Question

How do I formulate an XPath expression for obtaining the attribute value for the first input node following the a node (value A) for the following HTML:

<a href="http://www.link.com">link</a>
<input value="value A" />
<input value="value B" />

I tried:

//input[(preceding-sibling::a[1])]/@value

But this gets the value attribute for both input nodes when I only want the attribute for the node for which the a node is the nearest preceding sibling.

Answer 1

You’re using the [1] predicate on the wrong node. You want the first input that has a preciding-sibling::a, not all inputs that follow the first a element at a given level.

This should yield the right result:
//a/following-sibling::input[1]/@value

It means: for every element a found at any level in the document, select the first input element that follows it, while being its sibling. Then read its value attribute.

For this input:

<xml>
  <p>
    <input value="value A" />
    <a href="http://www.link.com">link</a>
    <input value="value B" />
    <input value="value C" />
  </p>
  <p> 
    <a href="http://www.link.com">link</a>
    <input value="value D" />
    <input value="value E" />
    <a href="http://www.link.com">link</a>
    <input value="value F" />
  </p>
</xml>

It returns:

• value B
• value D
• value F

EDIT

This works provided that there are no elements between a and the first input that follows it.

For instance, with input like this:

    <a href="http://www.link.com">link</a>
    <strong>OMG NOEZ!</strong>
    <input value="value A" />

value A still gets selected

If it’s OK, keep the above expression. If it’s not, modify it like this to filter other elements:

//a/following-sibling::*[1][self::input]/@value

This selects the first following sibling of a, no matter what the tag is and only after it’s selected does it check the actual tag. This way the input that’s further at the same level is not chosen.