I am building a new XML out of an XML. But first, I have to perform some validation/translation routines and db read that will build the target text() node. This extra process will "suspend" the XSLT processing so I will have to pass a number of times between XMLs to build a new XML. It's like an array handling routine where the array is populated and read back later.

Assuming, XML_01 contains a starting point element like

<value/>

XML_02 will be iterated. On each pass a new node/path is derived

First pass:

 <field name="field1">data 1</field>

Second pass:

 <field name="field2">data 2</field>

Third pass:

 <field name="field3">data 3</field>


On each pass above, XML_01 evolves into

First pass:

<value>
 <field name="field1">data 1</field>
</value>

Second pass:

<value>
 <field name="field1">data 1</field>
 <field name="field2">data 2</field>
</value>

Third pass - final:

<value>
 <field name="field1">data 1</field>
 <field name="field2">data 2</field>
 <field name="field3">data 3</field>
</value>


The reason I'm doing this is because I need to hold more than 40 or so fields and field values. Having a token to hold all of them individually maybe too cumbersome to maintain.

Thanks for the usual guidance!

I will answer my own inquiry. :D

Array like processing needs another xslt physically residing somewhere that needs to be imported to the one XSLT that builds the target XML.

Which is a luxury that I don't have.

So, I had to amend my validation lookup table by introducing

- use_flag if I need the field in the target XML
- mapping field
- and the node-set() instruction to find the data from the source XML

By putting the Xpath logic in the database, my XSLT code becomes simple and cleaner.