Everday XPath - Node Test

This blog post is part of a series on XPath. The content comes from my ebook EverydayXPath. Part of the content from the book will be released to the public as blog posts. In this post, we learn about the only mandatory part of the XPath expression: Node Test.

Node Test

The node test is used as a filter to select the nodes we want. It is the only mandatory part in a XPath expression. We break a node test into 3 types.

  • Name test: It makes use of the inherent properties of a node to select it. Properties such as the type of the node.

  • Node type test: Nodes can be generalized as a comment node, a text node, a processing instruction node or a catchall node type. The node type tests helps to filter along these lines.

  • Targeted processing instruction test: This test selects processing instruction nodes with a defined type.

For the following examples, we would use the below as the reference document.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

<table>
  <tr class="odd">
    <td>Alpha</td>
  </tr>
  <tr class="even" style="margin: 12px">
    <td>Bravo</td>
  </tr>
  <tr class="odd" style="margin: 12px">
    <td>Charlie</td>
  </tr>
  <tr class="even">
    <td>Delta</td>
  </tr>
  <tr class="odd">
    <td>
      <table style="margin: 12px">
        <tr>
          <td>Echo</td>
        </tr>
        <tr>
          <td>Foxtrot</td>
        </tr>
      </table>
    </td>
  </tr>
</table>

Name Test

The name test filter makes use of properties of the nodes.

Asterisk *

The asterisk or the glob, selects everything.

1

/table/tr/td/*

This expression selects all the nodes that come after the td nodes of the root level table node. It then selects the inner table as it came after the td nodes.

1
2
3
4
5
6
7
8

<table style="margin: 12px">
  <tr>
    <td>Echo</td>
  </tr>
  <tr>
    <td>Foxtrot</td>
  </tr>
</table>

QName

When we want to select something more specific, we use the QName. QName stands for qualified name and can be either a prefixed name or an unprefixed name.

1

QName = PrefixedName | UnprefixedName

A prefixed name is made up of 2 parts: a prefix and a localpart separated by a colon. The prefix is a name space prefix and is optional.

1

PrefixedName = Prefix ':' LocalPart

The unprefixed name and local part resolves to become a NCName which stands for non-colonized name: name without colons.

1
2
3

UnprefixedName = LocalPart
Prefix = NCName
LocalPart = NCName

td in the following expression below is a NCName. It means to select elements of the td type.

1

//td

The results returns a list of 7 td nodes.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

<td>Alpha</td> <1>
<td>Bravo</td> <2>
<td>Charlie</td> <3>
<td>Delta</td> <4>
<td> <5>
  <table style="margin: 12px">
    <tr>
      <td>Echo</td>
    </tr>
    <tr>
      <td>Foxtrot</td>
    </tr>
  </table>
</td>
<td>Echo</td> <6>
<td>Foxtrot</td> <7>

Node Type Test

There are times when you need a broader selection than a NCName and narrower than a glob. In such cases, there are 4 kinds of node types test which you can make use of.

  • comment(): Selects comment nodes.
  • node(): Selects all nodes irregardless of type.
  • text(): Selects the text from the current context.
  • processing-instruction(): Selects the processing instruction nodes.

For the following examples, we would use the below as the reference document.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

<table>
  <tr class="odd">
    <td>Alpha<img src="foo.jpg"></td>
  </tr>
  <tr class="even" style="margin: 12px">
    <td>Bravo</td>
  </tr>
  <tr class="odd" style="margin: 12px">
    <td>Charlie</td>
  </tr>
  <tr class="even">
    <!-- <td>Delta</td> -->
  </tr>
  <tr class="odd">
    <td>
      <table style="margin: 12px">
        <tr>
          <!-- <td>Echo</td> -->
        </tr>
        <tr>
          <td>Foxtrot</td>
        </tr>
      </table>
    </td>
  </tr>
</table>

comment()

Selects comment nodes from the current context.

1

//tr[@class="even"]/comment()

This expression selects the comment nodes from the context of the tr node with a class with value even.

1

<!-- <td>Delta</td> -->

node()

Selects all nodes irregardless of type, It even includes comment and processing instruction nodes.

1

//tr[@class="even"]/node()

This expression selects all the nodes from the context of the tr node with a class with value even.

1
2

<td>Bravo</td> <1>
<!-- <td>Delta</td> --> <2>

Compare this to the glob.

1

//tr[@class="even"]/*

node() captures everything, even more than the glob.

1

<td>Bravo</td>

text()

Selects the text from the current context.

1

//td/text()

This expression selects the inner text from the nodes selected. Notice that it returns an empty string if the context does not contain any text.

1
2
3
4
5
6

'Alpha' <1>
'Bravo' <2>
'Charlie' <3>
'' <4>
'Foxtrot' <5>
'' <7>

processing-instruction()

Selects the processing instruction nodes. Processing instruction nodes are usually found in XML documents. It usually carries meta information for applications. An example of a processing instruction node is below:

1

<?xml-stylesheet type="text/xsl" href="style.xsl"?>

It can be targeted. The example below selects only processing instructions with the target “xml-stylesheet”

1

processing-instruction("xml-stylesheet")

Comments