XSLT合并多个节点

Question

Recently I am trying to mess around with XSLT a little bid and am currently stuck with a problem, that is pretty specific I think.

I want to combine multiple nodes until a node with a special property appears and then Merge the node text into a single new node.

<paragraph>
    <row>
        <text>Text1</text>
    </row>
    <row>
        <properties>
            <vertAlign val="superscript"/>
        </properties>
        <text>2</text>
    </row>
    <row>
        <text>Text2</text>
    </row>
    <row>
        <text>Text3</text>
    </row>
    <row>
        <properties>
            <vertAlign val="superscript"/>
        </properties>
        <text>1</text>
    </row>
    <row>
        <text>Text4</text>
    </row>
    <row>
        <text>Text5</text>
    </row>
    <row>
        <text>Text6</text>
    </row>
    <row>
        <properties>
            <vertAlign val="superscript"/>
        </properties>
        <text>1</text>
    </row>
    <row>
        <text>Text7</text>
    </row>
    <row>
        <properties>
            <vertAlign val="superscript"/>
        </properties>
        <text>1</text>
    </row>
</paragraph>

In this example I want to collect all row/text until row/properties/vertAlign[@val="superscript"]. The output should be:

<root>
    <node>Text1</node>
    <node>Text2Text3</node>
    <node>Text4Text5Text6</node>
    <node>Text7</node>
</root>

Appreciate all of your Help, Asmo

Answer 1

Assuming XSLT 2.0 you can use

<xsl:template match="paragraph">
  <root>
    <xsl:for-each-group select="row" group-ending-with="row[properties/vertAlign/@val = 'superscript']">
      <node>
        <xsl:value-of select="current-group()[position() ne last()]/text" separator=""/>
      </node>
    </xsl:for-each-group>
  </root>
</xsl:template>

Answer 2

If you are currently using XSLT 1.0 then you need to read up on a technique called Muenchian Grouping . Assuming there will be always a "properties" row following a normal row at some point, you can "group" your row elements by the unique ID of the first following "properties" row. This means defining a key like so:

<xsl:key name="rows" 
        match="row[not(properties)]" 
        use="generate-id(following-sibling::row[properties][1])" />

You can then select the first row in each group like so....

<xsl:apply-templates 
     select="row[generate-id() = generate-id(key('rows', generate-id(following-sibling::row[properties][1]))[1])]" />

Then, in the template matching the row, you can get the text for each row in the group by again using the key:

<xsl:apply-templates select="key('rows', generate-id(following-sibling::row[properties][1]))/text" />

This will take advantage of XSLT's built-in templates which will iterate over (but not copy) elements, but output text nodes where it find them.

Try this XSLT

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

    <xsl:key name="rows" match="row[not(properties)]" use="generate-id(following-sibling::row[properties][1])" />

    <xsl:template match="paragraph">
        <root>
            <xsl:apply-templates select="row[generate-id() = generate-id(key('rows', generate-id(following-sibling::row[properties][1]))[1])]" />
        </root>
    </xsl:template>

    <xsl:template match="row">
        <node>
            <xsl:apply-templates select="key('rows', generate-id(following-sibling::row[properties][1]))/text" />
        </node>
    </xsl:template>
</xsl:stylesheet>