Added SAX-based parser for XmlMini, using LibXML

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

1 files changed, 84 insertions, 0 deletions

diff --git a/activesupport/lib/active_support/xml_mini/libxmlsax.rb b/activesupport/lib/active_support/xml_mini/libxmlsax.rb
new file mode 100644
index 0000000000..dba1f8755a
--- /dev/null
+++ b/activesupport/lib/active_support/xml_mini/libxmlsax.rb
@@ -0,0 +1,84 @@
+require 'libxml'
+
+# = XmlMini LibXML implementation using a SAX-based parser
+module ActiveSupport
+  module XmlMini_LibXMLSAX
+    extend self
+
+    # Class that will build the hash while the XML document
+    # is being parsed using SAX events.
+    class HashBuilder
+
+      include LibXML::XML::SaxParser::Callbacks
+
+      CONTENT_KEY   = '__content__'.freeze
+      HASH_SIZE_KEY = '__hash_size__'.freeze
+
+      attr_reader :hash
+
+      def current_hash
+        @hash_stack.last
+      end
+
+      def on_start_document
+        @hash = {}
+        @hash_stack = [@hash]
+      end
+
+      def on_end_document
+        raise "Parse stack not empty!" if @hash_stack.size > 1
+      end
+
+      def on_error(error_message)
+        raise LibXML::XML::Error, error_message
+      end
+
+      def on_start_element(name, attrs = {})
+        new_hash = { CONTENT_KEY => '' }.merge(attrs)
+        new_hash[HASH_SIZE_KEY] = new_hash.size + 1
+
+        case current_hash[name]
+          when Array then current_hash[name] << new_hash
+          when Hash  then current_hash[name] = [current_hash[name], new_hash]
+          when nil   then current_hash[name] = new_hash
+        end
+
+        @hash_stack.push(new_hash)
+      end
+
+      def on_end_element(name)
+        if current_hash.length > current_hash.delete(HASH_SIZE_KEY) && current_hash[CONTENT_KEY].blank? || current_hash[CONTENT_KEY] == ''
+          current_hash.delete(CONTENT_KEY)
+        end
+        @hash_stack.pop
+      end
+
+      def on_characters(string)
+        current_hash[CONTENT_KEY] << string
+      end
+
+      alias_method :on_cdata_block, :on_characters
+    end
+
+    attr_accessor :document_class
+    self.document_class = HashBuilder
+
+    def parse(data)
+      if !data.respond_to?(:read)
+        data = StringIO.new(data || '')
+      end
+
+      char = data.getc
+      if char.nil?
+        {}
+      else
+        data.ungetc(char)
+        document = self.document_class.new
+        parser = LibXML::XML::SaxParser.io(data)
+        parser.callbacks = document
+        parser.parse
+        document.hash
+      end
+    end
+  end
+end
\ No newline at end of file