xslt 1.0, select group of nodes with key

Question

I want to select nodes based on some variables. The XML code:

<data>
    <prot seq="AAA">
        <node num="1">1345</node>
        <node num="1">11245</node>
        <node num="2">88885</node>
    </prot>
    <prot seq="BBB">
        <node num="1">678</node>
        <node num="1">456</node>
        <node num="2">6666</node>
    </prot>
    <prot seq="CCC">
        <node num="1">111</node>
        <node num="1">222</node>
        <node num="2">333</node>
    </prot>
</data>

The XML that I want

<output>
    <prot seq="AAA">
        <node num="1">1345</node>
        <node num="2">88885</node>
    </prot>
    <prot seq="BBB">
        <node num="1">678</node>
        <node num="2">6666</node>
    </prot>
    <prot seq="CCC">
        <node num="1">111</node>
        <node num="2">333</node>
    </prot>
</data>

So, my idea has been to group the nodes with a xsl:key element, and then do a for-each of them. For example:

<xsl:stylesheet version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
    <xsl:output method="xml" indent="yes" omit-xml-declaration="yes"/>
    <xsl:key name="by" match="/data/prot" use="concat(@seq,'|',node/@num)"/>
    <xsl:template match="/">
        <root>
            <xsl:apply-templates select="/data/prot"/>
        </root>
    </xsl:template>
    <xsl:template match="/data/prot">
        <xsl:for-each select="./node">
            <xsl:for-each select="key('by',concat(current()/../@seq,'|',current()/@num))">
                node <xsl:value-of select="./node" />
            </xsl:for-each>
        </xsl:for-each>
    </xsl:template>
</xsl:stylesheet>

but the output is not what I expected, and I cannot see what I am doing wrong. I would prefer to keep the for-each structure. It is just as if I was not using properly the xsl:key grouping features.

the output that I get, unwanted

<root>
    node 1345
    node 1345
    node 678
    node 678
    node 111
node 111</root>

And the code as it to be tested http://www.xsltcake.com/slices/sgWUFu/20

Thanks!

Answer 1

The main problem in your code is that the key indexes prot elements, but what we want to de-duplicate (and need to index) is the node elements.

Here is a short and correct solution:

<xsl:stylesheet version="1.0" 
 xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
 <xsl:output omit-xml-declaration="yes" indent="yes"/>
 <xsl:strip-space elements="*"/>

 <xsl:key name="nodeByParentAndNum" match="node"
  use="concat(generate-id(..), '+', @num)"/>

 <xsl:template match="node()|@*">
  <xsl:copy>
   <xsl:apply-templates select="node()|@*"/>
  </xsl:copy>
 </xsl:template>

 <xsl:template match="/*">
  <data>
   <xsl:apply-templates/>
  </data>
 </xsl:template>

 <xsl:template match=
 "node
   [not(generate-id()
       =
        generate-id(key('nodeByParentAndNum',
                        concat(generate-id(..), '+', @num)
                        )
                         [1]
                    )
       )
   ]
 "/> 
</xsl:stylesheet>

when this transformation is applied on the provided XML document:

<data>
    <prot seq="AAA">
        <node num="1">1345</node>
        <node num="1">11245</node>
        <node num="2">88885</node>
    </prot>
    <prot seq="BBB">
        <node num="1">678</node>
        <node num="1">456</node>
        <node num="2">6666</node>
    </prot>
    <prot seq="CCC">
        <node num="1">111</node>
        <node num="1">222</node>
        <node num="2">333</node>
    </prot>
</data>

the wanted, correct result is produced:

<data>
   <prot seq="AAA">
      <node num="1">1345</node>
      <node num="2">88885</node>
   </prot>
   <prot seq="BBB">
      <node num="1">678</node>
      <node num="2">6666</node>
   </prot>
   <prot seq="CCC">
      <node num="1">111</node>
      <node num="2">333</node>
   </prot>
</data>