diff options
Diffstat (limited to 'spec')
| -rw-r--r-- | spec/controllers/request_controller_spec.rb | 6 | ||||
| -rw-r--r-- | spec/lib/alaveteli_text_masker_spec.rb | 7 | ||||
| -rw-r--r-- | spec/lib/basic_encoding_spec.rb | 24 | ||||
| -rw-r--r-- | spec/models/censor_rule_spec.rb | 58 | ||||
| -rw-r--r-- | spec/models/foi_attachment_spec.rb | 74 |
5 files changed, 144 insertions, 25 deletions
diff --git a/spec/controllers/request_controller_spec.rb b/spec/controllers/request_controller_spec.rb index a5534e9ff..9e2e1bff7 100644 --- a/spec/controllers/request_controller_spec.rb +++ b/spec/controllers/request_controller_spec.rb @@ -608,7 +608,7 @@ describe RequestController, "when showing one request" do response.body.should match('dull') end - it "should censor attachments downloaded as binary" do + it "should censor attachments downloaded directly" do ir = info_requests(:fancy_dog_request) censor_rule = CensorRule.new @@ -623,7 +623,7 @@ describe RequestController, "when showing one request" do get :get_attachment, :incoming_message_id => ir.incoming_messages[1].id, :id => ir.id, :part => 2, :file_name => 'hello world.txt', :skip_cache => 1 response.content_type.should == "text/plain" - response.should contain "xxxxxx hello" + response.should contain "Mouse hello" ensure ir.censor_rules.clear end @@ -645,7 +645,7 @@ describe RequestController, "when showing one request" do get :get_attachment, :incoming_message_id => ir.incoming_messages[1].id, :id => ir.id, :part => 2, :file_name => 'hello world.txt', :skip_cache => 1 response.content_type.should == "text/plain" - response.should contain "xxxxxx hello" + response.should contain "Mouse hello" ensure ir.user.censor_rules.clear end 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 diff --git a/spec/lib/basic_encoding_spec.rb b/spec/lib/basic_encoding_spec.rb index d77465ad8..6758d60a3 100644 --- a/spec/lib/basic_encoding_spec.rb +++ b/spec/lib/basic_encoding_spec.rb @@ -160,21 +160,24 @@ describe "convert_string_to_utf8" do describe "when passed uninterpretable character data" do - it "should return it as a utf8 string" do + it "should return it as a valid utf8 string with non-utf8 characters removed + and mark it as scrubbed" do converted = convert_string_to_utf8 random_string - converted.should == random_string if String.method_defined?(:encode) - converted.encoding.to_s.should == 'UTF-8' + converted.string.encoding.to_s.should == 'UTF-8' + converted.string.valid_encoding?.should == true end + converted.scrubbed?.should == true converted = convert_string_to_utf8 random_string,'UTF-8' - converted.should == random_string if String.method_defined?(:encode) - converted.encoding.to_s.should == 'UTF-8' + converted.string.encoding.to_s.should == 'UTF-8' + converted.string.valid_encoding?.should == true end + converted.scrubbed?.should == true end end @@ -185,11 +188,13 @@ describe "convert_string_to_utf8" do converted = convert_string_to_utf8 windows_1252_string - converted.should == "DASH – DASH" + converted.string.should == "DASH – DASH" if String.method_defined?(:encode) - converted.encoding.to_s.should == 'UTF-8' + converted.string.encoding.to_s.should == 'UTF-8' end + converted.scrubbed?.should == false + end end @@ -200,11 +205,12 @@ describe "convert_string_to_utf8" do converted = convert_string_to_utf8 gb_18030_spam_string - converted.should start_with("贵公司负责人") + converted.string.should start_with("贵公司负责人") if String.method_defined?(:encode) - converted.encoding.to_s.should == 'UTF-8' + converted.string.encoding.to_s.should == 'UTF-8' end + converted.scrubbed?.should == false end end 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 |