2 files changed, 121 insertions, 11 deletions
diff --git a/spec/models/censor_rule_spec.rb b/spec/models/censor_rule_spec.rb
index 314b060d2..d308ac1b9 100644
--- a/spec/models/censor_rule_spec.rb
+++ b/spec/models/censor_rule_spec.rb
@@ -64,19 +64,35 @@ describe CensorRule, "substituting things" do
             @censor_rule.replacement = "hello"
         end
 
-        it 'should do basic text substitution' do
-            body = "I don't know why you say goodbye"
-            @censor_rule.apply_to_text!(body)
-            body.should == "I don't know why you say hello"
+        describe :apply_to_text do
+
+            it 'should do basic text substitution' do
+                body = "I don't know why you say goodbye"
+                @censor_rule.apply_to_text!(body)
+                body.should == "I don't know why you say hello"
+            end
+
         end
 
-        it 'should keep size same for binary substitution' do
-            body = "I don't know why you say goodbye"
-            orig_body = body.dup
-            @censor_rule.apply_to_binary!(body)
-            body.size.should == orig_body.size
-            body.should == "I don't know why you say xxxxxxx"
-            body.should_not == orig_body # be sure duplicated as expected
+        describe :apply_to_binary do
+
+            it 'should keep size same for binary substitution' do
+                body = "I don't know why you say goodbye"
+                orig_body = body.dup
+                @censor_rule.apply_to_binary!(body)
+                body.size.should == orig_body.size
+                body.should == "I don't know why you say xxxxxxx"
+                body.should_not == orig_body # be sure duplicated as expected
+            end
+
+            it 'should handle a UTF-8 rule and ASCII-8BIT text' do
+                body = "I don't know why you say g‘oodbye"
+                body.force_encoding("ASCII-8BIT") if String.method_defined?(:encode)
+                @censor_rule.text = 'g‘oodbye'
+                @censor_rule.apply_to_binary!(body)
+                body.should == "I don't know why you say xxxxxxxxxx"
+            end
+
         end
 
     end
@@ -121,6 +137,26 @@ xxxxxxxxx
 BODY
         end
 
+        it "handles a UTF-8 rule with ASCII-8BIT text" do
+            @censor_rule.text = "--PRIVATE.*--P‘RIVATE"
+            @body =
+<<BODY
+Some public information
+--PRIVATE
+Some private information
+--P‘RIVATE
+BODY
+            @body.force_encoding('ASCII-8BIT') if String.method_defined?(:encode)
+            @censor_rule.apply_to_binary!(@body)
+            @body.should ==
+<<BODY
+Some public information
+xxxxxxxxx
+xxxxxxxxxxxxxxxxxxxxxxxx
+xxxxxxxxxxxx
+BODY
+        end
+
     end
 
 end
diff --git a/spec/models/foi_attachment_spec.rb b/spec/models/foi_attachment_spec.rb
index 9583f4c76..b383e5d09 100644
--- a/spec/models/foi_attachment_spec.rb
+++ b/spec/models/foi_attachment_spec.rb
@@ -50,6 +50,80 @@ describe FoiAttachment do
 
     end
 
+    describe :body do
+
+        it 'returns a binary encoded string when newly created' do
+            foi_attachment = FactoryGirl.create(:body_text)
+            if String.method_defined?(:encode)
+                expect(foi_attachment.body.encoding.to_s).to eq('ASCII-8BIT')
+            end
+        end
+
+
+        it 'returns a binary encoded string when saved' do
+            foi_attachment = FactoryGirl.create(:body_text)
+            foi_attachment = FoiAttachment.find(foi_attachment)
+            if String.method_defined?(:encode)
+                expect(foi_attachment.body.encoding.to_s).to eq('ASCII-8BIT')
+            end
+        end
+
+    end
+
+    describe :body_as_text do
+
+        it 'has a valid UTF-8 string when newly created' do
+            foi_attachment = FactoryGirl.create(:body_text)
+            if String.method_defined?(:encode)
+                expect(foi_attachment.body_as_text.string.encoding.to_s).to eq('UTF-8')
+                expect(foi_attachment.body_as_text.string.valid_encoding?).to be_true
+            end
+        end
+
+        it 'has a valid UTF-8 string when saved' do
+            foi_attachment = FactoryGirl.create(:body_text)
+            foi_attachment = FoiAttachment.find(foi_attachment)
+            if String.method_defined?(:encode)
+                expect(foi_attachment.body_as_text.string.encoding.to_s).to eq('UTF-8')
+                expect(foi_attachment.body_as_text.string.valid_encoding?).to be_true
+            end
+        end
+
+
+        it 'has a true scrubbed? value if the body has been coerced to valid UTF-8' do
+            foi_attachment = FactoryGirl.create(:body_text)
+            foi_attachment.body = "\x0FX\x1C\x8F\xA4\xCF\xF6\x8C\x9D\xA7\x06\xD9\xF7\x90lo"
+            expect(foi_attachment.body_as_text.scrubbed?).to be_true
+        end
+
+        it 'has a false scrubbed? value if the body has not been coerced to valid UTF-8' do
+            foi_attachment = FactoryGirl.create(:body_text)
+            foi_attachment.body = "κόσμε"
+            expect(foi_attachment.body_as_text.scrubbed?).to be_false
+        end
+
+    end
+
+    describe :default_body do
+
+        it 'returns valid UTF-8 for a text attachment' do
+            foi_attachment = FactoryGirl.create(:body_text)
+            if String.method_defined?(:encode)
+                expect(foi_attachment.default_body.encoding.to_s).to eq('UTF-8')
+                expect(foi_attachment.default_body.valid_encoding?).to be_true
+            end
+        end
+
+        it 'returns binary for a PDF attachment' do
+            foi_attachment = FactoryGirl.create(:pdf_attachment)
+            if String.method_defined?(:encode)
+                expect(foi_attachment.default_body.encoding.to_s).to eq('ASCII-8BIT')
+            end
+        end
+
+    end
+
+
     describe :ensure_filename! do
 
         it 'should create a filename for an instance with a blank filename' do