refdb-users

[Refdb-users] FO stylesheet problem #2: docbook xml book converts to invalid FO

[David N. <dav...@bi...>]

Hi all,

I've just attempted for the first time to produce pdf output from a 
docbook xml book (my previous efforts being with docbook xml _articles_).

Here is the basic document, without any refdb features:
........................................................................................
<?xml version="1.0"?>
<!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook XML V4.2//EN" 
"http://www.oasis-open.org/docbook/xml/4.2/docbookx.dtd">
<book>
    <bookinfo>
        <title>Test of Reference Name</title>
        <author>
            <surname>Nebauer</surname>
            <firstname>David</firstname>
        </author>
    </bookinfo>
    <chapter id="bogus">
        <title>Bogus Reference</title>
        <para>Here's a reference to [... citation here ...] involve the 
refdb machinery.</para>
    </chapter>
</book>
........................................................................................

This document easily converts to a 5-page pdf document with saxon and fop.

If I now add the bibliography entity declaration, a citation and the 
bibliography entity itself, I get the following document:
........................................................................................
<?xml version="1.0"?>
<!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook XML V4.2//EN" 
"http://www.oasis-open.org/docbook/xml/4.2/docbookx.dtd" [
<!ENTITY bibliography SYSTEM "bookref.bib.xml">
]>
<book>
    <bookinfo>
        <title>Test of Reference Name</title>
        <author>
            <surname>Nebauer</surname>
            <firstname>David</firstname>
        </author>
    </bookinfo>
    <chapter id="bogus">
        <title>Bogus Reference</title>
        <para>Here's a reference to <citation 
role="REFDB">lamport1994</citation> to involve the refdb machinery.</para>
    </chapter>
    &bibliography;
</book>
........................................................................................

Using the refdbnd Makefile to convert to pdf produces a valid .bib file:
........................................................................................
<!-- <?xml version="1.0" encoding="ISO-8859-1"?>
<!DOCTYPE bibliography PUBLIC "-//OASIS//DTD DocBook XML V4.1.2//EN" 
"http://www.oasis-open.org/docbook/xml/4.1.2/docbookx.dtd"> -->
<bibliography role="refdb">
<title>Reference List</title>
<bibliomixed id="IDlamport1994" role="BOOK">
<bibliomset role="intext" id="IDlamport1994X">(<bibliomset 
relation="author"><surname>Lamport</surname></bibliomset>, <pubdate 
role="primary">1994</pubdate>)</bibliomset>
<bibliomset role="intextsq" id="IDlamport1994S">(<bibliomset 
relation="author"><surname>Lamport</surname></bibliomset>, <pubdate 
role="primary">1994</pubdate>)</bibliomset>
<bibliomset role="authoronly" id="IDlamport1994A"><bibliomset 
relation="author"><surname>Lamport</surname></bibliomset></bibliomset>
<bibliomset role="authoronlysq" id="IDlamport1994Q"><bibliomset 
relation="author"><surname>Lamport</surname></bibliomset></bibliomset>
<bibliomset role="yearonly" id="IDlamport1994Y">(<pubdate 
role="primary">1994</pubdate>)</bibliomset>
<bibliomset role="bibliography" id="IDlamport1994B"><bibliomset 
relation="book"><bibliomset relation="author"><surname>Lamport</surname> 
<firstname>L.</firstname></bibliomset>, </bibliomset><bibliomset 
relation="book"><pubdate role="primary">1994</pubdate>, 
</bibliomset><bibliomset relation="book"><title role="BOOK">LaTeX: A 
Document Preparation System</title> (</bibliomset><bibliomset 
relation="book"><publishername>Addison-Wesley</publishername>, 
</bibliomset><bibliomset 
relation="book"><address><city>Massachusetts</city></address>) 
</bibliomset></bibliomset></bibliomixed>
<bibliomixed role="multixref"></bibliomixed></bibliography>
........................................................................................

It produces an FO file, but that file is not valid.  Neither passivetex, 
fop or xep would convert it to pdf.  passivetex became terminally 
confused and produced gibberish.  Here are the error messages for fop 
and xep:

FOP:
........................................................................................
[ERROR] 
file:/home/david/data/computing/docbook-xml/learn/bookref/bookref.fo:2:56273 
master-reference '' for fo:page-sequence matches no simple-page-master 
or page-sequence-master
........................................................................................

XEP:
........................................................................................
(document [system-id 
file:/home/david/data/computing/docbook-xml/learn/bookref/bookref.fo]
  (validate
    [error] 
file:/home/david/data/computing/docbook-xml/learn/bookref/bookref.fo: 
line 2: Attribute 'master-name' cannot occur at element 'fo:page-sequence'.
    [error] 
file:/home/david/data/computing/docbook-xml/learn/bookref/bookref.fo: 
line 2: Attribute 'master-reference' is required for 'fo:page-sequence'.
    [validation total: 2 errors]
Parse error: Invalid XSL FO source 
'file:/home/david/data/computing/docbook-xml/learn/bookref/bookref.fo': 
2 error(s) found during validation
.........................................................................................

Both of them are complaining about the same error: an illegal attribute 
('master-name') in element 'fo:page-sequence'.  xep also thinks a 
required attribute ('master-reference') is missing from the same element.

Here is the offending bit of the FO file.  Look for the 
<fo:page-sequence id="id2562418" ...> element and see the master-name 
attribute.  I have added some context to show that this element occurs 
in the FO file after the main text and before the reference list 
heading.  You can also see the duplication of the text: "Reference List".
........................................................................................................
Here's a reference to <fo:basic-link 
internal-destination="IDlamport1994">(Lamport, 1994)</fo:basic-link> to 
involve the refdb 
machinery.</fo:block></fo:flow></fo:page-sequence><fo:page-sequence 
id="id2562418" hyphenate="true" master-name="back" language="en"> [... 
snip ...] <fo:block>Reference List</fo:block> [... snip 
...]<fo:block>Reference List</fo:block> [... snip ...] <fo:block 
id="IDlamport1994" space-before.optimum="1em" 
space-before.minimum="0.8em" space-before.maximum="1.2em">Lamport L., 
1994, LaTeX: A Document Preparation System (Addison-Wesley, 
Massachusetts) </fo:block>
........................................................................................................

This error occurs with both saxon-xerces and xsltproc, which would seem 
to rule out an xslt processor-specific bug.

Since the source file converts to html perfectly the problem appears to 
lie with the refdb fo stylesheets.  Since docbook xml articles do not 
appear to produce the same error, it involves something specific to 
docbookx books -- or at least not affecting docbookx articles.  Once 
again, however, I am insufficiently knowledgeable about xsl (and fo) to 
determine the cause of the problem.

Does anybody have an idea what is happening here?  Has anyone ever 
produced a pdf file from a docbook xml book using refdb?

Regards,
David.

[Refdb-users] FO stylesheet problem #2: docbook xml book converts to invalid FO

[Markus H. <mar...@mh...>]

David Nebauer writes:
 > Since the source file converts to html perfectly the problem appears to 
 > lie with the refdb fo stylesheets.  Since docbook xml articles do not 
 > appear to produce the same error, it involves something specific to 
 > docbookx books -- or at least not affecting docbookx articles.  Once 
 > again, however, I am insufficiently knowledgeable about xsl (and fo) to 
 > determine the cause of the problem.
 > 

There's a few things we can do to address this problem. First of all,
your hand-coded test document #1 does not contain a bibliography. I
don't expect this to be the reason, but you should probably provide a
hand-coded bibliography to test the stock DocBook stylesheet
bibliography code.

Second, you could attempt to process the document using the stock
DocBook stylesheets. The result will be more than ugly, but this could
help to isolate the problem. If the RefDB driver files are the
culprit, the DocBook fo stylesheets should transform the document
without errors, regardless of how nasty the output is going to look.

Third, I suspect that a version mismatch might cause the problem. In
contrast to the more or less static DSSSL stylesheets, the XSLT
stylesheets have changed quite a bit since I implemented the RefDB
driver files. It is quite likely that the templates that the driver
files override have to be updated to a recent version of the stock
DocBook stylesheets. I'll try to find some time next week to look into
this.

In general, the bibliography code in the fo stylesheets is a lot more
complex than in the html stylesheets, as it has to take care of things
like recto/verso, proper pagination and such. Therefore it is not
surprising that it is easier to confuse than the html output.

regards,
Markus

-- 
Markus Hoenicka
mar...@ca...
(Spam-protected email: replace the quadrupeds with "mhoenicka")
http://www.mhoenicka.de

Re: [Refdb-users] FO stylesheet problem #2: docbook xml book converts to invalid FO

[David N. <dav...@bi...>]

Hi Markus,

<quote who="Markus Hoenicka">
> David Nebauer writes:
>  > Since the source file converts to html perfectly the problem appears
> to  lie with the refdb fo stylesheets.  Since docbook xml articles do
> not  appear to produce the same error, it involves something specific
> to  docbookx books -- or at least not affecting docbookx articles.
> Once  again, however, I am insufficiently knowledgeable about xsl
> (and fo) to  determine the cause of the problem.
>
> Second, you could attempt to process the document using the stock
> DocBook stylesheets. The result will be more than ugly, but this could
> help to isolate the problem. If the RefDB driver files are the
> culprit, the DocBook fo stylesheets should transform the document
> without errors, regardless of how nasty the output is going to look.

I didn't mention it in my message, but I did process the <filename>.xml,
file produced by refdbxml, using the stock docbookx stylesheets.  It
transformed the document into valid FO and subsequently to valid PDF. 
Although it was, as you say, more than ugly.

I guess that leaves us with the third option of a stylesheet version
mismatch.

It's so frustrating to have a workable toolchain just out of reach...

I'll keep prodding away at the problem, but for now it's beyond my skill
level to debug.

Regards,
David.

Re: [Refdb-users] FO stylesheet problem #2: docbook xml book converts to invalid FO

[Markus H. <mar...@mh...>]

David Nebauer writes:
 > I guess that leaves us with the third option of a stylesheet version
 > mismatch.
 > 

I've tried this option and updated the stylesheet code to match the
current DocBook-xsl verison 1.65.0. Seems to work alright now. I've
checked in the updated version (1.5). I've also sent you a copy by
private mail. Please try and see whether it fixes your problems as
well.

regards,
Markus

-- 
Markus Hoenicka
mar...@ca...
(Spam-protected email: replace the quadrupeds with "mhoenicka")
http://www.mhoenicka.de

Re: [Refdb-users] FO stylesheet problem #2: docbook xml book converts to invalid FO

[David N. <dav...@bi...>]

Hi Markus,

>David Nebauer writes:
> > I guess that leaves us with the third option of a stylesheet version
> > mismatch.
> > 
>
>I've tried this option and updated the stylesheet code to match the
>current DocBook-xsl verison 1.65.0. Seems to work alright now. I've
>checked in the updated version (1.5). I've also sent you a copy by
>private mail. Please try and see whether it fixes your problems as
>well.
>  
>

That certainly fixed the problem of the docbookx book not printing, 
which refdb is now doing just fine.

Amazingly, it also fixed the duplicate "Reference List" bug!

At this point every one of my major concerns has been met, and very 
promptly too.  Thanks for all your help.  For what it's worth, I now 
have a fully functional XML to PDF/HTML toolchain with full 
bibliographic support.  I don't know what inspired you to embark on the 
refdb project, or keeps you going, but long may it do so.

Time to tackle CVS and download an up-to-date version with all the 
changes rolled in.

Regards,
David.