Changeset 147

Timestamp:

06/04/2007 23:10:53 (18 months ago)

Author:

why

Message:

• lib/hpricot/parse.rb: well well well. a bunch of tags and attrs weren't getting properly downcased by the html cleaner. thanks aaronp!
  

Location:

trunk

Files:

• lib/hpricot/parse.rb (modified) (2 diffs)
• test/files/tenderlove.html (added)
• test/test_parser.rb (modified) (1 diff)

trunk/lib/hpricot/parse.rb

@@ -51 +51 @@
       Hpricot.scan(input) do |token|
         if stack.last[5] == :CDATA and ![:procins, :comment, :cdata].include?(token[0]) and
-            !(token[0] == :etag and token[1].send(conv) == stack.last[0])
+            !(token[0] == :etag and token[1].casecmp(stack.last[0]).zero?)
           token[0] = :text
           token[1] = token[3] if token[3]
         end
 
-        if token[0] == :emptytag and ElementContent[token[1].send(conv)] != :EMPTY and !opts[:xml]
+        if !opts[:xml] and token[0] == :emptytag and ElementContent[token[1].downcase] != :EMPTY
           token[0] = :stag
+        end
+
+        # TODO: downcase instead when parsing attributes?
+        if !opts[:xml] and token[2].is_a?(Hash)
+          token[2] = token[2].inject({}) { |hsh,(k,v)| hsh[k.downcase] = v; hsh }
         end
 
@@ -66 +71 @@
           end
 
-          stagname = token[0] = token[1].send(conv)
+          stagname = token[0] = token[1] = token[1].send(conv)
           if ElementContent[stagname] == :EMPTY and !opts[:xml]
             token[0] = :emptytag

trunk/test/test_parser.rb

@@ -174 +174 @@
   end
 
+  def test_tag_case 
+    @tenderlove = Hpricot.parse(TestFiles::TENDERLOVE) 
+    assert_equal 2, @tenderlove.search('//a').length 
+    assert_equal 2, @tenderlove.search('//area').length 
+    assert_equal 2, @tenderlove.search('//meta').length 
+  end 
+
   def test_alt_predicates
     @boingboing = Hpricot.parse(TestFiles::BOINGBOING)