Identitity Template Problem - Best Practices - What order to makechanges

J

johkar

I have source XML that is 5-10 MB in size. I am transforming it from
XML to XML. I am having some issues with overwriting changes that I
have already made. I have come to the realization when using template
match you need to do things in a certain order so as to not overwrite
changes already made. For example if you first remove all currency
formatting on the entire document and then copy nodes containing
formatted currency to a new level...the currency formatting is back.

There are five things I need to do with the XML...at various levels
and in more than one place...nothing too complicated:

1) Replace element names
2) Make elements child elements by adding a brand new parent node.
3) Remove elements entirely
4) Add new elements/values
5) Remove all currency formatting

My questions are:
1) Do I work from the root element and go deeper so that I don't
overwrite...in general what is the best strategy?
2) Do I have one identity template and one primary template match that
I put all my apply-templates within so that nothing is copied to the
output until all changes are made?
3) How can I use a common template to strip currency formatting for
any section of the XML that I might also be doing other changes too?
There is probably an easier template remove formatting than the one I
am using below.

<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
version="1.0">
<xsl:eek:utput method="xml" version="1.0" encoding="UTF-8" indent="yes"/
<!-- COPY EVERYTHING -->
<xsl:template match="node() | @*">
<xsl:copy>
<xsl:apply-templates select="@* | node()"/>
</xsl:copy>
</xsl:template>

<!-- REMOVE CURRENCY FORMATTING -->
<xsl:template match="text()[starts-with(.,'$') or starts-with(.,'($')
or starts-with(.,'-$') or contains(.,',')]">

<!-- REPLACE LEADING PARENTHESIS WITH MINUS SIGN AND REMOVE COMMAS --
<xsl:variable name="txt" select="translate(translate
(.,',',''),'(','-')"/>

<xsl:when test="starts-with($txt,'$') or starts-with($txt,'-$')">
<xsl:value-of select="translate($txt,translate
($txt,'0123456789.-',''),'')"/>
</xsl:when>
<!--FOR FIELDS WITHOUT DOLLAR SIGNS, TEST TO SEE IF NUMBER-->
<xsl:when test="number($txt) = number($txt)">
<xsl:value-of select="$txt"/>
</xsl:when>
<!--PASS THROUGH TEXT BECAUSE IT IS NOT A NUMBER-->
<xsl:eek:therwise>
<xsl:value-of select="."/>
</xsl:eek:therwise>
</xsl:choose>
</xsl:template>
</xsl:stylesheet>
 
J

Joe Kesselman

johkar said:
1) Do I work from the root element and go deeper so that I don't
overwrite...in general what is the best strategy?

You can't "overwrite" using XSLT. Things are output in the order, and
nesting, that the stylesheet outputs them in. Focus on what you want to
output and where you want to draw the information from. If the output
document will follow roughly the same structure as the input document, a
simple recursive tree walk via the xsl:apply-template operation and a
template for each node that needs special handling (typically with the
identity template as a catch-all for those nodes which can be copied
through unchanged) is best practice. In those places where the output
diverges from that structure, write the stylesheet logic "pull-style" --
generate the appropriately nested elements/attributes/text, with
xsl:value or attribute value templates used to explicitly retrieve the
information.
2) Do I have one identity template and one primary template match that
I put all my apply-templates within so that nothing is copied to the
output until all changes are made?

XSLT does not "make changes". It generates a new output document, which
is typically (but not guaranteed to be) written to the output as
stylesheet processing takes place. If that isn't acceptable, I'd suggest
you have the stylesheet write to a buffer of some variety -- eg, build a
DOM tree -- which you can then process atomically.
3) How can I use a common template to strip currency formatting for
any section of the XML that I might also be doing other changes too?

See the xsl:call-template operation, which essentially lets you use a
named template as a subroutine for other templates.
 
J

johkar

You can't "overwrite" using XSLT. Things are output in the order, and
nesting, that the stylesheet outputs them in. Focus on what you want to
output and where you want to draw the information from. If the output
document will follow roughly the same structure as the input document, a
simple recursive tree walk via the xsl:apply-template operation and a
template for each node that needs special handling (typically with the
identity template as a catch-all for those nodes which can be copied
through unchanged) is best practice. In those places where the output
diverges from that structure, write the stylesheet logic "pull-style" --
generate the appropriately nested elements/attributes/text, with
xsl:value or attribute value templates used to explicitly retrieve the
information.


XSLT does not "make changes". It generates a new output document, which
is typically (but not guaranteed to be) written to the output as
stylesheet processing takes place. If that isn't acceptable, I'd suggest
you have the stylesheet write to a buffer of some variety -- eg, build a
DOM tree -- which you can then process atomically.


See the xsl:call-template operation, which essentially lets you use a
named template as a subroutine for other templates.

Thanks for the reply. I think my main issue is that I was using push
style ALL the time and using separate templates out of the XML
structure sequence. I think I need to mix in push style when
appropriate. I am familiar with call-template but still have some
confusion on how one would use it to remove the currency formatting on
any text() that meets the criteria. Can you provide any more
guidance?
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Members online

Forum statistics

Threads
473,744
Messages
2,569,484
Members
44,903
Latest member
orderPeak8CBDGummies

Latest Threads

Top