XPath Expressions

What is XPath?

XPath is a language for navigating XML trees and querying data. Since the Publisher’s input data is in XML format, XPath is the central tool for accessing elements, attributes and text. Examples: “give me all articles with a certain attribute” or “how many articles are in this article group?”.

For a general introduction to XPath, see the tutorial at W3Schools.

Where are XPath expressions used?

The Publisher accepts XPath expressions in two places:

1. In the attributes select and test — the value is interpreted directly as XPath.
2. In all other attributes — curly braces ({ and }) force an XPath expression.

<Textblock width="{ $width }">
  <Paragraph>
    <Value select="."/>
  </Paragraph>
</Textblock>

Supported expressions

• Numbers and text: "5", 'hello world' — returned directly.
• Arithmetic operations: *, div, idiv, +, -, mod. Example: ( 6 + 4.5 ) * 2
• Variables: $column + 2
• Current node (dot operator): . + 2
• Child elements: productdata, *, foo/bar, node()
• Parent node: ../
• Attributes: @a, foo/@bar
• Filters: article[1] selects the first article child element.
• Comparisons: <, >, <=, >=, =, !=. Note: < must be written as < in XML.
• If/then/else: if (...) then ... else ...
• For expressions: for $i in (1,2,3) return $i * 2 or for $i in 1 to 3 return $i * 2
• Axes: preceding-sibling, parent, descendant-or-self, following-sibling etc.

XPath and namespaces

The speedata Publisher ignores XML namespaces by default. This is usually very helpful, as namespaces are not always easy to handle.

The XML file:

<bar:data xmlns:bar="somenamespace">
    <bar:child>
        <bar:sub>sub</bar:sub>
    </bar:child>
</bar:data>

can be addressed in the layout as follows:

<Layout xmlns="urn:speedata.de:2009/publisher/en"
    xmlns:sd="urn:speedata:2009/publisher/functions/en">

    <Record element="data">
      ...

although the namespace of the root element is somenamespace.

With <Options namespaces="strict" /> you can change this so that the namespace must be specified for <Record> and <ProcessNode>:

<Layout xmlns="urn:speedata.de:2009/publisher/en"
    xmlns:sd="urn:speedata:2009/publisher/functions/en"
    xmlns:sn="somenamespace">

    <Options namespaces="strict" />

    <Record element="sn:data">
        <PlaceObject>
            <Textblock>
                <Paragraph>
                    <Value select="local-name()"></Value>
                </Paragraph>
            </Textblock>
        </PlaceObject>
    </Record>
</Layout>

It is important that the namespace matches, the prefix (here: sn) is freely selectable as long as it is linked to the namespace.

<ProcessNode> works similar:

<ProcessNode select="*:child" />
<ProcessNode select="sn:child" />
<ProcessNode select="sn:*" />
<ProcessNode select="*" />

call the child element bar:child, as the namespaces of sn (from the layout) and bar (from the data) match. * is a ‘wildcard’ here, so it matches all names.

<ProcessNode select="child" /> will not work in the above case, as the empty namespace from the layout is urn:speedata.de:2009/publisher/en and does not match the namespace from the data.

You must also use namespaces in the XPath queries if they are specified in the data:

<Message select="count(sn:sub)" />

writes 1 to the log file. Without namespace a 0, as no element sub with the preset namespace is found.