Escape invalid byte sequence in Exception

aycabta · aycabta · commit 0815317d42c0 · 2020-12-28T10:45:17.000+09:00
This fixes #141.
diff --git a/lib/irb.rb b/lib/irb.rb
@@ -578,6 +578,29 @@ def eval_input
       end
     end
 
+    def convert_invalid_byte_sequence(str)
+      str = str.force_encoding(Encoding::ASCII_8BIT)
+      conv = Encoding::Converter.new(Encoding::ASCII_8BIT, Encoding::UTF_8)
+      dst = String.new
+      begin
+        ret = conv.primitive_convert(str, dst)
+        case ret
+        when :invalid_byte_sequence
+          conf.insert_output(conf.primitive_errinfo[3].dump[1..-2])
+          redo
+        when :undefined_conversion
+          c = conv.primitive_errinfo[3].dup.force_encoding(conv.primitive_errinfo[1])
+          conv.insert_output(c.dump[1..-2])
+          redo
+        when :incomplete_input
+          conv.insert_output(conv.primitive_errinfo[3].dump[1..-2])
+        when :finished
+        end
+        break
+      end while nil
+      dst
+    end
+
     def handle_exception(exc)
       if exc.backtrace && exc.backtrace[0] =~ /\/irb(2)?(\/.*|-.*|\.rb)?:/ && exc.class.to_s !~ /^IRB/ &&
          !(SyntaxError === exc) && !(EncodingError === exc)
@@ -621,7 +644,8 @@ def handle_exception(exc)
         end
         puts messages.reverse
       end
-      m = exc.to_s.split(/\n/)
+      converted_exc_s = convert_invalid_byte_sequence(exc.to_s.dup)
+      m = converted_exc_s.split(/\n/)
       print "#{attr[1]}#{exc.class} (#{attr[4]}#{m.shift}#{attr[0, 1]})#{attr[]}\n"
       puts m.map {|s| "#{attr[1]}#{s}#{attr[]}\n"}
       if attr == ATTR_PLAIN
diff --git a/test/irb/test_raise_no_backtrace_exception.rb b/test/irb/test_raise_no_backtrace_exception.rb
@@ -13,5 +13,13 @@ def e.backtrace; nil; end
       raise e
 IRB
     end
+
+    def test_raise_exception_with_invalid_byte_sequence
+      skip if RUBY_ENGINE == 'truffleruby'
+      bundle_exec = ENV.key?('BUNDLE_GEMFILE') ? ['-rbundler/setup'] : []
+      assert_in_out_err(bundle_exec + %w[-rirb -W0 -e IRB.start(__FILE__) -- -f --], <<~IRB, /StandardError \(A\\xF3B\)/, [])
+        raise StandardError, "A\\xf3B"
+      IRB
+    end
   end
 end
