codegen/vulkan/vulkan-docs-next/scripts/README.adoc

// Copyright 2014-2023 The Khronos Group Inc.
// SPDX-License-Identifier: CC-BY-4.0

= Vulkan^(R)^ Specification Repository Scripts

This directory contains scripts used in building the Vulkan API
specification and related artifacts. For the most part, these scripts are
invoked from the top-level directory or from the API Registry in
../xml to build generated components of the specification.

Scripts in this directory include:

  * `genvk.py` - Python script to generate Vulkan headers and some other
    targets, using the generators described below.
  * `reg.py` - Python tools to read a registry XML file and call into
    generators to create headers and other types of output.
  * `conventions.py`, `vkconventions.py`, `apiconventions.py` - API-specific
    parameters and formatting / style conventions used by generators.
  * `generator.py` - output generator base class.
  ** `cgenerator.py` - C header output generator.
  ** `docgenerator.py` - Asciidoc interface language include generator.
  ** `extensionmetadocgenerator.py` - Generator for Asciidoc extension
     descriptions in spec appendices.
  ** `hostsyncgenerator.py` - Asciidoc host sync table generator.
  ** `interfacedocgenerator.py` - Asciidoc extension appendix interface
     generator.
  ** `pygenerator.py, `rubygenerator.py`, `scriptgenerator.py` - generate
     Python and Ruby encodings of parts of the API description for use in
     other scripts and asciidoctor extensions.
  ** `spirvcapgenerator.py` - Asciidoc generator for SPIR-V capability
     tables.
  ** `validitygenerator.py` - Asciidoc validity language generator.

  * `check_spec_links.py` - validates a variety of markup and content in the
    Asciidoctor specification source.
  * `conventions.py`, `vkconventions.py` - API-specific options used by
    scripts shared with OpenXR and other APIs.
  * `extdependency.py` - generate extension dependencies for use when
    building the specification.
  * `genRelease`, `genspec.py` - build HTML and PDF Specifications with a
    variety of options to control target directories, extensions included
    while building, etc.
  * `genRef.py`, `reflib.py`, `makemanaliases.py` - extract API reference
    pages from specification source into single-page source documents, and
    create symbolic links in the generated refpage HTML for API aliases.
  * `indexExt.py` - generate HTML index of all extensions for inclusion into
    the Vulkan registry index page.
  * `reflow.py`, `reflow_count.py` - reflow specification source text to
    follow style guidelines, and insert Valid Usage statements where they
    are needed.
  * `test_check_spec_links.py`, `test_check_spec_links_api_specific.py`,
    `test_entity_db.py` - these are from another Khronos WG repository and
    are unused by Vulkan at present.
  * `xml_consistency.py` - perform some internal consistency checks on
    `vk.xml`.

  * `compImages.sh` - compare images in two branches.
  * `htmldiff/htmldiff` - HTML diff script (see below).
  * `Retired/` - contains obsolete, unused, or single-purpose scripts. Not
    maintained.

== HTML Diff Script for Vulkan

This is a first cut at a script to compare Vulkan HTML specifications. Usage
is simply 'htmldiff file1.html file2.html > diff.html'. The script does not
copy CSS and images required by the input specs, so it is best to generate
the output in the same directory as one of the inputs. However, the script
must be invoked from the directory it is located within.

The scripts used require Python and Perl. Additionally, the python
'utidylib' module and the underlying libtidy C library are required.
On Debian Linux, it may be necessary to install the 'python-utidylib' and
'libtidy' packages if they are not already present.

The scripts are taken from the code backing the

    http://services.w3.org/htmldiff

website. `htmldiff` is the Python driver script. `htmldiff.pl` is the
Perl script which generates the diff after preprocessing of the input
HTML by `htmldiff`. `htmldiff.orig` is the original Python script from
the website, modified to run at the command line instead of as a CGI
script.