Encode serialized output to avoid encoding errors. (#44)

lib/console/output/encoder.rb

@@ -0,0 +1,40 @@
+# frozen_string_literal: true
+
+# Released under the MIT License.
+# Copyright, 2023, by Samuel Williams.
+
+module Console
+	module Output
+		class Encoder
+			def initialize(output, encoding = ::Encoding::UTF_8)
+				@output = output
+				@encoding = encoding
+			end
+
+			attr :output
+
+			attr :encoding
+
+			def call(subject = nil, *arguments, **options, &block)
+				subject = encode(subject)
+				arguments = encode(arguments)
+				options = encode(options)
+
+				@output.call(subject, *arguments, **options, &block)
+			end
+
+			def encode(value)
+				case value
+				when String
+					value.encode(@encoding, invalid: :replace, undef: :replace)
+				when Array
+					value.map{|item| encode(item)}
+				when Hash
+					value.transform_values{|item| encode(item)}
+				else
+					value
+				end
+			end
+		end
+	end
+end

lib/console/output/json.rb

@@ -4,12 +4,16 @@
 # Copyright, 2021-2022, by Samuel Williams.
 
 require_relative '../serialized/logger'
+require_relative 'encoder'
 
 module Console
 	module Output
 		module JSON
 			def self.new(output, **options)
-				Serialized::Logger.new(output, format: ::JSON, **options)
+				# The output encoder can prevent encoding issues (e.g. invalid UTF-8):
+				Output::Encoder.new(
+					Serialized::Logger.new(output, format: ::JSON, **options)
+				)
 			end
 		end
 	end

test/console/output.rb

@@ -14,7 +14,7 @@
 			let(:output) {File.open('/tmp/console.log', 'w')}
 
 			it 'should use a serialized format' do
-				expect(Console::Output.new(output, env)).to be_a(Console::Serialized::Logger)
+				expect(Console::Output.new(output, env).output).to be_a(Console::Serialized::Logger)
 			end
 		end
 
@@ -31,7 +31,9 @@
 
 		it 'can set output to Serialized and format to JSON by ENV' do
 			output = Console::Output.new(StringIO.new, {'CONSOLE_OUTPUT' => 'JSON'})
+			expect(output).to be_a(Console::Output::Encoder)
 
+			output = output.output
 			expect(output).to be_a Console::Serialized::Logger
 			expect(output.format).to be == JSON
 		end
@@ -64,4 +66,18 @@
 			expect(output.terminal).to be_a Console::Terminal::Text
 		end
 	end
+
+	with "invalid UTF-8" do
+		let(:capture) {StringIO.new}
+
+		it "should replace invalid characters" do
+			expect(capture).to receive(:tty?).and_return(false)
+			output = Console::Output.new(capture, {})
+
+			output.call("Hello \xFF")
+
+			message = JSON.parse(capture.string)
+			expect(message['subject']).to be == "Hello \uFFFD"
+		end
+	end
 end

test/console/output/encoder.rb

@@ -0,0 +1,40 @@
+# frozen_string_literal: true
+
+# Released under the MIT License.
+# Copyright, 2023, by Samuel Williams.
+
+require 'console/output/encoder'
+require 'console/capture'
+
+describe Console::Output::Encoder do
+	let(:output) {Console::Capture.new}
+	let(:encoder) {subject.new(output)}
+
+	let(:invalid_string) {"hello \xc3\x28 world"}
+	it "is an invalid string" do
+		expect(invalid_string).not.to be(:valid_encoding?)
+	end
+
+	it "can fix encoding" do
+		valid_string = encoder.encode(invalid_string)
+		expect(valid_string).to be(:valid_encoding?)
+	end
+
+	it "can encode hashes" do
+		invalid = {key: invalid_string}
+		valid = encoder.encode(invalid)
+
+		expect(valid[:key]).to be(:valid_encoding?)
+	end
+
+	it "can encode arrays" do
+		invalid = [invalid_string]
+		valid = encoder.encode(invalid)
+
+		expect(valid.first).to be(:valid_encoding?)
+	end
+
+	it "ignores non-strings" do
+		expect(encoder.encode(1)).to be == 1
+	end
+end