|
|
|
@ -10,7 +10,7 @@ module HtmlToPlainText
|
|
|
|
|
|
|
|
|
|
|
|
doc = Nokogiri::HTML.fragment(cleaned)
|
|
|
|
doc = Nokogiri::HTML.fragment(cleaned)
|
|
|
|
|
|
|
|
|
|
|
|
doc.css('script').each(&:remove)
|
|
|
|
doc.xpath('.//script').each(&:remove)
|
|
|
|
|
|
|
|
|
|
|
|
result = process_nodes(doc, line_length)
|
|
|
|
result = process_nodes(doc, line_length)
|
|
|
|
|
|
|
|
|
|
|
|
@ -33,12 +33,9 @@ module HtmlToPlainText
|
|
|
|
result = +''
|
|
|
|
result = +''
|
|
|
|
|
|
|
|
|
|
|
|
node.children.each do |child|
|
|
|
|
node.children.each do |child|
|
|
|
|
case child
|
|
|
|
if child.text? || child.cdata?
|
|
|
|
when Nokogiri::XML::Text
|
|
|
|
|
|
|
|
result << child.text
|
|
|
|
result << child.text
|
|
|
|
when Nokogiri::XML::Comment
|
|
|
|
elsif child.element?
|
|
|
|
next
|
|
|
|
|
|
|
|
when Nokogiri::XML::Element
|
|
|
|
|
|
|
|
result << process_element(child, line_length)
|
|
|
|
result << process_element(child, line_length)
|
|
|
|
end
|
|
|
|
end
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|