code-srv-test/dita-ot-3.6/docsrc/reference/map-first-preprocessing.dita
2021-03-23 22:38:58 +00:00

131 lines
8.2 KiB
XML
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE concept PUBLIC "-//OASIS//DTD DITA Concept//EN" "concept.dtd">
<!-- This file is part of the DITA Open Toolkit project. See the accompanying LICENSE file for applicable license. -->
<concept id="ID" rev="3.0">
<title>Map-first preprocessing</title>
<titlealts>
<navtitle>Map-first preprocessing</navtitle>
</titlealts>
<shortdesc><ph id="map-first-preproc-desc">DITA-OT 3.0 provides a map-first preprocessing option as an alternative to
the default <codeph>preprocess</codeph> operation. The method, which was introduced in DITA-OT 2.5 as an
experimental feature, has been improved and is ready for use in many production scenarios. Map-first-preprocessing
provides the same functionality as the default <codeph>preprocess</codeph>, but takes a different
approach.</ph></shortdesc>
<prolog>
<metadata>
<keywords>
<indexterm>filters
<indexterm>map-first preprocessing</indexterm></indexterm>
<indexterm>preprocessing
<indexterm>map first</indexterm></indexterm>
<indexterm>DITA 1.0</indexterm>
<indexterm>DITA 1.2</indexterm>
<indexterm>DITA 1.3
<indexterm>effect on preprocessing</indexterm></indexterm>
<indexterm>pipelines
<indexterm>map first</indexterm></indexterm>
<indexterm>validate</indexterm>
<indexterm>maps
<index-see>DITA maps</index-see></indexterm>
<indexterm>DITA maps
<indexterm>validate</indexterm></indexterm>
</keywords>
</metadata>
</prolog>
<conbody>
<p id="map-first-preproc-gain">Whereas the default preprocessing routine handles both maps and topics at the same
time, often switching back and forth between map operations and topic operations, the map-first approach only
begins processing topics after nearly all map processing is complete. This simplifies the processing logic and
creates cleaner module responsibilities, which makes it easier to process only those topics that are actually
referenced after filtering, for example, or to only process the map to validate the map structure.</p>
<p>The current preprocessing architecture was established during the DITA 1.0 era when there were fewer DITA
features that operated on the map level. Initially, the difference between processing modes was not that great.
DITA 1.2 and 1.3 introduced many more map-level features, such as keys and key scopes, that make it difficult to
reliably work with topics before all map features have been resolved. </p>
<p>The original preprocessing operation already handles many map operations first, but this was not the original
design and requires regular refactoring to handle edge cases. The new map-first preprocessing is designed with
this model in mind, improving the overall processing flow and making it more formal about the map-first model. The
new model also takes advantage of hashed topic file names in the temporary directory, which simplifies many
processing steps, and is better able to handle topics referenced outside of the map directory (that case has
resulted in a variety of issues with the original model).</p>
<note id="map-first-preproc-note">The map-first preprocessing option is enabled by default in DITA-OT 3.0 for PDF
and HTML Help. These formats were chosen because they generate a compiled result file, so temporarily hashed file
names should all be invisible to the build. After further testing and feedback, the new option will most likely
become the default for other output formats in future versions. Because the DITA-OT development team cannot have
access to all varieties of DITA, all edge cases, or even all the ways DITA-OT itself is extended, the switch to
default map-first preprocessing for other output formats will be smoother for everyone if more people can test and
provide feedback.</note>
<section>
<title>How to use map-first preprocessing</title>
<p>To use (or test) map-first preprocessing, call the <codeph>preprocess2</codeph> Ant target in your custom
transformation types instead of using the default <codeph>preprocess</codeph> target.</p>
<p>For example, if you have a custom HTML5 transformation type named "myhtml", then you may have a plug-in
extension that looks this:</p>
<codeblock outputclass="language-xml">&lt;!-- Simple variant: set properties and call default HTML5 -->
&lt;target name="dita2myhtml" depends="myhtml.init,dita2html5"/>
</codeblock>
<p>This type of extension is quite common, and is used to set default properties for your environment followed by
a normal build to use those properties. In this case, you'll need to replace <codeph>dita2html5</codeph> with
the normal HTML5 steps, swapping out <codeph>preprocess</codeph> for <codeph>preprocess2</codeph>:</p>
<codeblock
outputclass="language-xml normalize-space show-line-numbers show-whitespace"
>&lt;!-- Simple variant: set properties and call default HTML5 -->
&lt;target name="dita2myhtml"
depends="myhtml.init,
html5.init,
build-init,
<b>preprocess2,</b>
html5.topic,
html5.map,
html5.css"/></codeblock>
<note>If you use this simple method for customized PDF or HTML Help builds, you will automatically be using
<codeph>preprocess2</codeph>.</note>
<p>Some custom transformation types already require you to repeat the default dependencies, in which case you
should already call <codeph>preprocess</codeph> directly, as in the following:</p>
<codeblock
outputclass="language-xml normalize-space show-line-numbers show-whitespace"
>&lt;!-- More complex variant: add processing steps to default HTML5 -->
&lt;target name="dita2myhtml"
depends="myhtml.init,
build-init,
preprocess,
local-extensions-after-preprocess,
html5.topic,
html5.map,
html5.css"/></codeblock>
<p>In such cases, the modification is much easier simply add a <codeph>2</codeph> to the existing
<codeph>preprocess</codeph> target.</p>
</section>
<section>
<title>How to test in a production environment</title>
<p>In some cases, you may be responsible for maintaining transformation types that are actually run by many people
on your team or around a company. In this case, you likely need to maintain your existing transformation types
based on the backwards-compatible <codeph>preprocess</codeph> modules, but also want to provide your colleagues
with a way to test their own documents using <codeph>preprocess2</codeph>.</p>
<p>There are several ways to do this. One fairly straightforward approach would be to create a new custom
transformation type that is exactly the same, except for preprocessing. For example, if you have a local HTML
variant called <codeph>myhtml</codeph> as above, instead of modifying that transformation directly, you could
create a second transformation type called <codeph>myhtml-beta</codeph> that provides exactly the same support,
but with the new map-first preprocessing:</p>
<codeblock outputclass="language-xml normalize-space show-line-numbers show-whitespace"><i
>&lt;!-- Original "myhtml" is not modified, used for production --></i>
&lt;target name="dita2myhtml5" depends="myhtml.init,dita2html5"/>
<i>&lt;!-- "myhtml-beta" used to test and provide feedback on preprocess2 --></i>
&lt;target name="dita2myhtml-beta"
depends="myhtml.init,
html5.init,
build-init,
<b>preprocess2,</b>
html5.topic,
html5.map,
html5.css"/></codeblock>
</section>
<section>
<title>Known limitations</title>
<p>The <codeph>preprocess2</codeph> implementation details are subject to change; dependencies within
<codeph>preprocess2</codeph> may be renamed or removed based on feedback.</p>
<p conkeyref="reusable-components/no-internal-preprocess2-ext"/>
</section>
</conbody>
</concept>