Re: [docbook-apps] DocBook to text

["W. Martin Borgert" <debacle@debian.org>]

On Tue, Jun 10, 2008 at 08:39:03AM +0200, Thomas Schraitle wrote:
> Maybe you could try it with a DB XSLT stylesheet below and go
> from the XHTML output to text:
...
> <!-- No callout graphics: -->
> <xsl:param name="callout.graphics" select="0"/>
> <!-- Let appear callouts as Unicode characters: -->
> <xsl:param name="callout.unicode" select="0"/>

Yes, this makes sense. Currently, I don't have callouts, but I
will add these parameters to my XSL file just in case.

> <!-- Navigation is useless for text output -->
> <xsl:param name="suppress.navigation" select="1"></xsl:param>

If I produce only one big HTML file in the first place, I don't
get navigation links anyway, right?

> Did you also try other text browsers? W3m is one, but there
> is also lynx (it seems to have some problems with tables).

As you can read in my original message:

On Thu, Jun 05, 2008 at 05:00:28AM +0200, W. Martin Borgert wrote:
> elinks (0.11.3), html2text (1.3.2a), links (2.1pre36) and lynx
> (2.8.7dev9) all fail on UTF-8 or tables or both, which is - sad.
> The only text based web browser, which both support UTF-8 (e.g.
> for Chinese, Hebrew etc.) and tables is w3m (0.5.1).

On Tue, Jun 10, 2008 at 08:39:03AM +0200, Thomas Schraitle wrote:
> You can even save the XHTML to text with Firefox and Opera.

IMHO, the output of w3m is better than Firefox, esp. for tables.
Further, I did not yet find out how to make firefox export text
from the command line (or Makefile) and without $DISPLAY. I did
not try Opera - it fails my "free software" requirement.

Thanks anyway for your help!

P.S. Another failed attempt: OpenOffice. The DocBook roundtrip
stylesheets (1.73.2) seem to be too limited. docbook2odf (0.244)
creates a file, which OpenOffice (2.4.0) and unoconv (0.3) fail
to open. odt2txt (0.3) creates text: no tables, broken Japanese.