Allow u-* properties to be invalid URIs

lib/microformats2.rb

@@ -4,6 +4,7 @@ require "json"
 require "active_support/inflector"
 
 require "microformats2/version"
+require "microformats2/absolute_uri"
 require "microformats2/parser"
 require "microformats2/format_parser"
 require "microformats2/property_parser"

lib/microformats2/absolute_uri.rb

@@ -0,0 +1,32 @@
+module Microformats2
+  class AbsoluteUri
+    attr_accessor :base, :relative
+
+    def initialize(base, relative)
+      @base = base
+      @relative = relative
+    end
+
+    def absolutize
+      return nil if relative.nil? or relative == ""
+
+      uri = URI.parse(relative)
+
+      if base && !uri.absolute?
+        uri = URI.join(base.to_s, relative.to_s)
+      end
+
+      uri.normalize!
+      uri.to_s
+    rescue URI::InvalidURIError => e
+      logger.warn e.message
+      relative.to_s
+    end
+
+    def logger
+     @logger ||= Logger.new(STDOUT)
+     @logger.level = Logger::WARN
+     @logger
+    end
+  end
+end

lib/microformats2/collection.rb

@@ -88,7 +88,7 @@ module Microformats2
 
         if rel_values.member?("alternate")
           alternate_inst = {}
-          alternate_inst["url"] = absolutize(rel.attribute("href").text)
+          alternate_inst["url"] = Microformats2::AbsoluteUri.new(@base, rel.attribute("href").text).absolutize
           alternate_inst["rel"] = (rel_values - ["alternate"]).join(" ")
           unless rel.attribute("media").nil?
             alternate_inst["media"] = rel.attribute("media").text
@@ -103,21 +103,10 @@ module Microformats2
         else
           rel_values.each do |rel_value|
             @rels[rel_value] = [] unless @rels.has_key?(rel_value)
-            @rels[rel_value] << absolutize(rel.attribute("href").text)
+            @rels[rel_value] << Microformats2::AbsoluteUri.new(@base, rel.attribute("href").text).absolutize
           end
         end
       end
     end
-
-    def absolutize(href)
-      uri = URI.parse(href)
-
-      if @base && !uri.absolute?
-        uri = URI.join(@base, href)
-      end
-
-      uri.normalize!
-      uri.to_s
-    end
   end
 end

lib/microformats2/implied_property/url.rb

@@ -7,19 +7,7 @@ module Microformats2
       end
 
       def to_s
-        @to_s = absolutize(super.to_s) if super.to_s != ""
+        @to_s = Microformats2::AbsoluteUri.new(@base, super.to_s).absolutize
-      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

lib/microformats2/property/url.rb

@@ -3,19 +3,7 @@ module Microformats2
     class Url < Foundation
 
       def to_s
-        @to_s = absolutize(super.to_s)
+        @to_s = Microformats2::AbsoluteUri.new(@base, super.to_s).absolutize
-      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