path: root/docs/manual/mod/mod_include.xml

<?xml version="1.0"?>
<!DOCTYPE modulesynopsis SYSTEM "../style/modulesynopsis.dtd">
<?xml-stylesheet type="text/xsl" href="../style/manual.en.xsl"?>
<!-- $LastChangedRevision$ -->

<!--
 Licensed to the Apache Software Foundation (ASF) under one or more
 contributor license agreements.  See the NOTICE file distributed with
 this work for additional information regarding copyright ownership.
 The ASF licenses this file to You under the Apache License, Version 2.0
 (the "License"); you may not use this file except in compliance with
 the License.  You may obtain a copy of the License at

     http://www.apache.org/licenses/LICENSE-2.0

 Unless required by applicable law or agreed to in writing, software
 distributed under the License is distributed on an "AS IS" BASIS,
 WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 See the License for the specific language governing permissions and
 limitations under the License.
-->

<modulesynopsis metafile="mod_include.xml.meta">

<name>mod_include</name>
<description>Server-parsed html documents (Server Side Includes)</description>
<status>Base</status>
<sourcefile>mod_include.c</sourcefile>
<identifier>include_module</identifier>

<summary>
    <p>This module provides a filter which will process files
    before they are sent to the client. The processing is
    controlled by specially formatted SGML comments, referred to as
    <dfn>elements</dfn>. These elements allow conditional text, the
    inclusion of other files or programs, as well as the setting and
    printing of environment variables.</p>
</summary>
<seealso><directive module="core">Options</directive></seealso>
<seealso><directive module="core">AcceptPathInfo</directive></seealso>
<seealso><a href="../filter.html">Filters</a></seealso>
<seealso><a href="../howto/ssi.html">SSI Tutorial</a></seealso>

<section id="enabling">
    <title>Enabling Server-Side Includes</title>

    <p>Server Side Includes are implemented by the
    <code>INCLUDES</code> <a href="../filter.html">filter</a>. If
    documents containing server-side include directives are given
    the extension .shtml, the following directives will make Apache
    parse them and assign the resulting document the mime type of
    <code>text/html</code>:</p>

    <highlight language="config">
AddType text/html .shtml
AddOutputFilter INCLUDES .shtml
    </highlight>

    <p>The following directive must be given for the directories
    containing the shtml files (typically in a
    <directive module="core" type="section">Directory</directive> section,
    but this directive is also valid in <code>.htaccess</code> files if
    <directive module="core">AllowOverride</directive> <code>Options</code>
    is set):</p>

    <highlight language="config">
Options +Includes
    </highlight>

    <p>For backwards compatibility, the <code>server-parsed</code>
    <a href="../handler.html">handler</a> also activates the
    INCLUDES filter. As well, Apache will activate the INCLUDES
    filter for any document with mime type
    <code>text/x-server-parsed-html</code> or
    <code>text/x-server-parsed-html3</code> (and the resulting
    output will have the mime type <code>text/html</code>).</p>

    <p>For more information, see our <a
    href="../howto/ssi.html">Tutorial on Server Side Includes</a>.</p>
</section> <!-- /enabling -->

<section id="pathinfo">
    <title>PATH_INFO with Server Side Includes</title>

    <p>Files processed for server-side includes no longer accept
    requests with <code>PATH_INFO</code> (trailing pathname information)
    by default.  You can use the <directive
    module="core">AcceptPathInfo</directive> directive to
    configure the server to accept requests with <code>PATH_INFO</code>.</p>
</section> <!-- /pathinfo -->

<section id="elements"><title>Available Elements</title>
    <p>The document is parsed as an HTML document, with special
    commands embedded as SGML comments. A command has the syntax: </p>

    <example>
      &lt;!--#<var>element</var> <var>attribute</var>=<var>value</var>
      <var>attribute</var>=<var>value</var> ... --&gt;
    </example>

    <p>The value will often be enclosed in double quotes, but single
    quotes (<code>'</code>) and backticks (<code>`</code>) are also
    possible. Many commands only allow a single attribute-value pair.
    Note that the comment terminator (<code>--&gt;</code>) should be
    preceded by whitespace to ensure that it isn't considered part of
    an SSI token. Note that the leading <code>&lt;!--#</code> is <em>one</em>
    token and may not contain any whitespaces.</p>

    <p>The allowed elements are listed in the following table:</p>

    <table border="1">
    <tr><th>Element</th><th>Description</th></tr>
    <tr><td><code><a href="#element.comment">comment</a></code></td>
        <td>SSI comment</td></tr>
    <tr><td><code><a href="#element.config">config</a></code></td>
        <td>configure output formats</td></tr>
    <tr><td><code><a href="#element.echo">echo</a></code></td>
        <td>print variables</td></tr>
    <tr><td><code><a href="#element.exec">exec</a></code></td>
        <td>execute external programs</td></tr>
    <tr><td><code><a href="#element.fsize">fsize</a></code></td>
        <td>print size of a file</td></tr>
    <tr><td><code><a href="#element.flastmod">flastmod</a></code></td>
        <td>print last modification time of a file</td></tr>
    <tr><td><code><a href="#element.include">include</a></code></td>
        <td>include a file</td></tr>
    <tr><td><code><a href="#element.printenv">printenv</a></code></td>
        <td>print all available variables</td></tr>
    <tr><td><code><a href="#element.set">set</a></code></td>
        <td>set a value of a variable</td></tr>
    </table>

    <p>SSI elements may be defined by modules other than
    <module>mod_include</module>. In fact, the <code><a
    href="#element.exec">exec</a></code> element is provided by
    <module>mod_cgi</module>, and will only be available if this
    module is loaded.</p>

    <section id="element.comment"><title>The comment Element</title>
      <p>This command doesn't output anything. Its only use is to
      add comments within a file. These comments are not printed.</p>

      <p>This syntax is available in version 2.5 and later.</p>

      <example>
      &lt;!--#comment Blah Blah Blah --&gt;
      </example>
    </section> <!-- /comment -->

    <section id="element.config"><title>The config Element</title>
      <p>This command controls various aspects of the parsing. The
      valid attributes are:</p>

      <dl>
      <dt><code>echomsg</code> (<em>Apache 2.1 and later</em>)</dt>
      <dd><p>The value is a message that is sent back to the
      client if the <code><a href="#element.echo">echo</a></code> element
      attempts to echo an undefined variable. This overrides any <directive
      module="mod_include">SSIUndefinedEcho</directive> directives.</p>

      <example>
        &lt;!--#config echomsg="[Value Undefined]" --&gt;
      </example>
      </dd>

      <dt><code>errmsg</code></dt>
      <dd><p>The value is a message that is sent back to the
      client if an error occurs while parsing the
      document. This overrides any <directive
      module="mod_include">SSIErrorMsg</directive> directives.</p>

      <example>
       &lt;!--#config errmsg="[Oops, something broke.]" --&gt;
      </example>
      </dd>

      <dt><code>sizefmt</code></dt>
      <dd><p>The value sets the format to be used when displaying
      the size of a file. Valid values are <code>bytes</code>
      for a count in bytes, or <code>abbrev</code> for a count
      in Kb or Mb as appropriate, for example a size of 1024 bytes
      will be printed as "1K".</p>

      <example>
      &lt;!--#config sizefmt="abbrev" --&gt;
      </example>

      </dd>

      <dt><code>timefmt</code></dt>
      <dd><p>The value is a string to be used by the
      <code>strftime(3)</code> library routine when printing
      dates.</p>

      <example>
      &lt;!--#config timefmt=""%R, %B %d, %Y"" --&gt;
      </example>

      </dd>
      </dl>
    </section> <!-- /config -->

    <section id="element.echo"><title>The echo Element</title>
      <p>This command prints one of the <a href="#includevars">include
      variables</a> defined below. If the variable is unset, the result is
      determined by the <directive module="mod_include"
      >SSIUndefinedEcho</directive> directive. Any dates printed are
      subject to the currently configured <code>timefmt</code>.</p>

      <p>Attributes:</p>

      <dl>
      <dt><code>var</code></dt>
      <dd>The value is the name of the variable to print.</dd>

      <dt><code>decoding</code></dt>
      <dd><p>Specifies whether Apache should strip an encoding from
      the variable before processing the variable further. The default
      is <code>none</code>, where no decoding will be done. If set to
      <code>url</code>, then URL decoding (also known as %-encoding;
      this is appropriate for use within URLs in links, etc.) will be
      performed. If set to <code>urlencoded</code>,
      application/x-www-form-urlencoded compatible encoding (found in
      query strings) will be stripped. If set to <code>base64</code>,
      base64 will be decoded, and if set to <code>entity</code>, HTML
      entity encoding will be stripped. Decoding is done prior to any
      further encoding on the variable. Multiple encodings can be
      stripped by specifying more than one comma separated encoding.
      The decoding setting will remain in effect until the next decoding
      attribute is encountered, or the element ends.</p>

      <p>The <code>decoding</code> attribute must <em>precede</em> the
      corresponding <code>var</code> attribute to be effective.</p>
      </dd>

      <dt><code>encoding</code></dt>
      <dd><p>Specifies how Apache should encode special characters
      contained in the variable before outputting them. If set
      to <code>none</code>, no encoding will be done. If set to
      <code>url</code>, then URL encoding (also known as %-encoding;
      this is appropriate for use within URLs in links, etc.) will be
      performed. If set to <code>urlencoded</code>,
      application/x-www-form-urlencoded compatible encoding will be
      performed instead, and should be used with query strings. If set
      to <code>base64</code>, base64 encoding will be performed. At
      the start of an <code>echo</code> element, the default is set to
      <code>entity</code>, resulting in entity encoding (which is
      appropriate in the context of a block-level HTML element,
      <em>e.g.</em> a paragraph of text). This can be changed by adding
      an <code>encoding</code> attribute, which will remain in effect
      until the next <code>encoding</code> attribute is encountered or
      the element ends, whichever comes first.</p>

      <p>The <code>encoding</code> attribute must <em>precede</em> the
      corresponding <code>var</code> attribute to be effective.</p>

      <note type="warning">
        In order to avoid cross-site scripting issues, you should
        <em>always</em> encode user supplied data.
      </note>

      <example><title>Example</title>
        &lt;!--#echo encoding="entity" var="QUERY_STRING" --&gt;
      </example>
      </dd>
      </dl>
    </section> <!-- /echo -->

    <section id="element.exec"><title>The exec Element</title>
      <p>The <code>exec</code> command executes a given shell command or
      CGI script. It requires <module>mod_cgi</module> to be present
      in the server. If <directive module="core">Options</directive>
      <code>IncludesNOEXEC</code> is set, this command is completely
      disabled. The valid attributes are:</p>

      <dl>
      <dt><code>cgi</code></dt>
      <dd><p>The value specifies a (%-encoded) URL-path to
      the CGI script. If the path does not begin with a slash (/),
      then it is taken to be relative to the current
      document. The document referenced by this path is
      invoked as a CGI script, even if the server would not
      normally recognize it as such. However, the directory
      containing the script must be enabled for CGI scripts
      (with <directive module="mod_alias">ScriptAlias</directive>
      or <directive module="core">Options</directive>
      <code>ExecCGI</code>).</p>

      <p>The CGI script is given the <code>PATH_INFO</code> and query
      string (<code>QUERY_STRING</code>) of the original request from the
      client; these <em>cannot</em> be specified in the URL path. The
      include variables will be available to the script in addition to
      the standard <a href="mod_cgi.html">CGI</a> environment.</p>

      <example><title>Example</title>
        &lt;!--#exec cgi="/cgi-bin/example.cgi" --&gt;
      </example>

      <p>If the script returns a <code>Location:</code> header instead of
      output, then this will be translated into an HTML anchor.</p>

      <p>The <code><a href="#includevirtual">include virtual</a></code>
      element should be used in preference to <code>exec cgi</code>. In
      particular, if you need to pass additional arguments to a CGI program,
      using the query string, this cannot be done with <code>exec
      cgi</code>, but can be done with <code>include virtual</code>, as
      shown here:</p>

      <example>
        &lt;!--#include virtual="/cgi-bin/example.cgi?argument=value" --&gt;
      </example>
      </dd>

      <dt><code>cmd</code></dt>
      <dd><p>The server will execute the given string using
      <code>/bin/sh</code>. The <a href="#includevars"
      >include variables</a> are available to the command, in addition
      to the usual set of CGI variables.</p>

      <p>The use of <code><a href="#includevirtual"
      >#include virtual</a></code> is almost always prefered to using
      either <code>#exec cgi</code> or <code>#exec cmd</code>. The former
      (<code>#include virtual</code>) uses the standard Apache sub-request
      mechanism to include files or scripts. It is much better tested and
      maintained.</p>

      <p>In addition, on some platforms, like Win32, and on unix when
      using <a href="../suexec.html">suexec</a>, you cannot pass arguments
      to a command in an <code>exec</code> directive, or otherwise include
      spaces in the command. Thus, while the following will work under a
      non-suexec configuration on unix, it will not produce the desired
      result under Win32, or when running suexec:</p>

      <example>
        &lt;!--#exec cmd="perl /path/to/perlscript arg1 arg2" --&gt;
      </example>
      </dd>
      </dl>
    </section> <!-- /exec -->

    <section id="element.fsize"><title>The fsize Element</title>
      <p>This command prints the size of the specified file, subject
      to the <code>sizefmt</code> format specification. Attributes:</p>

      <dl>
      <dt><code>file</code></dt>
      <dd>The value is a path relative to the directory
      containing the current document being parsed.

    <example>
        This file is &lt;!--#fsize file="mod_include.html" --&gt; bytes.
    </example>

    The value of <code>file</code> cannot start with a slash
    (<code>/</code>), nor can it contain <code>../</code> so as to
    refer to a file above the current directory or outside of the
    document root. Attempting to so will result in the error message:
    <code>The given path was above the root path</code>.
      </dd>

      <dt><code>virtual</code></dt>
      <dd>The value is a (%-encoded) URL-path. If it does not begin with
      a slash (/) then it is taken to be relative to the current document.
      Note, that this does <em>not</em> print the size of any CGI output,
      but the size of the CGI script itself.</dd>
      </dl>

    <example>
        This file is &lt;!--#fsize virtual="/docs/mod/mod_include.html" --&gt; bytes.
    </example>

      <p>Note that in many cases these two are exactly the same thing.
      However, the <code>file</code> attribute doesn't respect URL-space
      aliases.</p>
    </section> <!-- /fsize -->

    <section id="element.flastmod"><title>The flastmod Element</title>
      <p>This command prints the last modification date of the
      specified file, subject to the <code>timefmt</code> format
      specification. The attributes are the same as for the
      <code><a href="#element.fsize">fsize</a></code> command.</p>
    </section> <!-- /flastmod -->

    <section id="element.include"><title>The include Element</title>
      <p>This command inserts the text of another document or file
      into the parsed file. Any included file is subject to the usual
      access control. If the directory containing the parsed file has
      <a href="core.html#options">Options</a>
      <code>IncludesNOEXEC</code> set, then only documents with a text
      <glossary>MIME-type</glossary> (<code>text/plain</code>,
      <code>text/html</code> etc.) will be included. Otherwise CGI
      scripts are invoked as normal using the complete URL given in
      the command, including any query string.</p>

      <p>An attribute defines the location of the document, and may
      appear more than once in an include element; an inclusion is
      done for each attribute given to the include command in turn.
      The valid attributes are:</p>

      <dl>
      <dt><code>file</code></dt>
      <dd>The value is a path relative to the directory
      containing the current document being parsed. It cannot
      contain <code>../</code>, nor can it be an absolute path.
      Therefore, you cannot include files that are outside of the
      document root, or above the current document in the directory
      structure. The <code>virtual</code> attribute should always be
      used in preference to this one.</dd>

      <dt><code><a id="includevirtual" name="includevirtual"
      >virtual</a></code></dt>
      <dd><p>The value is a (%-encoded) URL-path. The URL cannot contain a
      scheme or hostname, only a path and an optional query string. If it
      does not begin with a slash (/) then it is taken to be relative to the
      current document.</p>

      <p>A URL is constructed from the attribute, and the output the
      server would return if the URL were accessed by the client is
      included in the parsed output. Thus included files can be nested.</p>

      <p>If the specified URL is a CGI program, the program will be
      executed and its output inserted in place of the directive in the
      parsed file. You may include a query string in a CGI url:</p>

      <example>
        &lt;!--#include virtual="/cgi-bin/example.cgi?argument=value" --&gt;
      </example>

      <p><code>include virtual</code> should be used in preference
      to <code>exec cgi</code> to include the output of CGI programs
      into an HTML document.</p>

      <p>If the <directive module="mod_request">KeptBodySize</directive>
      directive is correctly configured and valid for this included
      file, attempts to POST requests to the enclosing HTML document
      will be passed through to subrequests as POST requests as well.
      Without the directive, all subrequests are processed as GET
      requests.</p>

      </dd>

      <dt><code>onerror</code></dt>
      <dd><p>The value is a (%-encoded) URL-path which is shown should a
      previous attempt to include a file or virtual attribute failed.
      To be effective, this attribute must be specified after the
      file or virtual attributes being covered. If the attempt to
      include the onerror path fails, or if onerror is not specified, the
      default error message will be included.</p>

      <example>
        # Simple example<br />
        &lt;!--#include virtual="/not-exist.html" onerror="/error.html" --&gt;
      </example>

      <example>
        # Dedicated onerror paths<br />
        &lt;!--#include virtual="/path-a.html" onerror="/error-a.html" virtual="/path-b.html" onerror="/error-b.html" --&gt;
      </example>

      </dd>
      </dl>
    </section> <!-- /include -->

    <section id="element.printenv"><title>The printenv Element</title>
      <p>This prints out a plain text listing of all existing variables and
      their values. Special characters are entity encoded (see the <code><a
      href="#element.echo">echo</a></code> element for details)
      before being output. There are no attributes.</p>

      <example><title>Example</title>
        &lt;pre&gt;
          &lt;!--#printenv --&gt;
        &lt;/pre&gt;
      </example>
    </section> <!-- /printenv -->

    <section id="element.set"><title>The set Element</title>
      <p>This sets the value of a variable. Attributes:</p>

      <dl>
      <dt><code>var</code></dt>
      <dd>The name of the variable to set.</dd>

      <dt><code>value</code></dt>
      <dd>The value to give a variable.</dd>

      <dt><code>decoding</code></dt>
      <dd><p>Specifies whether Apache should strip an encoding from
      the variable before processing the variable further. The default
      is <code>none</code>, where no decoding will be done. If set to
      <code>url</code>, <code>urlencoded</code>, <code>base64</code>
      or <code>entity</code>, URL decoding,
      application/x-www-form-urlencoded decoding, base64 decoding or HTML
      entity decoding will be performed respectively. More than one
      decoding can be specified by separating with commas. The decoding
      setting will remain in effect until the next decoding attribute
      is encountered, or the element ends. The <code>decoding</code>
      attribute must <em>precede</em> the corresponding
      <code>var</code> attribute to be effective.</p>
      </dd>

      <dt><code>encoding</code></dt>
      <dd><p>Specifies how Apache should encode special characters
      contained in the variable before setting them. The default is
      <code>none</code>, where no encoding will be done. If set to
      <code>url</code>, <code>urlencoding</code>, <code>base64</code>
      or <code>entity</code>, URL encoding,
      application/x-www-form-urlencoded encoding, base64 encoding or
      HTML entity encoding will be performed respectively. More than
      one encoding can be specified by separating with commas. The
      encoding setting will remain in effect until the next encoding
      attribute is encountered, or the element ends. The
      <code>encoding</code> attribute must <em>precede</em> the
      corresponding <code>var</code> attribute to be effective.
      Encodings are applied after all decodings have been
      stripped.</p>
      </dd>
      </dl>

      <example><title>Example</title>
        &lt;!--#set var="category" value="help" --&gt;
      </example>
    </section> <!-- /set -->
</section> <!-- /basic elements -->

<section id="includevars">
    <title>Include Variables</title>

    <p>In addition to the variables in the standard CGI environment,
    these are available for the <code>echo</code> command, for
    <code>if</code> and <code>elif</code>, and to any program
    invoked by the document.</p>

    <dl>
      <dt><code>DATE_GMT</code></dt>
      <dd>The current date in Greenwich Mean Time.</dd>

      <dt><code>DATE_LOCAL</code></dt>
      <dd>The current date in the local time zone.</dd>

      <dt><code>DOCUMENT_ARGS</code></dt>
      <dd>This variable contains the query string of the active SSI
      document, or the empty string if a query string is not
      included.  For subrequests invoked through the
      <code>include</code> SSI directive, <code>QUERY_STRING</code>
      will represent the query string of the subrequest and
      <code>DOCUMENT_ARGS</code> will represent the query string of
      the SSI document.  (Available in Apache HTTP Server 2.4.19 and
      later.)</dd>

      <dt><code>DOCUMENT_NAME</code></dt>
      <dd>The filename (excluding directories) of the document
      requested by the user.</dd>

      <dt><code>DOCUMENT_URI</code></dt>
      <dd>The (%-decoded) URL path of the document requested by the
      user. Note that in the case of nested include files, this is
      <em>not</em> the URL for the current document.  Note also that
      if the URL is modified internally (e.g. by an <directive
      module="mod_alias">alias</directive> or <directive
      module="mod_dir">directoryindex</directive>), the modified
      URL is shown.</dd>

      <dt><code>LAST_MODIFIED</code></dt>
      <dd>The last modification date of the document requested by
      the user.</dd>

      <dt><code>QUERY_STRING_UNESCAPED</code></dt>
      <dd>If a query string is present in the request for the active
      SSI document, this variable contains the (%-decoded) query
      string, which is <em>escaped</em> for shell usage (special
      characters like <code>&amp;</code> etc. are preceded by
      backslashes).  It is not set if a query string is not
      present.  Use <code>DOCUMENT_ARGS</code> if shell escaping
      is not desired.</dd>
    </dl>
</section>

<section id="substitution"><title>Variable Substitution</title>

    <p>Variable substitution is done within quoted strings in most
    cases where they may reasonably occur as an argument to an SSI
    directive. This includes the <code>config</code>,
    <code>exec</code>, <code>flastmod</code>, <code>fsize</code>,
    <code>include</code>, <code>echo</code>, and <code>set</code>
    directives. If <directive module="mod_include"
    >SSILegacyExprParser</directive> is set to <code>on</code>,
    substitution also occurs in the arguments to conditional operators.
    You can insert a literal dollar sign into the string using backslash
    quoting:</p>

    <example>
      &lt;!--#set var="cur" value="\$test" --&gt;
    </example>

    <p>If a variable reference needs to be substituted in the
    middle of a character sequence that might otherwise be
    considered a valid identifier in its own right, it can be
    disambiguated by enclosing the reference in braces,
    <em>a la</em> shell substitution:</p>

    <example>
      &lt;!--#set var="Zed" value="${REMOTE_HOST}_${REQUEST_METHOD}" --&gt;
    </example>

    <p>This will result in the <code>Zed</code> variable being set
    to "<code>X_Y</code>" if <code>REMOTE_HOST</code> is
    "<code>X</code>" and <code>REQUEST_METHOD</code> is
    "<code>Y</code>".</p>
</section>

<section id="flowctrl">
    <title>Flow Control Elements</title>

    <p>The basic flow control elements are:</p>

    <example>
      &lt;!--#if expr="<var>test_condition</var>" --&gt;<br />
      &lt;!--#elif expr="<var>test_condition</var>" --&gt;<br />
      &lt;!--#else --&gt;<br />
      &lt;!--#endif --&gt;
    </example>

    <p>The <code>if</code> element works like an if statement in a
    programming language. The test condition is evaluated and if
    the result is true, then the text until the next <code>elif</code>,
    <code>else</code> or <code>endif</code> element is included in the
    output stream.</p>

    <p>The <code>elif</code> or <code>else</code> statements are used
    to put text into the output stream if the original
    <var>test_condition</var> was false. These elements are optional.</p>

    <p>The <code>endif</code> element ends the <code>if</code> element
    and is required.</p>

    <p><var>test_condition</var> is a boolean expression which follows the
    <a href="../expr.html">ap_expr</a> syntax. The syntax can be changed
    to be compatible with Apache HTTPD 2.2.x using <directive
    module="mod_include">SSILegacyExprParser</directive>.</p>

    <p>The SSI variables set with the <code>var</code> element are exported
    into the request environment and can be accessed with the
    <code>reqenv</code> function. As a short-cut, the function name
    <code>v</code> is also available inside <module>mod_include</module>.</p>

    <p>The below example will print "from local net" if client IP address
    belongs to the 10.0.0.0/8 subnet.</p>

    <example>
      &lt;!--#if expr='-R "10.0.0.0/8"' --&gt;<br />
      <indent>
        from local net<br />
      </indent>
      &lt;!--#else --&gt;<br />
      <indent>
        from somewhere else<br />
      </indent>
      &lt;!--#endif --&gt;
    </example>

    <p>The below example will print "foo is bar" if the variable
    <code>foo</code> is set to the value "bar".</p>

    <example>
      &lt;!--#if expr='v("foo") = "bar"' --&gt;<br />
      <indent>
        foo is bar<br />
      </indent>
      &lt;!--#endif --&gt;
    </example>

    <note><title>Reference Documentation</title>
    <p>See also: <a href="../expr.html">Expressions in Apache HTTP Server</a>,
    for a complete reference and examples. The <em>restricted</em> functions
    are not available inside <module>mod_include</module></p>
    </note>
</section>

<section id="legacyexpr">
    <title>Legacy expression syntax</title>

    <p>This section describes the syntax of the <code>#if expr</code>
    element if <directive module="mod_include">SSILegacyExprParser</directive>
    is set to <code>on</code>.</p>

    <dl>
      <dt><code><var>string</var></code></dt>
      <dd>true if <var>string</var> is not empty</dd>

      <dt><code><var>-A string</var></code></dt>
      <dd><p>true if the URL represented by the string is accessible by
      configuration, false otherwise. This is useful where content on a
      page is to be hidden from users who are not authorized to view the
      URL, such as a link to that URL. Note that the URL is only tested
      for whether access would be granted, not whether the URL exists.</p>

      <example><title>Example</title>
        &lt;!--#if expr="-A /private" --&gt;<br />
        <indent>
          Click &lt;a href="/private"&gt;here&lt;/a&gt; to access private
          information.<br />
        </indent>
        &lt;!--#endif --&gt;
      </example>
      </dd>

      <dt><code><var>string1</var> = <var>string2</var><br />
      <var>string1</var> == <var>string2</var><br />
      <var>string1</var> != <var>string2</var></code></dt>

      <dd><p>Compare <var>string1</var> with <var>string2</var>. If
      <var>string2</var> has the form <code>/<var>string2</var>/</code>
      then it is treated as a regular expression. Regular expressions are
      implemented by the <a href="http://www.pcre.org">PCRE</a> engine and
      have the same syntax as those in <a href="http://www.perl.com">perl
      5</a>. Note that <code>==</code> is just an alias for <code>=</code>
      and behaves exactly the same way.</p>

      <p>If you are matching positive (<code>=</code> or <code>==</code>), you
      can capture grouped parts of the regular expression. The captured parts
      are stored in the special variables <code>$1</code> ..
      <code>$9</code>. The whole string matched by the regular expression is
      stored in the special variable <code>$0</code></p>

      <example><title>Example</title>
        &lt;!--#if expr="$QUERY_STRING = /^sid=([a-zA-Z0-9]+)/" --&gt;<br />
        <indent>
          &lt;!--#set var="session" value="$1" --&gt;<br />
        </indent>
        &lt;!--#endif --&gt;
      </example>
      </dd>

      <dt><code><var>string1</var> &lt; <var>string2</var><br />
       <var>string1</var> &lt;= <var>string2</var><br />
       <var>string1</var> &gt; <var>string2</var><br />
       <var>string1</var> &gt;= <var>string2</var></code></dt>

      <dd>Compare <var>string1</var> with <var>string2</var>. Note, that
      strings are compared <em>literally</em> (using
      <code>strcmp(3)</code>). Therefore the string "100" is less than
      "20".</dd>

      <dt><code>( <var>test_condition</var> )</code></dt>
      <dd>true if <var>test_condition</var> is true</dd>

      <dt><code>! <var>test_condition</var></code></dt>
      <dd>true if <var>test_condition</var> is false</dd>

      <dt><code><var>test_condition1</var> &amp;&amp;
        <var>test_condition2</var></code></dt>
      <dd>true if both <var>test_condition1</var> and
      <var>test_condition2</var> are true</dd>

      <dt><code><var>test_condition1</var> ||
        <var>test_condition2</var></code></dt>
      <dd>true if either <var>test_condition1</var> or
      <var>test_condition2</var> is true</dd>
    </dl>

    <p>"<code>=</code>" and "<code>!=</code>" bind more tightly than
    "<code>&amp;&amp;</code>" and "<code>||</code>". "<code>!</code>" binds
    most tightly. Thus, the following are equivalent:</p>

    <example>
      &lt;!--#if expr="$a = test1 &amp;&amp; $b = test2" --&gt;<br />
      &lt;!--#if expr="($a = test1) &amp;&amp; ($b = test2)" --&gt;
    </example>

    <p>The boolean operators <code>&amp;&amp;</code> and <code>||</code>
    share the same priority. So if you want to bind such an operator more
    tightly, you should use parentheses.</p>

    <p>Anything that's not recognized as a variable or an operator
    is treated as a string. Strings can also be quoted:
    <code>'string'</code>. Unquoted strings can't contain whitespace
    (blanks and tabs) because it is used to separate tokens such as
    variables. If multiple strings are found in a row, they are
    concatenated using blanks. So,</p>

    <example>
      <p><code><var>string1</var>&nbsp;&nbsp;&nbsp;&nbsp;<var
      >string2</var></code> results in <code><var>string1</var>&nbsp;<var
      >string2</var></code><br />
      <br />
      and<br />
      <br />
      <code>'<var>string1</var>&nbsp;&nbsp;&nbsp;&nbsp;<var
      >string2</var>'</code> results in <code><var
      >string1</var>&nbsp;&nbsp;&nbsp;&nbsp;<var>string2</var></code>.</p>
    </example>

    <note><title>Optimization of Boolean Expressions</title>
      <p>If the expressions become more complex and slow down processing
      significantly, you can try to optimize them according to the
      evaluation rules:</p>
      <ul>
      <li>Expressions are evaluated from left to right</li>
      <li>Binary boolean operators (<code>&amp;&amp;</code> and <code>||</code>)
          are short circuited wherever possible. In conclusion with the rule
          above that means, <module>mod_include</module> evaluates at first
          the left expression. If the left result is sufficient to determine
          the end result, processing stops here. Otherwise it evaluates the
          right side and computes the end result from both left and right
          results.</li>
      <li>Short circuit evaluation is turned off as long as there are regular
          expressions to deal with. These must be evaluated to fill in the
          backreference variables (<code>$1</code> .. <code>$9</code>).</li>
      </ul>
      <p>If you want to look how a particular expression is handled, you can
      recompile <module>mod_include</module> using the
      <code>-DDEBUG_INCLUDE</code> compiler option. This inserts for every
      parsed expression tokenizer information, the parse tree and how it is
      evaluated into the output sent to the client.</p>
    </note>

    <note><title>Escaping slashes in regex strings</title>
     <p>All slashes which are not intended to act as delimiters in your regex must
     be escaped.  This is regardless of their meaning to the regex engine.</p>
    </note>

</section>

<directivesynopsis>
<name>SSIEndTag</name>
<description>String that ends an include element</description>
<syntax>SSIEndTag <var>tag</var></syntax>
<default>SSIEndTag &quot;--&gt;&quot;</default>
<contextlist><context>server config</context><context>virtual host</context>
</contextlist>

<usage>
    <p>This directive changes the string that <module>mod_include</module>
    looks for to mark the end of an include element.</p>

    <highlight language="config">
SSIEndTag "%&gt;"
    </highlight>

</usage>
<seealso><directive module="mod_include">SSIStartTag</directive></seealso>
</directivesynopsis>

<directivesynopsis>
<name>SSIUndefinedEcho</name>
<description>String displayed when an unset variable is echoed</description>
<syntax>SSIUndefinedEcho <var>string</var></syntax>
<default>SSIUndefinedEcho &quot;(none)&quot;</default>
<contextlist><context>server config</context><context>virtual host</context>
<context>directory</context><context>.htaccess</context></contextlist>
<override>All</override>

<usage>
    <p>This directive changes the string that <module>mod_include</module>
    displays when a variable is not set and &quot;echoed&quot;.</p>

    <highlight language="config">
SSIUndefinedEcho "&lt;!-- undef --&gt;"
    </highlight>
</usage>
</directivesynopsis>

<directivesynopsis>
<name>SSIErrorMsg</name>
<description>Error message displayed when there is an SSI
error</description>
<syntax>SSIErrorMsg <var>message</var></syntax>
<default>SSIErrorMsg &quot;[an error occurred while processing this
directive]&quot;</default>
<contextlist><context>server config</context><context>virtual host</context>
<context>directory</context><context>.htaccess</context></contextlist>
<override>All</override>

<usage>
    <p>The <directive>SSIErrorMsg</directive> directive changes the error
    message displayed when <module>mod_include</module> encounters an
    error. For production servers you may consider changing the default
    error message to <code>&quot;&lt;!-- Error --&gt;&quot;</code> so that
    the message is not presented to the user.</p>

    <p>This directive has the same effect as the <code>&lt;!--#config
    errmsg=<var>message</var> --&gt;</code> element.</p>

    <highlight language="config">
SSIErrorMsg "&lt;!-- Error --&gt;"
    </highlight>
</usage>
</directivesynopsis>

<directivesynopsis>
<name>SSIStartTag</name>
<description>String that starts an include element</description>
<syntax>SSIStartTag <var>tag</var></syntax>
<default>SSIStartTag &quot;&lt;!--#&quot;</default>
<contextlist><context>server config</context><context>virtual host</context>
</contextlist>

<usage>
    <p>This directive changes the string that <module>mod_include</module>
    looks for to mark an include element to process.</p>

    <p>You may want to use this option if you have 2 servers parsing the
    output of a file each processing different commands (possibly at
    different times).</p>

    <highlight language="config">
SSIStartTag "&lt;%"
SSIEndTag   "%&gt;"
    </highlight>

    <p>The example given above, which also specifies a matching
    <directive module="mod_include">SSIEndTag</directive>, will
    allow you to use SSI directives as shown in the example
    below:</p>

    <example><title>SSI directives with alternate start and end tags</title>
      &lt;%printenv %&gt;
    </example>
</usage>
<seealso><directive module="mod_include">SSIEndTag</directive></seealso>
</directivesynopsis>

<directivesynopsis>
<name>SSITimeFormat</name>
<description>Configures the format in which date strings are
displayed</description>
<syntax>SSITimeFormat <var>formatstring</var></syntax>
<default>SSITimeFormat &quot;%A, %d-%b-%Y %H:%M:%S %Z&quot;</default>
<contextlist>
<context>server config</context><context>virtual host</context>
<context>directory</context><context>.htaccess</context></contextlist>
<override>All</override>

<usage>
<p>This directive changes the format in which date strings are displayed
    when echoing <code>DATE</code> environment variables. The
    <var>formatstring</var> is as in <code>strftime(3)</code> from the
    C standard library.</p>

    <p>This directive has the same effect as the <code>&lt;!--#config
    timefmt=<var>formatstring</var> --&gt;</code> element.</p>

    <highlight language="config">
SSITimeFormat "%R, %B %d, %Y"
    </highlight>

    <p>The above directive would cause times to be displayed in the
    format "22:26, June 14, 2002".</p>
</usage>
</directivesynopsis>

<directivesynopsis>
<name>SSIETag</name>
<description>Controls whether ETags are generated by the server.</description>
<syntax>SSIETag on|off</syntax>
<default>SSIETag off</default>
<contextlist><context>directory</context><context>.htaccess</context></contextlist>

<usage>
      <p>Under normal circumstances, a file filtered by <module>mod_include</module>
        may contain elements that are either dynamically generated, or that may
        have changed independently of the original file. As a result, by default
        the server is asked not to generate an <code>ETag</code> header for the
        response by adding <code>no-etag</code> to the request notes.</p>

      <p>The <directive>SSIETag</directive> directive suppresses this
        behaviour, and allows the server to generate an <code>ETag</code> header.
        This can be used to enable caching of the output. Note that a backend server
        or dynamic content generator may generate an ETag of its own, ignoring
        <code>no-etag</code>, and this ETag will be passed by
        <module>mod_include</module> regardless of the value of this setting.
        <directive>SSIETag</directive> can take on the following values:</p>

      <dl>

        <dt><code>off</code></dt>
        <dd><code>no-etag</code> will be added to the request notes, and the server
          is asked not to generate an ETag. Where a server ignores the value of
          <code>no-etag</code> and generates an ETag anyway, the ETag will be
          respected.</dd>

        <dt><code>on</code></dt>
        <dd>Existing ETags will be respected, and ETags generated by the server will
          be passed on in the response.</dd>

      </dl>

</usage>
</directivesynopsis>

<directivesynopsis>
<name>SSILastModified</name>
<description>Controls whether <code>Last-Modified</code> headers are generated by the
server.</description>
<syntax>SSILastModified on|off</syntax>
<default>SSILastModified off</default>
<contextlist><context>directory</context><context>.htaccess</context></contextlist>

<usage>
      <p>Under normal circumstances, a file filtered by <module>mod_include</module>
        may contain elements that are either dynamically generated, or that may
        have changed independently of the original file. As a result, by default
        the <code>Last-Modified</code> header is stripped from the response.</p>

      <p>The <directive>SSILastModified</directive> directive overrides this
        behaviour, and allows the <code>Last-Modified</code> header to be respected
        if already present, or set if the header is not already present. This can
        be used to enable caching of the output. <directive>SSILastModified</directive>
        can take on the following values:</p>

      <dl>

        <dt><code>off</code></dt>
        <dd>The <code>Last-Modified</code> header will be stripped from responses,
          unless the <directive module="mod_include">XBitHack</directive> directive
          is set to <code>full</code> as described below.</dd>

        <dt><code>on</code></dt>
        <dd>The <code>Last-Modified</code> header will be respected if already
          present in a response, and added to the response if the response is a
          file and the header is missing. The
          <directive module="mod_include">SSILastModified</directive> directive
          takes precedence over <directive module="mod_include">XBitHack</directive>.</dd>

      </dl>

</usage>
</directivesynopsis>

<directivesynopsis>
<name>SSILegacyExprParser</name>
<description>Enable compatibility mode for conditional expressions.</description>
<syntax>SSILegacyExprParser on|off</syntax>
<default>SSILegacyExprParser off</default>
<contextlist><context>directory</context><context>.htaccess</context></contextlist>
<compatibility>Available in version 2.3.13 and later.</compatibility>

<usage>
    <p>As of version 2.3.13, <module>mod_include</module> has switched to the
    new <a href="../expr.html">ap_expr</a> syntax for conditional expressions
    in <code>#if</code> flow control elements.  This directive allows to
    switch to the <a href="#legacyexpr">old syntax</a> which is compatible
    with Apache HTTPD version 2.2.x and earlier.
    </p>
</usage>
</directivesynopsis>

<directivesynopsis>
<name>XBitHack</name>
<description>Parse SSI directives in files with the execute bit
set</description>
<syntax>XBitHack on|off|full</syntax>
<default>XBitHack off</default>
<contextlist><context>server config</context><context>virtual host</context>
<context>directory</context><context>.htaccess</context></contextlist>
<override>Options</override>

<usage>
    <p>The <directive>XBitHack</directive> directive controls the parsing
    of ordinary html documents. This directive only affects files associated
    with the <glossary>MIME-type</glossary> <code>text/html</code>. <directive
    >XBitHack</directive> can take on the following values:</p>

    <dl>
      <dt><code>off</code></dt>
      <dd>No special treatment of executable files.</dd>

      <dt><code>on</code></dt>
      <dd>Any <code>text/html</code> file that has the user-execute bit
      set will be treated as a server-parsed html document.</dd>

      <dt><code>full</code></dt>
      <dd>As for <code>on</code> but also test the group-execute bit.
      If it is set, then set the <code>Last-modified</code> date of the
      returned file to be the last modified time of the file. If
      it is not set, then no last-modified date is sent. Setting
      this bit allows clients and proxies to cache the result of
      the request.

      <note><title>Note</title>
      <p>You would not want to use the full option, unless you assure the
      group-execute bit is unset for every SSI script which might <code
      >#include</code> a CGI or otherwise produces different output on
      each hit (or could potentially change on subsequent requests).</p>

      <p>The <directive module="mod_include">SSILastModified</directive>
      directive takes precedence over the
      <directive module="mod_include">XBitHack</directive> directive when
      <directive module="mod_include">SSILastModified</directive> is set to
      <code>on</code>.</p>
      </note>

      </dd>
    </dl>

</usage>
</directivesynopsis>

</modulesynopsis>