Handle escaped question marks properly in python text-format.

conformance/conformance_test.cc

@@ -127,8 +127,8 @@ static TestStatus FormatFailureMessage(const TestStatus& input) {
   // Remove newlines
   Normalize(formatted_failure_message);
   // Truncate failure message if needed
-  if (formatted_failure_message.length() > 128) {
-    formatted_failure_message = formatted_failure_message.substr(0, 128);
+  if (formatted_failure_message.length() > 300) {
+    formatted_failure_message = formatted_failure_message.substr(0, 300);
   }
   TestStatus properly_formatted;
   properly_formatted.set_name(input.name());
@@ -303,8 +303,8 @@ std::string ConformanceTestSuite::ConformanceRequestSetting::OutputFormatString(
 }
 
 void ConformanceTestSuite::TruncateDebugPayload(string* payload) {
-  if (payload != nullptr && payload->size() > 200) {
-    payload->resize(200);
+  if (payload != nullptr && payload->size() > 500) {
+    payload->resize(500);
     payload->append("...(truncated)");
   }
 }

conformance/text_format_failure_list_python.txt

@@ -3,10 +3,6 @@
 # TODO: These should be fixed.
 Required.*.TextFormatInput.FloatFieldNoNegativeOctal                                                               # Should have failed to parse, but didn't.
 Required.*.TextFormatInput.FloatFieldNoOctal                                                                       # Should have failed to parse, but didn't.
-Required.*.TextFormatInput.StringLiteralBasicEscapesBytes.ProtobufOutput                                           # Output was not equivalent to reference message: modified: optional_bytes: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\t
-Required.*.TextFormatInput.StringLiteralBasicEscapesBytes.TextFormatOutput                                         # Output was not equivalent to reference message: modified: optional_bytes: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\t
-Required.*.TextFormatInput.StringLiteralBasicEscapesString.ProtobufOutput                                          # Output was not equivalent to reference message: modified: optional_string: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\
-Required.*.TextFormatInput.StringLiteralBasicEscapesString.TextFormatOutput                                        # Output was not equivalent to reference message: modified: optional_string: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\
 
 # Optional float interpreted as `inf`
 Required.*.TextFormatInput.FloatFieldMaxValue.ProtobufOutput                                                       # Output was not equivalent to reference message

conformance/text_format_failure_list_python_cpp.txt

@@ -1,6 +1,2 @@
 Required.*.TextFormatInput.FloatFieldNoNegativeOctal                                                               # Should have failed to parse, but didn't.
 Required.*.TextFormatInput.FloatFieldNoOctal                                                                       # Should have failed to parse, but didn't.
-Required.*.TextFormatInput.StringLiteralBasicEscapesBytes.ProtobufOutput                                           # Output was not equivalent to reference message: modified: optional_bytes: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\t
-Required.*.TextFormatInput.StringLiteralBasicEscapesBytes.TextFormatOutput                                         # Output was not equivalent to reference message: modified: optional_bytes: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\t
-Required.*.TextFormatInput.StringLiteralBasicEscapesString.ProtobufOutput                                          # Output was not equivalent to reference message: modified: optional_string: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\
-Required.*.TextFormatInput.StringLiteralBasicEscapesString.TextFormatOutput                                        # Output was not equivalent to reference message: modified: optional_string: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\

conformance/text_format_failure_list_python_upb.txt

@@ -3,7 +3,3 @@
 # TODO: These should be fixed.
 Required.*.TextFormatInput.FloatFieldNoNegativeOctal                                                               # Should have failed to parse, but didn't.
 Required.*.TextFormatInput.FloatFieldNoOctal                                                                       # Should have failed to parse, but didn't.
-Required.*.TextFormatInput.StringLiteralBasicEscapesBytes.ProtobufOutput                                           # Output was not equivalent to reference message: modified: optional_bytes: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\t
-Required.*.TextFormatInput.StringLiteralBasicEscapesBytes.TextFormatOutput                                         # Output was not equivalent to reference message: modified: optional_bytes: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\t
-Required.*.TextFormatInput.StringLiteralBasicEscapesString.ProtobufOutput                                          # Output was not equivalent to reference message: modified: optional_string: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\
-Required.*.TextFormatInput.StringLiteralBasicEscapesString.TextFormatOutput                                        # Output was not equivalent to reference message: modified: optional_string: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\

python/google/protobuf/text_encoding.py

@@ -97,6 +97,9 @@ def ReplaceHex(m):
   # allow single-digit hex escapes (like '\xf').
   result = _CUNESCAPE_HEX.sub(ReplaceHex, text)
 
+  # Special handling due to C++ trigraphs :(.
+  result = result.replace('\\?', '?')
+
   # Replaces Unicode escape sequences with their character equivalents.
   result = result.encode('raw_unicode_escape').decode('raw_unicode_escape')
   # Encode Unicode characters as UTF-8, then decode to Latin-1 escaping