Combining Kayessian intersection and Muenchian grouping

3
votes

I have a very flat document which contains implied groups of elements based on their positioning after a Heading item:

<Document>
    <Body>
        ...
        <Heading>Section 1</Heading>
        <Item Id="1.1">Alpha</Item>
        <Item Id="1.1">Bravo</Item>
        ...
        <Heading>Section 2</Heading>
        <Item Id="2.1">Alpha</Item>
        <Item Id="2.1">Bravo</Item>
        ...
    </Body>
</Document>

From this document, I want to extract the groups, but also filter the items in each group to take the first items with a given identifier. For example, where there are two items with the ID "1.1", only the first item is expected in the output. I intend to do additional processing to include the duplicates as children of the first item.

To achieve this grouping, I am using Muenchian grouping, where the key for the group is the identifier value:

<xsl:key
    name="ItemsById"
    match="/Document/Body/Item"
    use="@Id"/>

This works great, except that there's a number of Item elements defined as examples that happen to use the same identifiers and winds up in the node-set matched in the key.

As there is a range in the middle of the document that I care about, I am using the Kayessian method of intersection to restrict the node-set to just the section in the document I am interested in:

<xsl:variable
    name="section"
    select="(/Document/Body/Heading[text() = 'Example']
        /following-sibling::*[2]/following-sibling::*)[
    count(. | /Document/Body/Heading[text() = 'Appendix B']
        /preceding-sibling::*) 
    = count(/Document/Body/Heading[text() = 'Appendix B']
        /preceding-sibling::*)
    ]" />

This node-set is the intersection of two node-sets: all the elements after the Heading "Section 1" (including the heading itself) and all the elements before the Heading "Appendix B".

This matches the elements I care about, however since the key is unfiltered, the "first" value for a given identifier is sometimes outside of this node-set. I have tried using the variable in the key, but I've since discovered that there are numerous restrictions on the match in a key which prevent the use of variables.

Here is the full source document:

<Document>
    <Body>

        <Heading>Preamble</Heading>
        <Para>
            Lorem ipsum dolor sit amet, consectetur
            adipiscing elit, sed do eiusmod tempor incididunt
            ut labore et dolore magna aliqua.
        </Para>

        <Heading>Example</Heading>
        <Item Id="1.1">Example Alpha</Item>
        <Item Id="1.1">Example Bravo</Item>

        <Heading>Section 1</Heading>
        <Item Id="1.1">Alpha</Item>
        <Item Id="1.1">Bravo</Item>
        <Item Id="1.2">Charlie</Item>
        <Item Id="1.3">Delta</Item>
        <Item Id="1.3">Echo</Item>
        <Item Id="1.4">Foxtrot</Item>

        <Heading>Section 2</Heading>
        <Item Id="2.1">Alpha</Item>
        <Item Id="2.1">Bravo</Item>
        <Item Id="2.2">Charlie</Item>
        <Item Id="2.3">Delta</Item>
        <Item Id="2.3">Echo</Item>
        <Item Id="2.4">Foxtrot</Item>

        <Heading>Appendix A</Heading>
        <Item Id="A.1">Alpha</Item>
        <Item Id="A.1">Bravo</Item>
        <Item Id="A.2">Charlie</Item>
        <Item Id="A.3">Delta</Item>
        <Item Id="A.3">Echo</Item>
        <Item Id="A.4">Foxtrot</Item>

        <Heading>Appendix B</Heading>
        <Para>
            Lorem ipsum dolor sit amet, consectetur
            adipiscing elit, sed do eiusmod tempor incididunt
            ut labore et dolore magna aliqua.
        </Para>

    </Body>
</Document>

I'm apply the following stylesheet:

<xsl:stylesheet
    version="1.0"
    xmlns:xsl="http://www.w3.org/1999/XSL/Transform">

    <xsl:output method="xml" indent="yes"/>

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

    <!-- The node-set which covers the wanted section of elements. -->
    <xsl:variable
        name="section"
        select="(/Document/Body/Heading[text() = 'Example']
            /following-sibling::*[2]/following-sibling::*)[
        count(. | /Document/Body/Heading[text() = 'Appendix B']
            /preceding-sibling::*) 
        = count(/Document/Body/Heading[text() = 'Appendix B']
            /preceding-sibling::*)
        ]" />

    <!-- The items keyed by their ID. -->
    <xsl:key
        name="ItemsById"
        match="/Document/Body/Item"
        use="@Id"/>

    <!-- Matches the root to begin the output structure. -->
    <xsl:template match="/">
        <Document>
            <!-- Apply templates to the headings. -->
            <xsl:apply-templates select="$section[local-name() = 'Heading']" />
        </Document>
    </xsl:template>

    <xsl:template match="/Document/Body/Heading">
        <Section>
            <xsl:attribute name="Title">
                <xsl:value-of select="."/>
            </xsl:attribute>

            <xsl:variable
                name="heading"
                select="generate-id()" />

            <!-- Apply templates to the items in this set. -->
            <xsl:apply-templates
                select="$section[
                local-name() = 'Item'
                and
                generate-id() = generate-id(key('ItemsById', @Id)[1])
                and
                $heading = generate-id(preceding-sibling::Heading[1])
                ]" />
        </Section>
    </xsl:template>

</xsl:stylesheet>

This is the current output:

<Document>
  <Section Title="Section 1">
    <Item Id="1.2">Charlie</Item>
    <Item Id="1.3">Delta</Item>
    <Item Id="1.4">Foxtrot</Item>
  </Section>
  <Section Title="Section 2">
    <Item Id="2.1">Alpha</Item>
    <Item Id="2.2">Charlie</Item>
    <Item Id="2.3">Delta</Item>
    <Item Id="2.4">Foxtrot</Item>
  </Section>
  <Section Title="Appendix A">
    <Item Id="A.1">Alpha</Item>
    <Item Id="A.2">Charlie</Item>
    <Item Id="A.3">Delta</Item>
    <Item Id="A.4">Foxtrot</Item>
  </Section>
</Document>

The issue is that the Item 1.1 is missing from Section 1.

Is there anything different I can try to achieve the same grouping over the section I'm interested in?

1
xsltxslt-1.0
Good question, well asked. Sure that you're limited to XSLT 1.0? – Mathias Müller
I'm running under .NET, so sadly yes. – Paul Turner
How big are the documents? And how concerned are you about performance? There are ways without using keys, but they'd be much slower. (Btw, it's a shame .NET only supports 1.0...). – Mathias Müller
The documents are 8Mb in size, but performance isn't a priority. Fast is nice, but not necessary. – Paul Turner
Sorry, that's still not clear. Do you mean any section between Example and Appendix B, excluding the two? – michael.hor257k

1 Answers

2
votes

Couldn't this be (much) simpler? For example, the following stylesheet:

XSLT 1.0

<xsl:stylesheet version="1.0" 
xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
<xsl:output method="xml" version="1.0" encoding="UTF-8" indent="yes"/>

<xsl:key name="item-by-heading" match="Item" use="generate-id(preceding-sibling::Heading[1])" />
<xsl:key name="item-by-id" match="Item" use="concat(generate-id(preceding-sibling::Heading[1]), '|', @Id)" />

<xsl:template match="/Document">
    <xsl:copy>
        <xsl:apply-templates select="Body/Heading"/>
    </xsl:copy>
</xsl:template>

<xsl:template match="Heading">
    <Section Title="{.}">
        <xsl:copy-of select="key('item-by-heading', generate-id())[count(. | key('item-by-id', concat(generate-id(preceding-sibling::Heading[1]), '|', @Id))[1]) = 1]"/>
    </Section>
</xsl:template> 

</xsl:stylesheet>

when applied to your input, will return:

<?xml version="1.0" encoding="UTF-8"?>
<Document>
   <Section Title="Preamble"/>
   <Section Title="Example">
      <Item Id="1.1">Example Alpha</Item>
   </Section>
   <Section Title="Section 1">
      <Item Id="1.1">Alpha</Item>
      <Item Id="1.2">Charlie</Item>
      <Item Id="1.3">Delta</Item>
      <Item Id="1.4">Foxtrot</Item>
   </Section>
   <Section Title="Section 2">
      <Item Id="2.1">Alpha</Item>
      <Item Id="2.2">Charlie</Item>
      <Item Id="2.3">Delta</Item>
      <Item Id="2.4">Foxtrot</Item>
   </Section>
   <Section Title="Appendix A">
      <Item Id="A.1">Alpha</Item>
      <Item Id="A.2">Charlie</Item>
      <Item Id="A.3">Delta</Item>
      <Item Id="A.4">Foxtrot</Item>
   </Section>
   <Section Title="Appendix B"/>
</Document>

I couldn't understand how you determine which sections you want to include in (or exclude from) the output, but that too should be easy.

Edit:

The sections I want is Sections 1-2 and Appendix A; no other sections are relevant.

Well, then just do:

<xsl:template match="/Document">
    <xsl:copy>
        <xsl:apply-templates select="Body/Heading[.='Section 1' or .='Section 2'or .='Appendix A']"/>
    </xsl:copy>
</xsl:template>

Note that if the items ids are not duplicated across sections, then this could be even simpler. Ah, but I see that they are. That is the reason why item 1.1 is missing.

Edit 2:

This node-set is the intersection of two node-sets: all the elements after the Heading "Section 1" (including the heading itself) and all the elements before the Heading "Appendix B".

Okay, so:

<xsl:template match="/Document">
    <xsl:copy>
        <xsl:apply-templates select="Body/Heading[.='Section 1' or preceding-sibling::Heading[.='Section 1'] and following-sibling::Heading[.='Appendix B']]"/>
    </xsl:copy>
</xsl:template>

Or, even shorter:

<xsl:template match="/Document">
    <xsl:copy>
        <xsl:apply-templates select="Body/Heading[not(following-sibling::Heading[.='Section 1']) and following-sibling::Heading[.='Appendix B']]"/>
    </xsl:copy>
</xsl:template>