formatting

<?xml version="1.0"?>

<!DOCTYPE modulesynopsis SYSTEM "../style/modulesynopsis.dtd">

<?xml-stylesheet type="text/xsl" href="../style/manual.en.xsl"?>

<!-- $Revision: 1.4 $ -->

<!-- $Revision: 1.5 $ -->

<!--

 Copyright 2004 The Apache Software Foundation

<status>Extension</status>

<sourcefile>mod_filter.c</sourcefile>

<identifier>filter_module</identifier>

<compatibility>Apache 2.0 and higher</compatibility>

<compatibility>Version 2.1 and higher</compatibility>

<summary>

    <p>This module enables smart, context-sensitive configuration of

    output content filters.  For example, apache can be configured to

    process different content-types through different filters, even

when the content-type is not known in advance (e.g. in a proxy).

</p>

    when the content-type is not known in advance (e.g. in a proxy).</p>

    <p>mod_filter works by introducing indirection into the filter

    chain.  Instead of inserting filters in the chain, we insert

    a filter harness which in turn dispatches conditionally

    to a filter provider.  Any content filter may be used as a provider

    to mod_filter; no change to existing filter modules is required

(although it may be possible to simplify them).

</p>

    (although it may be possible to simplify them).</p>

</summary>

<section id="smart"><title>Smart Filtering</title>

    Each filter then needs to determine whether to run, and there is little

    flexibility available for server admins to allow the chain to be

    configured dynamically.</p>

    <p>mod_filter by contrast gives server administrators a great deal of

    flexibility in configuring the filter chain.  In fact, filters can be

    inserted based on any Request Header, Response Header or Environment

    content generator.  The ability to dispatch based on Environment

    Variables offers the full flexibility of configuration with

    <module>mod_rewrite</module> to anyone who needs it.</p>

</section>

<section id="terms"><title>Filter Declarations, Providers and Chains</title>

    <p class="figure">

    <img src="../images/mod_filter_old.gif" width="160" height="310"

    provided the filter chain can be correctly configured, but presents

    problems when the filters need to be configured dynamically based on

    the outcome of the handler.</p>

    <p class="figure">

    <img src="../images/mod_filter_new.gif" width="423" height="331"

    alt="[This image shows the mod_filter model]"/><br />

    <dfn>Figure 2:</dfn> The <module>mod_filter</module> model</p>

    <p>mod_filter works by introducing indirection into the filter

    chain.  Instead of inserting filters in the chain, we insert

    a filter harness which in turn dispatches conditionally

    to mod_filter; no change to existing filter modules is required

    (although it may be possible to simplify them).  There can be

    multiple providers for one filter, but no more than one provider will

run for any single request.

</p>

    run for any single request.</p>

    <p>A filter chain comprises any number of instances of the filter

    harness, each of which may have any number of providers.  A special

    case is that of a single provider with unconditional dispatch: this

is equivalent to inserting the provider filter directly into the chain.

</p>

    is equivalent to inserting the provider filter directly into the chain.</p>

</section>

<section id="config"><title>Configuring the Chain</title>

    <p>There are three stages to configuring a filter chain with mod_filter.

    For details of the directives, see below.</p>

    <dl>

    <dt>Declare Filters</dt>

    <dd>The <directive>FilterDeclare</directive> directive declares a filter,

    assigning it a name and a dispatch criterion.</dd>

    <dt>Register Providers</dt>

<dd>The <directive>FilterProvider</directive> directive registers a provider with a filter.

The filter must have been registered with <directive>FilterDeclare</directive>.

The provider must have been registered with

<code>ap_register_output_filter</code> by some module.  The final argument

to <directive>FilterProvider</directive> is a match string, that will be checked against

the filter's dispatch criterion to determine whether to run this provider.</dd>

    <dd>The <directive>FilterProvider</directive> directive registers a

    provider with a filter. The filter must have been registered with

    <directive>FilterDeclare</directive>. The provider must have been

    registered with <code>ap_register_output_filter</code> by some module.

    The final argument to <directive>FilterProvider</directive> is a match

    string, that will be checked against the filter's dispatch criterion

    to determine whether to run this provider.</dd>

    <dt>Configure the Chain</dt>

    <dd>The above directives build components of a smart filter chain,

but do not configure it to run.  The <directive>FilterChain</directive> directive

builds a filter chain from smart filters declared, offering the

    but do not configure it to run.  The <directive>FilterChain</directive>

    directive builds a filter chain from smart filters declared, offering the

    flexibility to insert filters at the beginning or end of the chain,

    remove a filter, or clear the chain.</dd>

</dl>

</section>

<section id="examples"><title>Examples</title>

    <dl>

    <dt>Server side Includes (SSI)</dt>

<dd>

<p>A simple case of using mod_filter in place of

<code>AddOutputFilterByType</code></p>

    <dd>A simple case of using mod_filter in place of

    <directive module="core">AddOutputFilterByType</directive>

    <example>

      FilterDeclare SSI Content-Type<br/>

      FilterProvider SSI INCLUDES $text/html<br/>

FilterChain SSI<br/>

      FilterChain SSI

    </example>

    </dd>

    <dt>Server side Includes (SSI)</dt>

<dd>

<p>The same as the above but dispatching on handler (classic

SSI behaviour; .shtml files get processed).</p>

    <dd>The same as the above but dispatching on handler (classic

    SSI behaviour; .shtml files get processed).

    <example>

      FilterDeclare SSI Handler<br/>

      FilterProvider SSI INCLUDES server-parsed<br/>

FilterChain SSI<br/>

      FilterChain SSI

    </example>

    </dd>

    <dt>Emulating mod_gzip with mod_deflate</dt>

<dd>

<p>Insert INFLATE filter only if "gzip" is NOT in the

Accept-Encoding header.</p>

    <dd>Insert INFLATE filter only if "gzip" is NOT in the

    Accept-Encoding header.

    <example>

      FilterDeclare gzip req=Accept-Encoding<br/>

      FilterProvider gzip inflate !$gzip<br/>

FilterChain gzip<br/>

      FilterChain gzip

    </example>

    </dd>

    <dt>Image Downsampling</dt>

<dd>

<p>Suppose we want to downsample all web images, and have filters

for GIF, JPEG and PNG.</p>

    <dd>Suppose we want to downsample all web images, and have filters

    for GIF, JPEG and PNG.

    <example>

      FilterDeclare unpack Content-Type<br/>

      FilterProvider unpack jpeg_unpack $image/jpeg<br/>

      FilterProvider unpack gif_unpack $image/gif<br/>

      FilterProvider unpack png_unpack $image/png<br/>

      <br />

      FilterDeclare downsample Content-Type<br/>

      FilterProvider downsample downsample_filter $image<br/>

      FilterProtocol downsample "change=yes"<br/>

      <br />

      FilterDeclare repack Content-Type<br/>

      FilterProvider repack jpeg_pack $image/jpeg<br/>

      FilterProvider repack gif_pack $image/gif<br/>

      FilterProvider repack png_pack $image/png<br/>

      &lt;Location /image-filter&gt;<br/>

      <indent>

        FilterChain unpack downsample repack<br/>

      </indent>

      </Location>

    </example>

    </dd>

    </dl>

</section>

<section id="protocol"><title>Protocol Handling</title>

    <p>Historically, each filter is responsible for ensuring that whatever

    changes it makes are correctly represented in the HTTP response headers,

    and that it does not run when it would make an illegal change.  This

    imposes a burden on filter authors to re-implement some common

    functionality in every filter:</p>

    <ul>

    <li>Many filters will change the content, invalidating existing content

    tags, checksums, hashes, and lengths.</li>

    <li>Filters that require an entire, unbroken response in input need to

    ensure they don't get byteranges from a backend.</li>

    <li>Filters that transform output in a filter need to ensure they don't

    violate a <code>Cache-Control: no-transform</code> header from the

    backend.</li>

    <li>Filters may make responses uncacheable.</li>

    </ul>

    <p>mod_filter aims to offer generic handling of these details of filter

    implementation, reducing the complexity required of content filter modules.

    This is work-in-progress; the <directive>FilterProtocol</directive>

    implements some of this functionality, but there are no API calls yet.</p>

    <p>At the same time, mod_filter should not interfere with a filter that

    wants to handle all aspects of the protocol.  By default (i.e. in the

    absence of any <directive>FilterProtocol</directive> directives), mod_filter

    for use in <directive>FilterProvider</directive>,

    <directive>FilterChain</directive> and

    <directive>FilterProtocol</directive> directives.</p>

    <p>The second is a string with optional <code>req=</code>,

    <code>resp=</code> or <code>env=</code> prefix causing it

    to dispatch on (respectively) the request header, response

    header, or environment variable named.  In the absence of a

    prefix, it defaults to a response header.  A special case is the

    word "handler", which causes mod_filter to dispatch on the handler.</p>

    <p>The final (optional) argument

    is the type of filter, and takes values of <var>ap_filter_type</var>

    - namely <var>RESOURCE</var> (the default), <var>CONTENT_SET</var>,

    <var>PROTOCOL</var>, <var>TRANSCODE</var>, <var>CONNECTION</var>

or <var>NETWORK</var>.

</p>

    or <var>NETWORK</var>.</p>

</usage>

</directivesynopsis>

<directivesynopsis>

<name>FilterProvider</name>

<description>Register a content filter</description>

<syntax>FilterProvider <var>filter-name</var> <var>provider-name</var> <var>match</var></syntax>

<syntax>FilterProvider <var>filter-name</var> <var>provider-name</var>

    <var>match</var></syntax>

<contextlist><context>server config</context><context>virtual host</context>

<context>directory</context><context>.htaccess</context></contextlist>

    here matches the value of the header or environment variable declared

    as <var>dispatch</var> in the <directive>FilterDeclare</directive>

    directive that declared <var>filter-name</var>.</p>

    <p><var>filter-name</var> must have been declared with 

    <directive>FilterDeclare</directive>.  <var>provider-name</var> must have

    been registered by loading a module that registers the name with

    <code>ap_register_output_filter</code>.</p>

    <p>The <var>match</var> argument specifies a match that will be applied to

    the filter's <var>dispatch</var> criterion.  The match may be a string

    match (exact match or substring), a regexp, an integer (greater, lessthan

    or equals), or unconditional.  The first characters of the <var>match</var>

    argument determines this:</p>

    <p><strong>First</strong>, if the first character is an exclamation mark

    <strong>!</strong>, this reverses the rule, so the provider will be used

    if and only if the match <em>fails</em>.</p>

    <p><strong>Second</strong>, it interprets the first character excluding

    any leading ! as follows:</p>

    <dl>

    <dt>default</dt>

    <dd>exact match</dd>

    <dt>$</dt>

    <dd>substring match</dd>

    <dt>/</dt>

    <dd>regexp match</dd>

    <dt>=</dt>

    <dd>integer equality</dd>

    <dt>&lt;</dt>

    <dd>integer less-than</dd>

    <dt>&gt;</dt>

    <dd>integer greater-than</dd>

    <dt>*</dt>

    <dd>Unconditional match</dd>

    </dl>

    <directive>FilterChain</directive> takes any number of arguments,

    each optionally preceded with a single-character control that

    determines what to do:</p>

    <dl>

    <dt>+filter-name</dt>

    <dd>Add filter-name to the end of the filter chain</dd>

    <dt>@filter-name</dt>

    <dd>Insert filter-name at the start of the filter chain</dd>

    <dt>-filter-name</dt>

    <dd>Remove filter-name from the filter chain</dd>

    <dt>=filter-name</dt>

    <dd>Empty the filter chain and insert filter-name</dd>

    <dt>!</dt>

    <dd>Empty the filter chain</dd>

    <dt>filter-name</dt>

    <dd>Equivalent to +filter-name</dd>

    </dl>

</usage>

</directivesynopsis>

<directivesynopsis>

<name>FilterProtocol</name>

<description>Deal with correct HTTP protocol handling</description>

<syntax>FilterProtocol filter-name [provider-name] "proto-flags"</syntax>

<contextlist><context>server config</context><context>virtual host</context>

<context>directory</context><context>.htaccess</context></contextlist>

<usage>

    <p>This directs mod_filter to deal with ensuring the filter doesn't run

    when it shouldn't, and that the HTTP response headers are correctly set

    taking into account the effects of the filter.</p>

    <p>There are two forms of this directive.  With three arguments, it

    applies specifically to a filter-name and a provider for that filter.

    With two arguments it applies to a filter-name whenever the filter runs

    <em>any</em> provider.</p>

<p>proto-flags is one or more of</p>

    <p><var>proto-flags</var> is one or more of</p>

    <dl>

    <dt>change=yes</dt>

<dd>The filter changes the content, including possibly the content length</dd>

    <dd>The filter changes the content, including possibly the content

    length</dd>

    <dt>change=1:1</dt>

<dd>The filter changes the content, but will not change the content length</dd>

    <dd>The filter changes the content, but will not change the content

    length</dd>

    <dt>byteranges=no</dt>

    <dd>The filter cannot work on byteranges and requires complete input</dd>

    <dt>proxy=no</dt>

    <dd>The filter should not run in a proxy context</dd>

    <dt>proxy=transform</dt>

    <dd>The filter transforms the response in a manner incompatible with

    the HTTP <code>Cache-Control: no-transform</code> header.</dd>

    <dt>cache=no</dt>

    <dd>The filter renders the output uncacheable (eg by introducing randomised

    content changes)</dd>

<syntax>FilterTrace filter-name level</syntax>

<contextlist><context>server config</context><context>virtual host</context>

<context>directory</context><context>.htaccess</context></contextlist>

<usage>

    <p>This directive generates debug information from mod_filter.

    It is designed to help test and debug providers (filter modules), although

    it may also help with mod_filter itself.</p>

    <p>The debug output depends on the level set:</p>

    <dl>

    <dt>0 (default)</dt>

    <dd>No debug information is generated.</dd>

    <dt>1</dt>

    <dd>mod_filter will record buckets and brigades passing through the filter

    to the error log, before the provider has processed them.

    This is similar to the information generated by

    <a href="http://apache.webthing.com/mod_diagnostics/">mod_diagnostics</a>.

    </dd>

    <dt>2 (not yet implemented)</dt>

<dd>Will dump the full data passing through to a tempfile before the provider.

<strong>For single-user debug only</strong>; this will not

    <dd>Will dump the full data passing through to a tempfile before the

    provider. <strong>For single-user debug only</strong>; this will not

    support concurrent hits.</dd>

    </dl>

</usage>