.. -*-RST-*-
=====================================
CL-DOCUTILS: Docutils for Common Lisp
=====================================
:Author: Dr John A.R. Williams
:Contact: J.A.R.Williams@jarw.org.uk
:created: 22 September 2009
:status: Draft
:version: 0.1
:copyright: © 2009 J.A.R. Williams
:Abstract: |CDU| is a Common Lisp implementation of the Docutils text
processing system for processing plaintext into presentational
formats such as HTML and LaTeX. It is based upon the Python Docutils
reference implementation but uses Common Lisp idioms making it
easier to extend and more flexible. As with the reference
implementation it includes a parser for the `reStructured text`_
plaintext markup syntax which is suitable for marking up
documentation and for use as user markup for collaborative web
sites. It is successfully used to support a
higher education peer-review assessment and online tutorial system.
.. meta::
:keywords: reStructuredText, Docutils, text, Common Lisp
.. contents:: Table of Contents
.. |CDU| replace:: CL-DOCUTILS
.. _restructured text: http://docutils.sourceforge.net/docs/ref/rst/
.. _Python: http://www.python.org/
.. _FlexiStreams: http://weitz.de/flexi-streams/
Download and Installation
=========================
|CDU| together with this documentation can be downloaded from
<http://www.jarw.org.uk/lisp/cl-docutils.tar.gz>. The
current version is 0.1.1.
|CDU| comes with a system definition for `ASDF
<http://www.cliki.net/asdf>`_ and is compiled and loaded in the usual
way. It depends upon `CL-DATA-FORMAT-VALIDATION
<http://www.jarw.org.uk/lisp/cl-data-format-validation.html>`_ and
`<CL-PPCRE <http://weitz.de/cl-ppcre/>`_. A git repository is
available at <http://www.jarw.org.uk/lisp/cl-docutils.git>. It has
been developed and tested using SBCL but should be portable to any
ANSI-compliant implementation of Common Lisp.
|CDU| is made available under the terms of the GPL v3 license - see
the file ``LICENSE.txt`` for details.
Support
=======
For questions, bug reports, feature requests, improvements, or patches
relating to this implementation please email <J.A.R.Williams@jarw.org.uk>.
See `reStructured text`_ for the wide range of documentation on the
restructured text syntax and `Docutils
<http://docutils.sourceforge.net/index.html>`_ for information on the
reference implementation on which this is based.
Convenience Functions For Users
===============================
Although this is intended as a library for embedding in other
applications the following functions in the DOCUTILS package are
provided as examples suitable for direct use:
- function **read-rst** `input => document`
Read and parse restructured text from an input (a stream, string or
pathname) and return the parsed document object.
- function **write-html** `output document`
Write the parsed document to output (a stream or pathname) in html
format.
- function **write-latex** `output document`
Write the parsed document to output (a stream or pathname) in LaTeX
format.
API For Developers
==================
This documentation is in early stages and those wishing to use the
library are encouraged to read the documentation for the Python Docutils
implementation to get a flavour and examine the code here to determine
differences. The generic interface is largely defined and documented
in the file ``publisher.lisp`` which is a good place to start. Queries
to the author as mentioned under support are also welcome.
---------
Data Flow
---------
The following diagram shows the components and data flow::
-----READER----
/ \
/ \
SOURCE ---> PARSE ---> TRANSFORM ---> WRITER ---> OUTPUT
------
Source
------
Specialised methods of the API are already provided for `stream`,
`pathname` and `string` source types. It is the responsibility of the
source and output streams to handle external character encodings.
- generic function **new-document** `source => document`
Implementations of this should create a new document instance with
the appropriate non-default document settings from source.
The default method loads no settings and the method specialised on
pathname source will load settings from `*standard-config-files*`
and `cl-docutils.conf` in the pathname directory.
- generic function **read-lines** `source => vector`
Given a source, return a vector of the text lines from source for
subsequent parsing.
- generic function **settings** `source => hash-table`
Return a hash table of processing settings to be associated with
documents with this source.
- function **register-settings-spec** `specifications => specifications`
Register a list of setting specifications. Each entry corresponds to
one setting and has the form `(name type default description)`
where `name` is a keyword symbol used to look up that setting,
`type` the data format validation type specification, `default` a
default value to be used if unspecified and `description` a string
description suitable for users.
Specifications are normally given in text files in internet header
style with the setting name, followed by a colon, followed by the
textual representation of the value. Continuation lines are supported.
-----
Parse
-----
- generic function **read-document** `source reader => document`
Read a document from `source` using `reader` and using the
transforms associated with this reader return a fully parsed
and transformed document.
The following restructured text specific API is exported from package
``docutils.parser.rst``.
- class **rst-reader**
Class name of the standard restructured text parser.
- macro **def-role** `((name textvar &rest argument-list) &body body)`
Define a new interpreted
role element handler for role with canonical name `name`.
`textvar` is the name of the variable holding the textual argument
to be referenced in `body` and `argument-list` is as follows::
lambda-list::= ({var | (var [[specializer] [default]])}*
[{{&content {var [[specializer] [default]]}}] )
`body` is the list of forms which should return a `docutils:text`
node instance to be substituted into the document
- macro **def-directive** `((name parent &rest argument-list) &body body)`
Define a new directive element handler with name `name`. Similar to
``def-role`` except the `body` is responsible for adding nodes to
the `parent` node.
---------
Transform
---------
- generic function **transforms** `reader => list`
Return a list of transform specifications (class names) to be
applied by this reader after parsing.
- generic function **transform** `transform`
Evaluate the given transform instance on its node.
- generic function **do-transforms** `transforms document => document`
Apply a list of transformation to the document in priority order.
------
Writer
------
- class **writer**
Base class for all writer instances.
- generic function **visit-node** `writer node`
Process the node updating `writer` parts accordingly. Visiting a
document node will clear all document parts first.
- generic function **write-document** `writer document destination`
Use `writer` to visit the `document` and write all parts in order
to `destination`
- generic function **write-part** `writer part destination`
Write a given named document part from a `writer` to `destination`.
- generic function **settings** `writer` => settings
Return an alist of settings associated with this specific writer
instance which will override the document settings during `writer`
processing.
- macro **with-part** `((part-name) &body)` => result
Specify dynamically the document part to be written to using
`part-append` and `part-prepend`.
- function **part-append** `&rest values`
function **part-prepend** `&rest values`
These function append or prepend their arguments to the currently
set part of the current writer.
- class **docutils.writer.html:html-writer**
The base class for writing a document out in html format. The
provided file ``docutils.css`` provides an example style sheet for
the generated documents.
- class **docutils.writer.latex:latex-writer**
The base class for writing a document out in latex format. The
provided file ``docutils.tex`` should be placed in the latex search
path as it is required by the produced latex documents.
---------
Variables
---------
- variable **`*unknown-reference-resolvers*`**
List of functions to try to resolve unknown references during
the transformation process. This can usefully be set to, for
example, check against a database of pages in a live online system and
return the relevant reference into the document.
- variable **`*standard-config-files*`**
List of pathnames to files to be parsed in order for document
configuration settings. Defaults include ``/etc/cl-docutils.conf``
and ``~/.cl-docutils.conf``. A pathname source will also check for
``cl-docutils.conf`` file in the same directory.
- variable **`*default-reader-package*`**
Value to bind `*package*` to when reading lisp forms from the
document.
- variable **`*evaluator*`**
Name of function to use to evaluate lisp forms in evaluateable
nodes/
- variable **`*evaluation-error-hook*`**
Function to be called if there are errors when calling `*evaluator*`
Differences from Reference Implementation
=========================================
Character Sets
This Common Lisp implementation uses standard Lisp string handling and
the CL-PPCRE regular expression library. Character set translation on
input and output is orthogonal to this processing and can be formed
using the appropriate specialised stream libraries such as
FlexiStreams_.
Nested Inline Elements
This implementation supports nested inline elements (where they make
sense) whereas the reference implementation did not.
Error Messages
Error messages is handled differently in this implementation with the
result that they may not be located in exactly the same position
relative the cause as in the Python implementation.
Python Enhancement Proposals
Parsing of Python Enhancement Proposals (PEPs) is not supported as
these have significance only for the Python community, however PEP
references are supported as to enable the use of the stand-alone
test documents provided with the reference Python implementation.
Customisation
=============
A fundamental requirement and important feature of restructured text
is that it can easily be customised and extended for use in embedded
applications (particularly web applications) without changing the core
language. Restructured text was explicitly designed with this in
mind, however the Common Lisp implementation is quite different,
and more powerful in this regard than the reference Python
implementation.
Directives
The major feature for language extension at the block level is by the
use of the named directive explicit markup. New directives can be
defined using the ``def-directive`` macro.
Roles
Roles provide a means of implementing new inline elements. They are
defined using the ``def-role`` macro.
Transformations
Transformations are associated with the parser element. If you wish
new transformations to be run, use a subclass of the parser which
return the appropriate list of transformations.
-------------------
Additional Elements
-------------------
- role **math**
This role allows the embedding of mathematical expressions using
LaTeX syntax. An implict syntax, delimiting the argument with '$$' is
also provided.
- directive **equation**
This directive allows the embedding of equations in LaTeX format.
- role **eval**
A role allowing the embedding of inline evaluateable expressions in
Lisp syntax
- directive **evaluation**
A directive allowing embedding of expressions to be evaluated -
Currently only Lisp expressions are supported.
Acknowledgements
================
The many contributers to the original Docutils project on which this based
and the many examples of code made available by Common Lisp community
which have inspired and taught me. Particular mention should go to Edi
Weitz for making available the CL-PPCRE used here and the Flex-Streams
library which can be used to deal with external character coding issue.