sourcepos verifier

verify.rb

#!/usr/bin/env ruby

require 'nokogiri'

doc = File.read(ARGV[0]).split("\n")
xml = Nokogiri::XML(`build/src/cmark-gfm -t xml --sourcepos #{ARGV[0]}`)

xml.css('text, code').each do |el|
  from, to = el['sourcepos'].split('-').map {|r| r.split(':').map(&:to_i)}
  if from[0] != to[0]
    puts "Cannot verify #{[from, to].inspect}"
    next
  end

  comp = doc[from[0] - 1].b[from[1] - 1..to[1] - 1].force_encoding('utf-8')
  if comp.nil?
    puts "#{[from, to].inspect}: got nil"
    next
  end

  if el.name == 'text'
    comp = comp.gsub(/\\(.)/, '\1')
  else
    comp = comp.strip
  end

  if el.inner_text != comp
    puts "#{[from, to].inspect}: Expecting #{el.inner_text.inspect}, actual doc has #{comp.inspect}"
  end
end