Use ASCII-8BIT for replacement patterns when handling binary data.

Also be explicit about using UTF-8 when handling text data.

1 files changed, 5 insertions, 2 deletions

diff --git a/spec/lib/alaveteli_text_masker_spec.rb b/spec/lib/alaveteli_text_masker_spec.rb
index f2d52c1cc..f8c22a849 100644
--- a/spec/lib/alaveteli_text_masker_spec.rb
+++ b/spec/lib/alaveteli_text_masker_spec.rb
@@ -31,10 +31,13 @@ describe AlaveteliTextMasker do
                 data.should == "There was a xxxxx called xxxxxxx, he wished that he was xxxx."
             end
 
-            it 'should handle multibyte characters correctly' do
+            it 'should handle multibyte characters in binary file types as binary data' do
                 data = 'á mouse'
+                if String.method_defined?(:encode)
+                    data = data.force_encoding("ASCII-8BIT")
+                end
                 @regex_censor_rule.text = 'á'
-                apply_masks!(data, "application/octet-stream", :censor_rules => @censor_rules).should == 'x mouse'
+                apply_masks!(data, "application/octet-stream", :censor_rules => @censor_rules).should == 'xx mouse'
             end
 
             it "should apply censor rules to HTML files" do