FMAS wrote:
> On 4 Feb., 00:03, Peter Flynn <peter.n...@m.silmaril.ie> wrote:
>> FMASwrote:
>>> I am transforming XML files which sometimes have untaged text between
>>> elements.
>>> It looks like this:
>>> <ut Style="external" DisplayText="bb">&lt;bb &quot;Document Numbering
>>> Formats&quot;&gt;</ut>
>>> <ut Style="external" DisplayText="bb">&lt;bb &quot;Custom Footnote
>>> Numbering&quot;&gt;</ut>*†‡
>>> <ut Style="external" DisplayText="bb">&lt;bb &quot;Volume
>>> Numbering&quot;&gt;</ut>
>>> No problem to transform the content of tags. But I do not find a way
>>> to copy untagged text such as "*†‡" in the example above.
>>> Any suggestion?
>> a) This looks like rather poor design by someone who really
>>     hasn't grasped the idea of XML properly.
>>
>> b) It also appears that they are trying to hide data marked in
>>     another language within the XML.
>>
>> c) Use the text() function to access character data in Mixed Content
>>     (which is what this resembles).
>>
>> ///Peter
>> --
>> XML FAQ:http://xml.silmaril.ie/- Zitierten Text ausblenden -
>>
>> - Zitierten Text anzeigen -
> 
> I agree that this is ugly XML but it get parsed witout errors. 

Well-formed documents will parse without errors. Assuming there is an 
outer containing element to your example, it's well-formed. But 
well-formed doesn't mean well-designed. I had a car once that was 
well-formed (it had four wheels and it went forwards and backwards) but 
I wouldn't have called it well-designed :-)

> The
> untaged text varies. It may be also numbers for chapters etc. I also
> thougt about the text() function, but it doesn' seem to help here as
> we are talking about input and not output. 

Not relevant. As Dmitre pointed out, text() isn't a function even though 
it looks like one: I was trying to oversimplify (usually a bad idea :-) 
It's actually a node type which will match any stretch of unmarked text.
[That too is an oversimplification, but it will do for the moment.]

> I have a structure like this:
> 
> <tag>text I can get</tag>text I can't get<tag>text2 I can get</tag>

The content of the two tag elements can -- as you have presumably found 
-- be accessed with a template matching "tag". [Please note the 
difference between a tag and an element: it's very difficult to explain 
anything using the wrong terms.]

As you don't give the name of the containing element it's harder to give 
an example of how to access the text between them, but let's assume that 
what you *really* have is approximately:

<root>
   <foo>text I can get</foo>text I can't get<foo>text2 I can get</foo>
</root>

There are three unmarked text nodes in the root element:

a) newline-space-space (before the first foo start-tag)
b) text I can't get (between the foo elements)
c) newline (after the last foo end-tag)

A template matching root/text() will access them. Making it 
root/text()[2] will give you (b) only.

> I wonder if I have to preprocess the XML file, maybe with perl. But if
> it can be avoided so much the better.

XSLT can do this unaided.

///Peter
-- 
XML FAQ: http://xml.silmaril.ie/