sparklemotion/nokogiri

CSS `foo~:nth-child(2)` gives incorrect XPath

Open

#707 opened on Jun 18, 2012

View on GitHub
 (3 comments) (0 reactions) (0 assignees)Ruby (5,615 stars) (806 forks)batch import
help wantedneeds/fix-for-failing-testtopic/css

Description

Hi,

I’m the maintainer of cssselect, which does in Python pretty much the same as Nokogiri for CSS selectors: translate them to XPath. It looks like the SimonSapin/cssselect#12 bug also applies to Nokogiri. Namely, the XPath translation of :nth-child() and similar pseudo-classes is wrong when used after the + or ~ combinator. Here is a test case:

require 'nokogiri'
doc = Nokogiri::XML('<root><child1/><child2/><child3/></root>')
puts doc.css(':nth-child(2)').map { |e| e.name }
puts doc.css('child1 ~ :nth-child(2)').map { |e| e.name }

Expected output: child2 child2. Actual output child2 child3.

The problem is in the XPath translation of the later selector: //child1/following-sibling::*[position() = 2 and self::*] gives the element at position 2 when counting from child1, while we want the position among the parent’s children.

I am not sure it is even possible to correctly translate this selector to XPath: the = XPath operator on node-sets compares the text content of elements, not their identity.

The issue is similar for SimonSapin/cssselect#4 and Nokogiri’s #394.

Contributor guide

Tech stack
ruby
Domain
backendapi
Issue type
bug
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
3
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
3-5 days
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
stale
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
clear
Prerequisites
RubyNokogiri usageCSS selectorsXPath
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
30
Research direction
Investigate the XPath translation logic in Nokogiri's CSS parser, likely in lib/nokogiri/css/parser.rb. The issue is similar to SimonSapin/cssselect#12 and Nokogiri #394. The core problem is that the `following sibling::*[position() = 2]` counts from the preceding sibling, not from the parent. Consider whether a correct translation is even possible using standard XPath 1.0, or if a workaround like using `preceding sibling` or a different approach is needed.