flat XML to hierarchical XML

eduijs · April 28th, 2006, 04:40 AM

I'm having a problem where I have to translate non hierarchical XML into hierarchical XML, and I can't get my head around to how to do it.

My input looks something like this:

Code:

<test>
 <header>
  <adres>adres 1</adres>
 </header>
 <line>
  <art>art 1a</art>
 </line>
 <line>
  <art>art 1b</art>
 </line>
 <header>
  <adres>adres 2</adres>
 </header>
 <line>
  <art>art 2a</art>
 </line>
 <line>
  <art>art 2b</art>
 </line>
 <line>
  <art>art 2c</art>
 </line>
</test>

and I want to transform this into something like this:

Code:

<?xml version="1.0" encoding="UTF-8"?>
<interchange>
 <message>
  <id>1</id>
  <adres>adres 1</adres>
  <details>
   <art>art 1a</art>
  </details>
  <details>
   <art>art 1b</art>
  </details>
 </message>
 <message>
  <id>2</id>
  <adres>adres 2</adres>
  <details>
   <art>art 2a</art>
  </details>
  <details>
   <art>art 2b</art>
  </details>
  <details>
   <art>art 2c</art>
  </details>
 </message>
</interchange>

My XSLT currently looks like this:

Code:

<?xml version="1.0"?>
<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">
  <xsl:output method="xml" indent="yes"/>

  <xsl:template match="/">
    <interchange>

      <xsl:for-each select="test/header">
        <xsl:variable name="curid" select="position()"/>
      
        <message>
          <id><xsl:value-of select="$curid"/></id>
          <adres><xsl:value-of select="adres"/></adres>
          
          
          <xsl:for-each select="following-sibling::line[ preceding-sibling::header[position() = $curid] ]">
          
            <details>
              <art><xsl:value-of select="art"/></art>
            </details>

          </xsl:for-each>
          
        </message>
      </xsl:for-each>
      
    </interchange>
  </xsl:template>

</xsl:stylesheet>

The for-each which is marked not to work, indeed doesn't work as expected.
The select should basically be something like
"SELECT all following <line> elements who's preceding <header> has the same position as the current <header>" (so basically select all following lines until the next header element).

Could someone please help to write this xpath expression correctly? What am I doing wrong?

Thanks

mhkay · April 28th, 2006, 05:43 AM

This problem is often referred to as "positional grouping". In XSLT 2.0 it can be done using <xsl:for-each-group group-starting-with="header">.

In 1.0 it's much trickier: one way is along the lines

for-each select="header"
for-each select="following-sibling::line[generate-id(preceding-sibling::header[1]) = generate-id(current())]

Other (more efficient) approaches are to reduce it to a value-based grouping problem, using generate-id(preceding-sibling::header[1]) as the grouping key (and then use Muenchian grouping in the usual way), or to to a recursive traversal of the siblings using recursive calls on apply-templates select="following-sibling::*[1]". Try googling for "XSLT positional grouping".

Michael Kay
http://www.saxonica.com/
Author, XSLT Programmer's Reference and XPath 2.0 Programmer's Reference