Merge branch 'bzr/golem' of /Users/distler/Sites/code/instiki

test/unit/chunks/nowiki_test.rb

@@ -42,4 +42,10 @@ class NoWikiTest < Test::Unit::TestCase
     )
   end
 
+  def test_sanitize_nowiki_bad_utf8
+    match(NoWiki, "<nowiki>\357elephant &AMP; \302ivory</nowiki>",
+                :plain_text => "\357\277\275elephant &amp; \357\277\275ivory"
+    )
+  end
+
 end

test/unit/sanitize_test.rb

@@ -39,6 +39,12 @@ class SanitizeTest < Test::Unit::TestCase
     assert_equal(output, my_rex(input))
     assert_equal(output2, input.to_utf8)
   end
+  
+  def test_sanitize_malformed_utf8
+    input = "<p>\357elephant &AMP; \302ivory</p>"
+    output = "<p>\357\277\275elephant &amp; \357\277\275ivory</p>"
+    check_sanitization(input, output, output, output)
+  end    
 
   Sanitizer::ALLOWED_ELEMENTS.each do |tag_name|
     define_method "test_should_allow_#{tag_name}_tag" do

test/unit/sanitizer_test.rb

@@ -30,6 +30,12 @@ class SanitizerTest < Test::Unit::TestCase
     assert_equal(output2, input.to_utf8)
   end
 
+  def test_sanitize_malformed_utf8
+    input = "<p>\357elephant &AMP; \302ivory</p>".purify
+    output = "<p>ephant &amp; vory</p>"
+    check_sanitization(input, output, output, output)
+  end    
+
   Sanitizer::ALLOWED_ELEMENTS.each do |tag_name|
     define_method "test_should_allow_#{tag_name}_tag" do
       input       = "<#{tag_name} title='1'>foo <bad>bar</bad> baz</#{tag_name}>"