Merge from google internal

14 files changed, 195 insertions, 38 deletions

diff --git a/python/google/protobuf/internal/any_test.proto b/python/google/protobuf/internal/any_test.proto
index 76a7ebd6..1a563fd9 100644
--- a/python/google/protobuf/internal/any_test.proto
+++ b/python/google/protobuf/internal/any_test.proto
@@ -39,6 +39,7 @@ import "google/protobuf/any.proto";
 message TestAny {
   optional google.protobuf.Any value = 1;
   optional int32 int_value = 2;
+  map<string,int32> map_value = 3;
   extensions 10 to max;
 }
 
diff --git a/python/google/protobuf/internal/api_implementation.py b/python/google/protobuf/internal/api_implementation.py
index bce71bb8..553fcdb6 100755
--- a/python/google/protobuf/internal/api_implementation.py
+++ b/python/google/protobuf/internal/api_implementation.py
@@ -159,7 +159,7 @@ if _implementation_type == 'cpp':
     # Unrecognized cpp implementation. Skipping the unknown fields APIs.
     pass
 else:
-  _python_proto3_preserve_unknowns_default = False
+  _python_proto3_preserve_unknowns_default = True
 
   def GetPythonProto3PreserveUnknownsDefault():
     return _python_proto3_preserve_unknowns_default
diff --git a/python/google/protobuf/internal/json_format_test.py b/python/google/protobuf/internal/json_format_test.py
index b2cf7622..19182b7f 100644
--- a/python/google/protobuf/internal/json_format_test.py
+++ b/python/google/protobuf/internal/json_format_test.py
@@ -50,6 +50,7 @@ from google.protobuf import struct_pb2
 from google.protobuf import timestamp_pb2
 from google.protobuf import wrappers_pb2
 from google.protobuf import unittest_mset_pb2
+from google.protobuf import unittest_pb2
 from google.protobuf.internal import well_known_types
 from google.protobuf import json_format
 from google.protobuf.util import json_format_proto3_pb2
@@ -159,15 +160,15 @@ class JsonFormatTest(JsonFormatBase):
     json_format.Parse(text, parsed_message)
     self.assertEqual(message, parsed_message)
 
-  def testUnknownEnumToJsonError(self):
+  def testUnknownEnumToJsonAndBack(self):
+    text = '{\n  "enumValue": 999\n}'
     message = json_format_proto3_pb2.TestMessage()
     message.enum_value = 999
-    # TODO(jieluo): should accept numeric unknown enum for proto3.
-    with self.assertRaises(json_format.SerializeToJsonError) as e:
-      json_format.MessageToJson(message)
-    self.assertEqual(str(e.exception),
-                     'Enum field contains an integer value which can '
-                     'not mapped to an enum value.')
+    self.assertEqual(json_format.MessageToJson(message),
+                     text)
+    parsed_message = json_format_proto3_pb2.TestMessage()
+    json_format.Parse(text, parsed_message)
+    self.assertEqual(message, parsed_message)
 
   def testExtensionToJsonAndBack(self):
     message = unittest_mset_pb2.TestMessageSetContainer()
@@ -757,11 +758,16 @@ class JsonFormatTest(JsonFormatBase):
         '{"enumValue": "baz"}',
         'Failed to parse enumValue field: Invalid enum value baz '
         'for enum type proto3.EnumType.')
-    # TODO(jieluo): fix json format to accept numeric unknown enum for proto3.
-    self.CheckError(
-        '{"enumValue": 12345}',
-        'Failed to parse enumValue field: Invalid enum value 12345 '
-        'for enum type proto3.EnumType.')
+    # Proto3 accepts numeric unknown enums.
+    text = '{"enumValue": 12345}'
+    json_format.Parse(text, message)
+    # Proto2 does not accept unknown enums.
+    message = unittest_pb2.TestAllTypes()
+    self.assertRaisesRegexp(
+        json_format.ParseError,
+        'Failed to parse optionalNestedEnum field: Invalid enum value 12345 '
+        'for enum type protobuf_unittest.TestAllTypes.NestedEnum.',
+        json_format.Parse, '{"optionalNestedEnum": 12345}', message)
 
   def testParseBadIdentifer(self):
     self.CheckError('{int32Value: 1}',
diff --git a/python/google/protobuf/internal/message_factory_test.py b/python/google/protobuf/internal/message_factory_test.py
index a1b6bb81..6df52ed2 100644
--- a/python/google/protobuf/internal/message_factory_test.py
+++ b/python/google/protobuf/internal/message_factory_test.py
@@ -107,8 +107,17 @@ class MessageFactoryTest(unittest.TestCase):
   def testGetMessages(self):
     # performed twice because multiple calls with the same input must be allowed
     for _ in range(2):
-      messages = message_factory.GetMessages([self.factory_test1_fd,
-                                              self.factory_test2_fd])
+      # GetMessage should work regardless of the order the FileDescriptorProto
+      # are provided. In particular, the function should succeed when the files
+      # are not in the topological order of dependencies.
+
+      # Assuming factory_test2_fd depends on factory_test1_fd.
+      self.assertIn(self.factory_test1_fd.name,
+                    self.factory_test2_fd.dependency)
+      # Get messages should work when a file comes before its dependencies:
+      # factory_test2_fd comes before factory_test1_fd.
+      messages = message_factory.GetMessages([self.factory_test2_fd,
+                                              self.factory_test1_fd])
       self.assertTrue(
           set(['google.protobuf.python.internal.Factory2Message',
                'google.protobuf.python.internal.Factory1Message'],
diff --git a/python/google/protobuf/internal/message_test.py b/python/google/protobuf/internal/message_test.py
index 4622f10f..a303b1aa 100755
--- a/python/google/protobuf/internal/message_test.py
+++ b/python/google/protobuf/internal/message_test.py
@@ -51,6 +51,7 @@ import operator
 import pickle
 import six
 import sys
+import warnings
 
 try:
   import unittest2 as unittest  # PY26
@@ -146,13 +147,22 @@ class MessageTest(BaseTestCase):
     msg = message_module.TestAllTypes()
     self.assertRaises(TypeError, msg.FromString, 0)
     self.assertRaises(Exception, msg.FromString, '0')
-    # TODO(jieluo): Fix cpp extension to check unexpected end-group tag.
+    # TODO(jieluo): Fix cpp extension to raise error instead of warning.
     # b/27494216
+    end_tag = encoder.TagBytes(1, 4)
     if api_implementation.Type() == 'python':
-      end_tag = encoder.TagBytes(1, 4)
       with self.assertRaises(message.DecodeError) as context:
         msg.FromString(end_tag)
       self.assertEqual('Unexpected end-group tag.', str(context.exception))
+    else:
+      with warnings.catch_warnings(record=True) as w:
+        # Cause all warnings to always be triggered.
+        warnings.simplefilter('always')
+        msg.FromString(end_tag)
+        assert len(w) == 1
+        assert issubclass(w[-1].category, RuntimeWarning)
+        self.assertEqual('Unexpected end-group tag: Not all data was converted',
+                         str(w[-1].message))
 
   def testDeterminismParameters(self, message_module):
     # This message is always deterministically serialized, even if determinism
diff --git a/python/google/protobuf/internal/text_format_test.py b/python/google/protobuf/internal/text_format_test.py
index 1214c3ea..a52f133f 100755
--- a/python/google/protobuf/internal/text_format_test.py
+++ b/python/google/protobuf/internal/text_format_test.py
@@ -1,4 +1,5 @@
 #! /usr/bin/env python
+# -*- coding: utf-8 -*-
 #
 # Protocol Buffers - Google's data interchange format
 # Copyright 2008 Google Inc.  All rights reserved.
@@ -298,6 +299,33 @@ class TextFormatTest(TextFormatBase):
     if message_module is unittest_pb2:
       test_util.ExpectAllFieldsSet(self, message)
 
+  def testParseAndMergeUtf8(self, message_module):
+    message = message_module.TestAllTypes()
+    test_util.SetAllFields(message)
+    ascii_text = text_format.MessageToString(message)
+    ascii_text = ascii_text.encode('utf-8')
+
+    parsed_message = message_module.TestAllTypes()
+    text_format.Parse(ascii_text, parsed_message)
+    self.assertEqual(message, parsed_message)
+    if message_module is unittest_pb2:
+      test_util.ExpectAllFieldsSet(self, message)
+
+    parsed_message.Clear()
+    text_format.Merge(ascii_text, parsed_message)
+    self.assertEqual(message, parsed_message)
+    if message_module is unittest_pb2:
+      test_util.ExpectAllFieldsSet(self, message)
+
+    if six.PY2:
+      msg2 = message_module.TestAllTypes()
+      text = (u'optional_string: "café"')
+      text_format.Merge(text, msg2)
+      self.assertEqual(msg2.optional_string, u'café')
+      msg2.Clear()
+      text_format.Parse(text, msg2)
+      self.assertEqual(msg2.optional_string, u'café')
+
   def testParseExotic(self, message_module):
     message = message_module.TestAllTypes()
     text = ('repeated_int64: -9223372036854775808\n'
@@ -399,13 +427,6 @@ class TextFormatTest(TextFormatBase):
                            r'has no value named BARR.'), text_format.Parse,
                           text, message)
 
-    message = message_module.TestAllTypes()
-    text = 'optional_nested_enum: 100'
-    six.assertRaisesRegex(self, text_format.ParseError,
-                          (r'1:23 : Enum type "\w+.TestAllTypes.NestedEnum" '
-                           r'has no value with number 100.'), text_format.Parse,
-                          text, message)
-
   def testParseBadIntValue(self, message_module):
     message = message_module.TestAllTypes()
     text = 'optional_int32: bork'
@@ -920,6 +941,14 @@ class Proto2Tests(TextFormatBase):
         '1:2 : Message type "protobuf_unittest.TestAllTypes" does not have '
         'extensions.'), text_format.Parse, text, message)
 
+  def testParseNumericUnknownEnum(self):
+    message = unittest_pb2.TestAllTypes()
+    text = 'optional_nested_enum: 100'
+    six.assertRaisesRegex(self, text_format.ParseError,
+                          (r'1:23 : Enum type "\w+.TestAllTypes.NestedEnum" '
+                           r'has no value with number 100.'), text_format.Parse,
+                          text, message)
+
   def testMergeDuplicateExtensionScalars(self):
     message = unittest_pb2.TestAllExtensions()
     text = ('[protobuf_unittest.optional_int32_extension]: 42 '
@@ -1113,13 +1142,8 @@ class Proto3Tests(unittest.TestCase):
     message2 = unittest_proto3_arena_pb2.TestAllTypes()
     message.optional_nested_enum = 999
     text_string = text_format.MessageToString(message)
-    # TODO(jieluo): proto3 should support numeric unknown enum.
-    with self.assertRaises(text_format.ParseError) as e:
-      text_format.Parse(text_string, message2)
-      self.assertEqual(999, message2.optional_nested_enum)
-    self.assertEqual(str(e.exception),
-                     '1:23 : Enum type "proto3_arena_unittest.TestAllTypes.'
-                     'NestedEnum" has no value with number 999.')
+    text_format.Parse(text_string, message2)
+    self.assertEqual(999, message2.optional_nested_enum)
 
   def testMergeExpandedAny(self):
     message = any_test_pb2.TestAny()
diff --git a/python/google/protobuf/internal/unknown_fields_test.py b/python/google/protobuf/internal/unknown_fields_test.py
index 9bdb6f27..8b7de2e7 100755
--- a/python/google/protobuf/internal/unknown_fields_test.py
+++ b/python/google/protobuf/internal/unknown_fields_test.py
@@ -92,7 +92,6 @@ class UnknownFieldsTest(BaseTestCase):
     # Verify that proto3 unknown fields behavior.
     default_preserve = (api_implementation
                         .GetPythonProto3PreserveUnknownsDefault())
-    self.assertEqual(False, default_preserve)
     self.expectSerializeProto3(default_preserve)
     api_implementation.SetPythonProto3PreserveUnknownsDefault(
         not default_preserve)
diff --git a/python/google/protobuf/internal/well_known_types.py b/python/google/protobuf/internal/well_known_types.py
index 57b96998..3573770b 100644
--- a/python/google/protobuf/internal/well_known_types.py
+++ b/python/google/protobuf/internal/well_known_types.py
@@ -40,6 +40,7 @@ This files defines well known classes which need extra maintenance including:
 
 __author__ = 'jieluo@google.com (Jie Luo)'
 
+import collections
 from datetime import datetime
 from datetime import timedelta
 import six
@@ -67,13 +68,14 @@ class ParseError(Error):
 class Any(object):
   """Class for Any Message type."""
 
-  def Pack(self, msg, type_url_prefix='type.googleapis.com/'):
+  def Pack(self, msg, type_url_prefix='type.googleapis.com/',
+           deterministic=None):
     """Packs the specified message into current Any message."""
     if len(type_url_prefix) < 1 or type_url_prefix[-1] != '/':
       self.type_url = '%s/%s' % (type_url_prefix, msg.DESCRIPTOR.full_name)
     else:
       self.type_url = '%s%s' % (type_url_prefix, msg.DESCRIPTOR.full_name)
-    self.value = msg.SerializeToString()
+    self.value = msg.SerializeToString(deterministic=deterministic)
 
   def Unpack(self, msg):
     """Unpacks the current Any message into specified message."""
@@ -734,9 +736,30 @@ class Struct(object):
   def __getitem__(self, key):
     return _GetStructValue(self.fields[key])
 
+  def __contains__(self, item):
+    return item in self.fields
+
   def __setitem__(self, key, value):
     _SetStructValue(self.fields[key], value)
 
+  def __delitem__(self, key):
+    del self.fields[key]
+
+  def __len__(self):
+    return len(self.fields)
+
+  def __iter__(self):
+    return iter(self.fields)
+
+  def keys(self):  # pylint: disable=invalid-name
+    return self.fields.keys()
+
+  def values(self):  # pylint: disable=invalid-name
+    return [self[key] for key in self]
+
+  def items(self):  # pylint: disable=invalid-name
+    return [(key, self[key]) for key in self]
+
   def get_or_create_list(self, key):
     """Returns a list for this key, creating if it didn't exist already."""
     if not self.fields[key].HasField('list_value'):
@@ -755,6 +778,8 @@ class Struct(object):
     for key, value in dictionary.items():
       _SetStructValue(self.fields[key], value)
 
+collections.MutableMapping.register(Struct)
+
 
 class ListValue(object):
   """Class for ListValue message type."""
@@ -776,6 +801,9 @@ class ListValue(object):
   def __setitem__(self, index, value):
     _SetStructValue(self.values.__getitem__(index), value)
 
+  def __delitem__(self, key):
+    del self.values[key]
+
   def items(self):
     for i in range(len(self)):
       yield self[i]
@@ -794,6 +822,8 @@ class ListValue(object):
     list_value.Clear()
     return list_value
 
+collections.MutableSequence.register(ListValue)
+
 
 WKTBASES = {
     'google.protobuf.Any': Any,
diff --git a/python/google/protobuf/internal/well_known_types_test.py b/python/google/protobuf/internal/well_known_types_test.py
index 70975da1..291fe4e8 100644
--- a/python/google/protobuf/internal/well_known_types_test.py
+++ b/python/google/protobuf/internal/well_known_types_test.py
@@ -34,6 +34,7 @@
 
 __author__ = 'jieluo@google.com (Jie Luo)'
 
+import collections
 from datetime import datetime
 
 try:
@@ -667,6 +668,8 @@ class StructTest(unittest.TestCase):
 
   def testStruct(self):
     struct = struct_pb2.Struct()
+    self.assertIsInstance(struct, collections.Mapping)
+    self.assertEqual(0, len(struct))
     struct_class = struct.__class__
 
     struct['key1'] = 5
@@ -674,11 +677,13 @@ class StructTest(unittest.TestCase):
     struct['key3'] = True
     struct.get_or_create_struct('key4')['subkey'] = 11.0
     struct_list = struct.get_or_create_list('key5')
+    self.assertIsInstance(struct_list, collections.Sequence)
     struct_list.extend([6, 'seven', True, False, None])
     struct_list.add_struct()['subkey2'] = 9
     struct['key6'] = {'subkey': {}}
     struct['key7'] = [2, False]
 
+    self.assertEqual(7, len(struct))
     self.assertTrue(isinstance(struct, well_known_types.Struct))
     self.assertEqual(5, struct['key1'])
     self.assertEqual('abc', struct['key2'])
@@ -696,6 +701,20 @@ class StructTest(unittest.TestCase):
     struct2.ParseFromString(serialized)
 
     self.assertEqual(struct, struct2)
+    for key, value in struct.items():
+      self.assertIn(key, struct)
+      self.assertIn(key, struct2)
+      self.assertEqual(value, struct2[key])
+
+    self.assertEqual(7, len(struct.keys()))
+    self.assertEqual(7, len(struct.values()))
+    for key in struct.keys():
+      self.assertIn(key, struct)
+      self.assertIn(key, struct2)
+      self.assertEqual(struct[key], struct2[key])
+
+    item = (next(iter(struct.keys())), next(iter(struct.values())))
+    self.assertEqual(item, next(iter(struct.items())))
 
     self.assertTrue(isinstance(struct2, well_known_types.Struct))
     self.assertEqual(5, struct2['key1'])
@@ -756,6 +775,16 @@ class StructTest(unittest.TestCase):
     empty_struct = list2[1]
     self.assertEqual({}, dict(empty_struct.fields))
 
+    self.assertEqual(9, len(struct))
+    del struct['key3']
+    del struct['key4']
+    self.assertEqual(7, len(struct))
+    self.assertEqual(6, len(struct['key5']))
+    del struct['key5'][1]
+    self.assertEqual(5, len(struct['key5']))
+    self.assertEqual([6, True, False, None, inner_struct],
+                     list(struct['key5'].items()))
+
   def testMergeFrom(self):
     struct = struct_pb2.Struct()
     struct_class = struct.__class__
@@ -863,6 +892,20 @@ class AnyTest(unittest.TestCase):
     self.assertTrue(msg.Unpack(unpacked_message))
     self.assertEqual(submessage, unpacked_message)
 
+  def testPackDeterministic(self):
+    submessage = any_test_pb2.TestAny()
+    for i in range(10):
+      submessage.map_value[str(i)] = i * 2
+    msg = any_pb2.Any()
+    msg.Pack(submessage, deterministic=True)
+    serialized = msg.SerializeToString(deterministic=True)
+    golden = (b'\n4type.googleapis.com/google.protobuf.internal.TestAny\x12F'
+              b'\x1a\x05\n\x010\x10\x00\x1a\x05\n\x011\x10\x02\x1a\x05\n\x01'
+              b'2\x10\x04\x1a\x05\n\x013\x10\x06\x1a\x05\n\x014\x10\x08\x1a'
+              b'\x05\n\x015\x10\n\x1a\x05\n\x016\x10\x0c\x1a\x05\n\x017\x10'
+              b'\x0e\x1a\x05\n\x018\x10\x10\x1a\x05\n\x019\x10\x12')
+    self.assertEqual(golden, serialized)
+
 
 if __name__ == '__main__':
   unittest.main()
diff --git a/python/google/protobuf/json_format.py b/python/google/protobuf/json_format.py
index 801eed60..878291db 100644
--- a/python/google/protobuf/json_format.py
+++ b/python/google/protobuf/json_format.py
@@ -251,6 +251,8 @@ class _Printer(object):
       if enum_value is not None:
         return enum_value.name
       else:
+        if field.file.syntax == 'proto3':
+          return value
         raise SerializeToJsonError('Enum field contains an integer value '
                                    'which can not mapped to an enum value.')
     elif field.cpp_type == descriptor.FieldDescriptor.CPPTYPE_STRING:
@@ -675,6 +677,9 @@ def _ConvertScalarFieldValue(value, field, require_str=False):
         raise ParseError('Invalid enum value {0} for enum type {1}.'.format(
             value, field.enum_type.full_name))
       if enum_value is None:
+        if field.file.syntax == 'proto3':
+          # Proto3 accepts unknown enums.
+          return number
         raise ParseError('Invalid enum value {0} for enum type {1}.'.format(
             value, field.enum_type.full_name))
     return enum_value.number
diff --git a/python/google/protobuf/message_factory.py b/python/google/protobuf/message_factory.py
index 15740280..e4fb065e 100644
--- a/python/google/protobuf/message_factory.py
+++ b/python/google/protobuf/message_factory.py
@@ -130,13 +130,22 @@ def GetMessages(file_protos):
   """Builds a dictionary of all the messages available in a set of files.
 
   Args:
-    file_protos: A sequence of file protos to build messages out of.
+    file_protos: Iterable of FileDescriptorProto to build messages out of.
 
   Returns:
     A dictionary mapping proto names to the message classes. This will include
     any dependent messages as well as any messages defined in the same file as
     a specified message.
   """
-  for file_proto in file_protos:
+  # The cpp implementation of the protocol buffer library requires to add the
+  # message in topological order of the dependency graph.
+  file_by_name = {file_proto.name: file_proto for file_proto in file_protos}
+  def _AddFile(file_proto):
+    for dependency in file_proto.dependency:
+      if dependency in file_by_name:
+        # Remove from elements to be visited, in order to cut cycles.
+        _AddFile(file_by_name.pop(dependency))
     _FACTORY.pool.Add(file_proto)
+  while file_by_name:
+    _AddFile(file_by_name.popitem()[1])
   return _FACTORY.GetMessages([file_proto.name for file_proto in file_protos])
diff --git a/python/google/protobuf/pyext/extension_dict.cc b/python/google/protobuf/pyext/extension_dict.cc
index 43ee5d15..6830b10d 100644
--- a/python/google/protobuf/pyext/extension_dict.cc
+++ b/python/google/protobuf/pyext/extension_dict.cc
@@ -32,6 +32,10 @@
 // Author: tibell@google.com (Johan Tibell)
 
 #include <google/protobuf/pyext/extension_dict.h>
+#include <memory>
+#ifndef _SHARED_PTR_H
+#include <google/protobuf/stubs/shared_ptr.h>
+#endif
 
 #include <google/protobuf/stubs/logging.h>
 #include <google/protobuf/stubs/common.h>
@@ -45,7 +49,6 @@
 #include <google/protobuf/pyext/repeated_composite_container.h>
 #include <google/protobuf/pyext/repeated_scalar_container.h>
 #include <google/protobuf/pyext/scoped_pyobject_ptr.h>
-#include <google/protobuf/stubs/shared_ptr.h>
 
 #if PY_MAJOR_VERSION >= 3
   #if PY_VERSION_HEX < 0x03030000
diff --git a/python/google/protobuf/pyext/message.cc b/python/google/protobuf/pyext/message.cc
index 0f54506b..ef75acab 100644
--- a/python/google/protobuf/pyext/message.cc
+++ b/python/google/protobuf/pyext/message.cc
@@ -2065,6 +2065,11 @@ static PyObject* MergeFromString(CMessage* self, PyObject* arg) {
   input.SetExtensionRegistry(factory->pool->pool, factory->message_factory);
   bool success = self->message->MergePartialFromCodedStream(&input);
   if (success) {
+    if (!input.ConsumedEntireMessage()) {
+      // TODO(jieluo): Raise error and return NULL instead.
+      // b/27494216
+      PyErr_Warn(NULL, "Unexpected end-group tag: Not all data was converted");
+    }
     return PyInt_FromLong(input.CurrentPosition());
   } else {
     PyErr_Format(DecodeError_class, "Error parsing message");
diff --git a/python/google/protobuf/text_format.py b/python/google/protobuf/text_format.py
index aaca78ad..6b12632e 100755
--- a/python/google/protobuf/text_format.py
+++ b/python/google/protobuf/text_format.py
@@ -485,7 +485,10 @@ def Parse(text,
     ParseError: On text parsing problems.
   """
   if not isinstance(text, str):
-    text = text.decode('utf-8')
+    if six.PY3:
+      text = text.decode('utf-8')
+    else:
+      text = text.encode('utf-8')
   return ParseLines(text.split('\n'),
                     message,
                     allow_unknown_extension,
@@ -517,6 +520,11 @@ def Merge(text,
   Raises:
     ParseError: On text parsing problems.
   """
+  if not isinstance(text, str):
+    if six.PY3:
+      text = text.decode('utf-8')
+    else:
+      text = text.encode('utf-8')
   return MergeLines(
       text.split('\n'),
       message,
@@ -1559,6 +1567,11 @@ def ParseEnum(field, value):
                        (enum_descriptor.full_name, value))
   else:
     # Numeric value.
+    if hasattr(field.file, 'syntax'):
+      # Attribute is checked for compatibility.
+      if field.file.syntax == 'proto3':
+        # Proto3 accept numeric unknown enums.
+        return number
     enum_value = enum_descriptor.values_by_number.get(number, None)
     if enum_value is None:
       raise ValueError('Enum type "%s" has no value with number %d.' %