XSLT every nth node with a filter

问题

I'm using XSLT for some output formatting, and I want a wrapper element around every N nodes of the output. I've read xslt - adding </tr><tr> every n node?, but my problem is that the source nodes have to come from a lookup:

<xsl:for-each select="key('items-by-product', $productid)">

rather than just a template match. All the examples I've found assume that the nodes you want are all next to each other, and they're just counting siblings.

I have a solution that works for me:

<xsl:stylesheet version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform" >
  <xsl:variable name='num_per_div' select='2' />
  <xsl:variable name='productid' select='1' />
  <xsl:output method="xml" indent="yes"/>
  <xsl:key name="items-by-product" match="item" use="productid"/>
  <xsl:template match="data">
    <output>
      <xsl:for-each select="key('items-by-product', $productid)">
        <xsl:variable name='pos' select='position()' />
        <xsl:if test="position() = 1 or not((position()-1) mod $num_per_div)">
          <outer pos="{$pos}">
            <xsl:for-each select="key('items-by-product', $productid)">
              <xsl:variable name='ipos' select='position()' />
              <xsl:if test="$ipos >= $pos and $ipos &lt; $pos + $num_per_div">
                <inner>
                  <xsl:value-of select="itemid"/>
                </inner>
              </xsl:if>
            </xsl:for-each>
          </outer>
        </xsl:if>
      </xsl:for-each>      
    </output>
  </xsl:template>
</xsl:stylesheet>

with data

<data>
  <item>
    <productid>1</productid>
    <itemid>1</itemid>
  </item>
  <item>
    <productid>1</productid>
    <itemid>2</itemid>
  </item>
  <item>
    <productid>2</productid>
    <itemid>A</itemid>
  </item>
  <item>
    <productid>1</productid>
    <itemid>3</itemid>
  </item>
  <item>
    <productid>2</productid>
    <itemid>B</itemid>
  </item>
  <item>
    <productid>1</productid>
    <itemid>4</itemid>
  </item>
</data>

which produces

<?xml version="1.0" encoding="utf-8"?>
<output>
  <outer pos="1">
    <inner>1</inner>
    <inner>2</inner>
  </outer>
  <outer pos="3">
    <inner>3</inner>
    <inner>4</inner>
  </outer>
</output>

But this is looping through all the nodes for each node, which strikes me as inefficient.

Is there a better approach that will produce the same output more efficiently? Can the following-sibling techniques work with a filter?

回答1:

You can use an outer loop with position() mod $num_per_div to get one "iteration" per chunk, then within that select out the members of that chunk out of the whole key(...) node set by their position:

<xsl:for-each select="key('items-by-product', $productid)
                       [position() mod $num_per_div = 1]">
  <xsl:variable name="iter" select="position()" />
  <xsl:variable name="first" select="($iter - 1) * $num_per_div + 1" />
  <xsl:variable name="last" select="$iter * $num_per_div" />
  <outer pos="{$first}">
    <xsl:for-each select="key('items-by-product', $productid)
                           [position() &gt;= $first and position() &lt;= $last]">
      <inner><xsl:value-of select="itemid"/></inner>
    </xsl:for-each>
  </outer>
</xsl:for-each>

The key thing here is to remember that the position() function is context-sensitive and means different things at different times. In the definition of the $iter variable, the current node list is the nodes selected by the outer for-each, i.e. the list with the first, third, fifth, etc. items returned by the key (so position() means the chunk number). But in the predicate on the select of the inner for-each the current node list is all the nodes returned from the key function call (so position() is the position of the node-under-test within the list of all nodes with the given productid).

回答2:

You could start by copying the nodes of interest into a variable; that would make them - and only them - siblings. However, in XSLT 1.0 such variable would contain a result-tree-fragment, which needs to be converted to a node-set before it can be processed further:

XSLT 1.0

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

<xsl:key name="items-by-product" match="item" use="productid"/>

<xsl:variable name="groupSize" select="2" />
<xsl:variable name="productid" select="1" />
<xsl:variable name="my-items">
    <xsl:copy-of select="key('items-by-product', $productid)"/>
</xsl:variable>

<xsl:template match="/">
    <output>
        <xsl:for-each select="exsl:node-set($my-items)/item[position() mod $groupSize = 1]">
            <outer pos="{position()}">
                <xsl:for-each select=". | following-sibling::item[position() &lt; $groupSize]" >
                    <inner>
                        <xsl:value-of select="itemid"/>
                    </inner>
                </xsl:for-each>
            </outer>
        </xsl:for-each>
    </output>
</xsl:template>

</xsl:stylesheet>

来源：https://stackoverflow.com/questions/28415867/xslt-every-nth-node-with-a-filter