Merge pull request #20 from G5/normalize-u-property-values

Normalize u-* property values

README.md

@@ -20,10 +20,10 @@ Implemented:
 * nested microformat with associated property
 * dynamic creation of properties
 * [rel](http://microformats.org/wiki/rel)
+* [normalize u-* property values](http://microformats.org/wiki/microformats2-parsing-faq#normalizing_u-.2A_property_values)
 
 Not Implemented:
 
-* [normalize u-* property values](http://microformats.org/wiki/microformats2-parsing-faq#normalizing_u-.2A_property_values)
 * nested microformat without associated property
 * [value-class-pattern](http://microformats.org/wiki/value-class-pattern)
 * [include-pattern](http://microformats.org/wiki/include-pattern)

lib/microformats2/collection.rb

@@ -25,7 +25,7 @@ module Microformats2
     end
 
     def all
-      @all ||= FormatParser.parse(@element).each do |format|
+      @all ||= FormatParser.parse(@element, @base).each do |format|
         save_format_name(format.method_name)
         define_method(format.method_name)
         set_value(format.method_name, format)

lib/microformats2/format.rb

@@ -4,8 +4,9 @@ module Microformats2
 
     attr_reader :method_name
 
-    def initialize(element)
+    def initialize(element, base)
       @element = element
+      @base = base
       @method_name = to_method_name(format_types.first)
       @property_names = []
     end
@@ -27,20 +28,20 @@ module Microformats2
     end
 
     def parse_properties
-      PropertyParser.parse(@element.children).each do |property|
+      PropertyParser.parse(@element.children, @base).each do |property|
         assign_property(property)
       end
     end
 
     def add_property(property_class, value)
-      property = Property.new(nil, property_class, value)
+      property = Property.new(nil, property_class, value, @base)
       assign_property(property)
     end
 
     def parse_implied_properties
       ip = []
       ip << ImpliedProperty::Name.new(@element).parse unless property_present?(:name)
-      ip << ImpliedProperty::Url.new(@element).parse unless property_present?(:url)
+      ip << ImpliedProperty::Url.new(@element, @base).parse unless property_present?(:url)
       ip << ImpliedProperty::Photo.new(@element).parse unless property_present?(:photo)
       ip.compact.each do |property|
         save_property_name(property.method_name)

lib/microformats2/format_parser.rb

@@ -1,7 +1,8 @@
 module Microformats2
   class FormatParser
     class << self
-      def parse(element)
+      def parse(element, base=nil)
+        @@base = base
         parse_node(element).flatten.compact
       end
 
@@ -31,7 +32,7 @@ module Microformats2
         const_name = constant_name(html_class)
         klass = find_or_create_ruby_class(const_name)
 
-        klass.new(element).parse
+        klass.new(element, @@base).parse
       end
 
       def format_classes(element)

lib/microformats2/implied_property/foundation.rb

@@ -2,8 +2,9 @@ module Microformats2
   module ImpliedProperty
     class Foundation
 
-      def initialize(element)
+      def initialize(element, base=nil)
         @element = element
+        @base = base
       end
 
       def parse

lib/microformats2/implied_property/url.rb

@@ -6,6 +6,22 @@ module Microformats2
         "url"
       end
 
+      def to_s
+        @to_s = absolutize(super.to_s) if super.to_s != ""
+      end
+
+      # TODO: make dry, repeated in Collection
+      def absolutize(href)
+        uri = URI.parse(href)
+
+        if @base && !uri.absolute?
+          uri = URI.join(@base, href)
+        end
+
+        uri.normalize!
+        uri.to_s
+      end
+
       protected
 
       def name_map

lib/microformats2/property.rb

@@ -8,13 +8,13 @@ module Microformats2
       "e" => Embedded }
 
     class << self
-	    def new(element, property_class, value=nil)
+	    def new(element, property_class, value=nil, base=nil)
 	      # p-class-name -> p
 	      prefix = property_class.split("-").first
 	      # find ruby class for kind of property
 	      klass = PREFIX_CLASS_MAP[prefix]
 	      raise InvalidPropertyPrefix unless klass
-	      klass.new(element, property_class, value)
+	      klass.new(element, property_class, value, base)
 	    end
 	  end
   end

lib/microformats2/property/foundation.rb

@@ -3,10 +3,11 @@ module Microformats2
     class Foundation
       attr_reader :method_name
 
-      def initialize(element, html_class, string_value=nil)
+      def initialize(element, html_class, string_value=nil, base=nil)
         @element = element
         @method_name = to_method_name(html_class)
         @string_value = string_value
+        @base = base
       end
 
       def parse

lib/microformats2/property/url.rb

@@ -2,6 +2,22 @@ module Microformats2
   module Property
     class Url < Foundation
 
+      def to_s
+        @to_s = absolutize(super.to_s)
+      end
+
+      # TODO: make dry, repeated in Collection
+      def absolutize(href)
+        uri = URI.parse(href)
+
+        if @base && !uri.absolute?
+          uri = URI.join(@base, href)
+        end
+
+        uri.normalize!
+        uri.to_s
+      end
+
       protected
 
       def attr_map

lib/microformats2/property_parser.rb

@@ -1,7 +1,8 @@
 module Microformats2
   class PropertyParser
     class << self
-      def parse(element)
+      def parse(element, base)
+        @@base = base
         parse_node(element).flatten.compact
       end
 
@@ -26,8 +27,8 @@ module Microformats2
 
       def parse_property(element)
         property_classes(element).map do |property_class|
-          property   = Property.new(element, property_class).parse
+          property   = Property.new(element, property_class, nil, @@base).parse
-          properties = format_classes(element).empty? ? PropertyParser.parse(element.children) : []
+          properties = format_classes(element).empty? ? PropertyParser.parse(element.children, @@base) : []
 
           [property].concat properties
         end

spec/lib/microformats2/collection_spec.rb

@@ -35,13 +35,13 @@ describe Microformats2::Collection do
           @collection.card.name.should be_kind_of Microformats2::Property::Text
         end
       end
-      describe "'.h-card .p-url'" do
+      describe "'.h-card .u-url'" do
         it "assigns all urls to HCard#urls" do
-          urls = ["http://flickr.com/jlsuttles", "http://twitter.com/jlsuttles"]
+          urls = ["http://example.org/", "http://flickr.com/", "http://twitter.com/jlsuttles"]
           @collection.card.urls.map(&:to_s).should == urls
         end
         it "assigns then first url to HCard#url" do
-          @collection.card.url.to_s.should == "http://flickr.com/jlsuttles"
+          @collection.card.url.to_s.should == "http://example.org/"
         end
         it "HCard#url is a Property::Url" do
           @collection.card.url.should be_kind_of Microformats2::Property::Url

spec/lib/microformats2/implied_property/url_spec.rb

@@ -15,6 +15,33 @@ describe Microformats2::ImpliedProperty::Url do
         end
       end
     end
+
+    describe "url-relative.html" do
+      html = "spec/support/lib/microformats2/implied_property/url-relative.html"
+      collection = Microformats2.parse(html)
+      it "should have 2 microformats" do
+        collection.all.length.should == 2
+      end
+      collection.all.each_with_index do |format, index|
+        it "implies url to be 'http://example.org/' in case #{index+1}" do
+          format.url.to_s.should == "http://example.org/"
+        end
+      end
+    end
+
+    describe "url-unnormalized.html" do
+      html = "spec/support/lib/microformats2/implied_property/url-unnormalized.html"
+      collection = Microformats2.parse(html)
+      it "should have 2 microformats" do
+        collection.all.length.should == 2
+      end
+      collection.all.each_with_index do |format, index|
+        it "implies url to be 'http://github.com/' in case #{index+1}" do
+          format.url.to_s.should == "http://github.com/"
+        end
+      end
+    end
+
     describe "url-fail.html" do
       html = "spec/support/lib/microformats2/implied_property/url-fail.html"
       collection = Microformats2.parse(html)
@@ -22,8 +49,8 @@ describe Microformats2::ImpliedProperty::Url do
         collection.all.length.should == 2
       end
       collection.all.each_with_index do |format, index|
-        it "implies url to be '' in case #{index+1}" do
+        it "does not imply url in case #{index+1}" do
-          expect {format.url.to_s.should == ""}.to raise_error(NoMethodError)
+          expect {format.url}.to raise_error(NoMethodError)
         end
       end
     end

spec/support/lib/microformats2/implied_property/url-relative.html

@@ -0,0 +1,13 @@
+<!DOCTYPE html>
+<html>
+<body>
+<base href="http://example.org/">
+
+<a class="h-card" href="/" />
+
+<div class="h-card">
+  <a href="/" />
+</div>
+
+</body>
+</html>

spec/support/lib/microformats2/implied_property/url-unnormalized.html

@@ -0,0 +1,12 @@
+<!DOCTYPE html>
+<html>
+<body>
+
+<a class="h-card" href="http://github.com" />
+
+<div class="h-card">
+  <a href="http://github.com" />
+</div>
+
+</body>
+</html>

spec/support/lib/microformats2/simple.html

@@ -1,10 +1,14 @@
 <!DOCTYPE html>
 <html>
+<base href="http://example.org/">
 <body>
   <div class="h-card">
-    <a href="http://flickr.com/jlsuttles" class="u-url p-name">
+    <a href="/" class="u-url p-name">
       Jessica Lynn Suttles
     </a>
+    <a href="http://flickr.com" class="u-url">
+      @jlsuttles
+    </a>
     <a href="http://twitter.com/jlsuttles" class="u-url">
       @jlsuttles
     </a>

spec/support/lib/microformats2/simple.js

@@ -1,7 +1,7 @@
 { "items": [{
     "type": ["h-card"],
     "properties": {
-      "url": ["http://flickr.com/jlsuttles", "http://twitter.com/jlsuttles"],
+      "url": ["http://example.org/", "http://flickr.com/", "http://twitter.com/jlsuttles"],
       "name": ["Jessica Lynn Suttles"],
       "bday": ["1990-10-15"],
       "content": ["<p>Vegan. Cat lover. Coder.</p>"]