Class: FeedParserUtilities::SanitizerDoc

Inherits:

Hpricot::Doc

• Object
• Hpricot::Doc
• FeedParserUtilities::SanitizerDoc

Defined in:

lib/rfeedparser/scrub.rb

Instance Method Summary

• #scrub ⇒ Object

Instance Method Details

#scrub ⇒ Object

# File 'lib/rfeedparser/scrub.rb', line 160

def scrub
  others = children.map do |e|
    if e.elem?
      if Acceptable_Elements.include?e.name
        e.strip_attributes
        e.inner_html = SanitizerDoc.new(e.children).scrub
        result = e
      else
        result = e
        
        if Unacceptable_Elements_With_End_Tag.include?e.name
          result = nil
        end
        
        if result 
          result = SanitizerDoc.new(result.children).scrub   # The important part
        end            
      end
      
    elsif e.doctype?
      result = nil

    elsif e.text?
      ets = e.to_html
      ets.gsub!(/'/, "'") 
      ets.gsub!(/"/, '"')
      ets.gsub!(/\r/,'')
      result = ets
    end
    result
  end
  
  unless $compatible # FIXME nonworking
    # yes, that '/' should be there. It's a search method. See the Hpricot docs.
    (self/tag).strip_style(@config[:allow_css_properties], @config[:allow_css_keywords])
  end
  return others.compact.join
end