DocBook Assembly Stylesheets
==============================
bobs@sagehill.net

This directory provides XSL stylesheets for working with
DocBook assemblies.  It is intended to enable working with
<topic> and <assembly> elements, as defined in DocBook 5.1
and later.

This kit currently supports most features of an assembly.
See the "Unsupported Features" section below for details
of what is not currently supported.  These more advanced
features will be supported as it is further developed.


Content of this directory:
--------------------------
topic-maker-chunk.xsl        - stylesheet to modularize an existing DB5 document.
topic-maker.xsl              - imported by topic-maker-chunk.xsl.
assemble.xsl                 - stylesheet to process an <assembly> into a document.


The toolkit consists of an assemble.xsl XSL stylesheet
to process a DocBook <assembly> element to convert it
to an assembled DocBook 5 document ready to be formatted.
This stylesheet will enable users to structure a book from
modular files.

To make it easy to initially create a modular book, this
kit also includes a topic-maker-chunk.xsl XSL stylesheet
to break apart an existing DocBook 5 book into modular
files, and also create the associated <assembly> document.
Then you can run the assemble.xsl stylesheet to put it
back together as a single DocBook document.


To create an assembly and topic files from a book or article document
=======================================================================

If you have an existing DocBook 5 book or article document,
you can convert it to an assembly and a collection of
modular topic files.  If you want to convert a DocBook 4
document, you must first convert it to version 5.

For example, to disassemble a DocBook 5 book document named book.xml:

xsltproc --xinclude \
         --stringparam assembly.filename myassembly.xml \
         --stringparam base.dir topics/  \
         topic-maker-chunk.xsl  \
         mybook.xml

This command will result in a master assembly file named
'myassembly.xml' with a root element of <assembly>, containing
a single <structure> element. It will also break up the
content of the book into modular chunks that are output
to the 'topics/' subdirectory as specified in the 'base.dir'
parameter.

Options
----------
The name of the assembly file is set by the stylesheet param
named 'assembly.filename', which should include the filename suffix.

Modular files are output to the directory location specified
by the 'base.dir' parameter.  If you want them in the current
directory, then don't set that  param.

By default the assembly element is output to the current
directory, *not* into base.dir with the modular files.
The <resources> element in the assembly has its xml:base
attribute set to the value of 'base.dir', so that it is
added to the paths to the modular files when processed.
If you set the stylesheet param 'manifest.in.base.dir'
to 1, then the assembly file is created in the base.dir
directory and the xml:base attribute is omitted (since
they are together in the same directory).

If you want the assembly file in 'base.dir' instead of
the current directory, then set the stylesheet param
'manifest.in.base.dir' to 1.

The stylesheet chunks a document into modules at the
same boundaries as the chunking XHTML stylesheet, because
it reuses many of the chunking stylesheet templates.
You can alter the chunking behavior with the same options
as for XHTML chunking.

For example, the stylesheet will chunk sections into topics
down to section level 3 by default.  To change that level,
change the stylesheet param 'chunk.section.depth' to
another value.

Finer control of chunking can be achieved by using
the <?dbhtml stop-chunking?> processing instruction in
the source file.

Many modular elements retain their original element name,
such as glossary, bibliography, index, and such.  By default, the
stylesheet converts chapter, article, preface and section elements
into <topic> modules.  To change that list of
converted element names, alter the stylesheet param named
'topic.elements'.  If that param is empty, then no elements
will be converted to <topic>, so they will all retain their
original element names.

Modular filenames use the same naming scheme as the chunking
XHTML stylesheet, and supports the same file naming options such as
the param 'use.id.as.filename', which is set to 1 by default.
Note that the stylesheet param 'html.ext' is set to '.xml'
because it is producing modular XML files, not HTML files.

Root element conversion
------------------------
By default, the root element of the original document is 
also converted to a module, and <structure> gets a resourceref
attribute to reference it.  If you set the stylesheet
param 'root.as.resourceref' to zero, then the root element
is handled differently, as described as follows.

If the structure element does not have a resourcref
attribute, the root element is constructed rather
than copied from a resource. The structure element must
have a renderas attribute (or its child output element must
have such) to select the output root element name.  

Any content between the root element start tag and the
first module is put into a resource with the original
root element.  To pull this content in, the first
module in the structure points to this resource but
uses a contentonly="yes" attribute. The effect of
that attribute is to pull in all content *except*
the root element of that resource.

In general, if you have content that does not logically
have its own container element, you can put the content
into a suitable container element and then deselect the
container element upon assembly with the contentonly="yes"
attribute.  That attribute can also be used to avoid
pulling in a resource's xml:id when you want to change it.


To process an <assembly> into an assembled DocBook document
==============================================================

To convert an <assembly> and its associated modular
files into a single DocBook document, process
your assembly document with the assemble.xsl stylesheet.
You should then be able to process the resulting
document with a DocBook XSL formatting stylesheet.


Useful params in assemble.xsl
-----------------------------
The $root.default.renderas param sets the name of the
root element of the assembled document, if it is not
otherwise specified with @renderas.  Its default value
is 'book'.

The $topic.default.renderas param sets the name of the
output element for any topic element included in the
assembly, if it is not otherwise specified with
@renderas.  It's default value is 'section'.

The $structure.id param lets you specify at runtime
the id value of the structure you want to reassemble.
This is only necessary if you have more than one
structure element in your assembly.

The $output.type param also lets you specify at runtime
which structure element to process.  In this case,
the value should match on an @type attribute on
the structure element.

The $output.format param lets you specify at runtime
which of several possible output formats are being generated.
The param value is compared to the @format
attribute on <output> elements to select specific properties
for a module.


Unsupported Features
-----------------------

The transforms and transform elements are currently ignored
by the assembly stylesheet.

The relationships and relationship elements are currently
ignored by the assembly stylesheet.

The filterin and filterout elements are not currently
supported.