julien51-sax-machine 0.0.14 → 0.0.18

data/lib/sax-machine/ns_stack.rb

@@ -0,0 +1,38 @@
+module SAXMachine
+  class NSStack < Hash
+    def initialize(parent=nil, attrs=[])
+      # Initialize
+      super()
+      @parent = parent
+
+      # Parse attributes
+      attrs.each do |attr|
+        if attr.kind_of?(Array)
+          k, v = attr
+          case k
+          when 'xmlns' then self[''] = v
+          when /^xmlns:(.+)/ then self[$1] = v
+          end
+        end
+      end
+    end
+
+    # Lookup
+    def [](name)
+      if (ns = super(name.to_s))
+        # I've got it
+        ns
+      elsif @parent
+        # Parent may have it
+        @parent[name]
+      else
+        # Undefined, empty namespace
+        ''
+      end
+    end
+
+    def pop
+      @parent
+    end
+  end
+end

data/lib/sax-machine/sax_collection_config.rb

@@ -8,16 +8,25 @@ module SAXMachine
         @name   = name.to_s
         @class  = options[:class]
         @as     = options[:as].to_s
+        @xmlns  = case options[:xmlns]
+                  when Array then options[:xmlns]
+                  when String then [options[:xmlns]]
+                  else nil
+                  end
       end
       
-      def handler
-        SAXHandler.new(@class.new)
+      def handler(nsstack)
+        SAXHandler.new(@class.new, nsstack)
       end
       
       def accessor
         as
       end
       
+      def xmlns_match?(ns)
+        @xmlns.nil? || @xmlns.include?(ns)
+      end
+      
     protected
       
       def as
@@ -30,4 +39,4 @@ module SAXMachine
     end
     
   end
-end
+end

data/lib/sax-machine/sax_config.rb

@@ -6,35 +6,48 @@ module SAXMachine
     attr_reader :top_level_elements, :collection_elements
 
     def initialize
-      @top_level_elements  = []
-      @collection_elements = []
+      @top_level_elements  = {}
+      @collection_elements = {}
     end
 
     def add_top_level_element(name, options)
-      @top_level_elements << ElementConfig.new(name, options)
+      @top_level_elements[name.to_s] ||= []
+      @top_level_elements[name.to_s] << ElementConfig.new(name, options)
     end
 
     def add_collection_element(name, options)
-      @collection_elements << CollectionConfig.new(name, options)
+      @collection_elements[name.to_s] ||= []
+      @collection_elements[name.to_s] << CollectionConfig.new(name, options)
     end
 
-    def collection_config(name)
-      @collection_elements.detect { |ce| ce.name.to_s == name.to_s }
+    def collection_config(name, nsstack)
+      prefix, name = name.split(':', 2)
+      prefix, name = nil, prefix unless name  # No prefix
+      namespace = nsstack[prefix]
+
+      (@collection_elements[name.to_s] || []).detect { |ce|
+        ce.name.to_s == name.to_s &&
+        ce.xmlns_match?(namespace)
+      }
     end
 
     def element_configs_for_attribute(name, attrs)
-      @top_level_elements.select do |element_config|
-        element_config.name == name &&
+      name = name.split(':', 2).last
+      (@top_level_elements[name.to_s] || []).select do |element_config|
         element_config.has_value_and_attrs_match?(attrs)
       end
     end
 
-    def element_config_for_tag(name, attrs)
-      @top_level_elements.detect do |element_config|
-        element_config.name == name &&
+    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[name.to_s] || []).detect do |element_config|
+        element_config.xmlns_match?(namespace) &&
         element_config.attrs_match?(attrs)
       end
     end
 
   end
-end
+end

data/lib/sax-machine/sax_document.rb

@@ -39,15 +39,19 @@ 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
-      sax_config.top_level_elements
+      r = []
+      sax_config.top_level_elements.each do |name, ecs|
+        r += ecs
+      end
+      r
     end
 
     def column(sym)
-      columns.select{|c| c.column == sym}[0]
+      sax_config.top_level_elements[sym.to_s][0]
     end
 
     def data_class(sym)
@@ -89,6 +93,14 @@ 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

@@ -1,12 +1,13 @@
 require "nokogiri"
+require "sax-machine/ns_stack"
 
 module SAXMachine
   class SAXHandler < Nokogiri::XML::SAX::Document
     attr_reader :object
 
-    def initialize(object)
+    def initialize(object, nsstack=nil)
       @object = object
-      @parsed_configs = {}
+      @nsstack = nsstack || NSStack.new
     end
 
     def characters(string)
@@ -22,15 +23,16 @@ module SAXMachine
     end
 
     def start_element(name, attrs = [])
-      
+
       @name   = name
       @attrs  = attrs.map { |a| SAXHandler.decode_xml(a) }
+      @nsstack  = NSStack.new(@nsstack, @attrs)
 
       if parsing_collection?
         @collection_handler.start_element(@name, @attrs)
 
-      elsif @collection_config = sax_config.collection_config(@name)
-        @collection_handler = @collection_config.handler
+      elsif @collection_config = sax_config.collection_config(@name, @nsstack)
+        @collection_handler = @collection_config.handler(@nsstack)
         @collection_handler.start_element(@name, @attrs)
 
       elsif (element_configs = sax_config.element_configs_for_attribute(@name, @attrs)).any?
@@ -44,6 +46,7 @@ module SAXMachine
 
     def end_element(name)
       if parsing_collection? && @collection_config.name == name
+        @collection_handler.end_element(name)
         @object.send(@collection_config.accessor) << @collection_handler.object
         reset_current_collection
 
@@ -51,11 +54,11 @@ 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 +78,14 @@ 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 +109,15 @@ 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

@@ -1,11 +1,11 @@
-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"
 
 module SAXMachine
-  VERSION = "0.0.14"
+  VERSION = "0.0.18"
 end

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 +281,57 @@ 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
+
+        it "should get the namespaced element even it's not first" do
+          document = @klass.parse("<root xmlns:a='urn:test'><a>foo</a><a>foo</a><a:a>bar</a:a></root>")
+          document.a.should == 'bar'
+        end
+      end
       
     end
   end
@@ -364,13 +415,16 @@ describe "SAXMachine" do
   end
   
   describe "full example" do
+    XMLNS_ATOM = "http://www.w3.org/2005/Atom"
+    XMLNS_FEEDBURNER = "http://rssnamespace.org/feedburner/ext/1.0"
+
     before :each do
       @xml = File.read('spec/sax-machine/atom.xml')
       class AtomEntry
         include SAXMachine
         element :title
         element :name, :as => :author
-        element "feedburner:origLink", :as => :url
+        element :origLink, :as => :orig_link, :xmlns => XMLNS_FEEDBURNER
         element :summary
         element :content
         element :published
@@ -381,7 +435,7 @@ describe "SAXMachine" do
         element :title
         element :link, :value => :href, :as => :url, :with => {:type => "text/html"}
         element :link, :value => :href, :as => :feed_url, :with => {:type => "application/atom+xml"}
-        elements :entry, :as => :entries, :class => AtomEntry
+        elements :entry, :as => :entries, :class => AtomEntry, :xmlns => XMLNS_ATOM
       end
     end # before
     
@@ -389,5 +443,15 @@ describe "SAXMachine" do
       f = Atom.parse(@xml)
       f.url.should == "http://www.pauldix.net/"
     end
+
+    it "should parse all entries" do
+      f = Atom.parse(@xml)
+      f.entries.length.should == 5
+    end
+
+    it "should parse the feedburner:origLink" do
+      f = Atom.parse(@xml)
+      f.entries[0].orig_link.should == 'http://www.pauldix.net/2008/09/marshal-data-to.html'
+    end
   end
 end

metadata

@@ -1,10 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: julien51-sax-machine
 version: !ruby/object:Gem::Version 
-  version: 0.0.14
+  version: 0.0.18
 platform: ruby
 authors: 
 - Paul Dix
+- astro
+- julien51
+- superfeedr
 autorequire: 
 bindir: bin
 cert_chain: []
@@ -37,6 +40,7 @@ files:
 - lib/sax-machine/sax_element_config.rb
 - lib/sax-machine/sax_document.rb
 - lib/sax-machine/sax_handler.rb
+- lib/sax-machine/ns_stack.rb
 - README.textile
 - Rakefile
 - spec/spec.opts