& XSLT On Friday 20 July 2007 17:34, Birgit Kellner wrote:
>  Here's a follow-up to my question from yesterday about parsing
>  footnotes with XML::LibXML.
> 
>  This time, I'm parsing verse lines, so it's a different
>  situation, but the replies to yesterday's query helped me to
>  build a string out of the text nodes in a mixed-content node that
>  is part of a larger structure.

First of all, why do you do that "by hand"? To get all text nodes 
from a subtree nicely concatenated, you can use e.g.

$text = $node-> findvalue('string(.)')

>  I'm recursively parsing from a higher level all the way down,
>  always passing a node to a subroutine. If it is a text node, the
>  subroutine appends its data content to a scalar and ends; if the
>  node has children, the subroutine is called again.
> 
>  This is the code (minus a few attributes):
> 
>  <lg><note><span><l>
>  <seg n="a">......</seg>
>  <seg n="b">......</seg></l><l>
>  <seg n="c">......</seg>
>  <seg n="d">......</seg>
>  </l></span><app>
>  ...
>  ...
>  </app></note></lg>
> 
>  The text nodes are inside the <seg>-elements, which may also
>  contain further <note>-elements, and whatnot.
> 
>  The <l>-elements are verse lines, and their data content needs to
>  be isolated.
>  This is done in two different ways: for the <l>-element
>  containing <seg>s "a" and "b", the trigger is the beginning of
>  <seg> "c". The scalar content is then copied and emptied out.
> 
>  For the <l>-element containing <seg>s "c" and "d", the routine
>  checks if the last <seg> parsed was "d", if  the node in question
>  is a text node and has no next sibling. This means it's the last
>  text node in the last segment of the verse, and thus the second
>  verse line is complete.
> 
>  This last check runs into problems, however, when there is
>  additional whitespace before the closing </lg>-tag:
> 
>  <lg><note><span><l>
>  <seg n="a">......</seg>
>  <seg n="b">......</seg></l><l>
>  <seg n="c">......</seg>
>  <seg n="d">......</seg>
>  </l></span><app>
>  ...
>  ...
>  </app></note>
>  </lg>
> 
>  The newline after the closing </note> tag results in a text node
>  with whitespace as its content. When the script arrives at that
>  node, it logically determines that this is the final text node,
>  has no siblings, and that the last <seg> parsed was "d".
> 
>  I can get around this problem by testing, in addition to checking
>  on the "n"-attribute of the last parsed <seg> and the absence of
>  a next sibling, whether the text node is actually inside <seg>
>  ($el->findnodes('ancestor::seg') - need not necessarily be a
>  parent).
> 
>  But still, I'm intrigued that this actually happens, and was
>  wondering whether there is any way to make XML::LibXML ignore
>  this whitespace.
> 
>  Birgit
> 

Without a DTD or schema, there is no reasonable way to distinguish 
ignorable whitespace and non-ignorable whitespace in XML. You can 
try setting $parser-> keep_blanks(0) before you parse the XML into 
memory and see if it solves you problem.

-- Petr
_______________________________________________
Perl-XML mailing list
Perl-XML@[...].com
To unsubscribe: http://listserv.ActiveState.com/mailman/mysubs