Merge pull request #55 from sanger/fix-logger-tagger-incompatible-character-encodings

Fix logger tagger incompatible character encodings

Author: yoldas

Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    sanger_warren (0.5.1)
+    sanger_warren (0.5.2)
       bunny (~> 2.17)
       connection_pool (~> 2.2)
       multi_json (>= 1.0, < 1.18.0)

lib/warren/log_tagger.rb

@@ -51,8 +51,14 @@ def self.level(name)
 
     private
 
+    # Applies the tag to the message, ensuring it is valid UTF-8 and safe for logging.
+    #
+    # @param message [Object] The message to tag
+    # @return [String] The tagged message
     def tag(message)
-      "#{@tag}: #{message}"
+      msg = message.to_s.dup.force_encoding('BINARY') # handle any encoding
+      msg = msg.encode('UTF-8', invalid: :replace, undef: :replace, replace: '?')
+      "#{@tag}: #{msg}"
     end
   end
 end

lib/warren/version.rb

@@ -2,5 +2,5 @@
 
 module Warren
   # Gem version number. Bump prior to release of new version
-  VERSION = '0.5.1'
+  VERSION = '0.5.2'
 end

spec/warren/log_tagger_spec.rb

@@ -16,24 +16,41 @@
     allow(standard_logger).to receive(:error) { |&block| block&.call }
   end
 
-  shared_examples 'a logger' do |method|
+  shared_examples 'a logger' do |method| # rubocop:disable Metrics/BlockLength
+    let(:program) { 'program' }
+    let(:message) { 'message' }
+
     context 'without a block' do
-      before { logger.send(method, 'message') }
+      before { logger.send(method, message) }
 
       it 'tags the message' do
         expect(standard_logger).to have_received(method).with('tag: message')
       end
+
+      context 'with invalid UTF-8 bytes' do
+        let(:message) { "message\xFF".dup.force_encoding('ASCII-8BIT') }
+
+        it 'tags the message, replacing invalid bytes' do
+          expect(standard_logger).to have_received(method).with('tag: message?')
+        end
+      end
     end
 
     context 'with a block' do
-      subject(:method_call) { logger.send(method, 'program') { 'message' } }
+      subject(:method_call) { logger.send(method, program) { message } }
 
       it 'calls the logger as normal' do
         method_call
-        expect(standard_logger).to have_received(method).with('program')
+        expect(standard_logger).to have_received(method).with(program)
       end
 
       it { is_expected.to eq 'tag: message' }
+
+      context 'with invalid UTF-8 bytes' do
+        let(:message) { "message\xFF".dup.force_encoding('ASCII-8BIT') }
+
+        it { is_expected.to eq 'tag: message?' }
+      end
     end
   end