[Scribus] DocBook/xml

James Gilmore james.d.gilmore
Fri Mar 16 04:58:05 CET 2007

On 3/15/07, Timo Stollenwerk <timo at zmag.de> wrote:

> > Hi Calum, Hi Steve,
> >
> > i've wrote some xslt stylesheets as a proof-of-concept to transform the
> > scribus fileformat into something a cms can handle easily:
> >
> > http://wiki.scribus.net/index.php/Draft_of_end-to-end_publishing_solution#Use_Case_.232_Publishing_Workflow_for_Magazine_Layout_.28Timo_Stollenwerk.29
> >
> > Until now this approach is independent from the choosen cms. Please
> > contribute your usecases to the wiki (if you don't have write access,
> > mail them to me), so that we can decide together what cms fits best for
> > our needs.

Yes! This is right on target. Barring a tool to import/parse/export docbook directly within Scribus, this is the best option. The ability to convert between .sla and xml in general is useful, but docbook is king, on account of bibliographies, tables of contents, glossaries, etc. 
I'm very interested in those stylesheets. Did you upload them or are they available? 

Recently an associate and I sat down to discuss this very issue. In formatting a series of books for print, we eventually decided that:
1. Docbook is the only format that makes reasonable sense, for at least 90% of publishing.
2. There is not currently a mechanism for importing Docbook into Scribus. By importing, I mean assigning the same styles to tags that are alike. If it doesn't make reasonable decisions regarding styles, it's no better than importing a flat text file.
3. Even if there were, if it didn't go through some sort of xslt-like filtering first, it would be pointless, since the real advantages of Docbook come from parsing tags to form the bibs, tocs, glossaries, sidebars, etc. Again, I'd love to be able to use scribus for styling, but I shouldn't have to go through the document setting styles--I already did that work in the docbook.
4. Once the file is in .sla format, it becomes very impractical to manage content any further, without being able to reliably export a docbook file and corresponding stylesheet.

In the end, we decided (currently) that our only reasonable solution, unfortunately, doesn't involve scribus except for special circumstance situations like children's books(i.e., books where illustrations are the meat of the content.) In the end, it's more practical to puzzle through the mire that is xslt than to try and kludge through a scribus publishing workflow. I believe that the lack of a reasonable workflow is the largest show-stopper where scribus adoption by publishing houses is concerned. Scribus rocks for one-offs and small projects, but without a good workflow that allows (two-way) interaction between scribus and a cms, it can't scale.

The absolute ideal situation would be if scribus could import and export docbook+stylesheet, or could actually parse a docbook file and create the necessary book parts on the fly, so that the docbook file was separate from the .sla file, or if scribus could use a "docbook" template to allow someone to design a style for the docbook elements using scribus' gui and output an .xslt stylesheet containing the formatting, so that the docbook could be run through fop/etc. to create print-ready pdfs.

A sort of "meet-me-halfway" option would be to create an xslt filter to spit out a scrapbook into a directory or file which contained each of the elements of the book, ready to drop into place. We attempted to test the feasibility of this but it seemed problematic. What makes a scrapbook a scrapbook? You seem not to be able to even transport a scrap from one directory to another--it doesn't show up in the new directory.

One thing that surprised me, when looking at a .sla file, was that text was inserted directly into, and separated by, individual text boxes. I expected the pages to be laid out, with container object coordinates, with a pointer to an xml tag containing the entirety of a block of text. I expected kerning and other styles to be applied by xml tags. That would have made creating a workflow easy as cake.

I know this is getting long--I'll try to wrap up. One last option, if importing and exporting docbook is just too problematic, is creating a new format for scribus with similar functionality to the docbook DTD but with the purpose of injecting itself into scribus with styles/kerning/etc. set, creating  bibs/etc. as Docbook parsers do, autogenerating textboxes where needed. The only reason I can see doing this rather than simply using a docbook file, though, is for naming styles.

James Gilmore
GnuPG Key ID: 0x54D1D809

More information about the scribus mailing list