On 6/16/2010 11:53 AM, David Link wrote:
> On a slightly different note, what about roundtripping from DocBook
> XML to HTML and back? Do people have experience with that, or know
> any problems therein?
As I've mentioned in my recent posts I'm a newb to DocBook and XSLT so
my answer maybe somewhat incomplete and naive. The XSLT Cookbook (ISBN
9780596009748) has a chapter on XML to HTML. From my limited knowledge
going from DocBook to HTML seems to be rather straigthforward and there
are a number of stylesheets 'out there' to perform comprehensive Doc
Book to HTML transformations. My preferred approach is to use CSS to
format the (X)HTML and let the DocBook XML <-> XHTML round-trip
transformation just deal with the content.
What I've struggled with more is HTML (which is mostly flat) to DocBook
(which is more hierarchic). I have attached my recent effort to develop
a route from InDesign exported PDF to DocBook XML (I hope the
attachments make it through the list management software). The process
I have at this point is far too simplistic to be useful at present, but
I'm hoping I can get into a state whereby I can downconvert several
thousand InDesign exported PDFs to DocBook with relative ease. My
process is:
InDesign created PDF -> Acrobat exported XML -> PDFXML2HTML XSLT ->
HTML2DBK XSLT -> DocBook XML
It is far from ready to make use of (and is only ever going to work with
my particular files as is), but the bit I found the most problematic was
the transformation from HTML to DocBook because of the flat to nested
hierarchy. Fortunately, XSLT 2.0 makes this much easier with the
for-each-group command, and a XSLT snippet on pages 340-342 of Michael
Kay's incredibly comprehensive XSLT 2.0 and XPath 2.0 4th Edition (ISBN
9780470192740) - I've barely scrapped the surface of the knowledge
contained in this book!
I'm still not using the templates properly yet as my transformation from
<p> to <para> is wrong, the <body> tags are still there - if anyone can
guide me on how to correct these problems that'd be great! The
PDFXML2HTML XSLT is also not ready, but as this is a transformation from
an already flat hierarchy (Acrobat generated XML) to another flat
hierarchy (HTML) it really is nothing much more than a one-to-one
mapping (I had to manually tweak the Eoyang_processed.xml - which is the
output of the PDFXML2HTML XSLT because it doesn't quite do what I want
yet). The flaws in my output files is just a reflection of my limited
knowledge of XSLT, but I'm already excited about the possibilities a
deeper understanding of XSLT is going to allow me to explore... so any
guidance is very much appreciated. At least the final DocBook XML
properly contains the nested <sectX> tags, which I was really struggling
with...
Bye for now, Kurt
