XmlMini supports different backend parsers, starting with libxml

[#2084 state:committed]

Signed-off-by: Jeremy Kemper <jeremy@bitsweat.net>

activesupport/CHANGELOG

@@ -1,3 +1,8 @@
+*Edge*
+
+* XmlMini supports libxml if available.  #2084 [Bart ten Brinke]
+
+
 *2.3.1 [RC2] (March 5, 2009)*
 
 * Vendorize i18n 0.1.3 gem (fixes issues with incompatible character encodings in Ruby 1.9) #2038 [Akira Matsuda]

activesupport/lib/active_support/xml_mini.rb

@@ -1,113 +1,21 @@
 # = XmlMini
-# This is a derivitive work of XmlSimple 1.0.11
-# Author::    Joseph Holsten <joseph@josephholsten.com>
-# Copyright:: Copyright (c) 2008 Joseph Holsten
-# Copyright:: Copyright (c) 2003-2006 Maik Schmidt <contact@maik-schmidt.de>
-# License::   Distributes under the same terms as Ruby.
 module ActiveSupport
   module XmlMini
     extend self
 
     CONTENT_KEY = '__content__'.freeze
 
-    # Parse an XML Document string into a simple hash
-    #
-    # Same as XmlSimple::xml_in but doesn't shoot itself in the foot,
-    # and uses the defaults from ActiveSupport
-    #
-    # string::
-    #   XML Document string to parse
-    def parse(string)
-      require 'rexml/document' unless defined?(REXML::Document)
-      doc = REXML::Document.new(string)
-      merge_element!({}, doc.root)
-    end
-
-    private
-      # Convert an XML element and merge into the hash
-      #
-      # hash::
-      #   Hash to merge the converted element into.
-      # element::
-      #   XML element to merge into hash
-      def merge_element!(hash, element)
-        merge!(hash, element.name, collapse(element))
-      end
-
-      # Actually converts an XML document element into a data structure.
-      #
-      # element::
-      #   The document element to be collapsed.
-      def collapse(element)
-        hash = get_attributes(element)
-
-        if element.has_elements?
-          element.each_element {|child| merge_element!(hash, child) }
-          merge_texts!(hash, element) unless empty_content?(element)
-          hash
-        else
-          merge_texts!(hash, element)
-        end
-      end
-
-      # Merge all the texts of an element into the hash
-      #
-      # hash::
-      #   Hash to add the converted emement to.
-      # element::
-      #   XML element whose texts are to me merged into the hash
-      def merge_texts!(hash, element)
-        unless element.has_text?
-          hash
-        else
-          # must use value to prevent double-escaping
-          merge!(hash, CONTENT_KEY, element.texts.sum(&:value))
-        end
-      end
-
-      # Adds a new key/value pair to an existing Hash. If the key to be added
-      # already exists and the existing value associated with key is not
-      # an Array, it will be wrapped in an Array. Then the new value is
-      # appended to that Array.
-      #
-      # hash::
-      #   Hash to add key/value pair to.
-      # key::
-      #   Key to be added.
-      # value::
-      #   Value to be associated with key.
-      def merge!(hash, key, value)
-        if hash.has_key?(key)
-          if hash[key].instance_of?(Array)
-            hash[key] << value
-          else
-            hash[key] = [hash[key], value]
-          end
-        elsif value.instance_of?(Array)
-          hash[key] = [value]
-        else
-          hash[key] = value
-        end
-        hash
+    # Hook the correct parser into XmlMini
+    def hook_parser
+      begin
+        require 'xml/libxml' unless defined? LibXML
+        require "active_support/xml_mini/libxml.rb"
+      rescue MissingSourceFile => e
+        require "active_support/xml_mini/rexml.rb"
       end
+    end
 
-      # Converts the attributes array of an XML element into a hash.
-      # Returns an empty Hash if node has no attributes.
-      #
-      # element::
-      #   XML element to extract attributes from.
-      def get_attributes(element)
-        attributes = {}
-        element.attributes.each { |n,v| attributes[n] = v }
-        attributes
-      end
+    hook_parser
 
-      # Determines if a document element has text content
-      #
-      # element::
-      #   XML element to be checked.
-      def empty_content?(element)
-        element.texts.join.blank?
-      end
   end
 end

activesupport/lib/active_support/xml_mini/libxml.rb

@@ -0,0 +1,131 @@
+# = XML Mini Libxml implementation
+module ActiveSupport
+  module XmlMini
+    extend self
+
+    # Parse an XML Document string into a simple hash using libxml.
+    # string::
+    #   XML Document string to parse
+    def parse(string)
+      require 'xml/libxml' unless defined? LibXML
+
+      string.strip!
+      XML.default_keep_blanks = false
+
+      return {} if string.blank?
+      return XML::Parser.string(string).parse.to_hash
+    end
+
+  end
+end
+
+module XML
+  module Conversions
+    module Document
+      def to_hash
+        root.to_hash
+      end
+    end
+
+    module Node
+      CONTENT_ROOT = '__content__'
+      LIB_XML_LIMIT = 30000000 # Hardcoded LibXML limit
+
+      # Convert XML document to hash
+      #
+      # hash::
+      #   Hash to merge the converted element into.
+      def to_hash(hash={})
+        if text?
+          raise RuntimeError if content.length >= LIB_XML_LIMIT
+          hash[CONTENT_ROOT] = content
+        else
+          sub_hash = insert_name_into_hash(hash, name)
+          attributes_to_hash(sub_hash)
+          if array?
+            children_array_to_hash(sub_hash)
+          elsif yaml?
+            children_yaml_to_hash(sub_hash)
+          else
+            children_to_hash(sub_hash)
+          end
+        end
+        hash
+      end
+
+      protected
+
+        # Insert name into hash
+        #
+        # hash::
+        #   Hash to merge the converted element into.
+        # name::
+        #   name to to merge into hash
+        def insert_name_into_hash(hash, name)
+          sub_hash = {}
+          if hash[name]
+            if !hash[name].kind_of? Array
+              hash[name] = [hash[name]]
+            end
+            hash[name] << sub_hash
+          else
+            hash[name] = sub_hash
+          end
+          sub_hash
+        end
+
+        # Insert children into hash
+        #
+        # hash::
+        #   Hash to merge the children into.
+        def children_to_hash(hash={})
+          each { |child| child.to_hash(hash) }
+          attributes_to_hash(hash)
+          hash
+        end
+
+        # Convert xml attributes to hash
+        #
+        # hash::
+        #   Hash to merge the attributes into
+        def attributes_to_hash(hash={})
+          each_attr { |attr| hash[attr.name] = attr.value }
+          hash
+        end
+
+        # Convert array into hash
+        #
+        # hash::
+        #   Hash to merge the array into
+        def children_array_to_hash(hash={})
+          hash[child.name] = map do |child|
+            returning({}) { |sub_hash| child.children_to_hash(sub_hash) }
+          end
+          hash
+        end
+
+        # Convert yaml into hash
+        #
+        # hash::
+        #   Hash to merge the yaml into
+        def children_yaml_to_hash(hash = {})
+          hash[CONTENT_ROOT] = content unless content.blank?
+          hash
+        end
+
+        # Check if child is of type array
+        def array?
+          child? && child.next? && child.name == child.next.name
+        end
+
+        # Check if child is of type yaml
+        def yaml?
+          attributes.collect{|x| x.value}.include?('yaml')
+        end
+
+    end
+  end
+end
+
+XML::Document.send(:include, XML::Conversions::Document)
+XML::Node.send(:include, XML::Conversions::Node)

activesupport/lib/active_support/xml_mini/rexml.rb

@@ -0,0 +1,106 @@
+# = XmlMini ReXML implementation
+module ActiveSupport
+  module XmlMini
+    extend self
+
+    # Parse an XML Document string into a simple hash
+    #
+    # Same as XmlSimple::xml_in but doesn't shoot itself in the foot,
+    # and uses the defaults from ActiveSupport
+    #
+    # string::
+    #   XML Document string to parse
+    def parse(string)
+      require 'rexml/document' unless defined?(REXML::Document)
+      doc = REXML::Document.new(string)
+      merge_element!({}, doc.root)
+    end
+
+    private
+      # Convert an XML element and merge into the hash
+      #
+      # hash::
+      #   Hash to merge the converted element into.
+      # element::
+      #   XML element to merge into hash
+      def merge_element!(hash, element)
+        merge!(hash, element.name, collapse(element))
+      end
+
+      # Actually converts an XML document element into a data structure.
+      #
+      # element::
+      #   The document element to be collapsed.
+      def collapse(element)
+        hash = get_attributes(element)
+
+        if element.has_elements?
+          element.each_element {|child| merge_element!(hash, child) }
+          merge_texts!(hash, element) unless empty_content?(element)
+          hash
+        else
+          merge_texts!(hash, element)
+        end
+      end
+
+      # Merge all the texts of an element into the hash
+      #
+      # hash::
+      #   Hash to add the converted emement to.
+      # element::
+      #   XML element whose texts are to me merged into the hash
+      def merge_texts!(hash, element)
+        unless element.has_text?
+          hash
+        else
+          # must use value to prevent double-escaping
+          merge!(hash, CONTENT_KEY, element.texts.sum(&:value))
+        end
+      end
+
+      # Adds a new key/value pair to an existing Hash. If the key to be added
+      # already exists and the existing value associated with key is not
+      # an Array, it will be wrapped in an Array. Then the new value is
+      # appended to that Array.
+      #
+      # hash::
+      #   Hash to add key/value pair to.
+      # key::
+      #   Key to be added.
+      # value::
+      #   Value to be associated with key.
+      def merge!(hash, key, value)
+        if hash.has_key?(key)
+          if hash[key].instance_of?(Array)
+            hash[key] << value
+          else
+            hash[key] = [hash[key], value]
+          end
+        elsif value.instance_of?(Array)
+          hash[key] = [value]
+        else
+          hash[key] = value
+        end
+        hash
+      end
+
+      # Converts the attributes array of an XML element into a hash.
+      # Returns an empty Hash if node has no attributes.
+      #
+      # element::
+      #   XML element to extract attributes from.
+      def get_attributes(element)
+        attributes = {}
+        element.attributes.each { |n,v| attributes[n] = v }
+        attributes
+      end
+
+      # Determines if a document element has text content
+      #
+      # element::
+      #   XML element to be checked.
+      def empty_content?(element)
+        element.texts.join.blank?
+      end
+  end
+end