default.rb in aranha-parsers-0.2.0

- old
+ new

@@ -7,16 +7,32 @@
     module Html
       module Node
         class Default < ::Aranha::Parsers::Html::Node::Base
           def string_value(node, xpath)
             if node.at_xpath(xpath)
-              node.at_xpath(xpath).text.to_s.tr("\u00A0", ' ').strip
+              sanitize_string(node.at_xpath(xpath).text)
             else
               ''
             end
           end
 
+          def string_recursive_value(node, xpath, required = true)
+            root = node.at_xpath(xpath)
+            if root.blank?
+              return nil unless required
+              raise "No node found (Xpath: #{xpath})"
+            end
+            result = string_recursive(root)
+            return result unless result.blank?
+            return nil unless required
+            raise "String blank (Xpath: #{xpath})"
+          end
+
+          def string_recursive_optional_value(node, xpath)
+            string_recursive_value(node, xpath, false)
+          end
+
           def quoted_value(node, xpath)
             s = string_value(node, xpath)
             return '' unless s
 
             m = /\"([^\"]+)\"/.match(s)
@@ -79,13 +95,27 @@
 
           def parse_float(node, xpath, required)
             s = string_value(node, xpath)
             m = /\d+(?:[\.\,](\d+))?/.match(s)
             if m
-              m[0].sub(',', '.').to_f
+              m[0].delete('.').tr(',', '.').to_f
             elsif required
               raise "Float value not found in \"#{s}\""
             end
+          end
+
+          def sanitize_string(obj)
+            obj.to_s.tr("\u00A0", ' ').strip
+          end
+
+          def string_recursive(node)
+            return sanitize_string(node.text) if node.is_a?(::Nokogiri::XML::Text)
+            s = ''
+            node.children.each do |child|
+              child_s = string_recursive(child)
+              s += ' ' + child_s if child_s.present?
+            end
+            sanitize_string(s)
           end
         end
       end
     end
   end