julien51-sax-machine 0.0.14 → 0.0.15

data/lib/sax-machine/sax_config.rb

@@ -29,12 +29,17 @@ module SAXMachine
       end
     end
 
-    def element_config_for_tag(name, attrs)
+    def element_config_for_tag(name, attrs, nsstack)
+      prefix, name = name.split(':', 2)
+      prefix, name = nil, prefix unless name  # No prefix
+      namespace = nsstack[prefix]
+
       @top_level_elements.detect do |element_config|
         element_config.name == name &&
+        element_config.xmlns_match?(namespace) &&
         element_config.attrs_match?(attrs)
       end
     end
 
   end
-end
+end

data/lib/sax-machine/sax_document.rb

@@ -39,7 +39,7 @@ module SAXMachine
       # this is how we allow custom parsing behavior. So you could define the setter
       # and have it parse the string into a date or whatever.
       attr_reader options[:as] unless instance_methods.include?(options[:as].to_s)
-      attr_writer options[:as] unless instance_methods.include?("#{options[:as]}=")
+      attr_writer_once options[:as] unless instance_methods.include?("#{options[:as]}=")
     end
 
     def columns
@@ -89,6 +89,14 @@ module SAXMachine
     def sax_config
       @sax_config ||= SAXConfig.new
     end
+
+    def attr_writer_once(attr)
+      class_eval <<-SRC
+          def #{attr}=(val)
+            @#{attr} ||= val
+          end
+        SRC
+    end
   end
   
 end

data/lib/sax-machine/sax_element_config.rb

@@ -30,6 +30,12 @@ module SAXMachine
         end
         @data_class = options[:class]
         @required = options[:required]
+
+        @xmlns = case options[:xmlns]
+                 when Array then options[:xmlns]
+                 when String then [options[:xmlns]]
+                 else nil
+                 end
       end
 
       def column
@@ -55,6 +61,10 @@ module SAXMachine
       def has_value_and_attrs_match?(attrs)
         !@value.nil? && attrs_match?(attrs)
       end
+
+      def xmlns_match?(ns)
+        @xmlns.nil? || @xmlns.include?(ns)
+      end
       
       def collection?
         @collection
@@ -62,4 +72,4 @@ module SAXMachine
     end
     
   end
-end
+end

data/lib/sax-machine/sax_handler.rb

@@ -1,4 +1,5 @@
 require "nokogiri"
+require "sax-machine/ns_stack"
 
 module SAXMachine
   class SAXHandler < Nokogiri::XML::SAX::Document
@@ -6,7 +7,7 @@ module SAXMachine
 
     def initialize(object)
       @object = object
-      @parsed_configs = {}
+      @nsstack = nil
     end
 
     def characters(string)
@@ -25,6 +26,7 @@ module SAXMachine
       
       @name   = name
       @attrs  = attrs.map { |a| SAXHandler.decode_xml(a) }
+      @nsstack  = NSStack.new(@nsstack, @attrs)
 
       if parsing_collection?
         @collection_handler.start_element(@name, @attrs)
@@ -51,11 +53,11 @@ module SAXMachine
         @collection_handler.end_element(name)
 
       elsif characaters_captured?
-        mark_as_parsed
         @object.send(@element_config.setter, @value)
       end
 
       reset_current_tag
+      @nsstack = @nsstack.pop
     end
 
     def characaters_captured?
@@ -75,27 +77,14 @@ module SAXMachine
 
     def parse_element_attributes(element_configs)
       element_configs.each do |ec|
-        unless parsed_config?(ec)
-          @object.send(ec.setter, ec.value_from_attrs(@attrs))
-          mark_as_parsed(ec)
-        end
+        @object.send(ec.setter, ec.value_from_attrs(@attrs))
       end
       @element_config = nil
     end
 
     def set_element_config_for_element_value
       @value = ""
-      @element_config = sax_config.element_config_for_tag(@name, @attrs)
-    end
-
-    def mark_as_parsed(element_config=nil)
-      element_config ||= @element_config
-      @parsed_configs[element_config] = true unless element_config.collection?
-    end
-
-    def parsed_config?(element_config=nil)
-      element_config ||= @element_config
-      @parsed_configs[element_config]
+      @element_config = sax_config.element_config_for_tag(@name, @attrs, @nsstack)
     end
 
     def reset_current_collection
@@ -119,14 +108,15 @@ module SAXMachine
     def self.decode_xml(str)
       return str.map &method(:decode_xml) if str.kind_of?(Array)
 
-      entities = {
-        '#38'   => '&amp;',
-        '#13'   => "\r",
-      } 
-      entities.keys.inject(str) { |string, key|
-        string.gsub(/&#{key};/, entities[key])
-      } 
+      # entities = {
+      #         '#38'   => '&amp;',
+      #         '#13'   => "\r",
+      #       } 
+      #       entities.keys.inject(str) { |string, key|
+      #         string.gsub(/&#{key};/, entities[key])
+      #       } 
+      CGI.unescapeHTML(str)
     end
   
   end
-end
+end

data/lib/sax-machine.rb

@@ -1,7 +1,7 @@
-require "rubygems"
-
 $LOAD_PATH.unshift(File.expand_path(File.dirname(__FILE__))) unless $LOAD_PATH.include?(File.expand_path(File.dirname(__FILE__)))
 
+require "cgi"
+
 require "sax-machine/sax_document"
 require "sax-machine/sax_handler"
 require "sax-machine/sax_config"

data/spec/sax-machine/sax_document_spec.rb

@@ -62,11 +62,11 @@ describe "SAXMachine" do
         document.title.should == "Title"
       end
 
-      it "should overwrite the value when the element is present" do
+      it "should *not* overwrite the value when the element is present (new behaviour!)" do
         document = @klass.new
         document.title = "Old title"
         document.parse("<title>New title</title>")
-        document.title.should == "New title"
+        document.title.should == "Old title"
       end
 
       it "should save the element text into an accessor" do
@@ -138,7 +138,7 @@ describe "SAXMachine" do
           
           it "should escape correctly the ampersand" do
             document = @klass.parse("<link href='http://api.flickr.com/services/feeds/photos_public.gne?id=49724566@N00&amp;lang=en-us&amp;format=atom' foo='bar'>asdf</link>")
-            document.link.should == "http://api.flickr.com/services/feeds/photos_public.gne?id=49724566@N00&amp;lang=en-us&amp;format=atom"
+            document.link.should == "http://api.flickr.com/services/feeds/photos_public.gne?id=49724566@N00&lang=en-us&format=atom"
           end
           
           it "should save the value of a matching element" do
@@ -281,6 +281,52 @@ describe "SAXMachine" do
           document.link_bar.should == 'test2'
         end
       end
+
+      describe "when specifying namespaces" do
+        before :all do
+          @klass = Class.new do
+            include SAXMachine
+            element :a, :xmlns => 'urn:test'
+            element :b, :xmlns => ['', 'urn:test']
+          end
+        end
+
+        it "should get the element with the xmlns" do
+          document = @klass.parse("<a xmlns='urn:test'>hello</a>")
+          document.a.should == 'hello'
+        end
+
+        it "shouldn't get the element without the xmlns" do
+          document = @klass.parse("<a>hello</a>")
+          document.a.should be_nil
+        end
+
+        it "shouldn't get the element with the wrong xmlns" do
+          document = @klass.parse("<a xmlns='urn:test2'>hello</a>")
+          document.a.should be_nil
+        end
+
+        it "should get an element without xmlns if the empty namespace is desired" do
+          document = @klass.parse("<b>hello</b>")
+          document.b.should == 'hello'
+        end
+
+        it "should get an element with the right prefix" do
+          document = @klass.parse("<p:a xmlns:p='urn:test'>hello</p:a>")
+          document.a.should == 'hello'
+        end
+
+        it "should not get an element with the wrong prefix" do
+          document = @klass.parse("<x:a xmlns:p='urn:test' xmlns:x='urn:test2'>hello</x:a>")
+          document.a.should be_nil
+        end
+
+        it "should get a prefixed element without xmlns if the empty namespace is desired" do
+          pending "this needs a less pickier nokogiri push parser"
+          document = @klass.parse("<x:b>hello</x:b>")
+          document.b.should == 'hello'
+        end
+      end
       
     end
   end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: julien51-sax-machine
 version: !ruby/object:Gem::Version 
-  version: 0.0.14
+  version: 0.0.15
 platform: ruby
 authors: 
 - Paul Dix