[XSL-LIST Mailing List Archive Home] [By Thread] [By Date]

Re: [xsl] Performance of predicate-based patterns

Subject: Re: [xsl] Performance of predicate-based patterns
From: "Eliot Kimber ekimber@xxxxxxxxxxxx" <xsl-list-service@xxxxxxxxxxxxxxxxxxxxxx>
Date: Wed, 4 Feb 2015 14:26:41 -0000

The DITA for Publishers Word-to-DITA framework
(https://github.com/dita4publishers/org.dita4publishers.word2dita) has a
generic WordML-to-SimpleML transform
er/xsl/wordml2simple.xsl). This generates a simplified and generic form of
"word processing markup" from the WordML.

But because it's handling the whole WordML in a generic way it doesn't
have many templates that use predicates, so not sure it helps here.

The intermediate format puts the style name and ID on the element to which
it applies, so the the templates that process that file are either also
simple or are handled within gnarly for-each-group loops that infer
hierarchy from the flat paragraph structures.

This process is also driven by a separately-defined style-to-tag mapping
document, so there's little or no need for templates that match on
variable properties of the input elements.


Eliot Kimber, Owner
Contrext, LLC

On 2/3/15, 5:34 PM, "Michael Kay mike@xxxxxxxxxxxx"
<xsl-list-service@xxxxxxxxxxxxxxxxxxxxxx> wrote:

>> You will end up with similar match pattern if you try to map Word
>> styles (saved in WordprocessingML) into some XML structure. Style name
>> is stored in a subelement which is two levels down from actual
>> paragraph element. And a lot of publishing companies is processing
>> Word input documents. You will have templates like:
>> <xsl:template match="p[pPr/pStyle/@val = 'Heading 1']">
>> <h1>
>> <xsl:apply-templates/>
>> <h1>
>> </xsl:template>
>We've got a precondition there that it will only match a <p> element, so
>that's a good start. It then depends how many other rules there are that
>also match <p> elements.
>But yes, it would be good to look at some Word-ML stylesheets if anyone
>knows of any. (I've come across a few over the years, but all specific to
>a particular client.)
>Michael Kay

Current Thread