already have another service running on port 8118, or if you want to
1684 serve requests from other machines (e.g. on your local network) as well, you
1685 will need to override the default.
1688 > IPv6 addresses containing colons have to be quoted by brackets.
1691 > If you leave out the IP address, <SPAN
1695 bind to all IPv4 interfaces (addresses) on your machine and may become reachable
1696 from the Internet. In that case, consider using <A
1697 HREF="config.html#ACLS"
1698 >access control lists</A
1699 > (ACL's, see below), and/or
1700 a firewall. If the hostname is localhost, <SPAN
1704 will explicitly try to bind to an IPv4 address. For other hostnames it depends
1705 on the operating system which IP version will be used.
1711 > to untrusted users, you will
1712 also want to make sure that the following actions are disabled: <TT
1715 HREF="config.html#ENABLE-EDIT-ACTIONS"
1716 >enable-edit-actions</A
1722 HREF="config.html#ENABLE-REMOTE-TOGGLE"
1723 >enable-remote-toggle</A
1732 > Suppose you are running <SPAN
1736 a machine which has the address 192.168.0.1 on your local private network
1737 (192.168.0.0) and has another outside connection with a different address.
1738 You want it to serve requests from inside only:
1748 CLASS="PROGRAMLISTING"
1749 > listen-address 192.168.0.1:8118</PRE
1756 > Suppose you are running <SPAN
1760 IPv6-capable machine and you want it to listen on the IPv6 address
1761 of the loopback device:
1771 CLASS="PROGRAMLISTING"
1772 > listen-address [::1]:8118</PRE
1793 CLASS="VARIABLELIST"
1799 > Initial state of "toggle" status
1815 >Effect if unset:</DT
1818 > Act as if toggled on
1825 > If set to 0, <SPAN
1831 >"toggled off"</SPAN
1832 > mode, i.e. mostly behave like a normal,
1833 content-neutral proxy with both ad blocking and content filtering
1836 >enable-remote-toggle</TT
1840 > The windows version will only display the toggle icon in the system tray
1841 if this option is present.
1852 NAME="ENABLE-REMOTE-TOGGLE"
1853 >7.4.3. enable-remote-toggle</A
1858 CLASS="VARIABLELIST"
1864 > Whether or not the <A
1865 HREF="http://config.privoxy.org/toggle"
1885 >Effect if unset:</DT
1888 > The web-based toggle feature is disabled.
1895 > When toggled off, <SPAN
1898 > mostly acts like a normal,
1899 content-neutral proxy, i.e. doesn't block ads or filter content.
1902 > Access to the toggle feature can <SPAN
1909 controlled separately by <SPAN
1912 > or HTTP authentication,
1913 so that everybody who can access <SPAN
1924 toggle it for all users. So this option is <SPAN
1931 for multi-user environments with untrusted users.
1934 > Note that malicious client side code (e.g Java) is also
1935 capable of using this option.
1942 documentation, this feature is disabled by default.
1945 > Note that you must have compiled <SPAN
1949 support for this feature, otherwise this option has no effect.
1960 NAME="ENABLE-REMOTE-HTTP-TOGGLE"
1961 >7.4.4. enable-remote-http-toggle</A
1966 CLASS="VARIABLELIST"
1972 > Whether or not Privoxy recognizes special HTTP headers to change its behaviour.
1988 >Effect if unset:</DT
1991 > Privoxy ignores special HTTP headers.
1998 > When toggled on, the client can change <SPAN
2002 behaviour by setting special HTTP headers. Currently the only supported
2003 special header is <SPAN
2005 >"X-Filter: No"</SPAN
2006 >, to disable filtering for
2007 the ongoing request, even if it is enabled in one of the action files.
2010 > This feature is disabled by default. If you are using
2014 > in a environment with trusted clients,
2015 you may enable this feature at your discretion. Note that malicious client
2016 side code (e.g Java) is also capable of using this feature.
2019 > This option will be removed in future releases as it has been obsoleted
2020 by the more general header taggers.
2031 NAME="ENABLE-EDIT-ACTIONS"
2032 >7.4.5. enable-edit-actions</A
2037 CLASS="VARIABLELIST"
2043 > Whether or not the <A
2044 HREF="http://config.privoxy.org/show-status"
2064 >Effect if unset:</DT
2067 > The web-based actions file editor is disabled.
2074 > Access to the editor can <SPAN
2081 controlled separately by <SPAN
2084 > or HTTP authentication,
2085 so that everybody who can access <SPAN
2096 modify its configuration for all users.
2099 > This option is <SPAN
2106 with untrusted users and as a lot of <SPAN
2110 users don't read documentation, this feature is disabled by default.
2113 > Note that malicious client side code (e.g Java) is also
2114 capable of using the actions editor and you shouldn't enable
2115 this options unless you understand the consequences and are
2116 sure your browser is configured correctly.
2119 > Note that you must have compiled <SPAN
2123 support for this feature, otherwise this option has no effect.
2134 NAME="ENFORCE-BLOCKS"
2135 >7.4.6. enforce-blocks</A
2140 CLASS="VARIABLELIST"
2146 > Whether the user is allowed to ignore blocks and can <SPAN
2148 >"go there anyway"</SPAN
2177 >Effect if unset:</DT
2180 > Blocks are not enforced.
2190 > is mainly used to block and filter
2191 requests as a service to the user, for example to block ads and other
2192 junk that clogs the pipes. <SPAN
2196 isn't perfect and sometimes innocent pages are blocked. In this situation it
2197 makes sense to allow the user to enforce the request and have
2204 > In the default configuration <SPAN
2211 > page contains a <SPAN
2213 >"go there anyway"</SPAN
2215 link to adds a special string (the force prefix) to the request URL.
2216 If that link is used, <SPAN
2220 detect the force prefix, remove it again and let the request pass.
2226 > can also be used to enforce
2227 a network policy. In that case the user obviously should not be able to
2228 bypass any blocks, and that's what the <SPAN
2230 >"enforce-blocks"</SPAN
2232 option is for. If it's enabled, <SPAN
2238 >"go there anyway"</SPAN
2239 > link. If the user adds the force
2240 prefix by hand, it will not be accepted and the circumvention attempt
2260 >7.4.7. ACLs: permit-access and deny-access</A
2263 NAME="PERMIT-ACCESS"
2271 CLASS="VARIABLELIST"
2277 > Who can access what.
2329 > are IPv4 addresses in dotted decimal notation or valid
2347 > are subnet masks in CIDR notation, i.e. integer
2348 values from 2 to 30 representing the length (in bits) of the network address. The masks and the whole
2349 destination part are optional.
2352 > If your system implements
2354 HREF="http://tools.ietf.org/html/rfc3493"
2368 > can be IPv6 addresses delimeted by
2375 or a service name, and
2409 any port will match. If no <TT
2420 > is given, the complete IP
2421 address has to match (i.e. 32 bits for IPv4 and 128 bits for IPv6).
2425 >Effect if unset:</DT
2428 > Don't restrict access further than implied by <TT
2438 > Access controls are included at the request of ISPs and systems
2439 administrators, and <SPAN
2443 >are not usually needed by individual users</I
2446 For a typical home user, it will normally suffice to ensure that
2450 > only listens on the localhost
2451 (127.0.0.1) or internal (home) network address by means of the
2453 HREF="config.html#LISTEN-ADDRESS"
2465 > Please see the warnings in the FAQ that <SPAN
2469 is not intended to be a substitute for a firewall or to encourage anyone
2470 to defer addressing basic security weaknesses.
2473 > Multiple ACL lines are OK.
2474 If any ACLs are specified, <SPAN
2478 to IP addresses that match at least one <TT
2482 and don't match any subsequent <TT
2485 > line. In other words, the
2486 last match wins, with the default being <TT
2495 > is using a forwarder (see <TT
2499 for a particular destination URL, the <TT
2505 that is examined is the address of the forwarder and <SPAN
2512 of the ultimate target. This is necessary because it may be impossible for the local
2516 > to determine the IP address of the
2517 ultimate target (that's often what gateways are used for).
2520 > You should prefer using IP addresses over DNS names, because the address lookups take
2521 time. All DNS names must resolve! You can <SPAN
2527 > use domain patterns
2531 > or partial domain names. If a DNS name resolves to multiple
2532 IP addresses, only the first one is used.
2535 > Some systems allows IPv4 client to connect to IPv6 server socket.
2536 Then the client's IPv4 address will be translated by system into
2537 IPv6 address space with special prefix ::ffff:0:0/96 (so called IPv4
2538 mapped IPv6 address). <SPAN
2542 and maps such ACL addresses automatically.
2545 > Denying access to particular sites by ACL may have undesired side effects
2546 if the site in question is hosted on a machine which also hosts other sites
2554 > Explicitly define the default behavior if no ACL and
2562 is OK. The absence of a <TT
2574 > destination addresses are OK:
2585 > permit-access localhost</PRE
2592 > Allow any host on the same class C subnet as www.privoxy.org access to
2593 nothing but www.example.com (or other domains hosted on the same system):
2604 > permit-access www.privoxy.org/24 www.example.com/32</PRE
2611 > Allow access from any host on the 26-bit subnet 192.168.45.64 to anywhere,
2612 with the exception that 192.168.45.73 may not access the IP address behind
2613 www.dirty-stuff.example.com:
2624 > permit-access 192.168.45.64/26
2625 deny-access 192.168.45.73 www.dirty-stuff.example.com</PRE
2632 > Allow access from the IPv4 network 192.0.2.0/24 even if listening on
2633 an IPv6 wild card address (not supported on all platforms):
2643 CLASS="PROGRAMLISTING"
2644 > permit-access 192.0.2.0/24</PRE
2651 > This is equivalent to the following line even if listening on an
2652 IPv4 address (not supported on all platforms):
2662 CLASS="PROGRAMLISTING"
2663 > permit-access [::ffff:192.0.2.0]/120</PRE
2679 >7.4.8. buffer-limit</A
2684 CLASS="VARIABLELIST"
2690 > Maximum size of the buffer for content filtering.
2706 >Effect if unset:</DT
2709 > Use a 4MB (4096 KB) limit.
2716 > For content filtering, i.e. the <TT
2723 > actions, it is necessary that
2727 > buffers the entire document body.
2728 This can be potentially dangerous, since a server could just keep sending
2729 data indefinitely and wait for your RAM to exhaust -- with nasty consequences.
2733 > When a document buffer size reaches the <TT
2737 flushed to the client unfiltered and no further attempt to
2738 filter the rest of the document is made. Remember that there may be multiple threads
2739 running, which might require up to <TT
2749 >, unless you have enabled <SPAN
2751 >"single-threaded"</SPAN
2769 > This feature allows routing of HTTP requests through a chain of
2770 multiple proxies.</P
2772 > Forwarding can be used to chain Privoxy with a caching proxy to speed
2773 up browsing. Using a parent proxy may also be necessary if the machine
2777 > runs on has no direct Internet access.</P
2779 > Note that parent proxies can severely decrease your privacy level.
2780 For example a parent proxy could add your IP address to the request
2781 headers and if it's a caching proxy it may add the <SPAN
2785 header to revalidation requests again, even though you configured Privoxy
2786 to remove it. It may also ignore Privoxy's header time randomization and use the
2787 original values which could be used by the server as cookie replacement
2788 to track your steps between visits.</P
2790 > Also specified here are SOCKS proxies. <SPAN
2794 supports the SOCKS 4 and SOCKS 4A protocols.</P
2806 CLASS="VARIABLELIST"
2812 > To which parent HTTP proxy specific requests should be routed.
2844 HREF="actions-file.html#AF-PATTERNS"
2847 that specifies to which requests (i.e. URLs) this forward rule shall apply. Use <TT
2866 is the DNS name or IP address of the parent HTTP proxy through which the requests should be forwarded,
2867 optionally followed by its listening port (default: 8000).
2868 Use a single dot (<TT
2873 >"no forwarding"</SPAN
2890 >Effect if unset:</DT
2893 > Don't use parent HTTP proxies.
2908 >, then requests are not
2909 forwarded to another HTTP proxy but are made directly to the web servers.
2918 numerical IPv6 address (if
2920 HREF="http://tools.ietf.org/html/rfc3493"
2924 implemented). To prevent clashes with the port delimiter, the whole IP
2925 address has to be put into brackets. On the other hand a <TT
2930 > containing an IPv6 address
2931 has to be put into angle brackets (normal brackets are reserved for
2932 regular expressions already).
2935 > Multiple lines are OK, they are checked in sequence, and the last match wins.
2942 > Everything goes to an example parent proxy, except SSL on port 443 (which it doesn't handle):
2953 > forward / parent-proxy.example.org:8080
2961 > Everything goes to our example ISP's caching proxy, except for requests
2962 to that ISP's sites:
2973 > forward / caching-proxy.isp.example.net:8000
2974 forward .isp.example.net .</PRE
2981 > Parent proxy specified by an IPv6 address:
2991 CLASS="PROGRAMLISTING"
2992 > foward / [2001:DB8::1]:8000</PRE
2999 > Suppose your parent proxy doesn't support IPv6:
3009 CLASS="PROGRAMLISTING"
3010 > forward / parent-proxy.example.org:8000
3011 forward ipv6-server.example.org .
3012 forward <[2-3][0-9a-f][0-9a-f][0-9a-f]:*> .</PRE
3028 >7.5.2. forward-socks4, forward-socks4a and forward-socks5</A
3031 NAME="FORWARD-SOCKS4"
3034 NAME="FORWARD-SOCKS4A"
3039 CLASS="VARIABLELIST"
3045 > Through which SOCKS proxy (and optionally to which parent HTTP proxy) specific requests should be routed.
3089 HREF="actions-file.html#AF-PATTERNS"
3091 > that specifies to which
3092 requests (i.e. URLs) this forward rule shall apply. Use <TT
3111 are IP addresses in dotted decimal notation or valid DNS names
3123 >"no HTTP forwarding"</SPAN
3124 >), and the optional
3130 > parameters are TCP ports,
3131 i.e. integer values from 1 to 65535
3147 >Effect if unset:</DT
3150 > Don't use SOCKS proxies.
3157 > Multiple lines are OK, they are checked in sequence, and the last match wins.
3160 > The difference between <TT
3165 >forward-socks4a</TT
3167 is that in the SOCKS 4A protocol, the DNS resolution of the target hostname happens on the SOCKS
3168 server, while in SOCKS 4 it happens locally.
3174 > the DNS resolution will happen on the remote server as well.
3189 numerical IPv6 address (if
3191 HREF="http://tools.ietf.org/html/rfc3493"
3195 implemented). To prevent clashes with the port delimiter, the whole IP
3196 address has to be put into brackets. On the other hand a <TT
3201 > containing an IPv6 address
3202 has to be put into angle brackets (normal brackets are reserved for
3203 regular expressions already).
3214 >, then requests are not
3215 forwarded to another HTTP proxy but are made (HTTP-wise) directly to the web servers, albeit through
3223 > From the company example.com, direct connections are made to all
3227 > domains, but everything outbound goes through
3228 their ISP's proxy by way of example.com's corporate SOCKS 4A gateway to
3240 > forward-socks4a / socks-gw.example.com:1080 www-cache.isp.example.net:8080
3241 forward .example.com .</PRE
3248 > A rule that uses a SOCKS 4 gateway for all destinations but no HTTP parent looks like this:
3259 > forward-socks4 / socks-gw.example.com:1080 .</PRE
3266 > To chain Privoxy and Tor, both running on the same system, you would use
3278 > forward-socks5 / 127.0.0.1:9050 .</PRE
3288 > network can't be used to
3289 reach your local network, if you need to access local servers you
3290 therefore might want to make some exceptions:
3301 > forward 192.168.*.*/ .
3303 forward 127.*.*.*/ .</PRE
3310 > Unencrypted connections to systems in these address ranges will
3311 be as (un)secure as the local network is, but the alternative is that you
3312 can't reach the local network through <SPAN
3316 at all. Of course this may actually be desired and there is no reason
3317 to make these exceptions if you aren't sure you need them.
3320 > If you also want to be able to reach servers in your local network by
3321 using their names, you will need additional exceptions that look like
3333 > forward localhost/ .</PRE
3348 NAME="ADVANCED-FORWARDING-EXAMPLES"
3349 >7.5.3. Advanced Forwarding Examples</A
3352 > If you have links to multiple ISPs that provide various special content
3353 only to their subscribers, you can configure multiple <SPAN
3357 which have connections to the respective ISPs to act as forwarders to each other, so that
3364 > users can see the internal content of all ISPs.</P
3366 > Assume that host-a has a PPP connection to isp-a.example.net. And host-b has a PPP connection to
3367 isp-b.example.org. Both run <SPAN
3371 configuration can look like this:</P
3384 forward .isp-b.example.net host-b:8118</PRE
3401 forward .isp-a.example.org host-a:8118</PRE
3407 > Now, your users can set their browser's proxy to use either
3408 host-a or host-b and be able to browse the internal content
3409 of both isp-a and isp-b.</P
3411 > If you intend to chain <SPAN
3418 > locally, then chaining as
3421 >browser -> squid -> privoxy</TT
3422 > is the recommended way. </P
3424 > Assuming that <SPAN
3431 run on the same box, your <SPAN
3434 > configuration could then look like this:</P
3444 > # Define Privoxy as parent proxy (without ICP)
3445 cache_peer 127.0.0.1 parent 8118 7 no-query
3447 # Define ACL for protocol FTP
3450 # Do not forward FTP requests to Privoxy
3451 always_direct allow ftp
3453 # Forward all the rest to Privoxy
3454 never_direct allow all</PRE
3460 > You would then need to change your browser's proxy settings to <SPAN
3463 >'s address and port.
3464 Squid normally uses port 3128. If unsure consult <TT
3472 > You could just as well decide to only forward requests you suspect
3473 of leading to Windows executables through a virus-scanning parent proxy,
3476 >antivir.example.com</TT
3488 forward /.*\.(exe|com|dll|zip)$ antivir.example.com:8010</PRE
3499 NAME="FORWARDED-CONNECT-RETRIES"
3500 >7.5.4. forwarded-connect-retries</A
3505 CLASS="VARIABLELIST"
3511 > How often Privoxy retries if a forwarded connection request fails.
3521 >Number of retries.</I
3539 >Effect if unset:</DT
3542 > Connections forwarded through other proxies are treated like direct connections and no retry attempts are made.
3552 >forwarded-connect-retries</I
3554 > is mainly interesting
3555 for socks4a connections, where <SPAN
3558 > can't detect why the connections failed.
3559 The connection might have failed because of a DNS timeout in which case a retry makes sense,
3560 but it might also have failed because the server doesn't exist or isn't reachable. In this
3561 case the retry will just delay the appearance of Privoxy's error message.
3564 > Note that in the context of this option, <SPAN
3566 >"forwarded connections"</SPAN
3567 > includes all connections
3568 that Privoxy forwards through other proxies. This option is not limited to the HTTP CONNECT method.
3571 > Only use this option, if you are getting lots of forwarding-related error messages
3572 that go away when you try again manually. Start with a small value and check Privoxy's
3573 logfile from time to time, to see how many retries are usually needed.
3580 > forwarded-connect-retries 1
3593 >7.6. Miscellaneous</A
3600 NAME="ACCEPT-INTERCEPTED-REQUESTS"
3601 >7.6.1. accept-intercepted-requests</A
3606 CLASS="VARIABLELIST"
3612 > Whether intercepted requests should be treated as valid.
3640 >Effect if unset:</DT
3643 > Only proxy requests are accepted, intercepted requests are treated as invalid.
3650 > If you don't trust your clients and want to force them
3655 option and configure your packet filter to redirect outgoing
3656 HTTP connections into <SPAN
3662 > Make sure that <SPAN
3666 aren't redirected as well. Additionally take care that
3670 > can't intentionally connect
3671 to itself, otherwise you could run into redirection loops if
3675 > listening port is reachable
3676 by the outside or an attacker has access to the pages you visit.
3683 > accept-intercepted-requests 1
3694 NAME="ALLOW-CGI-REQUEST-CRUNCHING"
3695 >7.6.2. allow-cgi-request-crunching</A
3700 CLASS="VARIABLELIST"
3706 > Whether requests to <SPAN
3709 > CGI pages can be blocked or redirected.
3737 >Effect if unset:</DT
3743 > ignores block and redirect actions for its CGI pages.
3753 > ignores block or redirect actions
3754 for its CGI pages. Intercepting these requests can be useful in multi-user
3755 setups to implement fine-grained access control, but it can also render the complete
3756 web interface useless and make debugging problems painful if done without care.
3759 > Don't enable this option unless you're sure that you really need it.
3766 > allow-cgi-request-crunching 1
3777 NAME="SPLIT-LARGE-FORMS"
3778 >7.6.3. split-large-forms</A
3783 CLASS="VARIABLELIST"
3789 > Whether the CGI interface should stay compatible with broken HTTP clients.
3817 >Effect if unset:</DT
3820 > The CGI form generate long GET URLs.
3830 > CGI forms can lead to
3831 rather long URLs. This isn't a problem as far as the HTTP
3832 standard is concerned, but it can confuse clients with arbitrary
3833 URL length limitations.
3836 > Enabling split-large-forms causes <SPAN
3840 to divide big forms into smaller ones to keep the URL length down.
3841 It makes editing a lot less convenient and you can no longer
3842 submit all changes at once, but at least it works around this
3846 > If you don't notice any editing problems, there is no reason
3847 to enable this option, but if one of the submit buttons appears
3848 to be broken, you should give it a try.
3855 > split-large-forms 1
3866 NAME="KEEP-ALIVE-TIMEOUT"
3867 >7.6.4. keep-alive-timeout</A
3872 CLASS="VARIABLELIST"
3878 > Number of seconds after which an open connection will no longer be reused.
3888 >Time in seconds.</I
3900 >Effect if unset:</DT
3903 > Connections are not kept alive.
3910 > This option allows clients to keep the connection to <SPAN
3914 alive. If the server supports it, <SPAN
3918 the connection to the server alive as well. Under certain
3919 circumstances this may result in speed-ups.
3925 > will close the connection to the server if
3926 the client connection gets closed, or if the specified timeout
3927 has been reached without a new request coming in. This behaviour
3928 can be changed with the <A
3929 HREF="#CONNECTION-SHARING"
3931 >connection-sharing</A
3935 > This option has no effect if <SPAN
3939 has been compiled without keep-alive support.
3946 > keep-alive-timeout 300
3957 NAME="CONNECTION-SHARING"
3958 >7.6.5. connection-sharing</A
3963 CLASS="VARIABLELIST"
3969 > Whether or not outgoing connections that have been kept alive
3970 should be shared between different incoming connections.
3992 >Effect if unset:</DT
3995 > Connections are not shared.
4002 > This option has no effect if <SPAN
4006 has been compiled without keep-alive support, or if it's disabled.
4013 > Note that reusing connections doesn't necessary cause speedups.
4014 There are also a few privacy implications you should be aware of.
4017 > If this option is effective, outgoing connections are shared between
4018 clients (if there are more than one) and closing the browser that initiated
4019 the outgoing connection does no longer affect the connection between <SPAN
4023 and the server unless the client's request hasn't been completed yet.
4026 > If the outgoing connection is idle, it will not be closed until either
4030 > or the server's timeout is reached.
4031 While it's open, the server knows that the system running <SPAN
4038 > If there are more than one client (maybe even belonging to multiple users),
4039 they will be able to reuse each others connections. This is potentially
4040 dangerous in case of authentication schemes like NTLM where only the
4041 connection is authenticated, instead of requiring authentication for
4045 > If there is only a single client, and if said client can keep connections
4046 alive on its own, enabling this option has next to no effect. If the client
4047 doesn't support connection keep-alive, enabling this option may make sense
4051 > to keep outgoing connections alive even if the client
4052 itself doesn't support it.
4055 > You should also be aware that enabling this option increases the likelihood
4056 of getting the "No server or forwarder data" error message, especially if you
4057 are using a slow connection to the Internet.
4060 > This option should only be used by experienced users who
4061 understand the risks and can weight them against the benefits.
4068 > connection-sharing 1
4079 NAME="SOCKET-TIMEOUT"
4080 >7.6.6. socket-timeout</A
4085 CLASS="VARIABLELIST"
4091 > Number of seconds after which a socket times out if
4092 no data is received.
4102 >Time in seconds.</I
4114 >Effect if unset:</DT
4117 > A default value of 300 seconds is used.
4124 > For SOCKS requests the timeout currently doesn't start until
4125 the SOCKS server accepted the request. This will be fixed in
4133 > socket-timeout 300
4144 NAME="MAX-CLIENT-CONNECTIONS"
4145 >7.6.7. max-client-connections</A
4150 CLASS="VARIABLELIST"
4156 > Maximum number of client connections that will be served.
4166 >Positive number.</I
4178 >Effect if unset:</DT
4181 > Connections are served until a resource limit is reached.
4191 > creates one thread (or process) for every incoming client
4192 connection that isn't rejected based on the access control settings.
4195 > If the system is powerful enough, <SPAN
4198 > can theoretically deal with
4199 several hundred (or thousand) connections at the same time, but some
4200 operating systems enforce resource limits by shutting down offending
4201 processes and their default limits may be below the ones <SPAN
4205 require under heavy load.
4211 > to enforce a connection limit below the thread
4212 or process limit used by the operating system makes sure this doesn't
4213 happen. Simply increasing the operating system's limit would work too,
4217 > isn't the only application running on the system,
4218 you may actually want to limit the resources used by <SPAN
4227 > is only used by a single trusted user, limiting the
4228 number of client connections is probably unnecessary. If there
4229 are multiple possibly untrusted users you probably still want to
4230 additionally use a packet filter to limit the maximal number of
4231 incoming connections per client. Otherwise a malicious user could
4232 intentionally create a high number of connections to prevent other
4233 users from using <SPAN
4239 > Obviously using this option only makes sense if you choose a limit
4240 below the one enforced by the operating system.
4247 > max-client-connections 256
4260 >7.7. Windows GUI Options</A
4266 > has a number of options specific to the
4267 Windows GUI interface:</P
4269 NAME="ACTIVITY-ANIMATION"
4274 >"activity-animation"</SPAN
4279 > icon will animate when
4283 > is active. To turn off, set to 0.</P
4288 CLASS="LITERALLAYOUT"
4293 >activity-animation 1</I
4296 </P
4306 >"log-messages"</SPAN
4311 > will log messages to the console
4317 CLASS="LITERALLAYOUT"
4325 </P
4330 NAME="LOG-BUFFER-SIZE"
4336 >"log-buffer-size"</SPAN
4337 > is set to 1, the size of the log buffer,
4338 i.e. the amount of memory used for the log messages displayed in the
4339 console window, will be limited to <SPAN
4341 >"log-max-lines"</SPAN
4344 > Warning: Setting this to 0 will result in the buffer to grow infinitely and
4345 eat up all your memory!</P
4350 CLASS="LITERALLAYOUT"
4355 >log-buffer-size 1</I
4358 </P
4363 NAME="LOG-MAX-LINES"
4368 >log-max-lines</SPAN
4369 > is the maximum number of lines held
4370 in the log buffer. See above.</P
4375 CLASS="LITERALLAYOUT"
4380 >log-max-lines 200</I
4383 </P
4388 NAME="LOG-HIGHLIGHT-MESSAGES"
4393 >"log-highlight-messages"</SPAN
4398 > will highlight portions of the log
4399 messages with a bold-faced font:</P
4404 CLASS="LITERALLAYOUT"
4409 >log-highlight-messages 1</I
4412 </P
4417 NAME="LOG-FONT-NAME"
4420 > The font used in the console window:</P
4425 CLASS="LITERALLAYOUT"
4430 >log-font-name Comic Sans MS</I
4433 </P
4438 NAME="LOG-FONT-SIZE"
4441 > Font size used in the console window:</P
4446 CLASS="LITERALLAYOUT"
4454 </P
4459 NAME="SHOW-ON-TASK-BAR"
4465 >"show-on-task-bar"</SPAN
4466 > controls whether or not
4470 > will appear as a button on the Task bar
4476 CLASS="LITERALLAYOUT"
4481 >show-on-task-bar 0</I
4484 </P
4489 NAME="CLOSE-BUTTON-MINIMIZES"
4494 >"close-button-minimizes"</SPAN
4495 > is set to 1, the Windows close
4496 button will minimize <SPAN
4499 > instead of closing
4500 the program (close with the exit option on the File menu).</P
4505 CLASS="LITERALLAYOUT"
4510 >close-button-minimizes 1</I
4513 </P
4523 >"hide-console"</SPAN
4524 > option is specific to the MS-Win console
4528 >. If this option is used,
4532 > will disconnect from and hide the
4538 CLASS="LITERALLAYOUT"
4546 </P
4557 SUMMARY="Footer navigation table"
4568 HREF="configuration.html"
4586 HREF="actions-file.html"
4596 >Privoxy Configuration</TD