Hi Gannon,

Am 09.04.2010 17:20, schrieb Gannon Dick:
Svante, I thought you had us Americans convinced "Vacation" is a default value 
:)

I wish it would be.. :)
When would you anticipate the switch-over to XSLT 2.0 ?  Does the free Saxon 
handle that or is the warning just to nag you to upgrade?  Dr. Kay might not 
admire my, um, frugality, as much as I do his work.
Yes, the free Saxon handles XSLT 2.0 and although I might need some vacation ahead to decide, I see here a strong opinion on the list to switch, which I do not want to block.
The<head>  meta data needs quite a bit of work. I do not think that the 
itemprop mechanism in HTML 5 is going to make the modeling issues (SKOS, OWL) with 
Linked Data any clearer.
You might reference to an external RDF file from HTML similar as referencing to an external CSS file.
I am working on an OpenOffice to Linked Data "Home Page" scheme for small scale Government Work.  
With 5-6 small steps I can switch to an XSD Schema and tag the meta data and citations important for linking, 
yet preserve the layout style which is easy to read.  The idea is that if you had a pile of forms (raw data) 
you could first redact out the private data then make a "Home Page" for the rest of the data, 
linked with deri's voiD or something similar. The hardest part is to convince *yourself* that commercial 
branding and style tricks behind the scenes are counter-productive.  As I am fond of saying, everybody lives 
near "Berlin", but only some of those Berlins are in Germany.

Looking forward to read your suggestions ;)

Have a nice week-end (mine will be about a week longer)... :)
Svante
--Gannon


--- On Fri, 4/9/10, Svante Schubert - Sun Germany - Software 
Engineer<[email protected]>  wrote:

From: Svante Schubert - Sun Germany - Software Engineer<[email protected]>
Subject: Re: [xml-dev] odf2xhtml
To: [email protected]
Cc: [email protected]
Date: Friday, April 9, 2010, 8:31 AM
Hi guys,

sorry for the belayed response. Coincidently I had been
working on the
XHTML filter lately, fixing some image alignment problem
(using CSS
float to get a similar layout as in OOO). When I bring in
this changes
into OOO, we can work together on the filter.

I like the idea of keeping the inheritance via multiple
styles, it will
lower the burden of resolving inheritance from ODF to HTML.
Years ago it
did not work in the main browsers, did you tested it if it
is now
working in the latest common browsers?

Regarding the switch from XSLT 1.0 to XSLT 2.0, we would
drop the
possibility to use XSLT 1.0 processors.
For instance, I once wrote a prototype of a Firefox viewer
using the on
board Mozilla XSLT processor.
What are the advantages of XSLT 2.0?

Note: When we upgrade Saxon - what I thinking over to do as
there are
some bugs lately fixed, that causes problem in some filter
- there will
no longer Java method access being supported by the free
Saxon Home edition.
As this Java method Access is a none standardized feature I
am willing
to switch to the latest version and dropping this
undocumented feature.

Going to be next week on vacation, soon after that I plan
to start a CWS
where we can work together on the HTML filter.
Martynas could you write please write an issue, that we
might use in the
CWS? Otherwise the QA might oversee your patch.

Thanks in advance,
Svante


On 03/21/10 16:56, Martynas Jusevicius wrote:
I forgot I'm running it on XSLT 2.0 processor...
<xsl:apply-templates select="$parentStyle">  will
probably not work in
XSLT 1.0, but you get the general idea.

On Sun, Mar 21, 2010 at 4:29 PM, Martynas Jusevicius
<[email protected]>
wrote:
Here's a partial fix for this issue, I don't know
if I should try
submitting it somewhere.

It requires modification of
/export/xhtml/body.xhtml stylesheet:
http://framework.openoffice.org/source/browse/*checkout*/framework/filter/source/xslt/odf2xhtml/export/xhtml/body.xsl?content-type=text%2Fplain&rev=1.3.16.3

@text:style-name template needs to be modified and
one recursive
template (inherited-classes) needs to be added:

<xsl:template match="style:style"
mode="inherited-classes">
     <xsl:param name="globalData"/>
     <xsl:variable name="styleFamily"
select="@style:family" />
     <xsl:variable
name="parentStyleName" select="@style:parent-style-name"
/>
     <xsl:variable name="parentStyle"

select="$globalData/office:styles/style:sty...@style:family
=
$styleFamily and @style:name =
$parentStyleName]"/>
     <xsl:value-of
select="translate(@style:name, '.,;: %()[]/\+',
'_____________')"/>
     <xsl:text>  </xsl:text>
     <xsl:apply-templates
select="$parentStyle" mode="inherited-classes">
         <xsl:with-param
name="globalData" select="$globalData"/>
     </xsl:apply-templates>
</xsl:template>

<xsl:template match="@text:style-name |
@draw:style-name |
@draw:text-style-name |
@table:style-name"><!-- |
@presentation:style-name-->
         <xsl:param
name="globalData"/>
         <xsl:variable
name="style"
select="$globalData/office:automatic-styles/style:sty...@style:name
=
current()]"/>
         <xsl:variable
name="styleFamily" select="$style/@style:family" />
         <xsl:variable
name="parentStyleName"
select="$style/@style:parent-style-name" />
         <xsl:variable
name="parentStyle"
select="$globalData/office:styles/style:sty...@style:family
=
$styleFamily and @style:name =
$parentStyleName]"/>
         <xsl:attribute
name="class">

   <xsl:value-of select="translate(., '.,;:
%()[]/\+',
'_____________')"/>

   <xsl:text>  </xsl:text>

   <xsl:apply-templates select="$parentStyle"
mode="inherited-classes">

       <xsl:with-param name="globalData"
select="$globalData"/>

   </xsl:apply-templates>
         </xsl:attribute>
</xsl:template>

There is also at least one other place in body.xsl
(in the text:h
template) where @text:style-name template needs to
be called (with the
$globalData param) for the fix to have effect
instead of simply
outputting style name to the @class attribute.

The full fix would be to prevent excessive
(overlapping in parent and
children) CSS properties from bein generated.

Martynas
odt2epub.com

On Fri, Mar 19, 2010 at 5:55 PM, Vyacheslav Sedov
<[email protected]>
wrote:
good idea - i can help with XSLT 2.0 - our
project have some odf2xhtml
code - our code is XSLT 2.0 - not XSLT 1.0
from OpenOffice
with best wishes,
Slav

2010/3/19 Martynas Jusevicius<[email protected]>:
Hey list,

I've been working with odf2xhtml XSLT
filter a lot lately, in relation
to ePub e-books. It does its job but it's
really showing its age. I
wonder if there are plans for a new
version? It could really benefit
from many of XSLT 2 features.

One of the most annoying things is the
generated CSS. If style A
defines bold text and style B is inherited
from A and additionally
defines right alignment, odf2xhtml
generates such code:
.A { font-weight: bold; }
.B { font-weight: bold; text-align: right;
}
<p class="A">  ...<p
class="B">
That shows up OK in the browser, but
removes the relationship between
A and B in the XHTML/CSS output. For
example, if I need to select all
elements with style inherited from A, I
cannot do that without looking
into the original ODT and examining the
styles.
I think much better output could be
achieved using multiple classes in XHTML:
.A { font-weight: bold; }
.B { text-align: right; }

<p class="A">  ...<p class="A
B">
What do you think?

Martynas
odt2epub.com


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]



---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]



---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]





---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]



---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to