Repro case for XML parsing issue with Nokogiri 1.8.1

nokogiri-repro.rb

require 'nokogiri'

def generate_xml(number_of_occurrences)
  lines = []
  lines << "<Root>\n"
  lines << '  <ModelFile Name="aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" ModellerType="aa" Unit="1" UnitFromCAD="1" VM="10" Vm="0">\n'
  number_of_occurrences.times do |i|
    occ = <<-XML
            <ProductOccurrence Id="#{i+1}" Name="aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" PersistentId="1111111111" Layer="65535" Style="65535" Behaviour="1" FilePath="aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" OriginalFilePath="aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" Children="1111111111111111111111111111111111111111111111111111111" ModellerType="22" ProductLoadStatus="4" ProductFlag="4" Unit="1" DensityVolumeUnit="1" DensityMassUnit="1" UnitFromCAD="1">
                <Transformation RelativeTransfo="1111111111111111111111111111111" />
                <A3DMiscMaterialPropertiesData m_dDensity="-1" />
            </ProductOccurrence>
    XML
    lines << occ
  end
  lines << "  </ModelFile>\n"
  lines << "</Root>\n"
  lines.join('')
end


def parse_last_occurrence_id(xml)
  doc = Nokogiri::XML(xml)
  model_file = doc.root.children.select {|c| c.name == "ModelFile"}.first
  last_occurrence = model_file.children.select {|c| c.name == "ProductOccurrence"}.last
  last_occurrence["Id"]
end

number_of_occurrences = 12000
loop do
  xml = generate_xml(number_of_occurrences)
  last_id = parse_last_occurrence_id(xml)

  if last_id.to_i != number_of_occurrences
    puts
    puts "expected last ProductOccurrence @Id to be #{number_of_occurrences}, was #{last_id}\n"
    last_lines = xml.split("\n").last(10).join("\n")
    puts "last tags in XML: \n\n#{last_lines}"
    # File.open("sample.xml", "w").write(xml)
    break
  end

  number_of_occurrences += 50
  print "#{number_of_occurrences}, " # show progress
end