Querying first result with chained xpth return entire array

[davidsulc]

Hopefully, this is in fact unexpected behavior and not me misunderstanding xpath...

This might in some way be related to #370

#!/usr/bin/ruby

require 'nokogiri'

html_doc = <<DOC
    <span>Stuffe we don't want</span>
    <h2>Start selection</h2>
        <p>Content we're interested in</p>
        <p>More content</p>
    <h2>End selection</h2>
        <p>Not intersted in this bit</p>
    <h2>Another title</h2>
        <p>Not intersted in this bit either</p>
DOC

page = Nokogiri::HTML(html_doc)
selection_start = page.xpath('//h2[ . = "Start selection"]/following::node()')
content = selection_start.xpath('./following-sibling::h2[1]')
puts content.inspect

The before last line returns the entire results array for following-sibling::h2 instead of (as I would expect) the first following sibling match. Interestingly, using xpath following-sibling::h2[2] behaves as expected and returns the second following sibling match.

The reason I'm attempting to use this, is I want to select the content from the h2 with text "Start selection", but only until the next h2 element.

[knu]

I think the current behavior is correct. //h2[ . = "Start selection"]/following::node() will match all following nodes of the first h2 element including the third p element, which has just one following sibling with a local-name of h2, which is the last h2 element. So, the last h2 element is indeed following-sibling:h2[1] for the third p element that matches //h2[ . = "Start selection"]/following::node().

I think changing the first XPath expression to //h2[ . = "Start selection"]/following::node()[1] will fix your problem.

[davidsulc]

Thanks for the clarification!