From baf52bd79e80b019370a1bfda70e46ff6a692e2b Mon Sep 17 00:00:00 2001 From: Chad Whipkey Date: Fri, 15 Apr 2016 09:23:50 -0700 Subject: [PATCH 1/8] Change protobuf CPP proto generator to support the 'lite' option in proto3. Added a couple unit test proto3 files, for arena_lite and lite. Cloned the proto3_arena_unittest to test some of the basics of generated code (and to ensure that the generated proto3 test files are used by some test). --- BUILD | 4 + src/Makefile.am | 8 + .../protobuf/compiler/cpp/cpp_message.cc | 21 +- src/google/protobuf/descriptor.cc | 5 - src/google/protobuf/descriptor_unittest.cc | 12 - .../protobuf/proto3_arena_lite_unittest.cc | 164 ++++++++++++++ src/google/protobuf/proto3_lite_unittest.cc | 145 ++++++++++++ .../protobuf/unittest_proto3_arena_lite.proto | 207 ++++++++++++++++++ .../protobuf/unittest_proto3_lite.proto | 206 +++++++++++++++++ 9 files changed, 743 insertions(+), 29 deletions(-) create mode 100644 src/google/protobuf/proto3_arena_lite_unittest.cc create mode 100644 src/google/protobuf/proto3_lite_unittest.cc create mode 100644 src/google/protobuf/unittest_proto3_arena_lite.proto create mode 100644 src/google/protobuf/unittest_proto3_lite.proto diff --git a/BUILD b/BUILD index 1615486fae..09d917b719 100644 --- a/BUILD +++ b/BUILD @@ -376,6 +376,8 @@ RELATIVE_TEST_PROTOS = [ "google/protobuf/unittest_preserve_unknown_enum.proto", "google/protobuf/unittest_preserve_unknown_enum2.proto", "google/protobuf/unittest_proto3_arena.proto", + "google/protobuf/unittest_proto3_arena_lite.proto", + "google/protobuf/unittest_proto3_lite.proto", "google/protobuf/unittest_well_known_types.proto", "google/protobuf/util/internal/testdata/anys.proto", "google/protobuf/util/internal/testdata/books.proto", @@ -461,6 +463,8 @@ cc_test( "src/google/protobuf/no_field_presence_test.cc", "src/google/protobuf/preserve_unknown_enum_test.cc", "src/google/protobuf/proto3_arena_unittest.cc", + "src/google/protobuf/proto3_arena_lite_unittest.cc", + "src/google/protobuf/proto3_lite_unittest.cc", "src/google/protobuf/reflection_ops_unittest.cc", "src/google/protobuf/repeated_field_reflection_unittest.cc", "src/google/protobuf/repeated_field_unittest.cc", diff --git a/src/Makefile.am b/src/Makefile.am index 2ba0ef9b8b..8006550621 100644 --- a/src/Makefile.am +++ b/src/Makefile.am @@ -507,6 +507,8 @@ protoc_inputs = \ google/protobuf/unittest_preserve_unknown_enum.proto \ google/protobuf/unittest.proto \ google/protobuf/unittest_proto3_arena.proto \ + google/protobuf/unittest_proto3_arena_lite.proto \ + google/protobuf/unittest_proto3_lite.proto \ google/protobuf/unittest_well_known_types.proto \ google/protobuf/util/internal/testdata/anys.proto \ google/protobuf/util/internal/testdata/books.proto \ @@ -609,6 +611,10 @@ protoc_outputs = \ google/protobuf/unittest_preserve_unknown_enum.pb.h \ google/protobuf/unittest_proto3_arena.pb.cc \ google/protobuf/unittest_proto3_arena.pb.h \ + google/protobuf/unittest_proto3_arena_lite.pb.cc \ + google/protobuf/unittest_proto3_arena_lite.pb.h \ + google/protobuf/unittest_proto3_lite.pb.cc \ + google/protobuf/unittest_proto3_lite.pb.h \ google/protobuf/unittest_well_known_types.pb.cc \ google/protobuf/unittest_well_known_types.pb.h \ google/protobuf/util/internal/testdata/anys.pb.cc \ @@ -710,6 +716,8 @@ protobuf_test_SOURCES = \ google/protobuf/no_field_presence_test.cc \ google/protobuf/preserve_unknown_enum_test.cc \ google/protobuf/proto3_arena_unittest.cc \ + google/protobuf/proto3_arena_lite_unittest.cc \ + google/protobuf/proto3_lite_unittest.cc \ google/protobuf/reflection_ops_unittest.cc \ google/protobuf/repeated_field_reflection_unittest.cc \ google/protobuf/repeated_field_unittest.cc \ diff --git a/src/google/protobuf/compiler/cpp/cpp_message.cc b/src/google/protobuf/compiler/cpp/cpp_message.cc index c31666113b..9bd5cf51ca 100644 --- a/src/google/protobuf/compiler/cpp/cpp_message.cc +++ b/src/google/protobuf/compiler/cpp/cpp_message.cc @@ -1777,7 +1777,8 @@ void MessageGenerator:: GenerateClassMethods(io::Printer* printer) { // mutable_unknown_fields wrapper function for LazyStringOutputStream // callback. - if (!UseUnknownFieldSet(descriptor_->file())) { + if (PreserveUnknownFields(descriptor_) && + !UseUnknownFieldSet(descriptor_->file())) { printer->Print( "static ::std::string* MutableUnknownFieldsFor$classname$(\n" " $classname$* ptr) {\n" @@ -2656,18 +2657,13 @@ GenerateSwap(io::Printer* printer) { } } - if (PreserveUnknownFields(descriptor_)) { - if (UseUnknownFieldSet(descriptor_->file())) { - printer->Print( + // Ignore PreserveUnknownFields here - always swap internal_metadata as it + // may contain more than just unknown fields. + if (UseUnknownFieldSet(descriptor_->file())) { + printer->Print( "_internal_metadata_.Swap(&other->_internal_metadata_);\n"); - } else { - printer->Print("_unknown_fields_.Swap(&other->_unknown_fields_);\n"); - } } else { - // Still swap internal_metadata as it may contain more than just - // unknown fields. - printer->Print( - "_internal_metadata_.Swap(&other->_internal_metadata_);\n"); + printer->Print("_unknown_fields_.Swap(&other->_unknown_fields_);\n"); } printer->Print("std::swap(_cached_size_, other->_cached_size_);\n"); if (descriptor_->extension_range_count() > 0) { @@ -2908,7 +2904,8 @@ GenerateMergeFromCodedStream(io::Printer* printer) { " ::google::protobuf::uint32 tag;\n", "classname", classname_); - if (!UseUnknownFieldSet(descriptor_->file())) { + if (PreserveUnknownFields(descriptor_) && + !UseUnknownFieldSet(descriptor_->file())) { // Use LazyStringOutputString to avoid initializing unknown fields string // unless it is actually needed. For the same reason, disable eager refresh // on the CodedOutputStream. diff --git a/src/google/protobuf/descriptor.cc b/src/google/protobuf/descriptor.cc index 78a34617fb..f46c7f2327 100644 --- a/src/google/protobuf/descriptor.cc +++ b/src/google/protobuf/descriptor.cc @@ -5113,11 +5113,6 @@ void DescriptorBuilder::ValidateProto3( for (int i = 0; i < file->enum_type_count(); ++i) { ValidateProto3Enum(file->enum_types_ + i, proto.enum_type(i)); } - if (IsLite(file)) { - AddError(file->name(), proto, - DescriptorPool::ErrorCollector::OTHER, - "Lite runtime is not supported in proto3."); - } } static string ToLowercaseWithoutUnderscores(const string& name) { diff --git a/src/google/protobuf/descriptor_unittest.cc b/src/google/protobuf/descriptor_unittest.cc index e4720a7e46..f937b9eac2 100644 --- a/src/google/protobuf/descriptor_unittest.cc +++ b/src/google/protobuf/descriptor_unittest.cc @@ -5740,18 +5740,6 @@ TEST_F(ValidationErrorTest, ValidateProto3Enum) { "}"); } -TEST_F(ValidationErrorTest, ValidateProto3LiteRuntime) { - // Lite runtime is not supported in proto3. - BuildFileWithErrors( - "name: 'foo.proto' " - "syntax: 'proto3' " - "options { " - " optimize_for: LITE_RUNTIME " - "} ", - "foo.proto: foo.proto: OTHER: Lite runtime is not supported " - "in proto3.\n"); -} - TEST_F(ValidationErrorTest, ValidateProto3Group) { BuildFileWithErrors( "name: 'foo.proto' " diff --git a/src/google/protobuf/proto3_arena_lite_unittest.cc b/src/google/protobuf/proto3_arena_lite_unittest.cc new file mode 100644 index 0000000000..0f18c027e8 --- /dev/null +++ b/src/google/protobuf/proto3_arena_lite_unittest.cc @@ -0,0 +1,164 @@ +// Protocol Buffers - Google's data interchange format +// Copyright 2008 Google Inc. All rights reserved. +// https://developers.google.com/protocol-buffers/ +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +#include +#include +#ifndef _SHARED_PTR_H +#include +#endif +#include + +#include +#include +#include +#include +#include +#include + +namespace google { +using proto3_arena_lite_unittest::TestAllTypes; + +namespace protobuf { +namespace { +// We selectively set/check a few representative fields rather than all fields +// as this test is only expected to cover the basics of arena support. +void SetAllFields(TestAllTypes* m) { + m->set_optional_int32(100); + m->set_optional_string("asdf"); + m->set_optional_bytes("jkl;"); + m->mutable_optional_nested_message()->set_bb(42); + m->mutable_optional_foreign_message()->set_c(43); + m->set_optional_nested_enum( + proto3_arena_lite_unittest::TestAllTypes_NestedEnum_BAZ); + m->set_optional_foreign_enum( + proto3_arena_lite_unittest::FOREIGN_BAZ); + m->mutable_optional_lazy_message()->set_bb(45); + m->add_repeated_int32(100); + m->add_repeated_string("asdf"); + m->add_repeated_bytes("jkl;"); + m->add_repeated_nested_message()->set_bb(46); + m->add_repeated_foreign_message()->set_c(47); + m->add_repeated_nested_enum( + proto3_arena_lite_unittest::TestAllTypes_NestedEnum_BAZ); + m->add_repeated_foreign_enum( + proto3_arena_lite_unittest::FOREIGN_BAZ); + m->add_repeated_lazy_message()->set_bb(49); + + m->set_oneof_uint32(1); + m->mutable_oneof_nested_message()->set_bb(50); + m->set_oneof_string("test"); // only this one remains set +} + +void ExpectAllFieldsSet(const TestAllTypes& m) { + EXPECT_EQ(100, m.optional_int32()); + EXPECT_EQ("asdf", m.optional_string()); + EXPECT_EQ("jkl;", m.optional_bytes()); + EXPECT_EQ(true, m.has_optional_nested_message()); + EXPECT_EQ(42, m.optional_nested_message().bb()); + EXPECT_EQ(true, m.has_optional_foreign_message()); + EXPECT_EQ(43, m.optional_foreign_message().c()); + EXPECT_EQ(proto3_arena_lite_unittest::TestAllTypes_NestedEnum_BAZ, + m.optional_nested_enum()); + EXPECT_EQ(proto3_arena_lite_unittest::FOREIGN_BAZ, + m.optional_foreign_enum()); + EXPECT_EQ(true, m.has_optional_lazy_message()); + EXPECT_EQ(45, m.optional_lazy_message().bb()); + + EXPECT_EQ(1, m.repeated_int32_size()); + EXPECT_EQ(100, m.repeated_int32(0)); + EXPECT_EQ(1, m.repeated_string_size()); + EXPECT_EQ("asdf", m.repeated_string(0)); + EXPECT_EQ(1, m.repeated_bytes_size()); + EXPECT_EQ("jkl;", m.repeated_bytes(0)); + EXPECT_EQ(1, m.repeated_nested_message_size()); + EXPECT_EQ(46, m.repeated_nested_message(0).bb()); + EXPECT_EQ(1, m.repeated_foreign_message_size()); + EXPECT_EQ(47, m.repeated_foreign_message(0).c()); + EXPECT_EQ(1, m.repeated_nested_enum_size()); + EXPECT_EQ(proto3_arena_lite_unittest::TestAllTypes_NestedEnum_BAZ, + m.repeated_nested_enum(0)); + EXPECT_EQ(1, m.repeated_foreign_enum_size()); + EXPECT_EQ(proto3_arena_lite_unittest::FOREIGN_BAZ, + m.repeated_foreign_enum(0)); + EXPECT_EQ(1, m.repeated_lazy_message_size()); + EXPECT_EQ(49, m.repeated_lazy_message(0).bb()); + + EXPECT_EQ(proto3_arena_lite_unittest::TestAllTypes::kOneofString, + m.oneof_field_case()); + EXPECT_EQ("test", m.oneof_string()); +} + +// In this file we only test some basic functionalities of arena support in +// proto3 and expect the arena support to be fully tested in proto2 unittests +// because proto3 shares most code with proto2. + +TEST(Proto3ArenaLiteTest, Parsing) { + TestAllTypes original; + SetAllFields(&original); + + Arena arena; + TestAllTypes* arena_message = Arena::CreateMessage(&arena); + arena_message->ParseFromString(original.SerializeAsString()); + ExpectAllFieldsSet(*arena_message); +} + +TEST(Proto3ArenaLiteTest, Swap) { + Arena arena1; + Arena arena2; + + // Test Swap(). + TestAllTypes* arena1_message = Arena::CreateMessage(&arena1); + TestAllTypes* arena2_message = Arena::CreateMessage(&arena2); + arena1_message->Swap(arena2_message); + EXPECT_EQ(&arena1, arena1_message->GetArena()); + EXPECT_EQ(&arena2, arena2_message->GetArena()); +} + +TEST(Proto3ArenaLiteTest, SetAllocatedMessage) { + Arena arena; + TestAllTypes *arena_message = Arena::CreateMessage(&arena); + TestAllTypes::NestedMessage* nested = new TestAllTypes::NestedMessage; + nested->set_bb(118); + arena_message->set_allocated_optional_nested_message(nested); + EXPECT_EQ(118, arena_message->optional_nested_message().bb()); +} + +TEST(Proto3ArenaLiteTest, ReleaseMessage) { + Arena arena; + TestAllTypes* arena_message = Arena::CreateMessage(&arena); + arena_message->mutable_optional_nested_message()->set_bb(118); + google::protobuf::scoped_ptr nested( + arena_message->release_optional_nested_message()); + EXPECT_EQ(118, nested->bb()); +} + +} // namespace +} // namespace protobuf +} // namespace google diff --git a/src/google/protobuf/proto3_lite_unittest.cc b/src/google/protobuf/proto3_lite_unittest.cc new file mode 100644 index 0000000000..2e2beea909 --- /dev/null +++ b/src/google/protobuf/proto3_lite_unittest.cc @@ -0,0 +1,145 @@ +// Protocol Buffers - Google's data interchange format +// Copyright 2008 Google Inc. All rights reserved. +// https://developers.google.com/protocol-buffers/ +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +#include +#include +#ifndef _SHARED_PTR_H +#include +#endif +#include + +#include +#include +#include +#include +#include +#include + +namespace google { +using proto3_lite_unittest::TestAllTypes; + +namespace protobuf { +namespace { +// We selectively set/check a few representative fields rather than all fields +// as this test is only expected to cover the basics of lite support. +void SetAllFields(TestAllTypes* m) { + m->set_optional_int32(100); + m->set_optional_string("asdf"); + m->set_optional_bytes("jkl;"); + m->mutable_optional_nested_message()->set_bb(42); + m->mutable_optional_foreign_message()->set_c(43); + m->set_optional_nested_enum( + proto3_lite_unittest::TestAllTypes_NestedEnum_BAZ); + m->set_optional_foreign_enum( + proto3_lite_unittest::FOREIGN_BAZ); + m->mutable_optional_lazy_message()->set_bb(45); + m->add_repeated_int32(100); + m->add_repeated_string("asdf"); + m->add_repeated_bytes("jkl;"); + m->add_repeated_nested_message()->set_bb(46); + m->add_repeated_foreign_message()->set_c(47); + m->add_repeated_nested_enum( + proto3_lite_unittest::TestAllTypes_NestedEnum_BAZ); + m->add_repeated_foreign_enum( + proto3_lite_unittest::FOREIGN_BAZ); + m->add_repeated_lazy_message()->set_bb(49); + + m->set_oneof_uint32(1); + m->mutable_oneof_nested_message()->set_bb(50); + m->set_oneof_string("test"); // only this one remains set +} + +void ExpectAllFieldsSet(const TestAllTypes& m) { + EXPECT_EQ(100, m.optional_int32()); + EXPECT_EQ("asdf", m.optional_string()); + EXPECT_EQ("jkl;", m.optional_bytes()); + EXPECT_EQ(true, m.has_optional_nested_message()); + EXPECT_EQ(42, m.optional_nested_message().bb()); + EXPECT_EQ(true, m.has_optional_foreign_message()); + EXPECT_EQ(43, m.optional_foreign_message().c()); + EXPECT_EQ(proto3_lite_unittest::TestAllTypes_NestedEnum_BAZ, + m.optional_nested_enum()); + EXPECT_EQ(proto3_lite_unittest::FOREIGN_BAZ, + m.optional_foreign_enum()); + EXPECT_EQ(true, m.has_optional_lazy_message()); + EXPECT_EQ(45, m.optional_lazy_message().bb()); + + EXPECT_EQ(1, m.repeated_int32_size()); + EXPECT_EQ(100, m.repeated_int32(0)); + EXPECT_EQ(1, m.repeated_string_size()); + EXPECT_EQ("asdf", m.repeated_string(0)); + EXPECT_EQ(1, m.repeated_bytes_size()); + EXPECT_EQ("jkl;", m.repeated_bytes(0)); + EXPECT_EQ(1, m.repeated_nested_message_size()); + EXPECT_EQ(46, m.repeated_nested_message(0).bb()); + EXPECT_EQ(1, m.repeated_foreign_message_size()); + EXPECT_EQ(47, m.repeated_foreign_message(0).c()); + EXPECT_EQ(1, m.repeated_nested_enum_size()); + EXPECT_EQ(proto3_lite_unittest::TestAllTypes_NestedEnum_BAZ, + m.repeated_nested_enum(0)); + EXPECT_EQ(1, m.repeated_foreign_enum_size()); + EXPECT_EQ(proto3_lite_unittest::FOREIGN_BAZ, + m.repeated_foreign_enum(0)); + EXPECT_EQ(1, m.repeated_lazy_message_size()); + EXPECT_EQ(49, m.repeated_lazy_message(0).bb()); + + EXPECT_EQ(proto3_lite_unittest::TestAllTypes::kOneofString, + m.oneof_field_case()); + EXPECT_EQ("test", m.oneof_string()); +} + +// In this file we only test some basic functionalities of in proto3 and expect +// the rest is fully tested in proto2 unittests because proto3 shares most code +// with proto2. + +TEST(Proto3LiteTest, Parsing) { + TestAllTypes original; + SetAllFields(&original); + + TestAllTypes msg; + msg.ParseFromString(original.SerializeAsString()); + ExpectAllFieldsSet(msg); +} + +TEST(Proto3LiteTest, Swap) { + // Test Swap(). + TestAllTypes msg1; + TestAllTypes msg2; + msg1.set_optional_string("123"); + msg2.set_optional_string("3456"); + msg1.Swap(&msg2); + EXPECT_EQ("3456", msg1.optional_string()); + EXPECT_EQ("123", msg2.optional_string()); + EXPECT_EQ(msg1.ByteSize(), msg2.ByteSize() + 1); +} + +} // namespace +} // namespace protobuf +} // namespace google diff --git a/src/google/protobuf/unittest_proto3_arena_lite.proto b/src/google/protobuf/unittest_proto3_arena_lite.proto new file mode 100644 index 0000000000..5a60b90f5c --- /dev/null +++ b/src/google/protobuf/unittest_proto3_arena_lite.proto @@ -0,0 +1,207 @@ +// Protocol Buffers - Google's data interchange format +// Copyright 2008 Google Inc. All rights reserved. +// https://developers.google.com/protocol-buffers/ +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +syntax = "proto3"; + +option cc_enable_arenas = true; +option optimize_for = LITE_RUNTIME; + +import "google/protobuf/unittest_import.proto"; + +package proto3_arena_lite_unittest; + +// This proto includes every type of field in both singular and repeated +// forms. +message TestAllTypes { + message NestedMessage { + // The field name "b" fails to compile in proto1 because it conflicts with + // a local variable named "b" in one of the generated methods. Doh. + // This file needs to compile in proto1 to test backwards-compatibility. + int32 bb = 1; + } + + enum NestedEnum { + ZERO = 0; + FOO = 1; + BAR = 2; + BAZ = 3; + NEG = -1; // Intentionally negative. + } + + // Singular + int32 optional_int32 = 1; + int64 optional_int64 = 2; + uint32 optional_uint32 = 3; + uint64 optional_uint64 = 4; + sint32 optional_sint32 = 5; + sint64 optional_sint64 = 6; + fixed32 optional_fixed32 = 7; + fixed64 optional_fixed64 = 8; + sfixed32 optional_sfixed32 = 9; + sfixed64 optional_sfixed64 = 10; + float optional_float = 11; + double optional_double = 12; + bool optional_bool = 13; + string optional_string = 14; + bytes optional_bytes = 15; + + // Groups are not allowed in proto3. + // optional group OptionalGroup = 16 { + // optional int32 a = 17; + // } + + NestedMessage optional_nested_message = 18; + ForeignMessage optional_foreign_message = 19; + protobuf_unittest_import.ImportMessage optional_import_message = 20; + + NestedEnum optional_nested_enum = 21; + ForeignEnum optional_foreign_enum = 22; + + // Omitted (compared to unittest.proto) because proto2 enums are not allowed + // inside proto2 messages. + // + // optional protobuf_unittest_import.ImportEnum optional_import_enum = 23; + + string optional_string_piece = 24 [ctype=STRING_PIECE]; + string optional_cord = 25 [ctype=CORD]; + + // Defined in unittest_import_public.proto + protobuf_unittest_import.PublicImportMessage + optional_public_import_message = 26; + + NestedMessage optional_lazy_message = 27 [lazy=true]; + + // Repeated + repeated int32 repeated_int32 = 31; + repeated int64 repeated_int64 = 32; + repeated uint32 repeated_uint32 = 33; + repeated uint64 repeated_uint64 = 34; + repeated sint32 repeated_sint32 = 35; + repeated sint64 repeated_sint64 = 36; + repeated fixed32 repeated_fixed32 = 37; + repeated fixed64 repeated_fixed64 = 38; + repeated sfixed32 repeated_sfixed32 = 39; + repeated sfixed64 repeated_sfixed64 = 40; + repeated float repeated_float = 41; + repeated double repeated_double = 42; + repeated bool repeated_bool = 43; + repeated string repeated_string = 44; + repeated bytes repeated_bytes = 45; + + // Groups are not allowed in proto3. + // repeated group RepeatedGroup = 46 { + // optional int32 a = 47; + // } + + repeated NestedMessage repeated_nested_message = 48; + repeated ForeignMessage repeated_foreign_message = 49; + repeated protobuf_unittest_import.ImportMessage repeated_import_message = 50; + + repeated NestedEnum repeated_nested_enum = 51; + repeated ForeignEnum repeated_foreign_enum = 52; + + // Omitted (compared to unittest.proto) because proto2 enums are not allowed + // inside proto2 messages. + // + // repeated protobuf_unittest_import.ImportEnum repeated_import_enum = 53; + + repeated string repeated_string_piece = 54 [ctype=STRING_PIECE]; + repeated string repeated_cord = 55 [ctype=CORD]; + + repeated NestedMessage repeated_lazy_message = 57 [lazy=true]; + + oneof oneof_field { + uint32 oneof_uint32 = 111; + NestedMessage oneof_nested_message = 112; + string oneof_string = 113; + bytes oneof_bytes = 114; + } +} + +// Test messages for packed fields + +message TestPackedTypes { + repeated int32 packed_int32 = 90 [packed = true]; + repeated int64 packed_int64 = 91 [packed = true]; + repeated uint32 packed_uint32 = 92 [packed = true]; + repeated uint64 packed_uint64 = 93 [packed = true]; + repeated sint32 packed_sint32 = 94 [packed = true]; + repeated sint64 packed_sint64 = 95 [packed = true]; + repeated fixed32 packed_fixed32 = 96 [packed = true]; + repeated fixed64 packed_fixed64 = 97 [packed = true]; + repeated sfixed32 packed_sfixed32 = 98 [packed = true]; + repeated sfixed64 packed_sfixed64 = 99 [packed = true]; + repeated float packed_float = 100 [packed = true]; + repeated double packed_double = 101 [packed = true]; + repeated bool packed_bool = 102 [packed = true]; + repeated ForeignEnum packed_enum = 103 [packed = true]; +} + +// Explicitly set packed to false +message TestUnpackedTypes { + repeated int32 repeated_int32 = 1 [packed = false]; + repeated int64 repeated_int64 = 2 [packed = false]; + repeated uint32 repeated_uint32 = 3 [packed = false]; + repeated uint64 repeated_uint64 = 4 [packed = false]; + repeated sint32 repeated_sint32 = 5 [packed = false]; + repeated sint64 repeated_sint64 = 6 [packed = false]; + repeated fixed32 repeated_fixed32 = 7 [packed = false]; + repeated fixed64 repeated_fixed64 = 8 [packed = false]; + repeated sfixed32 repeated_sfixed32 = 9 [packed = false]; + repeated sfixed64 repeated_sfixed64 = 10 [packed = false]; + repeated float repeated_float = 11 [packed = false]; + repeated double repeated_double = 12 [packed = false]; + repeated bool repeated_bool = 13 [packed = false]; + repeated TestAllTypes.NestedEnum repeated_nested_enum = 14 [packed = false]; +} + +// This proto includes a recusively nested message. +message NestedTestAllTypes { + NestedTestAllTypes child = 1; + TestAllTypes payload = 2; +} + +// Define these after TestAllTypes to make sure the compiler can handle +// that. +message ForeignMessage { + int32 c = 1; +} + +enum ForeignEnum { + FOREIGN_ZERO = 0; + FOREIGN_FOO = 4; + FOREIGN_BAR = 5; + FOREIGN_BAZ = 6; +} + +// TestEmptyMessage is used to test behavior of unknown fields. +message TestEmptyMessage { +} + diff --git a/src/google/protobuf/unittest_proto3_lite.proto b/src/google/protobuf/unittest_proto3_lite.proto new file mode 100644 index 0000000000..874ade6c0e --- /dev/null +++ b/src/google/protobuf/unittest_proto3_lite.proto @@ -0,0 +1,206 @@ +// Protocol Buffers - Google's data interchange format +// Copyright 2008 Google Inc. All rights reserved. +// https://developers.google.com/protocol-buffers/ +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +syntax = "proto3"; + +option optimize_for = LITE_RUNTIME; + +import "google/protobuf/unittest_import.proto"; + +package proto3_lite_unittest; + +// This proto includes every type of field in both singular and repeated +// forms. +message TestAllTypes { + message NestedMessage { + // The field name "b" fails to compile in proto1 because it conflicts with + // a local variable named "b" in one of the generated methods. Doh. + // This file needs to compile in proto1 to test backwards-compatibility. + int32 bb = 1; + } + + enum NestedEnum { + ZERO = 0; + FOO = 1; + BAR = 2; + BAZ = 3; + NEG = -1; // Intentionally negative. + } + + // Singular + int32 optional_int32 = 1; + int64 optional_int64 = 2; + uint32 optional_uint32 = 3; + uint64 optional_uint64 = 4; + sint32 optional_sint32 = 5; + sint64 optional_sint64 = 6; + fixed32 optional_fixed32 = 7; + fixed64 optional_fixed64 = 8; + sfixed32 optional_sfixed32 = 9; + sfixed64 optional_sfixed64 = 10; + float optional_float = 11; + double optional_double = 12; + bool optional_bool = 13; + string optional_string = 14; + bytes optional_bytes = 15; + + // Groups are not allowed in proto3. + // optional group OptionalGroup = 16 { + // optional int32 a = 17; + // } + + NestedMessage optional_nested_message = 18; + ForeignMessage optional_foreign_message = 19; + protobuf_unittest_import.ImportMessage optional_import_message = 20; + + NestedEnum optional_nested_enum = 21; + ForeignEnum optional_foreign_enum = 22; + + // Omitted (compared to unittest.proto) because proto2 enums are not allowed + // inside proto2 messages. + // + // optional protobuf_unittest_import.ImportEnum optional_import_enum = 23; + + string optional_string_piece = 24 [ctype=STRING_PIECE]; + string optional_cord = 25 [ctype=CORD]; + + // Defined in unittest_import_public.proto + protobuf_unittest_import.PublicImportMessage + optional_public_import_message = 26; + + NestedMessage optional_lazy_message = 27 [lazy=true]; + + // Repeated + repeated int32 repeated_int32 = 31; + repeated int64 repeated_int64 = 32; + repeated uint32 repeated_uint32 = 33; + repeated uint64 repeated_uint64 = 34; + repeated sint32 repeated_sint32 = 35; + repeated sint64 repeated_sint64 = 36; + repeated fixed32 repeated_fixed32 = 37; + repeated fixed64 repeated_fixed64 = 38; + repeated sfixed32 repeated_sfixed32 = 39; + repeated sfixed64 repeated_sfixed64 = 40; + repeated float repeated_float = 41; + repeated double repeated_double = 42; + repeated bool repeated_bool = 43; + repeated string repeated_string = 44; + repeated bytes repeated_bytes = 45; + + // Groups are not allowed in proto3. + // repeated group RepeatedGroup = 46 { + // optional int32 a = 47; + // } + + repeated NestedMessage repeated_nested_message = 48; + repeated ForeignMessage repeated_foreign_message = 49; + repeated protobuf_unittest_import.ImportMessage repeated_import_message = 50; + + repeated NestedEnum repeated_nested_enum = 51; + repeated ForeignEnum repeated_foreign_enum = 52; + + // Omitted (compared to unittest.proto) because proto2 enums are not allowed + // inside proto2 messages. + // + // repeated protobuf_unittest_import.ImportEnum repeated_import_enum = 53; + + repeated string repeated_string_piece = 54 [ctype=STRING_PIECE]; + repeated string repeated_cord = 55 [ctype=CORD]; + + repeated NestedMessage repeated_lazy_message = 57 [lazy=true]; + + oneof oneof_field { + uint32 oneof_uint32 = 111; + NestedMessage oneof_nested_message = 112; + string oneof_string = 113; + bytes oneof_bytes = 114; + } +} + +// Test messages for packed fields + +message TestPackedTypes { + repeated int32 packed_int32 = 90 [packed = true]; + repeated int64 packed_int64 = 91 [packed = true]; + repeated uint32 packed_uint32 = 92 [packed = true]; + repeated uint64 packed_uint64 = 93 [packed = true]; + repeated sint32 packed_sint32 = 94 [packed = true]; + repeated sint64 packed_sint64 = 95 [packed = true]; + repeated fixed32 packed_fixed32 = 96 [packed = true]; + repeated fixed64 packed_fixed64 = 97 [packed = true]; + repeated sfixed32 packed_sfixed32 = 98 [packed = true]; + repeated sfixed64 packed_sfixed64 = 99 [packed = true]; + repeated float packed_float = 100 [packed = true]; + repeated double packed_double = 101 [packed = true]; + repeated bool packed_bool = 102 [packed = true]; + repeated ForeignEnum packed_enum = 103 [packed = true]; +} + +// Explicitly set packed to false +message TestUnpackedTypes { + repeated int32 repeated_int32 = 1 [packed = false]; + repeated int64 repeated_int64 = 2 [packed = false]; + repeated uint32 repeated_uint32 = 3 [packed = false]; + repeated uint64 repeated_uint64 = 4 [packed = false]; + repeated sint32 repeated_sint32 = 5 [packed = false]; + repeated sint64 repeated_sint64 = 6 [packed = false]; + repeated fixed32 repeated_fixed32 = 7 [packed = false]; + repeated fixed64 repeated_fixed64 = 8 [packed = false]; + repeated sfixed32 repeated_sfixed32 = 9 [packed = false]; + repeated sfixed64 repeated_sfixed64 = 10 [packed = false]; + repeated float repeated_float = 11 [packed = false]; + repeated double repeated_double = 12 [packed = false]; + repeated bool repeated_bool = 13 [packed = false]; + repeated TestAllTypes.NestedEnum repeated_nested_enum = 14 [packed = false]; +} + +// This proto includes a recusively nested message. +message NestedTestAllTypes { + NestedTestAllTypes child = 1; + TestAllTypes payload = 2; +} + +// Define these after TestAllTypes to make sure the compiler can handle +// that. +message ForeignMessage { + int32 c = 1; +} + +enum ForeignEnum { + FOREIGN_ZERO = 0; + FOREIGN_FOO = 4; + FOREIGN_BAR = 5; + FOREIGN_BAZ = 6; +} + +// TestEmptyMessage is used to test behavior of unknown fields. +message TestEmptyMessage { +} + From 8052dc75a973253c9cd02148a19f56613bad667f Mon Sep 17 00:00:00 2001 From: Feng Xiao Date: Tue, 19 Apr 2016 15:29:18 -0700 Subject: [PATCH 2/8] Add a docs directory and move the third-party add-ons page here. --- docs/third_party.md | 147 ++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 147 insertions(+) create mode 100644 docs/third_party.md diff --git a/docs/third_party.md b/docs/third_party.md new file mode 100644 index 0000000000..69ed6a3868 --- /dev/null +++ b/docs/third_party.md @@ -0,0 +1,147 @@ +# Third-Party Add-ons for Protocol Buffers + +This page lists code related to Protocol Buffers which is developed and maintained by third parties. You may find this code useful, but note that **these projects are not affiliated with or endorsed by Google (unless explicitly marked)**; try them at your own risk. Also note that many projects here are in the early stages of development and not production-ready. + +If you have a project that should be listed here, please [send us a pull request](https://github.com/google/protobuf/pulls) to update this page. + +## Programming Languages + +These are projects we know about implementing Protocol Buffers for other programming languages: +* Action Script: http://code.google.com/p/protobuf-actionscript3/ +* Action Script: https://code.google.com/p/protoc-gen-as3/ +* Action Script: https://github.com/matrix3d/JProtoc +* C: https://github.com/protobuf-c/protobuf-c +* C: http://koti.kapsi.fi/jpa/nanopb/ +* C: https://github.com/cloudwu/pbc/ +* C: https://github.com/haberman/upb/wiki +* C: https://github.com/squidfunk/protobluff +* C++: https://github.com/google/protobuf (Google-official implementation) +* C/C++: http://spbc.sf.net/ +* C#: http://code.google.com/p/protobuf-csharp-port +* C#: http://code.google.com/p/protosharp/ +* C#: https://silentorbit.com/protobuf/ +* C#/.NET/WCF/VB: http://code.google.com/p/protobuf-net/ +* Clojure: http://github.com/ninjudd/clojure-protobuf +* Common Lisp: http://www.prism.gatech.edu/~ndantam3/docs/s-protobuf/ +* Common Lisp: http://github.com/brown/protobuf +* D: https://github.com/msoucy/dproto +* D: http://256.makerslocal.org/wiki/index.php/ProtocolBuffer +* D: https://github.com/opticron/ProtocolBuffer +* Dart: https://github.com/dart-lang/dart-protobuf (runtime) https://github.com/dart-lang/dart-protoc-plugin (code generator) +* Delphi: http://sourceforge.net/projects/protobuf-delphi/ +* Delphi: http://fundementals.sourceforge.net/dl.html +* Elixir: https://github.com/jeremyong/exprotoc +* Erlang: http://github.com/ngerakines/erlang_protobuffs/tree/master +* Erlang: http://piqi.org/ +* Erlang: https://code.google.com/p/protoc-gen-erl/ +* Erlang: https://github.com/basho/erlang_protobuffs +* Go: https://github.com/golang/protobuf (Google-official implementation) +* Go: http://code.google.com/p/goprotobuf/ +* Go: https://github.com/akunspy/gopbuf +* Haskell: http://hackage.haskell.org/package/hprotoc +* Haxe: https://github.com/Atry/protoc-gen-haxe +* Java: https://github.com/google/protobuf (Google-official implementation) +* Java/Android: https://github.com/square/wire +* Java ME: http://code.google.com/p/protobuf-javame/ +* Java ME: http://swingme.sourceforge.net/encode.shtml +* Java ME: http://github.com/ponderingpanda/protobuf-j2me +* Java ME: http://code.google.com/p/protobuf-j2me/ +* Javascript: http://code.google.com/p/protobuf-js/ +* Javascript: http://github.com/sirikata/protojs +* Javascript: https://github.com/dcodeIO/ProtoBuf.js +* Javascript: http://code.google.com/p/protobuf-for-node/ +* Javascript: http://code.google.com/p/protostuff/ +* Julia: https://github.com/tanmaykm/ProtoBuf.jl +* Lua: http://code.google.com/p/protoc-gen-lua/ +* Lua: http://github.com/indygreg/lua-protobuf +* Lua: https://github.com/Neopallium/lua-pb +* Matlab: http://code.google.com/p/protobuf-matlab/ +* Mercury: http://code.google.com/p/protobuf-mercury/ +* Objective C: http://code.google.com/p/protobuf-objc/ +* Objective C: https://github.com/alexeyxo/protobuf-objc +* OCaml: http://piqi.org/ +* Perl: http://groups.google.com/group/protobuf-perl +* Perl: http://search.cpan.org/perldoc?Google::ProtocolBuffers +* Perl/XS: http://code.google.com/p/protobuf-perlxs/ +* PHP: http://code.google.com/p/pb4php/ +* PHP: https://github.com/allegro/php-protobuf/ +* PHP: https://github.com/chobie/php-protocolbuffers +* PHP: http://drslump.github.com/Protobuf-PHP +* Prolog: http://www.swi-prolog.org/pldoc/package/protobufs.html +* Python: https://github.com/google/protobuf (Google-official implementation) +* Python: http://eigenein.github.com/protobuf/ +* R: http://cran.r-project.org/package=RProtoBuf +* Ruby: http://code.google.com/p/ruby-protobuf/ +* Ruby: http://github.com/mozy/ruby-protocol-buffers +* Ruby: https://github.com/bmizerany/beefcake/tree/master/lib/beefcake +* Ruby: https://github.com/localshred/protobuf +* Rust: https://github.com/stepancheg/rust-protobuf/ +* Scala: http://github.com/jeffplaisance/scala-protobuf +* Scala: http://code.google.com/p/protobuf-scala +* Scala: https://github.com/SandroGrzicic/ScalaBuff +* Scala: http://trueaccord.github.io/ScalaPB/ +* Swift: https://github.com/alexeyxo/protobuf-swift +* Vala: https://launchpad.net/protobuf-vala +* Visual Basic: http://code.google.com/p/protobuf-net/ + +## RPC Implementations + +GRPC (http://www.grpc.io/) is Google's RPC implementation for Protocol Buffers. There are other third-party RPC implementations as well. Some of these actually work with Protocol Buffers service definitions (defined using the `service` keyword in `.proto` files) while others just use Protocol Buffers message objects. + +* https://github.com/grpc/grpc (C++, Node.js, Python, Ruby, Objective-C, PHP, C#, Google-official implementation) +* http://zeroc.com/ice.html (Multiple languages) +* http://code.google.com/p/protobuf-net/ (C#/.NET/WCF/VB) +* https://launchpad.net/txprotobuf/ (Python) +* https://github.com/modeswitch/protobuf-rpc (Python) +* http://code.google.com/p/protobuf-socket-rpc/ (Java, Python) +* http://code.google.com/p/proto-streamer/ (Java) +* http://code.google.com/p/server1/ (C++) +* http://deltavsoft.com/RcfUserGuide/Protobufs (C++) +* http://code.google.com/p/protobuf-mina-rpc/ (Python client, Java server) +* http://code.google.com/p/casocklib/ (C++) +* http://code.google.com/p/cxf-protobuf/ (Java) +* http://code.google.com/p/protobuf-remote/ (C++/C#) +* http://code.google.com/p/protobuf-rpc-pro/ (Java) +* https://code.google.com/p/protorpc/ (Go/C++) +* https://code.google.com/p/eneter-protobuf-serializer/ (Java/.NET) +* http://www.deltavsoft.com/RCFProto.html (C++/Java/Python/C#) +* https://github.com/robbinfan/claire-protorpc (C++) +* https://github.com/BaiduPS/sofa-pbrpc (C++) +* https://github.com/ebencheung/arab (C++) +* http://code.google.com/p/protobuf-csharp-rpc/ (C#) +* https://github.com/thesamet/rpcz (C++/Python, based on ZeroMQ) +* https://github.com/w359405949/libmaid (C++, Python) +* https://github.com/madwyn/libpbrpc (C++) + +## Other Utilities + +There are miscellaneous other things you may find useful as a Protocol Buffers developer. + +* [NetBeans IDE plugin](http://code.google.com/p/protobuf-netbeans-plugin/) +* [Wireshark/Ethereal packet sniffer plugin](http://code.google.com/p/protobuf-wireshark/) +* [Alternate encodings (JSON, XML, HTML) for Java protobufs](http://code.google.com/p/protobuf-java-format/) +* [Another JSON encoder/decoder for Java](https://github.com/sijuv/protobuf-codec) +* [Editor for serialized protobufs](http://code.google.com/p/protobufeditor/) +* [Intellij IDEA plugin](http://github.com/nnmatveev/idea-plugin-protobuf) +* [TextMate syntax highlighting](http://github.com/michaeledgar/protobuf-tmbundle) +* [Oracle PL SQL plugin](http://code.google.com/p/protocol-buffer-plsql/) +* [Eclipse editor for protobuf (from Google)](http://code.google.com/p/protobuf-dt/) +* [C++ Builder compatible protobuf](https://github.com/saadware/protobuf-cppbuilder) +* Maven Protocol Compiler Plugin + * https://github.com/sergei-ivanov/maven-protoc-plugin/ + * http://igor-petruk.github.com/protobuf-maven-plugin/ + * http://code.google.com/p/maven-protoc-plugin/ + * https://github.com/os72/protoc-jar-maven-plugin +* [Documentation generator plugin (Markdown/HTML/DocBook/...)](https://github.com/estan/protoc-gen-doc) +* [DocBook generator for .proto files](http://code.google.com/p/protoc-gen-docbook/) +* [Protobuf for nginx module](https://github.com/dbcode/protobuf-nginx/) +* [RSpec matchers and Cucumber step defs for testing Protocol Buffers](https://github.com/connamara/protobuf_spec) +* [Sbt plugin for Protocol Buffers](https://github.com/Atry/sbt-cppp) +* [Gradle Protobuf Plugin](https://github.com/aantono/gradle-plugin-protobuf) +* [Multi-platform executable JAR and Java API for protoc](https://github.com/os72/protoc-jar) +* [Python scripts to convert between Protocol Buffers and JSON](https://github.com/NextTuesday/py-pb-converters) +* [Visual Studio Language Service support for Protocol Buffers](http://visualstudiogallery.msdn.microsoft.com/4bc0f38c-b058-4e05-ae38-155e053c19c5) +* [C++ library for serialization/de-serialization between Protocol Buffers and JSON.](https://github.com/yinqiwen/pbjson) +* [ProtoBuf with Java EE7 Expression Language 3.0; pure Java ProtoBuf Parser and Builder.](https://github.com/protobufel/protobuf-el) +* [Notepad++ Syntax Highlighting for .proto files](https://github.com/chai2010/notepadplus-protobuf) +* [Linter for .proto files](https://github.com/ckaznocha/protoc-gen-lint) From db93833bc656e7a0b8d851b1fefdb5c0b5d5b178 Mon Sep 17 00:00:00 2001 From: Jakub Kukul Date: Fri, 22 Apr 2016 16:04:27 +0200 Subject: [PATCH 3/8] Added serialVersionUID to ExtendableMessage. --- .../src/main/java/com/google/protobuf/GeneratedMessage.java | 2 ++ 1 file changed, 2 insertions(+) diff --git a/java/core/src/main/java/com/google/protobuf/GeneratedMessage.java b/java/core/src/main/java/com/google/protobuf/GeneratedMessage.java index a50afe559f..ea79df3e91 100644 --- a/java/core/src/main/java/com/google/protobuf/GeneratedMessage.java +++ b/java/core/src/main/java/com/google/protobuf/GeneratedMessage.java @@ -795,6 +795,8 @@ public abstract class GeneratedMessage extends AbstractMessage extends GeneratedMessage implements ExtendableMessageOrBuilder { + private static final long serialVersionUID = 1L; + private final FieldSet extensions; protected ExtendableMessage() { From 385755e13bd1574c5ff7e00a0546547ba4546d41 Mon Sep 17 00:00:00 2001 From: Tony Allevato Date: Fri, 22 Apr 2016 09:15:14 -0700 Subject: [PATCH 4/8] Add initial design document for Swift protocol buffers. (#1442) * Add initial design doc for Swift protocol buffers. --- docs/swift/DesignDoc.md | 674 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 674 insertions(+) create mode 100644 docs/swift/DesignDoc.md diff --git a/docs/swift/DesignDoc.md b/docs/swift/DesignDoc.md new file mode 100644 index 0000000000..364a4d3f1b --- /dev/null +++ b/docs/swift/DesignDoc.md @@ -0,0 +1,674 @@ +# Protocol Buffers in Swift + +## Objective + +This document describes the user-facing API and internal implementation of +proto2 and proto3 messages in Apple’s Swift programming language. + +One of the key goals of protobufs is to provide idiomatic APIs for each +language. In that vein, **interoperability with Objective-C is a non-goal of +this proposal.** Protobuf users who need to pass messages between Objective-C +and Swift code in the same application should use the existing Objective-C proto +library. The goal of the effort described here is to provide an API for protobuf +messages that uses features specific to Swift—optional types, algebraic +enumerated types, value types, and so forth—in a natural way that will delight, +rather than surprise, users of the language. + +## Naming + +* By convention, both typical protobuf message names and Swift structs/classes + are `UpperCamelCase`, so for most messages, the name of a message can be the + same as the name of its generated type. (However, see the discussion below + about prefixes under [Packages](#packages).) + +* Enum cases in protobufs typically are `UPPERCASE_WITH_UNDERSCORES`, whereas + in Swift they are `lowerCamelCase` (as of the Swift 3 API design + guidelines). We will transform the names to match Swift convention, using + a whitelist similar to the Objective-C compiler plugin to handle commonly + used acronyms. + +* Typical fields in proto messages are `lowercase_with_underscores`, while in + Swift they are `lowerCamelCase`. We will transform the names to match + Swift convention by removing the underscores and uppercasing the subsequent + letter. + +## Swift reserved words + +Swift has a large set of reserved words—some always reserved and some +contextually reserved (that is, they can be used as identifiers in contexts +where they would not be confused). As of Swift 2.2, the set of always-reserved +words is: + +``` +_, #available, #column, #else, #elseif, #endif, #file, #function, #if, #line, +#selector, as, associatedtype, break, case, catch, class, continue, default, +defer, deinit, do, dynamicType, else, enum, extension, fallthrough, false, for, +func, guard, if, import, in, init, inout, internal, is, let, nil, operator, +private, protocol, public, repeat, rethrows, return, self, Self, static, +struct, subscript, super, switch, throw, throws, true, try, typealias, var, +where, while +``` + +The set of contextually reserved words is: + +``` +associativity, convenience, dynamic, didSet, final, get, infix, indirect, +lazy, left, mutating, none, nonmutating, optional, override, postfix, +precedence, prefix, Protocol, required, right, set, Type, unowned, weak, +willSet +``` + +It is possible to use any reserved word as an identifier by escaping it with +backticks (for example, ``let `class` = 5``). Other name-mangling schemes would +require us to transform the names themselves (for example, by appending an +underscore), which requires us to then ensure that the new name does not collide +with something else in the same namespace. + +While the backtick feature may not be widely known by all Swift developers, a +small amount of user education can address this and it seems like the best +approach. We can unconditionally surround all property names with backticks to +simplify generation. + +Some remapping will still be required, though, to avoid collisions between +generated properties and the names of methods and properties defined in the base +protocol/implementation of messages. + +# Features of Protocol Buffers + +This section describes how the features of the protocol buffer syntaxes (proto2 +and proto3) map to features in Swift—what the code generated from a proto will +look like, and how it will be implemented in the underlying library. + +## Packages + +Modules are the main form of namespacing in Swift, but they are not declared +using syntactic constructs like namespaces in C++ or packages in Java. Instead, +they are tied to build targets in Xcode (or, in the future with open-source +Swift, declarations in a Swift Package Manager manifest). They also do not +easily support nesting submodules (Clang module maps support this, but pure +Swift does not yet provide a way to define submodules). + +We will generate types with fully-qualified underscore-delimited names. For +example, a message `Baz` in package `foo.bar` would generate a struct named +`Foo_Bar_Baz`. For each fully-qualified proto message, there will be exactly one +unique type symbol emitted in the generated binary. + +Users are likely to balk at the ugliness of underscore-delimited names for every +generated type. To improve upon this situation, we will add a new string file +level option, `swift_package_typealias`, that can be added to `.proto` files. +When present, this will cause `typealias`es to be added to the generated Swift +messages that replace the package name prefix with the provided string. For +example, the following `.proto` file: + +```protobuf +option swift_package_typealias = "FBP"; +package foo.bar; + +message Baz { + // Message fields +} +``` + +would generate the following Swift source: + +```swift +public struct Foo_Bar_Baz { + // Message fields and other methods +} + +typealias FBPBaz = Foo_Bar_Baz +``` + +It should be noted that this type alias is recorded in the generated +`.swiftmodule` so that code importing the module can refer to it, but it does +not cause a new symbol to be generated in the compiled binary (i.e., we do not +risk compiled size bloat by adding `typealias`es for every type). + +Other strategies to handle packages that were considered and rejected can be +found in [Appendix A](#appendix-a-rejected-strategies-to-handle-packages). + +## Messages + +Proto messages are natural value types and we will generate messages as structs +instead of classes. Users will benefit from Swift’s built-in behavior with +regard to mutability. We will define a `ProtoMessage` protocol that defines the +common methods and properties for all messages (such as serialization) and also +lets users treat messages polymorphically. Any shared method implementations +that do not differ between individual messages can be implemented in a protocol +extension. + +The backing storage itself for fields of a message will be managed by a +`ProtoFieldStorage` type that uses an internal dictionary keyed by field number, +and whose values are the value of the field with that number (up-cast to Swift’s +`Any` type). This class will provide type-safe getters and setters so that +generated messages can manipulate this storage, and core serialization logic +will live here as well. Furthermore, factoring the storage out into a separate +type, rather than inlining the fields as stored properties in the message +itself, lets us implement copy-on-write efficiently to support passing around +large messages. (Furthermore, because the messages themselves are value types, +inlining fields is not possible if the fields are submessages of the same type, +or a type that eventually includes a submessage of the same type.) + +### Required fields (proto2 only) + +Required fields in proto2 messages seem like they could be naturally represented +by non-optional properties in Swift, but this presents some problems/concerns. + +Serialization APIs permit partial serialization, which allows required fields to +remain unset. Furthermore, other language APIs still provide `has*` and `clear*` +methods for required fields, and knowing whether a property has a value when the +message is in memory is still useful. + +For example, an e-mail draft message may have the “to” address required on the +wire, but when the user constructs it in memory, it doesn’t make sense to force +a value until they provide one. We only want to force a value to be present when +the message is serialized to the wire. Using non-optional properties prevents +this use case, and makes client usage awkward because the user would be forced +to select a sentinel or placeholder value for any required fields at the time +the message was created. + +### Default values + +In proto2, fields can have a default value specified that may be a value other +than the default value for its corresponding language type (for example, a +default value of 5 instead of 0 for an integer). When reading a field that is +not explicitly set, the user expects to get that value. This makes Swift +optionals (i.e., `Foo?`) unsuitable for fields in general. Unfortunately, we +cannot implement our own “enhanced optional” type without severely complicating +usage (Swift’s use of type inference and its lack of implicit conversions would +require manual unwrapping of every property value). + +Instead, we can use **implicitly unwrapped optionals.** For example, a property +generated for a field of type `int32` would have Swift type `Int32!`. These +properties would behave with the following characteristics, which mirror the +nil-resettable properties used elsewhere in Apple’s SDKs (for example, +`UIView.tintColor`): + +* Assigning a non-nil value to a property sets the field to that value. +* Assigning nil to a property clears the field (its internal representation is + nilled out). +* Reading the value of a property returns its value if it is set, or returns + its default value if it is not set. Reading a property never returns nil. + +The final point in the list above implies that the optional cannot be checked to +determine if the field is set to a value other than its default: it will never +be nil. Instead, we must provide `has*` methods for each field to allow the user +to check this. These methods will be public in proto2. In proto3, these methods +will be private (if generated at all), since the user can test the returned +value against the zero value for that type. + +### Autocreation of nested messages + +For convenience, dotting into an unset field representing a nested message will +return an instance of that message with default values. As in the Objective-C +implementation, this does not actually cause the field to be set until the +returned message is mutated. Fortunately, thanks to the way mutability of value +types is implemented in Swift, the language automatically handles the +reassignment-on-mutation for us. A static singleton instance containing default +values can be associated with each message that can be returned when reading, so +copies are only made by the Swift runtime when mutation occurs. For example, +given the following proto: + +```protobuf +message Node { + Node child = 1; + string value = 2 [default = "foo"]; +} +``` + +The following Swift code would act as commented, where setting deeply nested +properties causes the copies and mutations to occur as the assignment statement +is unwound: + +```swift +var node = Node() + +let s = node.child.child.value +// 1. node.child returns the "default Node". +// 2. Reading .child on the result of (1) returns the same default Node. +// 3. Reading .value on the result of (2) returns the default value "foo". + +node.child.child.value = "bar" +// 4. Setting .value on the default Node causes a copy to be made and sets +// the property on that copy. Subsequently, the language updates the +// value of "node.child.child" to point to that copy. +// 5. Updating "node.child.child" in (4) requires another copy, because +// "node.child" was also the instance of the default node. The copy is +// assigned back to "node.child". +// 6. Setting "node.child" in (5) is a simple value reassignment, since +// "node" is a mutable var. +``` + +In other words, the generated messages do not internally have to manage parental +relationships to backfill the appropriate properties on mutation. Swift provides +this for free. + +## Scalar value fields + +Proto scalar value fields will map to Swift types in the following way: + +.proto Type | Swift Type +----------- | ------------------- +`double` | `Double` +`float` | `Float` +`int32` | `Int32` +`int64` | `Int64` +`uint32` | `UInt32` +`uint64` | `UInt64` +`sint32` | `Int32` +`sint64` | `Int64` +`fixed32` | `UInt32` +`fixed64` | `UInt64` +`sfixed32` | `Int32` +`sfixed64` | `Int64` +`bool` | `Bool` +`string` | `String` +`bytes` | `Foundation.NSData` + +The proto spec defines a number of integral types that map to the same Swift +type; for example, `intXX`, `sintXX`, and `sfixedXX` are all signed integers, +and `uintXX` and `fixedXX` are both unsigned integers. No other language +implementation distinguishes these further, so we do not do so either. The +rationale is that the various types only serve to distinguish how the value is +**encoded on the wire**; once loaded in memory, the user is not concerned about +these variations. + +Swift’s lack of implicit conversions among types will make it slightly annoying +to use these types in a context expecting an `Int`, or vice-versa, but since +this is a data-interchange format with explicitly-sized fields, we should not +hide that information from the user. Users will have to explicitly write +`Int(message.myField)`, for example. + +## Embedded message fields + +Embedded message fields can be represented using an optional variable of the +generated message type. Thus, the message + +```protobuf +message Foo { + Bar bar = 1; +} +``` + +would be represented in Swift as + +```swift +public struct Foo: ProtoMessage { + public var bar: Bar! { + get { ... } + set { ... } + } +} +``` + +If the user explicitly sets `bar` to nil, or if it was never set when read from +the wire, retrieving the value of `bar` would return a default, statically +allocated instance of `Bar` containing default values for its fields. This +achieves the desired behavior for default values in the same way that scalar +fields are designed, and also allows users to deep-drill into complex object +graphs to get or set fields without checking for nil at each step. + +## Enum fields + +The design and implementation of enum fields will differ somewhat drastically +depending on whether the message being generated is a proto2 or proto3 message. + +### proto2 enums + +For proto2, we do not need to be concerned about unknown enum values, so we can +use the simple raw-value enum syntax provided by Swift. So the following enum in +proto2: + +```protobuf +enum ContentType { + TEXT = 0; + IMAGE = 1; +} +``` + +would become this Swift enum: + +```swift +public enum ContentType: Int32, NilLiteralConvertible { + case text = 0 + case image = 1 + + public init(nilLiteral: ()) { + self = .text + } +} +``` + +See below for the discussion about `NilLiteralConvertible`. + +### proto3 enums + +For proto3, we need to be able to preserve unknown enum values that may come +across the wire so that they can be written back if unmodified. We can +accomplish this in Swift by using a case with an associated value for unknowns. +So the following enum in proto3: + +```protobuf +enum ContentType { + TEXT = 0; + IMAGE = 1; +} +``` + +would become this Swift enum: + +```swift +public enum ContentType: RawRepresentable, NilLiteralConvertible { + case text + case image + case UNKNOWN_VALUE(Int32) + + public typealias RawValue = Int32 + + public init(nilLiteral: ()) { + self = .text + } + + public init(rawValue: RawValue) { + switch rawValue { + case 0: self = .text + case 1: self = .image + default: self = .UNKNOWN_VALUE(rawValue) + } + + public var rawValue: RawValue { + switch self { + case .text: return 0 + case .image: return 1 + case .UNKNOWN_VALUE(let value): return value + } + } +} +``` + +Note that the use of a parameterized case prevents us from inheriting from the +raw `Int32` type; Swift does not allow an enum with a raw type to have cases +with arguments. Instead, we must implement the raw value initializer and +computed property manually. The `UNKNOWN_VALUE` case is explicitly chosen to be +"ugly" so that it stands out and does not conflict with other possible case +names. + +Using this approach, proto3 consumers must always have a default case or handle +the `.UNKNOWN_VALUE` case to satisfy case exhaustion in a switch statement; the +Swift compiler considers it an error if switch statements are not exhaustive. + +### NilLiteralConvertible conformance + +This is required to clean up the usage of enum-typed properties in switch +statements. Unlike other field types, enum properties cannot be +implicitly-unwrapped optionals without requiring that uses in switch statements +be explicitly unwrapped. For example, if we consider a message with the enum +above, this usage will fail to compile: + +```swift +// Without NilLiteralConvertible conformance on ContentType +public struct SomeMessage: ProtoMessage { + public var contentType: ContentType! { ... } +} + +// ERROR: no case named text or image +switch someMessage.contentType { + case .text: { ... } + case .image: { ... } +} +``` + +Even though our implementation guarantees that `contentType` will never be nil, +if it is an optional type, its cases would be `some` and `none`, not the cases +of the underlying enum type. In order to use it in this context, the user must +write `someMessage.contentType!` in their switch statement. + +Making the enum itself `NilLiteralConvertible` permits us to make the property +non-optional, so the user can still set it to nil to clear it (i.e., reset it to +its default value), while eliminating the need to explicitly unwrap it in a +switch statement. + +```swift +// With NilLiteralConvertible conformance on ContentType +public struct SomeMessage: ProtoMessage { + // Note that the property type is no longer optional + public var contentType: ContentType { ... } +} + +// OK: Compiles and runs as expected +switch someMessage.contentType { + case .text: { ... } + case .image: { ... } +} + +// The enum can be reset to its default value this way +someMessage.contentType = nil +``` + +One minor oddity with this approach is that nil will be auto-converted to the +default value of the enum in any context, not just field assignment. In other +words, this is valid: + +```swift +func foo(contentType: ContentType) { ... } +foo(nil) // Inside foo, contentType == .text +``` + +That being said, the advantage of being able to simultaneously support +nil-resettability and switch-without-unwrapping outweighs this side effect, +especially if appropriately documented. It is our hope that a new form of +resettable properties will be added to Swift that eliminates this inconsistency. +Some community members have already drafted or sent proposals for review that +would benefit our designs: + +* [SE-0030: Property Behaviors] + (https://github.com/apple/swift-evolution/blob/master/proposals/0030-property-behavior-decls.md) +* [Drafted: Resettable Properties] + (https://github.com/patters/swift-evolution/blob/master/proposals/0000-resettable-properties.md) + +### Enum aliases + +The `allow_alias` option in protobuf slightly complicates the use of Swift enums +to represent that type, because raw values of cases in an enum must be unique. +Swift lets us define static variables in an enum that alias actual cases. For +example, the following protobuf enum: + +```protobuf +enum Foo { + option allow_alias = true; + BAR = 0; + BAZ = 0; +} +``` + +will be represented in Swift as: + +```swift +public enum Foo: Int32, NilLiteralConvertible { + case bar = 0 + static public let baz = bar + + // ... etc. +} + +// Can still use .baz shorthand to reference the alias in contexts +// where the type is inferred +``` + +That is, we use the first name as the actual case and use static variables for +the other aliases. One drawback to this approach is that the static aliases +cannot be used as cases in a switch statement (the compiler emits the error +*“Enum case ‘baz’ not found in type ‘Foo’”*). However, in our own code bases, +there are only a few places where enum aliases are not mere renamings of an +older value, but they also don’t appear to be the type of value that one would +expect to switch on (for example, a group of named constants representing +metrics rather than a set of options), so this restriction is not significant. + +This strategy also implies that changing the name of an enum and adding the old +name as an alias below the new name will be a breaking change in the generated +Swift code. + +## Oneof types + +The `oneof` feature represents a “variant/union” data type that maps nicely to +Swift enums with associated values (algebraic types). These fields can also be +accessed independently though, and, specifically in the case of proto2, it’s +reasonable to expect access to default values when accessing a field that is not +explicitly set. + +Taking all this into account, we can represent a `oneof` in Swift with two sets +of constructs: + +* Properties in the message that correspond to the `oneof` fields. +* A nested enum named after the `oneof` and which provides the corresponding + field values as case arguments. + +This approach fulfills the needs of proto consumers by providing a +Swift-idiomatic way of simultaneously checking which field is set and accessing +its value, providing individual properties to access the default values +(important for proto2), and safely allows a field to be moved into a `oneof` +without breaking clients. + +Consider the following proto: + +```protobuf +message MyMessage { + oneof record { + string name = 1 [default = "unnamed"]; + int32 id_number = 2 [default = 0]; + } +} +``` + +In Swift, we would generate an enum, a property for that enum, and properties +for the fields themselves: + +```swift +public struct MyMessage: ProtoMessage { + public enum Record: NilLiteralConvertible { + case name(String) + case idNumber(Int32) + case NOT_SET + + public init(nilLiteral: ()) { self = .NOT_SET } + } + + // This is the "Swifty" way of accessing the value + public var record: Record { ... } + + // Direct access to the underlying fields + public var name: String! { ... } + public var idNumber: Int32! { ... } +} +``` + +This makes both usage patterns possible: + +```swift +// Usage 1: Case-based dispatch +switch message.record { + case .name(let name): + // Do something with name if it was explicitly set + case .idNumber(let id): + // Do something with id_number if it was explicitly set + case .NOT_SET: + // Do something if it’s not set +} + +// Usage 2: Direct access for default value fallback +// Sets the label text to the name if it was explicitly set, or to +// "unnamed" (the default value for the field) if id_number was set +// instead +let myLabel = UILabel() +myLabel.text = message.name +``` + +As with proto enums, the generated `oneof` enum conforms to +`NilLiteralConvertible` to avoid switch statement issues. Setting the property +to nil will clear it (i.e., reset it to `NOT_SET`). + +## Unknown Fields (proto2 only) + +To be written. + +## Extensions (proto2 only) + +To be written. + +## Reflection and Descriptors + +We will not include reflection or descriptors in the first version of the Swift +library. The use cases for reflection on mobile are not as strong and the static +data to represent the descriptors would add bloat when we wish to keep the code +size small. + +In the future, we will investigate whether they can be included as extensions +which might be able to be excluded from a build and/or automatically dead +stripped by the compiler if they are not used. + +## Appendix A: Rejected strategies to handle packages + +### Each package is its own Swift module + +Each proto package could be declared as its own Swift module, replacing dots +with underscores (e.g., package `foo.bar` becomes module `Foo_Bar`). Then, users +would simply import modules containing whatever proto modules they want to use +and refer to the generated types by their short names. + +**This solution is simply not possible, however.** Swift modules cannot +circularly reference each other, but there is no restriction against proto +packages doing so. Circular imports are forbidden (e.g., `foo.proto` importing +`bar.proto` importing `foo.proto`), but nothing prevents package `foo` from +using a type in package `bar` which uses a different type in package `foo`, as +long as there is no import cycle. If these packages were generated as Swift +modules, then `Foo` would contain an `import Bar` statement and `Bar` would +contain an `import Foo` statement, and there is no way to compile this. + +### Ad hoc namespacing with structs + +We can “fake” namespaces in Swift by declaring empty structs with private +initializers. Since modules are constructed based on compiler arguments, not by +syntactic constructs, and because there is no pure Swift way to define +submodules (even though Clang module maps support this), there is no +source-drive way to group generated code into namespaces aside from this +approach. + +Types can be added to those intermediate package structs using Swift extensions. +For example, a message `Baz` in package `foo.bar` could be represented in Swift +as follows: + +```swift +public struct Foo { + private init() {} +} + +public extension Foo { + public struct Bar { + private init() {} + } +} + +public extension Foo.Bar { + public struct Baz { + // Message fields and other methods + } +} + +let baz = Foo.Bar.Baz() +``` + +Each of these constructs would actually be defined in a separate file; Swift +lets us keep them separate and add multiple structs to a single “namespace” +through extensions. + +Unfortunately, these intermediate structs generate symbols of their own +(metatype information in the data segment). This becomes problematic if multiple +build targets contain Swift sources generated from different messages in the +same package. At link time, these symbols would collide, resulting in multiple +definition errors. + +This approach also has the disadvantage that there is no automatic “short” way +to refer to the generated messages at the deepest nesting levels; since this use +of structs is a hack around the lack of namespaces, there is no equivalent to +import (Java) or using (C++) to simplify this. Users would have to declare type +aliases to make this cleaner, or we would have to generate them for users. From d419ca10b4521e87d0be62f6df4ede97d63c7ee2 Mon Sep 17 00:00:00 2001 From: Josh Haberman Date: Fri, 22 Apr 2016 14:11:39 -0700 Subject: [PATCH 5/8] Updated upb and simplified ruby code a bit with new upb method. --- ruby/ext/google/protobuf_c/encode_decode.c | 7 +- ruby/ext/google/protobuf_c/message.c | 38 +- ruby/ext/google/protobuf_c/upb.c | 1412 ++++++++++--------- ruby/ext/google/protobuf_c/upb.h | 1429 +++++++++++--------- 4 files changed, 1608 insertions(+), 1278 deletions(-) diff --git a/ruby/ext/google/protobuf_c/encode_decode.c b/ruby/ext/google/protobuf_c/encode_decode.c index 9bc7273e06..f6bea50f39 100644 --- a/ruby/ext/google/protobuf_c/encode_decode.c +++ b/ruby/ext/google/protobuf_c/encode_decode.c @@ -656,7 +656,6 @@ static const upb_json_parsermethod *msgdef_jsonparsermethod(Descriptor* desc) { #define STACK_ENV_STACKBYTES 4096 typedef struct { upb_env env; - upb_seededalloc alloc; const char* ruby_error_template; char allocbuf[STACK_ENV_STACKBYTES]; } stackenv; @@ -681,16 +680,12 @@ static bool env_error_func(void* ud, const upb_status* status) { static void stackenv_init(stackenv* se, const char* errmsg) { se->ruby_error_template = errmsg; - upb_env_init(&se->env); - upb_seededalloc_init(&se->alloc, &se->allocbuf, STACK_ENV_STACKBYTES); - upb_env_setallocfunc( - &se->env, upb_seededalloc_getallocfunc(&se->alloc), &se->alloc); + upb_env_init2(&se->env, se->allocbuf, sizeof(se->allocbuf), NULL); upb_env_seterrorfunc(&se->env, env_error_func, se); } static void stackenv_uninit(stackenv* se) { upb_env_uninit(&se->env); - upb_seededalloc_uninit(&se->alloc); } /* diff --git a/ruby/ext/google/protobuf_c/message.c b/ruby/ext/google/protobuf_c/message.c index 3a51fe4711..e16250f3b0 100644 --- a/ruby/ext/google/protobuf_c/message.c +++ b/ruby/ext/google/protobuf_c/message.c @@ -151,32 +151,30 @@ VALUE Message_method_missing(int argc, VALUE* argv, VALUE _self) { name_len--; } - // Check for a oneof name first. - o = upb_msgdef_ntoo(self->descriptor->msgdef, - name, name_len); + // See if this name corresponds to either a oneof or field in this message. + if (!upb_msgdef_lookupname(self->descriptor->msgdef, name, name_len, &f, + &o)) { + return rb_call_super(argc, argv); + } + if (o != NULL) { + // This is a oneof -- return which field inside the oneof is set. if (setter) { rb_raise(rb_eRuntimeError, "Oneof accessors are read-only."); } return which_oneof_field(self, o); - } - - // Otherwise, check for a field with that name. - f = upb_msgdef_ntof(self->descriptor->msgdef, - name, name_len); - - if (f == NULL) { - return rb_call_super(argc, argv); - } - - if (setter) { - if (argc < 2) { - rb_raise(rb_eArgError, "No value provided to setter."); - } - layout_set(self->descriptor->layout, Message_data(self), f, argv[1]); - return Qnil; } else { - return layout_get(self->descriptor->layout, Message_data(self), f); + // This is a field -- get or set the field's value. + assert(f); + if (setter) { + if (argc < 2) { + rb_raise(rb_eArgError, "No value provided to setter."); + } + layout_set(self->descriptor->layout, Message_data(self), f, argv[1]); + return Qnil; + } else { + return layout_get(self->descriptor->layout, Message_data(self), f); + } } } diff --git a/ruby/ext/google/protobuf_c/upb.c b/ruby/ext/google/protobuf_c/upb.c index 212f1e0e6b..74a2a1db30 100644 --- a/ruby/ext/google/protobuf_c/upb.c +++ b/ruby/ext/google/protobuf_c/upb.c @@ -12,7 +12,7 @@ typedef struct { } str_t; static str_t *newstr(const char *data, size_t len) { - str_t *ret = malloc(sizeof(*ret) + len); + str_t *ret = upb_gmalloc(sizeof(*ret) + len); if (!ret) return NULL; ret->len = len; memcpy(ret->str, data, len); @@ -20,7 +20,7 @@ static str_t *newstr(const char *data, size_t len) { return ret; } -static void freestr(str_t *s) { free(s); } +static void freestr(str_t *s) { upb_gfree(s); } /* isalpha() etc. from are locale-dependent, which we don't want. */ static bool upb_isbetween(char c, char low, char high) { @@ -65,6 +65,22 @@ static bool upb_isident(const char *str, size_t len, bool full, upb_status *s) { return !start; } +static bool upb_isoneof(const upb_refcounted *def) { + return def->vtbl == &upb_oneofdef_vtbl; +} + +static bool upb_isfield(const upb_refcounted *def) { + return def->vtbl == &upb_fielddef_vtbl; +} + +static const upb_oneofdef *upb_trygetoneof(const upb_refcounted *def) { + return upb_isoneof(def) ? (const upb_oneofdef*)def : NULL; +} + +static const upb_fielddef *upb_trygetfield(const upb_refcounted *def) { + return upb_isfield(def) ? (const upb_fielddef*)def : NULL; +} + /* upb_def ********************************************************************/ @@ -88,9 +104,18 @@ const char *upb_def_name(const upb_def *d) { bool upb_def_setfullname(upb_def *def, const char *fullname, upb_status *s) { assert(!upb_def_isfrozen(def)); - if (!upb_isident(fullname, strlen(fullname), true, s)) return false; - free((void*)def->fullname); - def->fullname = upb_strdup(fullname); + if (!upb_isident(fullname, strlen(fullname), true, s)) { + return false; + } + + fullname = upb_gstrdup(fullname); + if (!fullname) { + upb_upberr_setoom(s); + return false; + } + + upb_gfree((void*)def->fullname); + def->fullname = fullname; return true; } @@ -123,7 +148,7 @@ static bool upb_def_init(upb_def *def, upb_deftype_t type, } static void upb_def_uninit(upb_def *def) { - free((void*)def->fullname); + upb_gfree((void*)def->fullname); } static const char *msgdef_name(const upb_msgdef *m) { @@ -260,8 +285,19 @@ static bool assign_msg_indices(upb_msgdef *m, upb_status *s) { int i; uint32_t selector; int n = upb_msgdef_numfields(m); - upb_fielddef **fields = malloc(n * sizeof(*fields)); - if (!fields) return false; + upb_fielddef **fields; + + if (n == 0) { + m->selector_count = UPB_STATIC_SELECTOR_COUNT; + m->submsg_field_count = 0; + return true; + } + + fields = upb_gmalloc(n * sizeof(*fields)); + if (!fields) { + upb_upberr_setoom(s); + return false; + } m->submsg_field_count = 0; for(i = 0, upb_msg_field_begin(&j, m); @@ -270,7 +306,7 @@ static bool assign_msg_indices(upb_msgdef *m, upb_status *s) { upb_fielddef *f = upb_msg_iter_field(&j); assert(f->msg.def == m); if (!upb_validate_field(f, s)) { - free(fields); + upb_gfree(fields); return false; } if (upb_fielddef_issubmsg(f)) { @@ -330,7 +366,7 @@ static bool assign_msg_indices(upb_msgdef *m, upb_status *s) { #undef TRY #endif - free(fields); + upb_gfree(fields); return true; } @@ -407,21 +443,26 @@ static void upb_enumdef_free(upb_refcounted *r) { upb_inttable_iter i; upb_inttable_begin(&i, &e->iton); for( ; !upb_inttable_done(&i); upb_inttable_next(&i)) { - /* To clean up the upb_strdup() from upb_enumdef_addval(). */ - free(upb_value_getcstr(upb_inttable_iter_value(&i))); + /* To clean up the upb_gstrdup() from upb_enumdef_addval(). */ + upb_gfree(upb_value_getcstr(upb_inttable_iter_value(&i))); } upb_strtable_uninit(&e->ntoi); upb_inttable_uninit(&e->iton); upb_def_uninit(upb_enumdef_upcast_mutable(e)); - free(e); + upb_gfree(e); } +const struct upb_refcounted_vtbl upb_enumdef_vtbl = {NULL, &upb_enumdef_free}; + upb_enumdef *upb_enumdef_new(const void *owner) { - static const struct upb_refcounted_vtbl vtbl = {NULL, &upb_enumdef_free}; - upb_enumdef *e = malloc(sizeof(*e)); + upb_enumdef *e = upb_gmalloc(sizeof(*e)); if (!e) return NULL; - if (!upb_def_init(upb_enumdef_upcast_mutable(e), UPB_DEF_ENUM, &vtbl, owner)) + + if (!upb_def_init(upb_enumdef_upcast_mutable(e), UPB_DEF_ENUM, + &upb_enumdef_vtbl, owner)) { goto err2; + } + if (!upb_strtable_init(&e->ntoi, UPB_CTYPE_INT32)) goto err2; if (!upb_inttable_init(&e->iton, UPB_CTYPE_CSTR)) goto err1; return e; @@ -429,7 +470,7 @@ upb_enumdef *upb_enumdef_new(const void *owner) { err1: upb_strtable_uninit(&e->ntoi); err2: - free(e); + upb_gfree(e); return NULL; } @@ -468,27 +509,36 @@ bool upb_enumdef_setfullname(upb_enumdef *e, const char *fullname, bool upb_enumdef_addval(upb_enumdef *e, const char *name, int32_t num, upb_status *status) { + char *name2; + if (!upb_isident(name, strlen(name), false, status)) { return false; } + if (upb_enumdef_ntoiz(e, name, NULL)) { upb_status_seterrf(status, "name '%s' is already defined", name); return false; } + if (!upb_strtable_insert(&e->ntoi, name, upb_value_int32(num))) { upb_status_seterrmsg(status, "out of memory"); return false; } - if (!upb_inttable_lookup(&e->iton, num, NULL) && - !upb_inttable_insert(&e->iton, num, upb_value_cstr(upb_strdup(name)))) { - upb_status_seterrmsg(status, "out of memory"); - upb_strtable_remove(&e->ntoi, name, NULL); - return false; + + if (!upb_inttable_lookup(&e->iton, num, NULL)) { + name2 = upb_gstrdup(name); + if (!name2 || !upb_inttable_insert(&e->iton, num, upb_value_cstr(name2))) { + upb_status_seterrmsg(status, "out of memory"); + upb_strtable_remove(&e->ntoi, name, NULL); + return false; + } } + if (upb_enumdef_numvals(e) == 1) { bool ok = upb_enumdef_setdefault(e, num, NULL); UPB_ASSERT_VAR(ok, ok); } + return true; } @@ -575,9 +625,9 @@ static void freefield(upb_refcounted *r) { upb_fielddef *f = (upb_fielddef*)r; upb_fielddef_uninit_default(f); if (f->subdef_is_symbolic) - free(f->sub.name); + upb_gfree(f->sub.name); upb_def_uninit(upb_fielddef_upcast_mutable(f)); - free(f); + upb_gfree(f); } static const char *enumdefaultstr(const upb_fielddef *f) { @@ -633,12 +683,14 @@ static bool enumdefaultint32(const upb_fielddef *f, int32_t *val) { return false; } +const struct upb_refcounted_vtbl upb_fielddef_vtbl = {visitfield, freefield}; + upb_fielddef *upb_fielddef_new(const void *o) { - static const struct upb_refcounted_vtbl vtbl = {visitfield, freefield}; - upb_fielddef *f = malloc(sizeof(*f)); + upb_fielddef *f = upb_gmalloc(sizeof(*f)); if (!f) return NULL; - if (!upb_def_init(upb_fielddef_upcast_mutable(f), UPB_DEF_FIELD, &vtbl, o)) { - free(f); + if (!upb_def_init(upb_fielddef_upcast_mutable(f), UPB_DEF_FIELD, + &upb_fielddef_vtbl, o)) { + upb_gfree(f); return NULL; } f->msg.def = NULL; @@ -689,7 +741,7 @@ upb_fielddef *upb_fielddef_dup(const upb_fielddef *f, const void *owner) { srcname = f->sub.def ? upb_def_fullname(f->sub.def) : NULL; } if (srcname) { - char *newname = malloc(strlen(f->sub.def->fullname) + 2); + char *newname = upb_gmalloc(strlen(f->sub.def->fullname) + 2); if (!newname) { upb_fielddef_unref(newf, owner); return NULL; @@ -697,7 +749,7 @@ upb_fielddef *upb_fielddef_dup(const upb_fielddef *f, const void *owner) { strcpy(newname, "."); strcat(newname, f->sub.def->fullname); upb_fielddef_setsubdefname(newf, newname, NULL); - free(newname); + upb_gfree(newname); } return newf; @@ -804,11 +856,12 @@ const char *upb_fielddef_containingtypename(upb_fielddef *f) { } static void release_containingtype(upb_fielddef *f) { - if (f->msg_is_symbolic) free(f->msg.name); + if (f->msg_is_symbolic) upb_gfree(f->msg.name); } bool upb_fielddef_setcontainingtypename(upb_fielddef *f, const char *name, upb_status *s) { + char *name_copy; assert(!upb_fielddef_isfrozen(f)); if (upb_fielddef_containingtype(f)) { upb_status_seterrmsg(s, "field has already been added to a message."); @@ -816,8 +869,15 @@ bool upb_fielddef_setcontainingtypename(upb_fielddef *f, const char *name, } /* TODO: validate name (upb_isident() doesn't quite work atm because this name * may have a leading "."). */ + + name_copy = upb_gstrdup(name); + if (!name_copy) { + upb_upberr_setoom(s); + return false; + } + release_containingtype(f); - f->msg.name = upb_strdup(name); + f->msg.name = name_copy; f->msg_is_symbolic = true; return true; } @@ -1218,7 +1278,7 @@ static bool upb_subdef_typecheck(upb_fielddef *f, const upb_def *subdef, static void release_subdef(upb_fielddef *f) { if (f->subdef_is_symbolic) { - free(f->sub.name); + upb_gfree(f->sub.name); } else if (f->sub.def) { upb_unref2(f->sub.def, f); } @@ -1248,15 +1308,23 @@ bool upb_fielddef_setenumsubdef(upb_fielddef *f, const upb_enumdef *subdef, bool upb_fielddef_setsubdefname(upb_fielddef *f, const char *name, upb_status *s) { + char *name_copy; assert(!upb_fielddef_isfrozen(f)); if (!upb_fielddef_hassubdef(f)) { upb_status_seterrmsg(s, "field type does not accept a subdef"); return false; } + + name_copy = upb_gstrdup(name); + if (!name_copy) { + upb_upberr_setoom(s); + return false; + } + /* TODO: validate name (upb_isident() doesn't quite work atm because this name * may have a leading "."). */ release_subdef(f); - f->sub.name = upb_strdup(name); + f->sub.name = name_copy; f->subdef_is_symbolic = true; return true; } @@ -1332,32 +1400,33 @@ static void visitmsg(const upb_refcounted *r, upb_refcounted_visit *visit, static void freemsg(upb_refcounted *r) { upb_msgdef *m = (upb_msgdef*)r; - upb_strtable_uninit(&m->ntoo); upb_strtable_uninit(&m->ntof); upb_inttable_uninit(&m->itof); upb_def_uninit(upb_msgdef_upcast_mutable(m)); - free(m); + upb_gfree(m); } +const struct upb_refcounted_vtbl upb_msgdef_vtbl = {visitmsg, freemsg}; + upb_msgdef *upb_msgdef_new(const void *owner) { - static const struct upb_refcounted_vtbl vtbl = {visitmsg, freemsg}; - upb_msgdef *m = malloc(sizeof(*m)); + upb_msgdef *m = upb_gmalloc(sizeof(*m)); if (!m) return NULL; - if (!upb_def_init(upb_msgdef_upcast_mutable(m), UPB_DEF_MSG, &vtbl, owner)) + + if (!upb_def_init(upb_msgdef_upcast_mutable(m), UPB_DEF_MSG, &upb_msgdef_vtbl, + owner)) { goto err2; - if (!upb_inttable_init(&m->itof, UPB_CTYPE_PTR)) goto err3; - if (!upb_strtable_init(&m->ntof, UPB_CTYPE_PTR)) goto err2; - if (!upb_strtable_init(&m->ntoo, UPB_CTYPE_PTR)) goto err1; + } + + if (!upb_inttable_init(&m->itof, UPB_CTYPE_PTR)) goto err2; + if (!upb_strtable_init(&m->ntof, UPB_CTYPE_PTR)) goto err1; m->map_entry = false; m->syntax = UPB_SYNTAX_PROTO2; return m; err1: - upb_strtable_uninit(&m->ntof); -err2: upb_inttable_uninit(&m->itof); -err3: - free(m); +err2: + upb_gfree(m); return NULL; } @@ -1415,6 +1484,19 @@ bool upb_msgdef_setfullname(upb_msgdef *m, const char *fullname, return upb_def_setfullname(upb_msgdef_upcast_mutable(m), fullname, s); } +bool upb_msgdef_setsyntax(upb_msgdef *m, upb_syntax_t syntax) { + if (syntax != UPB_SYNTAX_PROTO2 && syntax != UPB_SYNTAX_PROTO3) { + return false; + } + + m->syntax = syntax; + return true; +} + +upb_syntax_t upb_msgdef_syntax(const upb_msgdef *m) { + return m->syntax; +} + /* Helper: check that the field |f| is safe to add to msgdef |m|. Set an error * on status |s| and return false if not. */ static bool check_field_add(const upb_msgdef *m, const upb_fielddef *f, @@ -1425,9 +1507,11 @@ static bool check_field_add(const upb_msgdef *m, const upb_fielddef *f, } else if (upb_fielddef_name(f) == NULL || upb_fielddef_number(f) == 0) { upb_status_seterrmsg(s, "field name or number were not set"); return false; - } else if (upb_msgdef_ntofz(m, upb_fielddef_name(f)) || - upb_msgdef_itof(m, upb_fielddef_number(f))) { - upb_status_seterrmsg(s, "duplicate field name or number for field"); + } else if (upb_msgdef_itof(m, upb_fielddef_number(f))) { + upb_status_seterrmsg(s, "duplicate field number"); + return false; + } else if (upb_strtable_lookup(&m->ntof, upb_fielddef_name(f), NULL)) { + upb_status_seterrmsg(s, "name conflicts with existing field or oneof"); return false; } return true; @@ -1488,8 +1572,8 @@ bool upb_msgdef_addoneof(upb_msgdef *m, upb_oneofdef *o, const void *ref_donor, } else if (upb_oneofdef_name(o) == NULL) { upb_status_seterrmsg(s, "oneofdef name was not set"); return false; - } else if (upb_msgdef_ntooz(m, upb_oneofdef_name(o))) { - upb_status_seterrmsg(s, "duplicate oneof name"); + } else if (upb_strtable_lookup(&m->ntof, upb_oneofdef_name(o), NULL)) { + upb_status_seterrmsg(s, "name conflicts with existing field or oneof"); return false; } @@ -1506,7 +1590,7 @@ bool upb_msgdef_addoneof(upb_msgdef *m, upb_oneofdef *o, const void *ref_donor, /* Add oneof itself first. */ o->parent = m; - upb_strtable_insert(&m->ntoo, upb_oneofdef_name(o), upb_value_ptr(o)); + upb_strtable_insert(&m->ntof, upb_oneofdef_name(o), upb_value_ptr(o)); upb_ref2(o, m); upb_ref2(m, o); @@ -1530,23 +1614,47 @@ const upb_fielddef *upb_msgdef_itof(const upb_msgdef *m, uint32_t i) { const upb_fielddef *upb_msgdef_ntof(const upb_msgdef *m, const char *name, size_t len) { upb_value val; - return upb_strtable_lookup2(&m->ntof, name, len, &val) ? - upb_value_getptr(val) : NULL; + + if (!upb_strtable_lookup2(&m->ntof, name, len, &val)) { + return NULL; + } + + return upb_trygetfield(upb_value_getptr(val)); } const upb_oneofdef *upb_msgdef_ntoo(const upb_msgdef *m, const char *name, size_t len) { upb_value val; - return upb_strtable_lookup2(&m->ntoo, name, len, &val) ? - upb_value_getptr(val) : NULL; + + if (!upb_strtable_lookup2(&m->ntof, name, len, &val)) { + return NULL; + } + + return upb_trygetoneof(upb_value_getptr(val)); +} + +bool upb_msgdef_lookupname(const upb_msgdef *m, const char *name, size_t len, + const upb_fielddef **f, const upb_oneofdef **o) { + upb_value val; + + if (!upb_strtable_lookup2(&m->ntof, name, len, &val)) { + return false; + } + + *o = upb_trygetoneof(upb_value_getptr(val)); + *f = upb_trygetfield(upb_value_getptr(val)); + assert((*o != NULL) ^ (*f != NULL)); /* Exactly one of the two should be set. */ + return true; } int upb_msgdef_numfields(const upb_msgdef *m) { - return upb_strtable_count(&m->ntof); + /* The number table contains only fields. */ + return upb_inttable_count(&m->itof); } int upb_msgdef_numoneofs(const upb_msgdef *m) { - return upb_strtable_count(&m->ntoo); + /* The name table includes oneofs, and the number table does not. */ + return upb_strtable_count(&m->ntof) - upb_inttable_count(&m->itof); } void upb_msgdef_setmapentry(upb_msgdef *m, bool map_entry) { @@ -1577,10 +1685,21 @@ void upb_msg_field_iter_setdone(upb_msg_field_iter *iter) { } void upb_msg_oneof_begin(upb_msg_oneof_iter *iter, const upb_msgdef *m) { - upb_strtable_begin(iter, &m->ntoo); + upb_strtable_begin(iter, &m->ntof); + /* We need to skip past any initial fields. */ + while (!upb_strtable_done(iter) && + !upb_isoneof(upb_value_getptr(upb_strtable_iter_value(iter)))) { + upb_strtable_next(iter); + } } -void upb_msg_oneof_next(upb_msg_oneof_iter *iter) { upb_strtable_next(iter); } +void upb_msg_oneof_next(upb_msg_oneof_iter *iter) { + /* We need to skip past fields to return only oneofs. */ + do { + upb_strtable_next(iter); + } while (!upb_strtable_done(iter) && + !upb_isoneof(upb_value_getptr(upb_strtable_iter_value(iter)))); +} bool upb_msg_oneof_done(const upb_msg_oneof_iter *iter) { return upb_strtable_done(iter); @@ -1613,26 +1732,36 @@ static void freeoneof(upb_refcounted *r) { upb_oneofdef *o = (upb_oneofdef*)r; upb_strtable_uninit(&o->ntof); upb_inttable_uninit(&o->itof); - free((void*)o->name); - free(o); + upb_gfree((void*)o->name); + upb_gfree(o); } +const struct upb_refcounted_vtbl upb_oneofdef_vtbl = {visitoneof, freeoneof}; + upb_oneofdef *upb_oneofdef_new(const void *owner) { - static const struct upb_refcounted_vtbl vtbl = {visitoneof, freeoneof}; - upb_oneofdef *o = malloc(sizeof(*o)); + upb_oneofdef *o = upb_gmalloc(sizeof(*o)); + + if (!o) { + return NULL; + } + o->parent = NULL; - if (!o) return NULL; - if (!upb_refcounted_init(upb_oneofdef_upcast_mutable(o), &vtbl, owner)) - goto err2; o->name = NULL; + + if (!upb_refcounted_init(upb_oneofdef_upcast_mutable(o), &upb_oneofdef_vtbl, + owner)) { + goto err2; + } + if (!upb_inttable_init(&o->itof, UPB_CTYPE_PTR)) goto err2; if (!upb_strtable_init(&o->ntof, UPB_CTYPE_PTR)) goto err1; + return o; err1: upb_inttable_uninit(&o->itof); err2: - free(o); + upb_gfree(o); return NULL; } @@ -1661,9 +1790,19 @@ bool upb_oneofdef_setname(upb_oneofdef *o, const char *name, upb_status *s) { upb_status_seterrmsg(s, "oneof already added to a message"); return false; } - if (!upb_isident(name, strlen(name), true, s)) return false; - free((void*)o->name); - o->name = upb_strdup(name); + + if (!upb_isident(name, strlen(name), true, s)) { + return false; + } + + name = upb_gstrdup(name); + if (!name) { + upb_status_seterrmsg(s, "One of memory"); + return false; + } + + upb_gfree((void*)o->name); + o->name = name; return true; } @@ -1805,14 +1944,15 @@ static void freefiledef(upb_refcounted *r) { upb_inttable_uninit(&f->defs); upb_inttable_uninit(&f->deps); - free((void*)f->name); - free((void*)f->package); - free(f); + upb_gfree((void*)f->name); + upb_gfree((void*)f->package); + upb_gfree(f); } +const struct upb_refcounted_vtbl upb_filedef_vtbl = {visitfiledef, freefiledef}; + upb_filedef *upb_filedef_new(const void *owner) { - static const struct upb_refcounted_vtbl vtbl = {visitfiledef, freefiledef}; - upb_filedef *f = malloc(sizeof(*f)); + upb_filedef *f = upb_gmalloc(sizeof(*f)); if (!f) { return NULL; @@ -1822,7 +1962,8 @@ upb_filedef *upb_filedef_new(const void *owner) { f->name = NULL; f->syntax = UPB_SYNTAX_PROTO2; - if (!upb_refcounted_init(upb_filedef_upcast_mutable(f), &vtbl, owner)) { + if (!upb_refcounted_init(upb_filedef_upcast_mutable(f), &upb_filedef_vtbl, + owner)) { goto err; } @@ -1841,7 +1982,7 @@ err2: upb_inttable_uninit(&f->defs); err: - free(f); + upb_gfree(f); return NULL; } @@ -1886,12 +2027,12 @@ const upb_filedef *upb_filedef_dep(const upb_filedef *f, size_t i) { } bool upb_filedef_setname(upb_filedef *f, const char *name, upb_status *s) { - name = upb_strdup(name); + name = upb_gstrdup(name); if (!name) { - upb_status_seterrmsg(s, "Out of memory"); + upb_upberr_setoom(s); return false; } - free((void*)f->name); + upb_gfree((void*)f->name); f->name = name; return true; } @@ -1899,12 +2040,12 @@ bool upb_filedef_setname(upb_filedef *f, const char *name, upb_status *s) { bool upb_filedef_setpackage(upb_filedef *f, const char *package, upb_status *s) { if (!upb_isident(package, strlen(package), true, s)) return false; - package = upb_strdup(package); + package = upb_gstrdup(package); if (!package) { - upb_status_seterrmsg(s, "Out of memory"); + upb_upberr_setoom(s); return false; } - free((void*)f->package); + upb_gfree((void*)f->package); f->package = package; return true; } @@ -1953,7 +2094,7 @@ bool upb_filedef_adddef(upb_filedef *f, upb_def *def, const void *ref_donor, } return true; } else { - upb_status_seterrmsg(s, "Out of memory."); + upb_upberr_setoom(s); return false; } } @@ -1967,288 +2108,22 @@ bool upb_filedef_adddep(upb_filedef *f, const upb_filedef *dep) { return false; } } - - -#include -#include -#include - -typedef struct cleanup_ent { - upb_cleanup_func *cleanup; - void *ud; - struct cleanup_ent *next; -} cleanup_ent; - -static void *seeded_alloc(void *ud, void *ptr, size_t oldsize, size_t size); - -/* Default allocator **********************************************************/ - -/* Just use realloc, keeping all allocated blocks in a linked list to destroy at - * the end. */ - -typedef struct mem_block { - /* List is doubly-linked, because in cases where realloc() moves an existing - * block, we need to be able to remove the old pointer from the list - * efficiently. */ - struct mem_block *prev, *next; -#ifndef NDEBUG - size_t size; /* Doesn't include mem_block structure. */ -#endif -} mem_block; - -typedef struct { - mem_block *head; -} default_alloc_ud; - -static void *default_alloc(void *_ud, void *ptr, size_t oldsize, size_t size) { - default_alloc_ud *ud = _ud; - mem_block *from, *block; - void *ret; - UPB_UNUSED(oldsize); - - from = ptr ? (void*)((char*)ptr - sizeof(mem_block)) : NULL; - -#ifndef NDEBUG - if (from) { - assert(oldsize <= from->size); - } -#endif - - /* TODO(haberman): we probably need to provide even better alignment here, - * like 16-byte alignment of the returned data pointer. */ - block = realloc(from, size + sizeof(mem_block)); - if (!block) return NULL; - ret = (char*)block + sizeof(*block); - -#ifndef NDEBUG - block->size = size; -#endif - - if (from) { - if (block != from) { - /* The block was moved, so pointers in next and prev blocks must be - * updated to its new location. */ - if (block->next) block->next->prev = block; - if (block->prev) block->prev->next = block; - if (ud->head == from) ud->head = block; - } - } else { - /* Insert at head of linked list. */ - block->prev = NULL; - block->next = ud->head; - if (block->next) block->next->prev = block; - ud->head = block; - } - - return ret; -} - -static void default_alloc_cleanup(void *_ud) { - default_alloc_ud *ud = _ud; - mem_block *block = ud->head; - - while (block) { - void *to_free = block; - block = block->next; - free(to_free); - } -} - - -/* Standard error functions ***************************************************/ - -static bool default_err(void *ud, const upb_status *status) { - UPB_UNUSED(ud); - UPB_UNUSED(status); - return false; -} - -static bool write_err_to(void *ud, const upb_status *status) { - upb_status *copy_to = ud; - upb_status_copy(copy_to, status); - return false; -} - - -/* upb_env ********************************************************************/ - -void upb_env_init(upb_env *e) { - default_alloc_ud *ud = (default_alloc_ud*)&e->default_alloc_ud; - e->ok_ = true; - e->bytes_allocated = 0; - e->cleanup_head = NULL; - - ud->head = NULL; - - /* Set default functions. */ - upb_env_setallocfunc(e, default_alloc, ud); - upb_env_seterrorfunc(e, default_err, NULL); -} - -void upb_env_uninit(upb_env *e) { - cleanup_ent *ent = e->cleanup_head; - - while (ent) { - ent->cleanup(ent->ud); - ent = ent->next; - } - - /* Must do this after running cleanup functions, because this will delete - the memory we store our cleanup entries in! */ - if (e->alloc == default_alloc) { - default_alloc_cleanup(e->alloc_ud); - } -} - -UPB_FORCEINLINE void upb_env_setallocfunc(upb_env *e, upb_alloc_func *alloc, - void *ud) { - e->alloc = alloc; - e->alloc_ud = ud; -} - -UPB_FORCEINLINE void upb_env_seterrorfunc(upb_env *e, upb_error_func *func, - void *ud) { - e->err = func; - e->err_ud = ud; -} - -void upb_env_reporterrorsto(upb_env *e, upb_status *status) { - e->err = write_err_to; - e->err_ud = status; -} - -bool upb_env_ok(const upb_env *e) { - return e->ok_; -} - -bool upb_env_reporterror(upb_env *e, const upb_status *status) { - e->ok_ = false; - return e->err(e->err_ud, status); -} - -bool upb_env_addcleanup(upb_env *e, upb_cleanup_func *func, void *ud) { - cleanup_ent *ent = upb_env_malloc(e, sizeof(cleanup_ent)); - if (!ent) return false; - - ent->cleanup = func; - ent->ud = ud; - ent->next = e->cleanup_head; - e->cleanup_head = ent; - - return true; -} - -void *upb_env_malloc(upb_env *e, size_t size) { - e->bytes_allocated += size; - if (e->alloc == seeded_alloc) { - /* This is equivalent to the next branch, but allows inlining for a - * measurable perf benefit. */ - return seeded_alloc(e->alloc_ud, NULL, 0, size); - } else { - return e->alloc(e->alloc_ud, NULL, 0, size); - } -} - -void *upb_env_realloc(upb_env *e, void *ptr, size_t oldsize, size_t size) { - char *ret; - assert(oldsize <= size); - ret = e->alloc(e->alloc_ud, ptr, oldsize, size); - -#ifndef NDEBUG - /* Overwrite non-preserved memory to ensure callers are passing the oldsize - * that they truly require. */ - memset(ret + oldsize, 0xff, size - oldsize); -#endif - - return ret; -} - -size_t upb_env_bytesallocated(const upb_env *e) { - return e->bytes_allocated; -} - - -/* upb_seededalloc ************************************************************/ - -/* Be conservative and choose 16 in case anyone is using SSE. */ -static const size_t maxalign = 16; - -static size_t align_up(size_t size) { - return ((size + maxalign - 1) / maxalign) * maxalign; -} - -UPB_FORCEINLINE static void *seeded_alloc(void *ud, void *ptr, size_t oldsize, - size_t size) { - upb_seededalloc *a = ud; - - size = align_up(size); - - assert(a->mem_limit >= a->mem_ptr); - - if (oldsize == 0 && size <= (size_t)(a->mem_limit - a->mem_ptr)) { - /* Fast path: we can satisfy from the initial allocation. */ - void *ret = a->mem_ptr; - a->mem_ptr += size; - return ret; - } else { - char *chptr = ptr; - /* Slow path: fallback to other allocator. */ - a->need_cleanup = true; - /* Is `ptr` part of the user-provided initial block? Don't pass it to the - * default allocator if so; otherwise, it may try to realloc() the block. */ - if (chptr >= a->mem_base && chptr < a->mem_limit) { - void *ret; - assert(chptr + oldsize <= a->mem_limit); - ret = a->alloc(a->alloc_ud, NULL, 0, size); - if (ret) memcpy(ret, ptr, oldsize); - return ret; - } else { - return a->alloc(a->alloc_ud, ptr, oldsize, size); - } - } -} - -void upb_seededalloc_init(upb_seededalloc *a, void *mem, size_t len) { - default_alloc_ud *ud = (default_alloc_ud*)&a->default_alloc_ud; - a->mem_base = mem; - a->mem_ptr = mem; - a->mem_limit = (char*)mem + len; - a->need_cleanup = false; - a->returned_allocfunc = false; - - ud->head = NULL; - - upb_seededalloc_setfallbackalloc(a, default_alloc, ud); -} - -void upb_seededalloc_uninit(upb_seededalloc *a) { - if (a->alloc == default_alloc && a->need_cleanup) { - default_alloc_cleanup(a->alloc_ud); - } -} - -UPB_FORCEINLINE void upb_seededalloc_setfallbackalloc(upb_seededalloc *a, - upb_alloc_func *alloc, - void *ud) { - assert(!a->returned_allocfunc); - a->alloc = alloc; - a->alloc_ud = ud; -} - -upb_alloc_func *upb_seededalloc_getallocfunc(upb_seededalloc *a) { - a->returned_allocfunc = true; - return seeded_alloc; -} /* ** TODO(haberman): it's unclear whether a lot of the consistency checks should ** assert() or return false. */ -#include #include +static void *upb_calloc(size_t size) { + void *mem = upb_gmalloc(size); + if (mem) { + memset(mem, 0, size); + } + return mem; +} /* Defined for the sole purpose of having a unique pointer value for * UPB_NO_CLOSURE. */ @@ -2268,8 +2143,8 @@ static void freehandlers(upb_refcounted *r) { upb_inttable_uninit(&h->cleanup_); upb_msgdef_unref(h->msg, h); - free(h->sub); - free(h); + upb_gfree(h->sub); + upb_gfree(h); } static void visithandlers(const upb_refcounted *r, upb_refcounted_visit *visit, @@ -2519,14 +2394,20 @@ upb_handlers *upb_handlers_new(const upb_msgdef *md, const void *owner) { assert(upb_msgdef_isfrozen(md)); extra = sizeof(upb_handlers_tabent) * (md->selector_count - 1); - h = calloc(sizeof(*h) + extra, 1); + h = upb_calloc(sizeof(*h) + extra); if (!h) return NULL; h->msg = md; upb_msgdef_ref(h->msg, h); upb_status_clear(&h->status_); - h->sub = calloc(md->submsg_field_count, sizeof(*h->sub)); - if (!h->sub) goto oom; + + if (md->submsg_field_count > 0) { + h->sub = upb_calloc(md->submsg_field_count * sizeof(*h->sub)); + if (!h->sub) goto oom; + } else { + h->sub = 0; + } + if (!upb_refcounted_init(upb_handlers_upcast_mutable(h), &vtbl, owner)) goto oom; if (!upb_inttable_init(&h->cleanup_, UPB_CTYPE_FPTR)) goto oom; @@ -2941,7 +2822,6 @@ bool upb_byteshandler_setendstr(upb_byteshandler *h, #include -#include static void freeobj(upb_refcounted *o); @@ -3012,13 +2892,36 @@ static void upb_unlock() {} void upb_lock(); void upb_unlock(); -#endif +#endif + +/* UPB_DEBUG_REFS mode counts on being able to malloc() memory in some + * code-paths that can normally never fail, like upb_refcounted_ref(). Since + * we have no way to propagage out-of-memory errors back to the user, and since + * these errors can only occur in UPB_DEBUG_REFS mode, we use an allocator that + * immediately aborts on failure (avoiding the global allocator, which might + * inject failures). */ + +#include + +static void *upb_debugrefs_allocfunc(upb_alloc *alloc, void *ptr, + size_t oldsize, size_t size) { + UPB_UNUSED(alloc); + UPB_UNUSED(oldsize); + if (size == 0) { + free(ptr); + return NULL; + } else { + void *ret = realloc(ptr, size); + + if (!ret) { + abort(); + } -/* UPB_DEBUG_REFS mode counts on being able to malloc() memory in some - * code-paths that can normally never fail, like upb_refcounted_ref(). Since - * we have no way to propagage out-of-memory errors back to the user, and since - * these errors can only occur in UPB_DEBUG_REFS mode, we immediately fail. */ -#define CHECK_OOM(predicate) if (!(predicate)) { assert(predicate); exit(1); } + return ret; + } +} + +upb_alloc upb_alloc_debugrefs = {&upb_debugrefs_allocfunc}; typedef struct { int count; /* How many refs there are (duplicates only allowed for ref2). */ @@ -3026,8 +2929,7 @@ typedef struct { } trackedref; static trackedref *trackedref_new(bool is_ref2) { - trackedref *ret = malloc(sizeof(*ret)); - CHECK_OOM(ret); + trackedref *ret = upb_malloc(&upb_alloc_debugrefs, sizeof(*ret)); ret->count = 1; ret->is_ref2 = is_ref2; return ret; @@ -3052,15 +2954,15 @@ static void track(const upb_refcounted *r, const void *owner, bool ref2) { ref->count++; } else { trackedref *ref = trackedref_new(ref2); - bool ok = upb_inttable_insertptr(r->refs, owner, upb_value_ptr(ref)); - CHECK_OOM(ok); + upb_inttable_insertptr2(r->refs, owner, upb_value_ptr(ref), + &upb_alloc_debugrefs); if (ref2) { /* We know this cast is safe when it is a ref2, because it's coming from * another refcounted object. */ const upb_refcounted *from = owner; assert(!upb_inttable_lookupptr(from->ref2s, r, NULL)); - ok = upb_inttable_insertptr(from->ref2s, r, upb_value_ptr(NULL)); - CHECK_OOM(ok); + upb_inttable_insertptr2(from->ref2s, r, upb_value_ptr(NULL), + &upb_alloc_debugrefs); } } upb_unlock(); @@ -3118,7 +3020,6 @@ static void getref2s(const upb_refcounted *owner, upb_inttable *tab) { upb_value v; upb_value count; trackedref *ref; - bool ok; bool found; upb_refcounted *to = (upb_refcounted*)upb_inttable_iter_key(&i); @@ -3129,8 +3030,7 @@ static void getref2s(const upb_refcounted *owner, upb_inttable *tab) { ref = upb_value_getptr(v); count = upb_value_int32(ref->count); - ok = upb_inttable_insertptr(tab, to, count); - CHECK_OOM(ok); + upb_inttable_insertptr2(tab, to, count, &upb_alloc_debugrefs); } upb_unlock(); } @@ -3156,21 +3056,19 @@ static void visit_check(const upb_refcounted *obj, const upb_refcounted *subobj, assert(removed); newcount = upb_value_getint32(v) - 1; if (newcount > 0) { - upb_inttable_insert(ref2, (uintptr_t)subobj, upb_value_int32(newcount)); + upb_inttable_insert2(ref2, (uintptr_t)subobj, upb_value_int32(newcount), + &upb_alloc_debugrefs); } } static void visit(const upb_refcounted *r, upb_refcounted_visit *v, void *closure) { - bool ok; - /* In DEBUG_REFS mode we know what existing ref2 refs there are, so we know * exactly the set of nodes that visit() should visit. So we verify visit()'s * correctness here. */ check_state state; state.obj = r; - ok = upb_inttable_init(&state.ref2, UPB_CTYPE_INT32); - CHECK_OOM(ok); + upb_inttable_init2(&state.ref2, UPB_CTYPE_INT32, &upb_alloc_debugrefs); getref2s(r, &state.ref2); /* This should visit any children in the ref2 table. */ @@ -3178,32 +3076,22 @@ static void visit(const upb_refcounted *r, upb_refcounted_visit *v, /* This assertion will fail if the visit() function missed any children. */ assert(upb_inttable_count(&state.ref2) == 0); - upb_inttable_uninit(&state.ref2); + upb_inttable_uninit2(&state.ref2, &upb_alloc_debugrefs); if (r->vtbl->visit) r->vtbl->visit(r, v, closure); } -static bool trackinit(upb_refcounted *r) { - r->refs = malloc(sizeof(*r->refs)); - r->ref2s = malloc(sizeof(*r->ref2s)); - if (!r->refs || !r->ref2s) goto err1; - - if (!upb_inttable_init(r->refs, UPB_CTYPE_PTR)) goto err1; - if (!upb_inttable_init(r->ref2s, UPB_CTYPE_PTR)) goto err2; - return true; - -err2: - upb_inttable_uninit(r->refs); -err1: - free(r->refs); - free(r->ref2s); - return false; +static void trackinit(upb_refcounted *r) { + r->refs = upb_malloc(&upb_alloc_debugrefs, sizeof(*r->refs)); + r->ref2s = upb_malloc(&upb_alloc_debugrefs, sizeof(*r->ref2s)); + upb_inttable_init2(r->refs, UPB_CTYPE_PTR, &upb_alloc_debugrefs); + upb_inttable_init2(r->ref2s, UPB_CTYPE_PTR, &upb_alloc_debugrefs); } static void trackfree(const upb_refcounted *r) { - upb_inttable_uninit(r->refs); - upb_inttable_uninit(r->ref2s); - free(r->refs); - free(r->ref2s); + upb_inttable_uninit2(r->refs, &upb_alloc_debugrefs); + upb_inttable_uninit2(r->ref2s, &upb_alloc_debugrefs); + upb_free(&upb_alloc_debugrefs, r->refs); + upb_free(&upb_alloc_debugrefs, r->ref2s); } #else @@ -3226,9 +3114,8 @@ static void checkref(const upb_refcounted *r, const void *owner, bool ref2) { UPB_UNUSED(ref2); } -static bool trackinit(upb_refcounted *r) { +static void trackinit(upb_refcounted *r) { UPB_UNUSED(r); - return true; } static void trackfree(const upb_refcounted *r) { @@ -3338,12 +3225,12 @@ static upb_refcounted *pop(tarjan *t) { } static void tarjan_newgroup(tarjan *t) { - uint32_t *group = malloc(sizeof(*group)); + uint32_t *group = upb_gmalloc(sizeof(*group)); if (!group) oom(t); /* Push group and empty group leader (we'll fill in leader later). */ if (!upb_inttable_push(&t->groups, upb_value_ptr(group)) || !upb_inttable_push(&t->groups, upb_value_ptr(NULL))) { - free(group); + upb_gfree(group); oom(t); } *group = 0; @@ -3518,7 +3405,7 @@ static bool freeze(upb_refcounted *const*roots, int n, upb_status *s, if (obj == move) { /* Removing the last object from a group. */ assert(*obj->group == obj->individual_count); - free(obj->group); + upb_gfree(obj->group); } else { obj->next = move->next; /* This may decrease to zero; we'll collect GRAY objects (if any) that @@ -3574,7 +3461,7 @@ static bool freeze(upb_refcounted *const*roots, int n, upb_status *s, /* We eagerly free() the group's count (since we can't easily determine * the group's remaining size it's the easiest way to ensure it gets * done). */ - free(obj->group); + upb_gfree(obj->group); /* Visit to release ref2's (done in a separate pass since release_ref2 * depends on o->group being unmodified so it can test merged()). */ @@ -3594,7 +3481,7 @@ err4: if (!ret) { upb_inttable_begin(&iter, &t.groups); for(; !upb_inttable_done(&iter); upb_inttable_next(&iter)) - free(upb_value_getptr(upb_inttable_iter_value(&iter))); + upb_gfree(upb_value_getptr(upb_inttable_iter_value(&iter))); } upb_inttable_uninit(&t.groups); err3: @@ -3618,7 +3505,7 @@ static void merge(upb_refcounted *r, upb_refcounted *from) { if (merged(r, from)) return; *r->group += *from->group; - free(from->group); + upb_gfree(from->group); base = from; /* Set all refcount pointers in the "from" chain to the merged refcount. @@ -3652,7 +3539,7 @@ static void unref(const upb_refcounted *r) { if (unrefgroup(r->group)) { const upb_refcounted *o; - free(r->group); + upb_gfree(r->group); /* In two passes, since release_ref2 needs a guarantee that any subobjs * are alive. */ @@ -3696,13 +3583,10 @@ bool upb_refcounted_init(upb_refcounted *r, r->vtbl = vtbl; r->individual_count = 0; r->is_frozen = false; - r->group = malloc(sizeof(*r->group)); + r->group = upb_gmalloc(sizeof(*r->group)); if (!r->group) return false; *r->group = 0; - if (!trackinit(r)) { - free(r->group); - return false; - } + trackinit(r); upb_refcounted_ref(r, owner); return true; } @@ -3771,8 +3655,6 @@ bool upb_refcounted_freeze(upb_refcounted *const*roots, int n, upb_status *s, } -#include - /* Fallback implementation if the shim is not specialized by the JIT. */ #define SHIM_WRITER(type, ctype) \ bool upb_shim_set ## type (void *c, const void *hd, ctype val) { \ @@ -3798,14 +3680,14 @@ bool upb_shim_set(upb_handlers *h, const upb_fielddef *f, size_t offset, upb_handlerattr attr = UPB_HANDLERATTR_INITIALIZER; bool ok; - upb_shim_data *d = malloc(sizeof(*d)); + upb_shim_data *d = upb_gmalloc(sizeof(*d)); if (!d) return false; d->offset = offset; d->hasbit = hasbit; upb_handlerattr_sethandlerdata(&attr, d); upb_handlerattr_setalwaysok(&attr, true); - upb_handlers_addcleanup(h, d, free); + upb_handlers_addcleanup(h, d, upb_gfree); #define TYPE(u, l) \ case UPB_TYPE_##u: \ @@ -3856,7 +3738,6 @@ const upb_shim_data *upb_shim_getdata(const upb_handlers *h, upb_selector_t s, } -#include #include static void upb_symtab_free(upb_refcounted *r) { @@ -3868,13 +3749,17 @@ static void upb_symtab_free(upb_refcounted *r) { upb_def_unref(def, s); } upb_strtable_uninit(&s->symtab); - free(s); + upb_gfree(s); } - upb_symtab *upb_symtab_new(const void *owner) { static const struct upb_refcounted_vtbl vtbl = {NULL, &upb_symtab_free}; - upb_symtab *s = malloc(sizeof(*s)); + + upb_symtab *s = upb_gmalloc(sizeof(*s)); + if (!s) { + return NULL; + } + upb_refcounted_init(upb_symtab_upcast_mutable(s), &vtbl, owner); upb_strtable_init(&s->symtab, UPB_CTYPE_PTR); return s; @@ -4061,6 +3946,10 @@ static bool symtab_add(upb_symtab *s, upb_def *const*defs, size_t n, upb_strtable addtab; upb_inttable seen; + if (n == 0 && !freeze_also) { + return true; + } + assert(!upb_symtab_isfrozen(s)); if (!upb_strtable_init(&addtab, UPB_CTYPE_PTR)) { upb_status_seterrmsg(status, "out of memory"); @@ -4209,7 +4098,7 @@ static bool symtab_add(upb_symtab *s, upb_def *const*defs, size_t n, add_objs_size++; } - add_defs = malloc(sizeof(void*) * add_objs_size); + add_defs = upb_gmalloc(sizeof(void*) * add_objs_size); if (add_defs == NULL) goto oom_err; upb_strtable_begin(&iter, &addtab); for (add_n = 0; !upb_strtable_done(&iter); upb_strtable_next(&iter)) { @@ -4254,7 +4143,7 @@ static bool symtab_add(upb_symtab *s, upb_def *const*defs, size_t n, success = upb_strtable_insert(&s->symtab, name, upb_value_ptr(def)); UPB_ASSERT_VAR(success, success == true); } - free(add_objs); + upb_gfree(add_defs); return true; oom_err: @@ -4275,7 +4164,7 @@ err: { } } upb_strtable_uninit(&addtab); - free(add_objs); + upb_gfree(add_defs); assert(!upb_ok(status)); return false; } @@ -4292,7 +4181,7 @@ bool upb_symtab_addfile(upb_symtab *s, upb_filedef *file, upb_status *status) { bool ret; n = upb_filedef_defcount(file); - defs = malloc(sizeof(*defs) * n); + defs = upb_gmalloc(sizeof(*defs) * n); if (defs == NULL) { upb_status_seterrmsg(status, "Out of memory"); @@ -4305,7 +4194,7 @@ bool upb_symtab_addfile(upb_symtab *s, upb_filedef *file, upb_status *status) { ret = symtab_add(s, defs, n, NULL, upb_filedef_upcast_mutable(file), status); - free(defs); + upb_gfree(defs); return ret; } @@ -4347,7 +4236,6 @@ const upb_def *upb_symtab_iter_def(const upb_symtab_iter *iter) { */ -#include #include #define UPB_MAXARRSIZE 16 /* 64k. */ @@ -4356,6 +4244,17 @@ const upb_def *upb_symtab_iter_def(const upb_symtab_iter *iter) { #define ARRAY_SIZE(x) \ ((sizeof(x)/sizeof(0[x])) / ((size_t)(!(sizeof(x) % sizeof(0[x]))))) +#ifdef NDEBUG +static void upb_check_alloc(upb_table *t, upb_alloc *a) { + UPB_UNUSED(t); + UPB_UNUSED(a); +} +#else +static void upb_check_alloc(upb_table *t, upb_alloc *a) { + assert(t->alloc == a); +} +#endif + static const double MAX_LOAD = 0.85; /* The minimum utilization of the array part of a mixed hash/array table. This @@ -4373,11 +4272,11 @@ int log2ceil(uint64_t v) { return UPB_MIN(UPB_MAXARRSIZE, ret); } -char *upb_strdup(const char *s) { - return upb_strdup2(s, strlen(s)); +char *upb_strdup(const char *s, upb_alloc *a) { + return upb_strdup2(s, strlen(s), a); } -char *upb_strdup2(const char *s, size_t len) { +char *upb_strdup2(const char *s, size_t len, upb_alloc *a) { size_t n; char *p; @@ -4386,7 +4285,7 @@ char *upb_strdup2(const char *s, size_t len) { /* Always null-terminate, even if binary data; but don't rely on the input to * have a null-terminating byte since it may be a raw binary buffer. */ n = len + 1; - p = malloc(n); + p = upb_malloc(a, n); if (p) { memcpy(p, s, len); p[len] = 0; @@ -4434,16 +4333,20 @@ static bool isfull(upb_table *t) { } } -static bool init(upb_table *t, upb_ctype_t ctype, uint8_t size_lg2) { +static bool init(upb_table *t, upb_ctype_t ctype, uint8_t size_lg2, + upb_alloc *a) { size_t bytes; t->count = 0; t->ctype = ctype; t->size_lg2 = size_lg2; t->mask = upb_table_size(t) ? upb_table_size(t) - 1 : 0; +#ifndef NDEBUG + t->alloc = a; +#endif bytes = upb_table_size(t) * sizeof(upb_tabent); if (bytes > 0) { - t->entries = malloc(bytes); + t->entries = upb_malloc(a, bytes); if (!t->entries) return false; memset(mutable_entries(t), 0, bytes); } else { @@ -4452,7 +4355,10 @@ static bool init(upb_table *t, upb_ctype_t ctype, uint8_t size_lg2) { return true; } -static void uninit(upb_table *t) { free(mutable_entries(t)); } +static void uninit(upb_table *t, upb_alloc *a) { + upb_check_alloc(t, a); + upb_free(a, mutable_entries(t)); +} static upb_tabent *emptyent(upb_table *t) { upb_tabent *e = mutable_entries(t) + upb_table_size(t); @@ -4605,8 +4511,8 @@ static size_t begin(const upb_table *t) { /* A simple "subclass" of upb_table that only adds a hash function for strings. */ -static upb_tabkey strcopy(lookupkey_t k2) { - char *str = malloc(k2.str.len + sizeof(uint32_t) + 1); +static upb_tabkey strcopy(lookupkey_t k2, upb_alloc *a) { + char *str = upb_malloc(a, k2.str.len + sizeof(uint32_t) + 1); if (str == NULL) return 0; memcpy(str, &k2.str.len, sizeof(uint32_t)); memcpy(str + sizeof(uint32_t), k2.str.str, k2.str.len + 1); @@ -4625,51 +4531,56 @@ static bool streql(upb_tabkey k1, lookupkey_t k2) { return len == k2.str.len && memcmp(str, k2.str.str, len) == 0; } -bool upb_strtable_init(upb_strtable *t, upb_ctype_t ctype) { - return init(&t->t, ctype, 2); +bool upb_strtable_init2(upb_strtable *t, upb_ctype_t ctype, upb_alloc *a) { + return init(&t->t, ctype, 2, a); } -void upb_strtable_uninit(upb_strtable *t) { +void upb_strtable_uninit2(upb_strtable *t, upb_alloc *a) { size_t i; for (i = 0; i < upb_table_size(&t->t); i++) - free((void*)t->t.entries[i].key); - uninit(&t->t); + upb_free(a, (void*)t->t.entries[i].key); + uninit(&t->t, a); } -bool upb_strtable_resize(upb_strtable *t, size_t size_lg2) { +bool upb_strtable_resize(upb_strtable *t, size_t size_lg2, upb_alloc *a) { upb_strtable new_table; upb_strtable_iter i; - if (!init(&new_table.t, t->t.ctype, size_lg2)) + upb_check_alloc(&t->t, a); + + if (!init(&new_table.t, t->t.ctype, size_lg2, a)) return false; upb_strtable_begin(&i, t); for ( ; !upb_strtable_done(&i); upb_strtable_next(&i)) { - upb_strtable_insert2( + upb_strtable_insert3( &new_table, upb_strtable_iter_key(&i), upb_strtable_iter_keylength(&i), - upb_strtable_iter_value(&i)); + upb_strtable_iter_value(&i), + a); } - upb_strtable_uninit(t); + upb_strtable_uninit2(t, a); *t = new_table; return true; } -bool upb_strtable_insert2(upb_strtable *t, const char *k, size_t len, - upb_value v) { +bool upb_strtable_insert3(upb_strtable *t, const char *k, size_t len, + upb_value v, upb_alloc *a) { lookupkey_t key; upb_tabkey tabkey; uint32_t hash; + upb_check_alloc(&t->t, a); + if (isfull(&t->t)) { /* Need to resize. New table of double the size, add old elements to it. */ - if (!upb_strtable_resize(t, t->t.size_lg2 + 1)) { + if (!upb_strtable_resize(t, t->t.size_lg2 + 1, a)) { return false; } } key = strkey2(k, len); - tabkey = strcopy(key); + tabkey = strcopy(key, a); if (tabkey == 0) return false; hash = MurmurHash2(key.str.str, key.str.len, 0); @@ -4683,12 +4594,12 @@ bool upb_strtable_lookup2(const upb_strtable *t, const char *key, size_t len, return lookup(&t->t, strkey2(key, len), v, hash, &streql); } -bool upb_strtable_remove2(upb_strtable *t, const char *key, size_t len, - upb_value *val) { +bool upb_strtable_remove3(upb_strtable *t, const char *key, size_t len, + upb_value *val, upb_alloc *alloc) { uint32_t hash = MurmurHash2(key, strlen(key), 0); upb_tabkey tabkey; if (rm(&t->t, strkey2(key, len), val, &tabkey, hash, &streql)) { - free((void*)tabkey); + upb_free(alloc, (void*)tabkey); return true; } else { return false; @@ -4715,12 +4626,12 @@ bool upb_strtable_done(const upb_strtable_iter *i) { upb_tabent_isempty(str_tabent(i)); } -const char *upb_strtable_iter_key(upb_strtable_iter *i) { +const char *upb_strtable_iter_key(const upb_strtable_iter *i) { assert(!upb_strtable_done(i)); return upb_tabstr(str_tabent(i)->key, NULL); } -size_t upb_strtable_iter_keylength(upb_strtable_iter *i) { +size_t upb_strtable_iter_keylength(const upb_strtable_iter *i) { uint32_t len; assert(!upb_strtable_done(i)); upb_tabstr(str_tabent(i)->key, &len); @@ -4795,18 +4706,18 @@ static void check(upb_inttable *t) { } bool upb_inttable_sizedinit(upb_inttable *t, upb_ctype_t ctype, - size_t asize, int hsize_lg2) { + size_t asize, int hsize_lg2, upb_alloc *a) { size_t array_bytes; - if (!init(&t->t, ctype, hsize_lg2)) return false; + if (!init(&t->t, ctype, hsize_lg2, a)) return false; /* Always make the array part at least 1 long, so that we know key 0 * won't be in the hash part, which simplifies things. */ t->array_size = UPB_MAX(1, asize); t->array_count = 0; array_bytes = t->array_size * sizeof(upb_value); - t->array = malloc(array_bytes); + t->array = upb_malloc(a, array_bytes); if (!t->array) { - uninit(&t->t); + uninit(&t->t, a); return false; } memset(mutable_array(t), 0xff, array_bytes); @@ -4814,22 +4725,23 @@ bool upb_inttable_sizedinit(upb_inttable *t, upb_ctype_t ctype, return true; } -bool upb_inttable_init(upb_inttable *t, upb_ctype_t ctype) { - return upb_inttable_sizedinit(t, ctype, 0, 4); +bool upb_inttable_init2(upb_inttable *t, upb_ctype_t ctype, upb_alloc *a) { + return upb_inttable_sizedinit(t, ctype, 0, 4, a); } -void upb_inttable_uninit(upb_inttable *t) { - uninit(&t->t); - free(mutable_array(t)); +void upb_inttable_uninit2(upb_inttable *t, upb_alloc *a) { + uninit(&t->t, a); + upb_free(a, mutable_array(t)); } -bool upb_inttable_insert(upb_inttable *t, uintptr_t key, upb_value val) { - /* XXX: Table can't store value (uint64_t)-1. Need to somehow statically - * guarantee that this is not necessary, or fix the limitation. */ +bool upb_inttable_insert2(upb_inttable *t, uintptr_t key, upb_value val, + upb_alloc *a) { upb_tabval tabval; tabval.val = val.val; UPB_UNUSED(tabval); - assert(upb_arrhas(tabval)); + assert(upb_arrhas(tabval)); /* This will reject (uint64_t)-1. Fix this. */ + + upb_check_alloc(&t->t, a); if (key < t->array_size) { assert(!upb_arrhas(t->array[key])); @@ -4840,8 +4752,11 @@ bool upb_inttable_insert(upb_inttable *t, uintptr_t key, upb_value val) { /* Need to resize the hash part, but we re-use the array part. */ size_t i; upb_table new_table; - if (!init(&new_table, t->t.ctype, t->t.size_lg2 + 1)) + + if (!init(&new_table, t->t.ctype, t->t.size_lg2 + 1, a)) { return false; + } + for (i = begin(&t->t); i < upb_table_size(&t->t); i = next(&t->t, i)) { const upb_tabent *e = &t->t.entries[i]; uint32_t hash; @@ -4854,7 +4769,7 @@ bool upb_inttable_insert(upb_inttable *t, uintptr_t key, upb_value val) { assert(t->t.count == new_table.count); - uninit(&t->t); + uninit(&t->t, a); t->t = new_table; } insert(&t->t, intkey(key), key, val, upb_inthash(key), &inthash, &inteql); @@ -4900,8 +4815,9 @@ bool upb_inttable_remove(upb_inttable *t, uintptr_t key, upb_value *val) { return success; } -bool upb_inttable_push(upb_inttable *t, upb_value val) { - return upb_inttable_insert(t, upb_inttable_count(t), val); +bool upb_inttable_push2(upb_inttable *t, upb_value val, upb_alloc *a) { + upb_check_alloc(&t->t, a); + return upb_inttable_insert2(t, upb_inttable_count(t), val, a); } upb_value upb_inttable_pop(upb_inttable *t) { @@ -4911,8 +4827,10 @@ upb_value upb_inttable_pop(upb_inttable *t) { return val; } -bool upb_inttable_insertptr(upb_inttable *t, const void *key, upb_value val) { - return upb_inttable_insert(t, (uintptr_t)key, val); +bool upb_inttable_insertptr2(upb_inttable *t, const void *key, upb_value val, + upb_alloc *a) { + upb_check_alloc(&t->t, a); + return upb_inttable_insert2(t, (uintptr_t)key, val, a); } bool upb_inttable_lookupptr(const upb_inttable *t, const void *key, @@ -4924,7 +4842,7 @@ bool upb_inttable_removeptr(upb_inttable *t, const void *key, upb_value *val) { return upb_inttable_remove(t, (uintptr_t)key, val); } -void upb_inttable_compact(upb_inttable *t) { +void upb_inttable_compact2(upb_inttable *t, upb_alloc *a) { /* A power-of-two histogram of the table keys. */ size_t counts[UPB_MAXARRSIZE + 1] = {0}; @@ -4936,6 +4854,8 @@ void upb_inttable_compact(upb_inttable *t) { int size_lg2; upb_inttable new_t; + upb_check_alloc(&t->t, a); + upb_inttable_begin(&i, t); for (; !upb_inttable_done(&i); upb_inttable_next(&i)) { uintptr_t key = upb_inttable_iter_key(&i); @@ -4968,16 +4888,16 @@ void upb_inttable_compact(upb_inttable *t) { size_t hash_size = hash_count ? (hash_count / MAX_LOAD) + 1 : 0; size_t hashsize_lg2 = log2ceil(hash_size); - upb_inttable_sizedinit(&new_t, t->t.ctype, arr_size, hashsize_lg2); + upb_inttable_sizedinit(&new_t, t->t.ctype, arr_size, hashsize_lg2, a); upb_inttable_begin(&i, t); for (; !upb_inttable_done(&i); upb_inttable_next(&i)) { uintptr_t k = upb_inttable_iter_key(&i); - upb_inttable_insert(&new_t, k, upb_inttable_iter_value(&i)); + upb_inttable_insert2(&new_t, k, upb_inttable_iter_value(&i), a); } assert(new_t.array_size == arr_size); assert(new_t.t.size_lg2 == hashsize_lg2); } - upb_inttable_uninit(t); + upb_inttable_uninit2(t, a); *t = new_t; } @@ -5253,6 +5173,19 @@ static void nullz(upb_status *status) { memcpy(status->msg + sizeof(status->msg) - len, ellipsis, len); } + +/* upb_upberr *****************************************************************/ + +upb_errorspace upb_upberr = {"upb error"}; + +void upb_upberr_setoom(upb_status *status) { + status->error_space_ = &upb_upberr; + upb_status_seterrmsg(status, "Out of memory"); +} + + +/* upb_status *****************************************************************/ + void upb_status_clear(upb_status *status) { if (!status) return; status->ok_ = true; @@ -5291,20 +5224,258 @@ void upb_status_vseterrf(upb_status *status, const char *fmt, va_list args) { nullz(status); } -void upb_status_seterrcode(upb_status *status, upb_errorspace *space, - int code) { - if (!status) return; - status->ok_ = false; - status->error_space_ = space; - status->code_ = code; - space->set_message(status, code); -} - void upb_status_copy(upb_status *to, const upb_status *from) { if (!to) return; *to = *from; } -/* This file was generated by upbc (the upb compiler). + + +/* upb_alloc ******************************************************************/ + +static void *upb_global_allocfunc(upb_alloc *alloc, void *ptr, size_t oldsize, + size_t size) { + UPB_UNUSED(alloc); + UPB_UNUSED(oldsize); + if (size == 0) { + free(ptr); + return NULL; + } else { + return realloc(ptr, size); + } +} + +upb_alloc upb_alloc_global = {&upb_global_allocfunc}; + + +/* upb_arena ******************************************************************/ + +/* Be conservative and choose 16 in case anyone is using SSE. */ +static const size_t maxalign = 16; + +static size_t align_up(size_t size) { + return ((size + maxalign - 1) / maxalign) * maxalign; +} + +typedef struct mem_block { + struct mem_block *next; + size_t size; + size_t used; + bool owned; + /* Data follows. */ +} mem_block; + +typedef struct cleanup_ent { + struct cleanup_ent *next; + upb_cleanup_func *cleanup; + void *ud; +} cleanup_ent; + +static void upb_arena_addblock(upb_arena *a, void *ptr, size_t size, + bool owned) { + mem_block *block = ptr; + + block->next = a->block_head; + block->size = size; + block->used = align_up(sizeof(mem_block)); + block->owned = owned; + + a->block_head = block; + + /* TODO(haberman): ASAN poison. */ +} + + +static mem_block *upb_arena_allocblock(upb_arena *a, size_t size) { + size_t block_size = UPB_MAX(size, a->next_block_size) + sizeof(mem_block); + mem_block *block = upb_malloc(a->block_alloc, block_size); + + if (!block) { + return NULL; + } + + upb_arena_addblock(a, block, block_size, true); + a->next_block_size = UPB_MIN(block_size * 2, a->max_block_size); + + return block; +} + +static void *upb_arena_doalloc(upb_alloc *alloc, void *ptr, size_t oldsize, + size_t size) { + upb_arena *a = (upb_arena*)alloc; /* upb_alloc is initial member. */ + mem_block *block = a->block_head; + void *ret; + + if (size == 0) { + return NULL; /* We are an arena, don't need individual frees. */ + } + + size = align_up(size); + + /* TODO(haberman): special-case if this is a realloc of the last alloc? */ + + if (!block || block->size - block->used < size) { + /* Slow path: have to allocate a new block. */ + block = upb_arena_allocblock(a, size); + + if (!block) { + return NULL; /* Out of memory. */ + } + } + + ret = (char*)block + block->used; + block->used += size; + + if (oldsize > 0) { + memcpy(ret, ptr, oldsize); /* Preserve existing data. */ + } + + /* TODO(haberman): ASAN unpoison. */ + + a->bytes_allocated += size; + return ret; +} + +/* Public Arena API ***********************************************************/ + +void upb_arena_init(upb_arena *a) { + a->alloc.func = &upb_arena_doalloc; + a->block_alloc = &upb_alloc_global; + a->bytes_allocated = 0; + a->next_block_size = 256; + a->max_block_size = 16384; + a->cleanup_head = NULL; + a->block_head = NULL; +} + +void upb_arena_init2(upb_arena *a, void *mem, size_t size, upb_alloc *alloc) { + upb_arena_init(a); + + if (size > sizeof(mem_block)) { + upb_arena_addblock(a, mem, size, false); + } + + if (alloc) { + a->block_alloc = alloc; + } +} + +void upb_arena_uninit(upb_arena *a) { + cleanup_ent *ent = a->cleanup_head; + mem_block *block = a->block_head; + + while (ent) { + ent->cleanup(ent->ud); + ent = ent->next; + } + + /* Must do this after running cleanup functions, because this will delete + * the memory we store our cleanup entries in! */ + while (block) { + mem_block *next = block->next; + + if (block->owned) { + upb_free(a->block_alloc, block); + } + + block = next; + } +} + +bool upb_arena_addcleanup(upb_arena *a, upb_cleanup_func *func, void *ud) { + cleanup_ent *ent = upb_malloc(&a->alloc, sizeof(cleanup_ent)); + if (!ent) { + return false; /* Out of memory. */ + } + + ent->cleanup = func; + ent->ud = ud; + ent->next = a->cleanup_head; + a->cleanup_head = ent; + + return true; +} + +size_t upb_arena_bytesallocated(const upb_arena *a) { + return a->bytes_allocated; +} + + +/* Standard error functions ***************************************************/ + +static bool default_err(void *ud, const upb_status *status) { + UPB_UNUSED(ud); + UPB_UNUSED(status); + return false; +} + +static bool write_err_to(void *ud, const upb_status *status) { + upb_status *copy_to = ud; + upb_status_copy(copy_to, status); + return false; +} + + +/* upb_env ********************************************************************/ + +void upb_env_initonly(upb_env *e) { + e->ok_ = true; + e->error_func_ = &default_err; + e->error_ud_ = NULL; +} + +void upb_env_init(upb_env *e) { + upb_arena_init(&e->arena_); + upb_env_initonly(e); +} + +void upb_env_init2(upb_env *e, void *mem, size_t n, upb_alloc *alloc) { + upb_arena_init2(&e->arena_, mem, n, alloc); + upb_env_initonly(e); +} + +void upb_env_uninit(upb_env *e) { + upb_arena_uninit(&e->arena_); +} + +void upb_env_seterrorfunc(upb_env *e, upb_error_func *func, void *ud) { + e->error_func_ = func; + e->error_ud_ = ud; +} + +void upb_env_reporterrorsto(upb_env *e, upb_status *s) { + e->error_func_ = &write_err_to; + e->error_ud_ = s; +} + +bool upb_env_reporterror(upb_env *e, const upb_status *status) { + e->ok_ = false; + return e->error_func_(e->error_ud_, status); +} + +void *upb_env_malloc(upb_env *e, size_t size) { + return upb_malloc(&e->arena_.alloc, size); +} + +void *upb_env_realloc(upb_env *e, void *ptr, size_t oldsize, size_t size) { + return upb_realloc(&e->arena_.alloc, ptr, oldsize, size); +} + +void upb_env_free(upb_env *e, void *ptr) { + upb_free(&e->arena_.alloc, ptr); +} + +bool upb_env_addcleanup(upb_env *e, upb_cleanup_func *func, void *ud) { + return upb_arena_addcleanup(&e->arena_, func, ud); +} + +size_t upb_env_bytesallocated(const upb_env *e) { + return upb_arena_bytesallocated(&e->arena_); +} +/* This file was generated by upbc (the upb compiler) from the input + * file: + * + * upb/descriptor/descriptor.proto + * * Do not edit -- your changes will be discarded when the file is * regenerated. */ @@ -5323,28 +5494,28 @@ static upb_inttable reftables[264]; #endif static const upb_msgdef msgs[22] = { - UPB_MSGDEF_INIT("google.protobuf.DescriptorProto", 40, 8, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[0], 11, 10), UPB_STRTABLE_INIT(10, 15, UPB_CTYPE_PTR, 4, &strentries[0]),&reftables[0], &reftables[1]), - UPB_MSGDEF_INIT("google.protobuf.DescriptorProto.ExtensionRange", 4, 0, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[11], 3, 2), UPB_STRTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &strentries[16]),&reftables[2], &reftables[3]), - UPB_MSGDEF_INIT("google.protobuf.DescriptorProto.ReservedRange", 4, 0, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[14], 3, 2), UPB_STRTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &strentries[20]),&reftables[4], &reftables[5]), - UPB_MSGDEF_INIT("google.protobuf.EnumDescriptorProto", 11, 2, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[17], 4, 3), UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_PTR, 2, &strentries[24]),&reftables[6], &reftables[7]), - UPB_MSGDEF_INIT("google.protobuf.EnumOptions", 8, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &intentries[0], &arrays[21], 4, 2), UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_PTR, 2, &strentries[28]),&reftables[8], &reftables[9]), - UPB_MSGDEF_INIT("google.protobuf.EnumValueDescriptorProto", 8, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[25], 4, 3), UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_PTR, 2, &strentries[32]),&reftables[10], &reftables[11]), - UPB_MSGDEF_INIT("google.protobuf.EnumValueOptions", 7, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &intentries[2], &arrays[29], 2, 1), UPB_STRTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &strentries[36]),&reftables[12], &reftables[13]), - UPB_MSGDEF_INIT("google.protobuf.FieldDescriptorProto", 23, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[31], 11, 10), UPB_STRTABLE_INIT(10, 15, UPB_CTYPE_PTR, 4, &strentries[40]),&reftables[14], &reftables[15]), - UPB_MSGDEF_INIT("google.protobuf.FieldOptions", 12, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &intentries[4], &arrays[42], 11, 6), UPB_STRTABLE_INIT(7, 15, UPB_CTYPE_PTR, 4, &strentries[56]),&reftables[16], &reftables[17]), - UPB_MSGDEF_INIT("google.protobuf.FileDescriptorProto", 42, 6, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[53], 13, 12), UPB_STRTABLE_INIT(12, 15, UPB_CTYPE_PTR, 4, &strentries[72]),&reftables[18], &reftables[19]), - UPB_MSGDEF_INIT("google.protobuf.FileDescriptorSet", 6, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[66], 2, 1), UPB_STRTABLE_INIT(1, 3, UPB_CTYPE_PTR, 2, &strentries[88]),&reftables[20], &reftables[21]), - UPB_MSGDEF_INIT("google.protobuf.FileOptions", 31, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &intentries[6], &arrays[68], 39, 15), UPB_STRTABLE_INIT(16, 31, UPB_CTYPE_PTR, 5, &strentries[92]),&reftables[22], &reftables[23]), - UPB_MSGDEF_INIT("google.protobuf.MessageOptions", 10, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &intentries[8], &arrays[107], 8, 4), UPB_STRTABLE_INIT(5, 7, UPB_CTYPE_PTR, 3, &strentries[124]),&reftables[24], &reftables[25]), - UPB_MSGDEF_INIT("google.protobuf.MethodDescriptorProto", 15, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[115], 7, 6), UPB_STRTABLE_INIT(6, 7, UPB_CTYPE_PTR, 3, &strentries[132]),&reftables[26], &reftables[27]), - UPB_MSGDEF_INIT("google.protobuf.MethodOptions", 7, 1, UPB_INTTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &intentries[10], &arrays[122], 1, 0), UPB_STRTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &strentries[140]),&reftables[28], &reftables[29]), - UPB_MSGDEF_INIT("google.protobuf.OneofDescriptorProto", 5, 0, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[123], 2, 1), UPB_STRTABLE_INIT(1, 3, UPB_CTYPE_PTR, 2, &strentries[144]),&reftables[30], &reftables[31]), - UPB_MSGDEF_INIT("google.protobuf.ServiceDescriptorProto", 11, 2, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[125], 4, 3), UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_PTR, 2, &strentries[148]),&reftables[32], &reftables[33]), - UPB_MSGDEF_INIT("google.protobuf.ServiceOptions", 7, 1, UPB_INTTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &intentries[14], &arrays[129], 1, 0), UPB_STRTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &strentries[152]),&reftables[34], &reftables[35]), - UPB_MSGDEF_INIT("google.protobuf.SourceCodeInfo", 6, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[130], 2, 1), UPB_STRTABLE_INIT(1, 3, UPB_CTYPE_PTR, 2, &strentries[156]),&reftables[36], &reftables[37]), - UPB_MSGDEF_INIT("google.protobuf.SourceCodeInfo.Location", 19, 0, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[132], 7, 5), UPB_STRTABLE_INIT(5, 7, UPB_CTYPE_PTR, 3, &strentries[160]),&reftables[38], &reftables[39]), - UPB_MSGDEF_INIT("google.protobuf.UninterpretedOption", 18, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[139], 9, 7), UPB_STRTABLE_INIT(7, 15, UPB_CTYPE_PTR, 4, &strentries[168]),&reftables[40], &reftables[41]), - UPB_MSGDEF_INIT("google.protobuf.UninterpretedOption.NamePart", 6, 0, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[148], 3, 2), UPB_STRTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &strentries[184]),&reftables[42], &reftables[43]), + UPB_MSGDEF_INIT("google.protobuf.DescriptorProto", 40, 8, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[0], 11, 10), UPB_STRTABLE_INIT(10, 15, UPB_CTYPE_PTR, 4, &strentries[0]), false, UPB_SYNTAX_PROTO2, &reftables[0], &reftables[1]), + UPB_MSGDEF_INIT("google.protobuf.DescriptorProto.ExtensionRange", 4, 0, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[11], 3, 2), UPB_STRTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &strentries[16]), false, UPB_SYNTAX_PROTO2, &reftables[2], &reftables[3]), + UPB_MSGDEF_INIT("google.protobuf.DescriptorProto.ReservedRange", 4, 0, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[14], 3, 2), UPB_STRTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &strentries[20]), false, UPB_SYNTAX_PROTO2, &reftables[4], &reftables[5]), + UPB_MSGDEF_INIT("google.protobuf.EnumDescriptorProto", 11, 2, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[17], 4, 3), UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_PTR, 2, &strentries[24]), false, UPB_SYNTAX_PROTO2, &reftables[6], &reftables[7]), + UPB_MSGDEF_INIT("google.protobuf.EnumOptions", 8, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &intentries[0], &arrays[21], 4, 2), UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_PTR, 2, &strentries[28]), false, UPB_SYNTAX_PROTO2, &reftables[8], &reftables[9]), + UPB_MSGDEF_INIT("google.protobuf.EnumValueDescriptorProto", 8, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[25], 4, 3), UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_PTR, 2, &strentries[32]), false, UPB_SYNTAX_PROTO2, &reftables[10], &reftables[11]), + UPB_MSGDEF_INIT("google.protobuf.EnumValueOptions", 7, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &intentries[2], &arrays[29], 2, 1), UPB_STRTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &strentries[36]), false, UPB_SYNTAX_PROTO2, &reftables[12], &reftables[13]), + UPB_MSGDEF_INIT("google.protobuf.FieldDescriptorProto", 23, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[31], 11, 10), UPB_STRTABLE_INIT(10, 15, UPB_CTYPE_PTR, 4, &strentries[40]), false, UPB_SYNTAX_PROTO2, &reftables[14], &reftables[15]), + UPB_MSGDEF_INIT("google.protobuf.FieldOptions", 12, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &intentries[4], &arrays[42], 11, 6), UPB_STRTABLE_INIT(7, 15, UPB_CTYPE_PTR, 4, &strentries[56]), false, UPB_SYNTAX_PROTO2, &reftables[16], &reftables[17]), + UPB_MSGDEF_INIT("google.protobuf.FileDescriptorProto", 42, 6, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[53], 13, 12), UPB_STRTABLE_INIT(12, 15, UPB_CTYPE_PTR, 4, &strentries[72]), false, UPB_SYNTAX_PROTO2, &reftables[18], &reftables[19]), + UPB_MSGDEF_INIT("google.protobuf.FileDescriptorSet", 6, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[66], 2, 1), UPB_STRTABLE_INIT(1, 3, UPB_CTYPE_PTR, 2, &strentries[88]), false, UPB_SYNTAX_PROTO2, &reftables[20], &reftables[21]), + UPB_MSGDEF_INIT("google.protobuf.FileOptions", 31, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &intentries[6], &arrays[68], 39, 15), UPB_STRTABLE_INIT(16, 31, UPB_CTYPE_PTR, 5, &strentries[92]), false, UPB_SYNTAX_PROTO2, &reftables[22], &reftables[23]), + UPB_MSGDEF_INIT("google.protobuf.MessageOptions", 10, 1, UPB_INTTABLE_INIT(1, 1, UPB_CTYPE_PTR, 1, &intentries[8], &arrays[107], 8, 4), UPB_STRTABLE_INIT(5, 7, UPB_CTYPE_PTR, 3, &strentries[124]), false, UPB_SYNTAX_PROTO2, &reftables[24], &reftables[25]), + UPB_MSGDEF_INIT("google.protobuf.MethodDescriptorProto", 15, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[115], 7, 6), UPB_STRTABLE_INIT(6, 7, UPB_CTYPE_PTR, 3, &strentries[132]), false, UPB_SYNTAX_PROTO2, &reftables[26], &reftables[27]), + UPB_MSGDEF_INIT("google.protobuf.MethodOptions", 7, 1, UPB_INTTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &intentries[10], &arrays[122], 1, 0), UPB_STRTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &strentries[140]), false, UPB_SYNTAX_PROTO2, &reftables[28], &reftables[29]), + UPB_MSGDEF_INIT("google.protobuf.OneofDescriptorProto", 5, 0, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[123], 2, 1), UPB_STRTABLE_INIT(1, 3, UPB_CTYPE_PTR, 2, &strentries[144]), false, UPB_SYNTAX_PROTO2, &reftables[30], &reftables[31]), + UPB_MSGDEF_INIT("google.protobuf.ServiceDescriptorProto", 11, 2, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[125], 4, 3), UPB_STRTABLE_INIT(3, 3, UPB_CTYPE_PTR, 2, &strentries[148]), false, UPB_SYNTAX_PROTO2, &reftables[32], &reftables[33]), + UPB_MSGDEF_INIT("google.protobuf.ServiceOptions", 7, 1, UPB_INTTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &intentries[14], &arrays[129], 1, 0), UPB_STRTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &strentries[152]), false, UPB_SYNTAX_PROTO2, &reftables[34], &reftables[35]), + UPB_MSGDEF_INIT("google.protobuf.SourceCodeInfo", 6, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[130], 2, 1), UPB_STRTABLE_INIT(1, 3, UPB_CTYPE_PTR, 2, &strentries[156]), false, UPB_SYNTAX_PROTO2, &reftables[36], &reftables[37]), + UPB_MSGDEF_INIT("google.protobuf.SourceCodeInfo.Location", 19, 0, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[132], 7, 5), UPB_STRTABLE_INIT(5, 7, UPB_CTYPE_PTR, 3, &strentries[160]), false, UPB_SYNTAX_PROTO2, &reftables[38], &reftables[39]), + UPB_MSGDEF_INIT("google.protobuf.UninterpretedOption", 18, 1, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[139], 9, 7), UPB_STRTABLE_INIT(7, 15, UPB_CTYPE_PTR, 4, &strentries[168]), false, UPB_SYNTAX_PROTO2, &reftables[40], &reftables[41]), + UPB_MSGDEF_INIT("google.protobuf.UninterpretedOption.NamePart", 6, 0, UPB_INTTABLE_INIT(0, 0, UPB_CTYPE_PTR, 0, NULL, &arrays[148], 3, 2), UPB_STRTABLE_INIT(2, 3, UPB_CTYPE_PTR, 2, &strentries[184]), false, UPB_SYNTAX_PROTO2, &reftables[42], &reftables[43]), }; static const upb_fielddef fields[105] = { @@ -6277,7 +6448,7 @@ struct upb_descreader { }; static char *upb_strndup(const char *buf, size_t n) { - char *ret = malloc(n + 1); + char *ret = upb_gmalloc(n + 1); if (!ret) return NULL; memcpy(ret, buf, n); ret[n] = '\0'; @@ -6291,9 +6462,12 @@ static char *upb_strndup(const char *buf, size_t n) { * Caller owns a ref on the returned string. */ static char *upb_join(const char *base, const char *name) { if (!base || strlen(base) == 0) { - return upb_strdup(name); + return upb_gstrdup(name); } else { - char *ret = malloc(strlen(base) + strlen(name) + 2); + char *ret = upb_gmalloc(strlen(base) + strlen(name) + 2); + if (!ret) { + return NULL; + } ret[0] = '\0'; strcat(ret, base); strcat(ret, "."); @@ -6303,14 +6477,20 @@ static char *upb_join(const char *base, const char *name) { } /* Qualify the defname for all defs starting with offset "start" with "str". */ -static void upb_descreader_qualify(upb_filedef *f, char *str, int32_t start) { +static bool upb_descreader_qualify(upb_filedef *f, char *str, int32_t start) { size_t i; for (i = start; i < upb_filedef_defcount(f); i++) { upb_def *def = upb_filedef_mutabledef(f, i); char *name = upb_join(str, upb_def_fullname(def)); + if (!name) { + /* Need better logic here; at this point we've qualified some names but + * not others. */ + return false; + } upb_def_setfullname(def, name, NULL); - free(name); + upb_gfree(name); } + return true; } @@ -6336,16 +6516,19 @@ void upb_descreader_startcontainer(upb_descreader *r) { f->name = NULL; } -void upb_descreader_endcontainer(upb_descreader *r) { +bool upb_descreader_endcontainer(upb_descreader *r) { upb_descreader_frame *f = &r->stack[--r->stack_len]; - upb_descreader_qualify(r->file, f->name, f->start); - free(f->name); + if (!upb_descreader_qualify(r->file, f->name, f->start)) { + return false; + } + upb_gfree(f->name); f->name = NULL; + return true; } void upb_descreader_setscopename(upb_descreader *r, char *str) { upb_descreader_frame *f = &r->stack[r->stack_len-1]; - free(f->name); + upb_gfree(f->name); f->name = str; } @@ -6372,8 +6555,7 @@ static bool file_end(void *closure, const void *hd, upb_status *status) { upb_descreader *r = closure; UPB_UNUSED(hd); UPB_UNUSED(status); - upb_descreader_endcontainer(r); - return true; + return upb_descreader_endcontainer(r); } static size_t file_onname(void *closure, const void *hd, const char *buf, @@ -6387,6 +6569,7 @@ static size_t file_onname(void *closure, const void *hd, const char *buf, name = upb_strndup(buf, n); /* XXX: see comment at the top of the file. */ ok = upb_filedef_setname(r->file, name, NULL); + upb_gfree(name); UPB_ASSERT_VAR(ok, ok); return n; } @@ -6470,7 +6653,7 @@ static size_t enumval_onname(void *closure, const void *hd, const char *buf, UPB_UNUSED(hd); UPB_UNUSED(handle); /* XXX: see comment at the top of the file. */ - free(r->name); + upb_gfree(r->name); r->name = upb_strndup(buf, n); r->saw_name = true; return n; @@ -6495,7 +6678,7 @@ static bool enumval_endmsg(void *closure, const void *hd, upb_status *status) { } e = upb_downcast_enumdef_mutable(upb_descreader_last(r)); upb_enumdef_addval(e, r->name, r->number, status); - free(r->name); + upb_gfree(r->name); r->name = NULL; return true; } @@ -6527,7 +6710,7 @@ static size_t enum_onname(void *closure, const void *hd, const char *buf, UPB_UNUSED(handle); /* XXX: see comment at the top of the file. */ upb_def_setfullname(upb_descreader_last(r), fullname, NULL); - free(fullname); + upb_gfree(fullname); return n; } @@ -6537,7 +6720,7 @@ static bool field_startmsg(void *closure, const void *hd) { upb_descreader *r = closure; UPB_UNUSED(hd); assert(r->f); - free(r->default_string); + upb_gfree(r->default_string); r->default_string = NULL; /* fielddefs default to packed, but descriptors default to non-packed. */ @@ -6696,7 +6879,7 @@ static size_t field_onname(void *closure, const void *hd, const char *buf, /* XXX: see comment at the top of the file. */ upb_fielddef_setname(r->f, name, NULL); - free(name); + upb_gfree(name); return n; } @@ -6709,7 +6892,7 @@ static size_t field_ontypename(void *closure, const void *hd, const char *buf, /* XXX: see comment at the top of the file. */ upb_fielddef_setsubdefname(r->f, name, NULL); - free(name); + upb_gfree(name); return n; } @@ -6722,7 +6905,7 @@ static size_t field_onextendee(void *closure, const void *hd, const char *buf, /* XXX: see comment at the top of the file. */ upb_fielddef_setcontainingtypename(r->f, name, NULL); - free(name); + upb_gfree(name); return n; } @@ -6735,7 +6918,7 @@ static size_t field_ondefaultval(void *closure, const void *hd, const char *buf, /* Have to convert from string to the correct type, but we might not know the * type yet, so we save it as a string until the end of the field. * XXX: see comment at the top of the file. */ - free(r->default_string); + upb_gfree(r->default_string); r->default_string = upb_strndup(buf, n); return n; } @@ -6759,8 +6942,7 @@ static bool msg_end(void *closure, const void *hd, upb_status *status) { upb_status_seterrmsg(status, "Encountered message with no name."); return false; } - upb_descreader_endcontainer(r); - return true; + return upb_descreader_endcontainer(r); } static size_t msg_name(void *closure, const void *hd, const char *buf, @@ -6814,6 +6996,17 @@ static bool msg_endfield(void *closure, const void *hd) { return true; } +static bool msg_onmapentry(void *closure, const void *hd, bool mapentry) { + upb_descreader *r = closure; + upb_msgdef *m = upb_descreader_top(r); + UPB_UNUSED(hd); + + upb_msgdef_setmapentry(m, mapentry); + r->f = NULL; + return true; +} + + /** Code to register handlers *************************************************/ @@ -6884,6 +7077,8 @@ static void reghandlers(const void *closure, upb_handlers *h) { } else if (upbdefs_google_protobuf_FieldOptions_is(m)) { upb_handlers_setbool(h, F(FieldOptions, lazy), &field_onlazy, NULL); upb_handlers_setbool(h, F(FieldOptions, packed), &field_onpacked, NULL); + } else if (upbdefs_google_protobuf_MessageOptions_is(m)) { + upb_handlers_setbool(h, F(MessageOptions, map_entry), &msg_onmapentry, NULL); } assert(upb_ok(upb_handlers_status(h))); @@ -6899,12 +7094,12 @@ void descreader_cleanup(void *_r) { upb_filedef_unref(upb_descreader_file(r, i), &r->files); } - free(r->name); + upb_gfree(r->name); upb_inttable_uninit(&r->files); - free(r->default_string); + upb_gfree(r->default_string); while (r->stack_len > 0) { upb_descreader_frame *f = &r->stack[--r->stack_len]; - free(f->name); + upb_gfree(f->name); } } @@ -6980,8 +7175,8 @@ static void freegroup(upb_refcounted *r) { #ifdef UPB_USE_JIT_X64 upb_pbdecoder_freejit(g); #endif - free(g->bytecode); - free(g); + upb_gfree(g->bytecode); + upb_gfree(g); } static void visitgroup(const upb_refcounted *r, upb_refcounted_visit *visit, @@ -6996,7 +7191,7 @@ static void visitgroup(const upb_refcounted *r, upb_refcounted_visit *visit, } mgroup *newgroup(const void *owner) { - mgroup *g = malloc(sizeof(*g)); + mgroup *g = upb_gmalloc(sizeof(*g)); static const struct upb_refcounted_vtbl vtbl = {visitgroup, freegroup}; upb_refcounted_init(mgroup_upcast_mutable(g), &vtbl, owner); upb_inttable_init(&g->methods, UPB_CTYPE_PTR); @@ -7016,7 +7211,7 @@ static void freemethod(upb_refcounted *r) { } upb_inttable_uninit(&method->dispatch); - free(method); + upb_gfree(method); } static void visitmethod(const upb_refcounted *r, upb_refcounted_visit *visit, @@ -7028,7 +7223,7 @@ static void visitmethod(const upb_refcounted *r, upb_refcounted_visit *visit, static upb_pbdecodermethod *newmethod(const upb_handlers *dest_handlers, mgroup *group) { static const struct upb_refcounted_vtbl vtbl = {visitmethod, freemethod}; - upb_pbdecodermethod *ret = malloc(sizeof(*ret)); + upb_pbdecodermethod *ret = upb_gmalloc(sizeof(*ret)); upb_refcounted_init(upb_pbdecodermethod_upcast_mutable(ret), &vtbl, &ret); upb_byteshandler_init(&ret->input_handler_); @@ -7091,7 +7286,7 @@ typedef struct { } compiler; static compiler *newcompiler(mgroup *group, bool lazy) { - compiler *ret = malloc(sizeof(*ret)); + compiler *ret = upb_gmalloc(sizeof(*ret)); int i; ret->group = group; @@ -7104,7 +7299,7 @@ static compiler *newcompiler(mgroup *group, bool lazy) { } static void freecompiler(compiler *c) { - free(c); + upb_gfree(c); } const size_t ptr_words = sizeof(void*) / sizeof(uint32_t); @@ -7208,7 +7403,8 @@ static void put32(compiler *c, uint32_t v) { size_t oldsize = g->bytecode_end - g->bytecode; size_t newsize = UPB_MAX(oldsize * 2, 64); /* TODO(haberman): handle OOM. */ - g->bytecode = realloc(g->bytecode, newsize * sizeof(uint32_t)); + g->bytecode = upb_grealloc(g->bytecode, oldsize * sizeof(uint32_t), + newsize * sizeof(uint32_t)); g->bytecode_end = g->bytecode + newsize; c->pc = g->bytecode + ofs; } @@ -9073,7 +9269,6 @@ bool upb_pbdecoder_setmaxnesting(upb_pbdecoder *d, size_t max) { */ -#include /* The output buffer is divided into segments; a segment is a string of data * that is "ready to go" -- it does not need any varint lengths inserted into @@ -9318,12 +9513,12 @@ static void new_tag(upb_handlers *h, const upb_fielddef *f, upb_wiretype_t wt, upb_handlerattr *attr) { uint32_t n = upb_fielddef_number(f); - tag_t *tag = malloc(sizeof(tag_t)); + tag_t *tag = upb_gmalloc(sizeof(tag_t)); tag->bytes = upb_vencode64((n << 3) | wt, tag->tag); upb_handlerattr_init(attr); upb_handlerattr_sethandlerdata(attr, tag); - upb_handlers_addcleanup(h, tag, free); + upb_handlers_addcleanup(h, tag, upb_gfree); } static bool encode_tag(upb_pb_encoder *e, const tag_t *tag) { @@ -9582,9 +9777,6 @@ upb_pb_encoder *upb_pb_encoder_create(upb_env *env, const upb_handlers *h, upb_sink *upb_pb_encoder_input(upb_pb_encoder *e) { return &e->input_; } -#include -#include -#include upb_filedef **upb_loaddescriptor(const char *buf, size_t n, const void *owner, upb_status *status) { @@ -9615,7 +9807,7 @@ upb_filedef **upb_loaddescriptor(const char *buf, size_t n, const void *owner, goto cleanup; } - ret = malloc(sizeof (*ret) * (upb_descreader_filecount(reader) + 1)); + ret = upb_gmalloc(sizeof (*ret) * (upb_descreader_filecount(reader) + 1)); if (!ret) { goto cleanup; @@ -9647,7 +9839,6 @@ cleanup: #include #include #include -#include #include @@ -9743,14 +9934,14 @@ bool putf(upb_textprinter *p, const char *fmt, ...) { va_end(args_copy); /* + 1 for NULL terminator (vsprintf() requires it even if we don't). */ - str = malloc(len + 1); + str = upb_gmalloc(len + 1); if (!str) return false; written = vsprintf(str, fmt, args); va_end(args); UPB_ASSERT_VAR(written, written == len); ok = upb_bytessink_putbuf(p->output_, p->subc, str, len, NULL); - free(str); + upb_gfree(str); return ok; } @@ -10119,12 +10310,11 @@ upb_decoderet upb_vdecode_max8_wright(upb_decoderet r) { ** - handling of keys/escape-sequences/etc that span input buffers. */ -#include -#include #include -#include -#include #include +#include +#include +#include #define UPB_JSON_MAX_DEPTH 64 @@ -11247,11 +11437,11 @@ static void end_object(upb_json_parser *p) { * final state once, when the closing '"' is seen. */ -#line 1246 "upb/json/parser.rl" +#line 1245 "upb/json/parser.rl" -#line 1158 "upb/json/parser.c" +#line 1157 "upb/json/parser.c" static const char _json_actions[] = { 0, 1, 0, 1, 2, 1, 3, 1, 5, 1, 6, 1, 7, 1, 8, 1, @@ -11400,7 +11590,7 @@ static const int json_en_value_machine = 27; static const int json_en_main = 1; -#line 1249 "upb/json/parser.rl" +#line 1248 "upb/json/parser.rl" size_t parse(void *closure, const void *hd, const char *buf, size_t size, const upb_bufhandle *handle) { @@ -11422,7 +11612,7 @@ size_t parse(void *closure, const void *hd, const char *buf, size_t size, capture_resume(parser, buf); -#line 1329 "upb/json/parser.c" +#line 1328 "upb/json/parser.c" { int _klen; unsigned int _trans; @@ -11497,118 +11687,118 @@ _match: switch ( *_acts++ ) { case 0: -#line 1161 "upb/json/parser.rl" +#line 1160 "upb/json/parser.rl" { p--; {cs = stack[--top]; goto _again;} } break; case 1: -#line 1162 "upb/json/parser.rl" +#line 1161 "upb/json/parser.rl" { p--; {stack[top++] = cs; cs = 10; goto _again;} } break; case 2: -#line 1166 "upb/json/parser.rl" +#line 1165 "upb/json/parser.rl" { start_text(parser, p); } break; case 3: -#line 1167 "upb/json/parser.rl" +#line 1166 "upb/json/parser.rl" { CHECK_RETURN_TOP(end_text(parser, p)); } break; case 4: -#line 1173 "upb/json/parser.rl" +#line 1172 "upb/json/parser.rl" { start_hex(parser); } break; case 5: -#line 1174 "upb/json/parser.rl" +#line 1173 "upb/json/parser.rl" { hexdigit(parser, p); } break; case 6: -#line 1175 "upb/json/parser.rl" +#line 1174 "upb/json/parser.rl" { CHECK_RETURN_TOP(end_hex(parser)); } break; case 7: -#line 1181 "upb/json/parser.rl" +#line 1180 "upb/json/parser.rl" { CHECK_RETURN_TOP(escape(parser, p)); } break; case 8: -#line 1187 "upb/json/parser.rl" +#line 1186 "upb/json/parser.rl" { p--; {cs = stack[--top]; goto _again;} } break; case 9: -#line 1190 "upb/json/parser.rl" +#line 1189 "upb/json/parser.rl" { {stack[top++] = cs; cs = 19; goto _again;} } break; case 10: -#line 1192 "upb/json/parser.rl" +#line 1191 "upb/json/parser.rl" { p--; {stack[top++] = cs; cs = 27; goto _again;} } break; case 11: -#line 1197 "upb/json/parser.rl" +#line 1196 "upb/json/parser.rl" { start_member(parser); } break; case 12: -#line 1198 "upb/json/parser.rl" +#line 1197 "upb/json/parser.rl" { CHECK_RETURN_TOP(end_membername(parser)); } break; case 13: -#line 1201 "upb/json/parser.rl" +#line 1200 "upb/json/parser.rl" { end_member(parser); } break; case 14: -#line 1207 "upb/json/parser.rl" +#line 1206 "upb/json/parser.rl" { start_object(parser); } break; case 15: -#line 1210 "upb/json/parser.rl" +#line 1209 "upb/json/parser.rl" { end_object(parser); } break; case 16: -#line 1216 "upb/json/parser.rl" +#line 1215 "upb/json/parser.rl" { CHECK_RETURN_TOP(start_array(parser)); } break; case 17: -#line 1220 "upb/json/parser.rl" +#line 1219 "upb/json/parser.rl" { end_array(parser); } break; case 18: -#line 1225 "upb/json/parser.rl" +#line 1224 "upb/json/parser.rl" { start_number(parser, p); } break; case 19: -#line 1226 "upb/json/parser.rl" +#line 1225 "upb/json/parser.rl" { CHECK_RETURN_TOP(end_number(parser, p)); } break; case 20: -#line 1228 "upb/json/parser.rl" +#line 1227 "upb/json/parser.rl" { CHECK_RETURN_TOP(start_stringval(parser)); } break; case 21: -#line 1229 "upb/json/parser.rl" +#line 1228 "upb/json/parser.rl" { CHECK_RETURN_TOP(end_stringval(parser)); } break; case 22: -#line 1231 "upb/json/parser.rl" +#line 1230 "upb/json/parser.rl" { CHECK_RETURN_TOP(parser_putbool(parser, true)); } break; case 23: -#line 1233 "upb/json/parser.rl" +#line 1232 "upb/json/parser.rl" { CHECK_RETURN_TOP(parser_putbool(parser, false)); } break; case 24: -#line 1235 "upb/json/parser.rl" +#line 1234 "upb/json/parser.rl" { /* null value */ } break; case 25: -#line 1237 "upb/json/parser.rl" +#line 1236 "upb/json/parser.rl" { CHECK_RETURN_TOP(start_subobject(parser)); } break; case 26: -#line 1238 "upb/json/parser.rl" +#line 1237 "upb/json/parser.rl" { end_subobject(parser); } break; case 27: -#line 1243 "upb/json/parser.rl" +#line 1242 "upb/json/parser.rl" { p--; {cs = stack[--top]; goto _again;} } break; -#line 1515 "upb/json/parser.c" +#line 1514 "upb/json/parser.c" } } @@ -11621,7 +11811,7 @@ _again: _out: {} } -#line 1270 "upb/json/parser.rl" +#line 1269 "upb/json/parser.rl" if (p != pe) { upb_status_seterrf(&parser->status, "Parse error at '%.*s'\n", pe - p, p); @@ -11662,13 +11852,13 @@ static void json_parser_reset(upb_json_parser *p) { /* Emit Ragel initialization of the parser. */ -#line 1569 "upb/json/parser.c" +#line 1568 "upb/json/parser.c" { cs = json_start; top = 0; } -#line 1310 "upb/json/parser.rl" +#line 1309 "upb/json/parser.rl" p->current_state = cs; p->parser_top = top; accumulate_clear(p); @@ -11694,12 +11884,12 @@ static void free_json_parsermethod(upb_refcounted *r) { upb_value val = upb_inttable_iter_value(&i); upb_strtable *t = upb_value_getptr(val); upb_strtable_uninit(t); - free(t); + upb_gfree(t); } upb_inttable_uninit(&method->name_tables); - free(r); + upb_gfree(r); } static void add_jsonname_table(upb_json_parsermethod *m, const upb_msgdef* md) { @@ -11716,7 +11906,7 @@ static void add_jsonname_table(upb_json_parsermethod *m, const upb_msgdef* md) { } /* TODO(haberman): handle malloc failure. */ - t = malloc(sizeof(*t)); + t = upb_gmalloc(sizeof(*t)); upb_strtable_init(t, UPB_CTYPE_CONSTPTR); upb_inttable_insertptr(&m->name_tables, md, upb_value_ptr(t)); @@ -11729,7 +11919,7 @@ static void add_jsonname_table(upb_json_parsermethod *m, const upb_msgdef* md) { size_t field_len = upb_fielddef_getjsonname(f, buf, len); if (field_len > len) { size_t len2; - buf = realloc(buf, field_len); + buf = upb_grealloc(buf, 0, field_len); len = field_len; len2 = upb_fielddef_getjsonname(f, buf, len); UPB_ASSERT_VAR(len2, len == len2); @@ -11748,7 +11938,7 @@ static void add_jsonname_table(upb_json_parsermethod *m, const upb_msgdef* md) { } } - free(buf); + upb_gfree(buf); } /* Public API *****************************************************************/ @@ -11788,7 +11978,7 @@ upb_json_parsermethod *upb_json_parsermethod_new(const upb_msgdef* md, const void* owner) { static const struct upb_refcounted_vtbl vtbl = {visit_json_parsermethod, free_json_parsermethod}; - upb_json_parsermethod *ret = malloc(sizeof(*ret)); + upb_json_parsermethod *ret = upb_gmalloc(sizeof(*ret)); upb_refcounted_init(upb_json_parsermethod_upcast_mutable(ret), &vtbl, owner); ret->msg = md; @@ -11815,8 +12005,6 @@ const upb_byteshandler *upb_json_parsermethod_inputhandler( */ -#include -#include #include #include @@ -11849,22 +12037,22 @@ typedef struct { void freestrpc(void *ptr) { strpc *pc = ptr; - free(pc->ptr); - free(pc); + upb_gfree(pc->ptr); + upb_gfree(pc); } /* Convert fielddef name to JSON name and return as a string piece. */ strpc *newstrpc(upb_handlers *h, const upb_fielddef *f, bool preserve_fieldnames) { /* TODO(haberman): handle malloc failure. */ - strpc *ret = malloc(sizeof(*ret)); + strpc *ret = upb_gmalloc(sizeof(*ret)); if (preserve_fieldnames) { - ret->ptr = upb_strdup(upb_fielddef_name(f)); + ret->ptr = upb_gstrdup(upb_fielddef_name(f)); ret->len = strlen(ret->ptr); } else { size_t len; ret->len = upb_fielddef_getjsonname(f, NULL, 0); - ret->ptr = malloc(ret->len); + ret->ptr = upb_gmalloc(ret->len); len = upb_fielddef_getjsonname(f, ret->ptr, ret->len); UPB_ASSERT_VAR(len, len == ret->len); ret->len--; /* NULL */ @@ -12376,10 +12564,10 @@ static void set_enum_hd(upb_handlers *h, const upb_fielddef *f, bool preserve_fieldnames, upb_handlerattr *attr) { - EnumHandlerData *hd = malloc(sizeof(EnumHandlerData)); + EnumHandlerData *hd = upb_gmalloc(sizeof(EnumHandlerData)); hd->enumdef = (const upb_enumdef *)upb_fielddef_subdef(f); hd->keyname = newstrpc(h, f, preserve_fieldnames); - upb_handlers_addcleanup(h, hd, free); + upb_handlers_addcleanup(h, hd, upb_gfree); upb_handlerattr_sethandlerdata(attr, hd); } diff --git a/ruby/ext/google/protobuf_c/upb.h b/ruby/ext/google/protobuf_c/upb.h index b00581618e..2faf74e502 100644 --- a/ruby/ext/google/protobuf_c/upb.h +++ b/ruby/ext/google/protobuf_c/upb.h @@ -80,6 +80,18 @@ #include #include +#ifdef __cplusplus +namespace upb { +class Allocator; +class Arena; +class Environment; +class ErrorSpace; +class Status; +template class InlinedArena; +template class InlinedEnvironment; +} +#endif + /* UPB_INLINE: inline if possible, emit standalone code if required. */ #ifdef __cplusplus #define UPB_INLINE inline @@ -147,6 +159,7 @@ #define UPB_ASSERT_STDLAYOUT(type) \ static_assert(std::is_standard_layout::value, \ #type " must be standard layout"); +#define UPB_FINAL final #else /* !defined(UPB_CXX11) */ #define UPB_DISALLOW_COPY_AND_ASSIGN(class_name) \ class_name(const class_name&); \ @@ -156,6 +169,7 @@ ~class_name(); \ UPB_DISALLOW_COPY_AND_ASSIGN(class_name) #define UPB_ASSERT_STDLAYOUT(type) +#define UPB_FINAL #endif /* UPB_DECLARE_TYPE() @@ -257,6 +271,7 @@ /* Generic function type. */ typedef void upb_func(); + /* C++ Casts ******************************************************************/ #ifdef __cplusplus @@ -334,248 +349,430 @@ class PointerBase2 : public PointerBase { #endif -/* upb::reffed_ptr ************************************************************/ +/* upb::ErrorSpace ************************************************************/ + +/* A upb::ErrorSpace represents some domain of possible error values. This lets + * upb::Status attach specific error codes to operations, like POSIX/C errno, + * Win32 error codes, etc. Clients who want to know the very specific error + * code can check the error space and then know the type of the integer code. + * + * NOTE: upb::ErrorSpace is currently not used and should be considered + * experimental. It is important primarily in cases where upb is performing + * I/O, but upb doesn't currently have any components that do this. */ + +UPB_DECLARE_TYPE(upb::ErrorSpace, upb_errorspace) #ifdef __cplusplus +class upb::ErrorSpace { +#else +struct upb_errorspace { +#endif + const char *name; +}; -#include /* For std::swap(). */ -namespace upb { +/* upb::Status ****************************************************************/ -/* Provides RAII semantics for upb refcounted objects. Each reffed_ptr owns a - * ref on whatever object it points to (if any). */ -template class reffed_ptr { +/* upb::Status represents a success or failure status and error message. + * It owns no resources and allocates no memory, so it should work + * even in OOM situations. */ +UPB_DECLARE_TYPE(upb::Status, upb_status) + +/* The maximum length of an error message before it will get truncated. */ +#define UPB_STATUS_MAX_MESSAGE 128 + +UPB_BEGIN_EXTERN_C + +const char *upb_status_errmsg(const upb_status *status); +bool upb_ok(const upb_status *status); +upb_errorspace *upb_status_errspace(const upb_status *status); +int upb_status_errcode(const upb_status *status); + +/* Any of the functions that write to a status object allow status to be NULL, + * to support use cases where the function's caller does not care about the + * status message. */ +void upb_status_clear(upb_status *status); +void upb_status_seterrmsg(upb_status *status, const char *msg); +void upb_status_seterrf(upb_status *status, const char *fmt, ...); +void upb_status_vseterrf(upb_status *status, const char *fmt, va_list args); +void upb_status_copy(upb_status *to, const upb_status *from); + +UPB_END_EXTERN_C + +#ifdef __cplusplus + +class upb::Status { public: - reffed_ptr() : ptr_(NULL) {} + Status() { upb_status_clear(this); } - /* If ref_donor is NULL, takes a new ref, otherwise adopts from ref_donor. */ - template - reffed_ptr(U* val, const void* ref_donor = NULL) - : ptr_(upb::upcast(val)) { - if (ref_donor) { - assert(ptr_); - ptr_->DonateRef(ref_donor, this); - } else if (ptr_) { - ptr_->Ref(this); - } - } + /* Returns true if there is no error. */ + bool ok() const { return upb_ok(this); } - template - reffed_ptr(const reffed_ptr& other) - : ptr_(upb::upcast(other.get())) { - if (ptr_) ptr_->Ref(this); - } + /* Optional error space and code, useful if the caller wants to + * programmatically check the specific kind of error. */ + ErrorSpace* error_space() { return upb_status_errspace(this); } + int error_code() const { return upb_status_errcode(this); } - reffed_ptr(const reffed_ptr& other) - : ptr_(upb::upcast(other.get())) { - if (ptr_) ptr_->Ref(this); + /* The returned string is invalidated by any other call into the status. */ + const char *error_message() const { return upb_status_errmsg(this); } + + /* The error message will be truncated if it is longer than + * UPB_STATUS_MAX_MESSAGE-4. */ + void SetErrorMessage(const char* msg) { upb_status_seterrmsg(this, msg); } + void SetFormattedErrorMessage(const char* fmt, ...) { + va_list args; + va_start(args, fmt); + upb_status_vseterrf(this, fmt, args); + va_end(args); } - ~reffed_ptr() { if (ptr_) ptr_->Unref(this); } + /* Resets the status to a successful state with no message. */ + void Clear() { upb_status_clear(this); } - template - reffed_ptr& operator=(const reffed_ptr& other) { - reset(other.get()); - return *this; - } + void CopyFrom(const Status& other) { upb_status_copy(this, &other); } - reffed_ptr& operator=(const reffed_ptr& other) { - reset(other.get()); - return *this; - } + private: + UPB_DISALLOW_COPY_AND_ASSIGN(Status) +#else +struct upb_status { +#endif + bool ok_; - /* TODO(haberman): add C++11 move construction/assignment for greater - * efficiency. */ + /* Specific status code defined by some error space (optional). */ + int code_; + upb_errorspace *error_space_; - void swap(reffed_ptr& other) { - if (ptr_ == other.ptr_) { - return; - } + /* TODO(haberman): add file/line of error? */ - if (ptr_) ptr_->DonateRef(this, &other); - if (other.ptr_) other.ptr_->DonateRef(&other, this); - std::swap(ptr_, other.ptr_); - } + /* Error message; NULL-terminated. */ + char msg[UPB_STATUS_MAX_MESSAGE]; +}; - T& operator*() const { - assert(ptr_); - return *ptr_; - } +#define UPB_STATUS_INIT {true, 0, NULL, {0}} - T* operator->() const { - assert(ptr_); - return ptr_; - } - T* get() const { return ptr_; } +/** Built-in error spaces. ****************************************************/ - /* If ref_donor is NULL, takes a new ref, otherwise adopts from ref_donor. */ - template - void reset(U* ptr = NULL, const void* ref_donor = NULL) { - reffed_ptr(ptr, ref_donor).swap(*this); - } +/* Errors raised by upb that we want to be able to detect programmatically. */ +typedef enum { + UPB_NOMEM /* Can't reuse ENOMEM because it is POSIX, not ISO C. */ +} upb_errcode_t; - template - reffed_ptr down_cast() { - return reffed_ptr(upb::down_cast(get())); - } +extern upb_errorspace upb_upberr; - template - reffed_ptr dyn_cast() { - return reffed_ptr(upb::dyn_cast(get())); - } +void upb_upberr_setoom(upb_status *s); - /* Plain release() is unsafe; if we were the only owner, it would leak the - * object. Instead we provide this: */ - T* ReleaseTo(const void* new_owner) { - T* ret = NULL; - ptr_->DonateRef(this, new_owner); - std::swap(ret, ptr_); - return ret; - } +/* Since errno is defined by standard C, we define an error space for it in + * core upb. Other error spaces should be defined in other, platform-specific + * modules. */ + +extern upb_errorspace upb_errnoerr; + + +/** upb::Allocator ************************************************************/ + +/* A upb::Allocator is a possibly-stateful allocator object. + * + * It could either be an arena allocator (which doesn't require individual + * free() calls) or a regular malloc() (which does). The client must therefore + * free memory unless it knows that the allocator is an arena allocator. */ +UPB_DECLARE_TYPE(upb::Allocator, upb_alloc) + +/* A malloc()/free() function. + * If "size" is 0 then the function acts like free(), otherwise it acts like + * realloc(). Only "oldsize" bytes from a previous allocation are preserved. */ +typedef void *upb_alloc_func(upb_alloc *alloc, void *ptr, size_t oldsize, + size_t size); + +#ifdef __cplusplus + +class upb::Allocator UPB_FINAL { + public: + Allocator() {} private: - T* ptr_; + UPB_DISALLOW_COPY_AND_ASSIGN(Allocator) + + public: +#else +struct upb_alloc { +#endif /* __cplusplus */ + upb_alloc_func *func; }; -} /* namespace upb */ +UPB_INLINE void *upb_malloc(upb_alloc *alloc, size_t size) { + assert(size > 0); + return alloc->func(alloc, NULL, 0, size); +} -#endif /* __cplusplus */ +UPB_INLINE void *upb_realloc(upb_alloc *alloc, void *ptr, size_t oldsize, + size_t size) { + assert(size > 0); + return alloc->func(alloc, ptr, oldsize, size); +} +UPB_INLINE void upb_free(upb_alloc *alloc, void *ptr) { + alloc->func(alloc, ptr, 0, 0); +} -/* upb::Status ****************************************************************/ +/* The global allocator used by upb. Uses the standard malloc()/free(). */ -#ifdef __cplusplus -namespace upb { -class ErrorSpace; -class Status; +extern upb_alloc upb_alloc_global; + +/* Functions that hard-code the global malloc. + * + * We still get benefit because we can put custom logic into our global + * allocator, like injecting out-of-memory faults in debug/testing builds. */ + +UPB_INLINE void *upb_gmalloc(size_t size) { + return upb_malloc(&upb_alloc_global, size); } -#endif -UPB_DECLARE_TYPE(upb::ErrorSpace, upb_errorspace) -UPB_DECLARE_TYPE(upb::Status, upb_status) +UPB_INLINE void *upb_grealloc(void *ptr, size_t oldsize, size_t size) { + return upb_realloc(&upb_alloc_global, ptr, oldsize, size); +} -/* The maximum length of an error message before it will get truncated. */ -#define UPB_STATUS_MAX_MESSAGE 128 +UPB_INLINE void upb_gfree(void *ptr) { + upb_free(&upb_alloc_global, ptr); +} + +/* upb::Arena *****************************************************************/ + +/* upb::Arena is a specific allocator implementation that uses arena allocation. + * The user provides an allocator that will be used to allocate the underlying + * arena blocks. Arenas by nature do not require the individual allocations + * to be freed. However the Arena does allow users to register cleanup + * functions that will run when the arena is destroyed. + * + * A upb::Arena is *not* thread-safe. + * + * You could write a thread-safe arena allocator that satisfies the + * upb::Allocator interface, but it would not be as efficient for the + * single-threaded case. */ +UPB_DECLARE_TYPE(upb::Arena, upb_arena) + +typedef void upb_cleanup_func(void *ud); -/* An error callback function is used to report errors from some component. - * The function can return "true" to indicate that the component should try - * to recover and proceed, but this is not always possible. */ -typedef bool upb_errcb_t(void *closure, const upb_status* status); +#define UPB_ARENA_BLOCK_OVERHEAD (sizeof(size_t)*4) + +UPB_BEGIN_EXTERN_C + +void upb_arena_init(upb_arena *a); +void upb_arena_init2(upb_arena *a, void *mem, size_t n, upb_alloc *alloc); +void upb_arena_uninit(upb_arena *a); +upb_alloc *upb_arena_alloc(upb_arena *a); +bool upb_arena_addcleanup(upb_arena *a, upb_cleanup_func *func, void *ud); +size_t upb_arena_bytesallocated(const upb_arena *a); +void upb_arena_setnextblocksize(upb_arena *a, size_t size); +void upb_arena_setmaxblocksize(upb_arena *a, size_t size); + +UPB_END_EXTERN_C #ifdef __cplusplus -class upb::ErrorSpace { + +class upb::Arena { + public: + /* A simple arena with no initial memory block and the default allocator. */ + Arena() { upb_arena_init(this); } + + /* Constructs an arena with the given initial block which allocates blocks + * with the given allocator. The given allocator must outlive the Arena. + * + * If you pass NULL for the allocator it will default to the global allocator + * upb_alloc_global, and NULL/0 for the initial block will cause there to be + * no initial block. */ + Arena(void *mem, size_t len, Allocator* a) { + upb_arena_init2(this, mem, len, a); + } + + ~Arena() { upb_arena_uninit(this); } + + /* Sets the size of the next block the Arena will request (unless the + * requested allocation is larger). Each block will double in size until the + * max limit is reached. */ + void SetNextBlockSize(size_t size) { upb_arena_setnextblocksize(this, size); } + + /* Sets the maximum block size. No blocks larger than this will be requested + * from the underlying allocator unless individual arena allocations are + * larger. */ + void SetMaxBlockSize(size_t size) { upb_arena_setmaxblocksize(this, size); } + + /* Allows this arena to be used as a generic allocator. + * + * The arena does not need free() calls so when using Arena as an allocator + * it is safe to skip them. However they are no-ops so there is no harm in + * calling free() either. */ + Allocator* allocator() { return upb_arena_alloc(this); } + + /* Add a cleanup function to run when the arena is destroyed. + * Returns false on out-of-memory. */ + bool AddCleanup(upb_cleanup_func* func, void* ud) { + return upb_arena_addcleanup(this, func, ud); + } + + /* Total number of bytes that have been allocated. It is undefined what + * Realloc() does to this counter. */ + size_t BytesAllocated() const { + return upb_arena_bytesallocated(this); + } + + private: + UPB_DISALLOW_COPY_AND_ASSIGN(Arena) + #else -struct upb_errorspace { -#endif - const char *name; - /* Should the error message in the status object according to this code. */ - void (*set_message)(upb_status* status, int code); +struct upb_arena { +#endif /* __cplusplus */ + /* We implement the allocator interface. + * This must be the first member of upb_arena! */ + upb_alloc alloc; + + /* Allocator to allocate arena blocks. We are responsible for freeing these + * when we are destroyed. */ + upb_alloc *block_alloc; + + size_t bytes_allocated; + size_t next_block_size; + size_t max_block_size; + + /* Linked list of blocks. Points to an arena_block, defined in env.c */ + void *block_head; + + /* Cleanup entries. Pointer to a cleanup_ent, defined in env.c */ + void *cleanup_head; + + /* For future expansion, since the size of this struct is exposed to users. */ + void *future1; + void *future2; }; -#ifdef __cplusplus -/* Object representing a success or failure status. - * It owns no resources and allocates no memory, so it should work - * even in OOM situations. */ +/* upb::Environment ***********************************************************/ -class upb::Status { - public: - Status(); +/* A upb::Environment provides a means for injecting malloc and an + * error-reporting callback into encoders/decoders. This allows them to be + * independent of nearly all assumptions about their actual environment. + * + * It is also a container for allocating the encoders/decoders themselves that + * insulates clients from knowing their actual size. This provides ABI + * compatibility even if the size of the objects change. And this allows the + * structure definitions to be in the .c files instead of the .h files, making + * the .h files smaller and more readable. + * + * We might want to consider renaming this to "Pipeline" if/when the concept of + * a pipeline element becomes more formalized. */ +UPB_DECLARE_TYPE(upb::Environment, upb_env) - /* Returns true if there is no error. */ - bool ok() const; +/* A function that receives an error report from an encoder or decoder. The + * callback can return true to request that the error should be recovered, but + * if the error is not recoverable this has no effect. */ +typedef bool upb_error_func(void *ud, const upb_status *status); - /* Optional error space and code, useful if the caller wants to - * programmatically check the specific kind of error. */ - ErrorSpace* error_space(); - int code() const; +UPB_BEGIN_EXTERN_C - const char *error_message() const; +void upb_env_init(upb_env *e); +void upb_env_init2(upb_env *e, void *mem, size_t n, upb_alloc *alloc); +void upb_env_uninit(upb_env *e); - /* The error message will be truncated if it is longer than - * UPB_STATUS_MAX_MESSAGE-4. */ - void SetErrorMessage(const char* msg); - void SetFormattedErrorMessage(const char* fmt, ...); +void upb_env_initonly(upb_env *e); - /* If there is no error message already, this will use the ErrorSpace to - * populate the error message for this code. The caller can still call - * SetErrorMessage() to give a more specific message. */ - void SetErrorCode(ErrorSpace* space, int code); +upb_arena *upb_env_arena(upb_env *e); +bool upb_env_ok(const upb_env *e); +void upb_env_seterrorfunc(upb_env *e, upb_error_func *func, void *ud); - /* Resets the status to a successful state with no message. */ - void Clear(); +/* Convenience wrappers around the methods of the contained arena. */ +void upb_env_reporterrorsto(upb_env *e, upb_status *s); +bool upb_env_reporterror(upb_env *e, const upb_status *s); +void *upb_env_malloc(upb_env *e, size_t size); +void *upb_env_realloc(upb_env *e, void *ptr, size_t oldsize, size_t size); +void upb_env_free(upb_env *e, void *ptr); +bool upb_env_addcleanup(upb_env *e, upb_cleanup_func *func, void *ud); +size_t upb_env_bytesallocated(const upb_env *e); + +UPB_END_EXTERN_C + +#ifdef __cplusplus + +class upb::Environment { + public: + /* The given Arena must outlive this environment. */ + Environment() { upb_env_initonly(this); } + + Environment(void *mem, size_t len, Allocator *a) : arena_(mem, len, a) { + upb_env_initonly(this); + } - void CopyFrom(const Status& other); + Arena* arena() { return upb_env_arena(this); } + + /* Set a custom error reporting function. */ + void SetErrorFunction(upb_error_func* func, void* ud) { + upb_env_seterrorfunc(this, func, ud); + } + + /* Set the error reporting function to simply copy the status to the given + * status and abort. */ + void ReportErrorsTo(Status* status) { upb_env_reporterrorsto(this, status); } + + /* Returns true if all allocations and AddCleanup() calls have succeeded, + * and no errors were reported with ReportError() (except ones that recovered + * successfully). */ + bool ok() const { return upb_env_ok(this); } + + /* Reports an error to this environment's callback, returning true if + * the caller should try to recover. */ + bool ReportError(const Status* status) { + return upb_env_reporterror(this, status); + } private: - UPB_DISALLOW_COPY_AND_ASSIGN(Status) + UPB_DISALLOW_COPY_AND_ASSIGN(Environment) + #else -struct upb_status { -#endif +struct upb_env { +#endif /* __cplusplus */ + upb_arena arena_; + upb_error_func *error_func_; + void *error_ud_; bool ok_; +}; - /* Specific status code defined by some error space (optional). */ - int code_; - upb_errorspace *error_space_; - /* Error message; NULL-terminated. */ - char msg[UPB_STATUS_MAX_MESSAGE]; -}; +/* upb::InlinedArena **********************************************************/ +/* upb::InlinedEnvironment ****************************************************/ -#define UPB_STATUS_INIT {true, 0, NULL, {0}} +/* upb::InlinedArena and upb::InlinedEnvironment seed their arenas with a + * predefined amount of memory. No heap memory will be allocated until the + * initial block is exceeded. + * + * These types only exist in C++ */ #ifdef __cplusplus -extern "C" { -#endif -/* The returned string is invalidated by any other call into the status. */ -const char *upb_status_errmsg(const upb_status *status); -bool upb_ok(const upb_status *status); -upb_errorspace *upb_status_errspace(const upb_status *status); -int upb_status_errcode(const upb_status *status); +template class upb::InlinedArena : public upb::Arena { + public: + InlinedArena() : Arena(initial_block_, N, NULL) {} + explicit InlinedArena(Allocator* a) : Arena(initial_block_, N, a) {} -/* Any of the functions that write to a status object allow status to be NULL, - * to support use cases where the function's caller does not care about the - * status message. */ -void upb_status_clear(upb_status *status); -void upb_status_seterrmsg(upb_status *status, const char *msg); -void upb_status_seterrf(upb_status *status, const char *fmt, ...); -void upb_status_vseterrf(upb_status *status, const char *fmt, va_list args); -void upb_status_seterrcode(upb_status *status, upb_errorspace *space, int code); -void upb_status_copy(upb_status *to, const upb_status *from); + private: + UPB_DISALLOW_COPY_AND_ASSIGN(InlinedArena) -#ifdef __cplusplus -} /* extern "C" */ + char initial_block_[N + UPB_ARENA_BLOCK_OVERHEAD]; +}; -namespace upb { +template class upb::InlinedEnvironment : public upb::Environment { + public: + InlinedEnvironment() : Environment(initial_block_, N, NULL) {} + explicit InlinedEnvironment(Allocator *a) + : Environment(initial_block_, N, a) {} -/* C++ Wrappers */ -inline Status::Status() { Clear(); } -inline bool Status::ok() const { return upb_ok(this); } -inline const char* Status::error_message() const { - return upb_status_errmsg(this); -} -inline void Status::SetErrorMessage(const char* msg) { - upb_status_seterrmsg(this, msg); -} -inline void Status::SetFormattedErrorMessage(const char* fmt, ...) { - va_list args; - va_start(args, fmt); - upb_status_vseterrf(this, fmt, args); - va_end(args); -} -inline void Status::SetErrorCode(ErrorSpace* space, int code) { - upb_status_seterrcode(this, space, code); -} -inline void Status::Clear() { upb_status_clear(this); } -inline void Status::CopyFrom(const Status& other) { - upb_status_copy(this, &other); -} + private: + UPB_DISALLOW_COPY_AND_ASSIGN(InlinedEnvironment) + + char initial_block_[N + UPB_ARENA_BLOCK_OVERHEAD]; +}; + +#endif /* __cplusplus */ -} /* namespace upb */ -#endif #endif /* UPB_H_ */ @@ -617,10 +814,14 @@ typedef struct { #endif /* Like strdup(), which isn't always available since it's not ANSI C. */ -char *upb_strdup(const char *s); +char *upb_strdup(const char *s, upb_alloc *a); /* Variant that works with a length-delimited rather than NULL-delimited string, * as supported by strtable. */ -char *upb_strdup2(const char *s, size_t len); +char *upb_strdup2(const char *s, size_t len, upb_alloc *a); + +UPB_INLINE char *upb_gstrdup(const char *s) { + return upb_strdup(s, &upb_alloc_global); +} UPB_INLINE void _upb_value_setval(upb_value *v, uint64_t val, upb_ctype_t ctype) { @@ -797,14 +998,40 @@ typedef struct { * initialize const hash tables. Then we cast away const when we have to. */ const upb_tabent *entries; + +#ifndef NDEBUG + /* This table's allocator. We make the user pass it in to every relevant + * function and only use this to check it in debug mode. We do this solely + * to keep upb_table as small as possible. This might seem slightly paranoid + * but the plan is to use upb_table for all map fields and extension sets in + * a forthcoming message representation, so there could be a lot of these. + * If this turns out to be too annoying later, we can change it (since this + * is an internal-only header file). */ + upb_alloc *alloc; +#endif } upb_table; +#ifdef NDEBUG +# define UPB_TABLE_INIT(count, mask, ctype, size_lg2, entries) \ + {count, mask, ctype, size_lg2, entries} +#else +# ifdef UPB_DEBUG_REFS +/* At the moment the only mutable tables we statically initialize are debug + * ref tables. */ +# define UPB_TABLE_INIT(count, mask, ctype, size_lg2, entries) \ + {count, mask, ctype, size_lg2, entries, &upb_alloc_debugrefs} +# else +# define UPB_TABLE_INIT(count, mask, ctype, size_lg2, entries) \ + {count, mask, ctype, size_lg2, entries, NULL} +# endif +#endif + typedef struct { upb_table t; } upb_strtable; #define UPB_STRTABLE_INIT(count, mask, ctype, size_lg2, entries) \ - {{count, mask, ctype, size_lg2, entries}} + {UPB_TABLE_INIT(count, mask, ctype, size_lg2, entries)} #define UPB_EMPTY_STRTABLE_INIT(ctype) \ UPB_STRTABLE_INIT(0, 0, ctype, 0, NULL) @@ -817,7 +1044,7 @@ typedef struct { } upb_inttable; #define UPB_INTTABLE_INIT(count, mask, ctype, size_lg2, ent, a, asize, acount) \ - {{count, mask, ctype, size_lg2, ent}, a, asize, acount} + {UPB_TABLE_INIT(count, mask, ctype, size_lg2, ent), a, asize, acount} #define UPB_EMPTY_INTTABLE_INIT(ctype) \ UPB_INTTABLE_INIT(0, 0, ctype, 0, NULL, NULL, 0, 0) @@ -857,10 +1084,26 @@ UPB_INLINE bool upb_arrhas(upb_tabval key) { /* Initialize and uninitialize a table, respectively. If memory allocation * failed, false is returned that the table is uninitialized. */ -bool upb_inttable_init(upb_inttable *table, upb_ctype_t ctype); -bool upb_strtable_init(upb_strtable *table, upb_ctype_t ctype); -void upb_inttable_uninit(upb_inttable *table); -void upb_strtable_uninit(upb_strtable *table); +bool upb_inttable_init2(upb_inttable *table, upb_ctype_t ctype, upb_alloc *a); +bool upb_strtable_init2(upb_strtable *table, upb_ctype_t ctype, upb_alloc *a); +void upb_inttable_uninit2(upb_inttable *table, upb_alloc *a); +void upb_strtable_uninit2(upb_strtable *table, upb_alloc *a); + +UPB_INLINE bool upb_inttable_init(upb_inttable *table, upb_ctype_t ctype) { + return upb_inttable_init2(table, ctype, &upb_alloc_global); +} + +UPB_INLINE bool upb_strtable_init(upb_strtable *table, upb_ctype_t ctype) { + return upb_strtable_init2(table, ctype, &upb_alloc_global); +} + +UPB_INLINE void upb_inttable_uninit(upb_inttable *table) { + upb_inttable_uninit2(table, &upb_alloc_global); +} + +UPB_INLINE void upb_strtable_uninit(upb_strtable *table) { + upb_strtable_uninit2(table, &upb_alloc_global); +} /* Returns the number of values in the table. */ size_t upb_inttable_count(const upb_inttable *t); @@ -875,9 +1118,20 @@ UPB_INLINE size_t upb_strtable_count(const upb_strtable *t) { * * If a table resize was required but memory allocation failed, false is * returned and the table is unchanged. */ -bool upb_inttable_insert(upb_inttable *t, uintptr_t key, upb_value val); -bool upb_strtable_insert2(upb_strtable *t, const char *key, size_t len, - upb_value val); +bool upb_inttable_insert2(upb_inttable *t, uintptr_t key, upb_value val, + upb_alloc *a); +bool upb_strtable_insert3(upb_strtable *t, const char *key, size_t len, + upb_value val, upb_alloc *a); + +UPB_INLINE bool upb_inttable_insert(upb_inttable *t, uintptr_t key, + upb_value val) { + return upb_inttable_insert2(t, key, val, &upb_alloc_global); +} + +UPB_INLINE bool upb_strtable_insert2(upb_strtable *t, const char *key, + size_t len, upb_value val) { + return upb_strtable_insert3(t, key, len, val, &upb_alloc_global); +} /* For NULL-terminated strings. */ UPB_INLINE bool upb_strtable_insert(upb_strtable *t, const char *key, @@ -900,8 +1154,13 @@ UPB_INLINE bool upb_strtable_lookup(const upb_strtable *t, const char *key, /* Removes an item from the table. Returns true if the remove was successful, * and stores the removed item in *val if non-NULL. */ bool upb_inttable_remove(upb_inttable *t, uintptr_t key, upb_value *val); -bool upb_strtable_remove2(upb_strtable *t, const char *key, size_t len, - upb_value *val); +bool upb_strtable_remove3(upb_strtable *t, const char *key, size_t len, + upb_value *val, upb_alloc *alloc); + +UPB_INLINE bool upb_strtable_remove2(upb_strtable *t, const char *key, + size_t len, upb_value *val) { + return upb_strtable_remove3(t, key, len, val, &upb_alloc_global); +} /* For NULL-terminated strings. */ UPB_INLINE bool upb_strtable_remove(upb_strtable *t, const char *key, @@ -916,19 +1175,33 @@ bool upb_inttable_replace(upb_inttable *t, uintptr_t key, upb_value val); /* Handy routines for treating an inttable like a stack. May not be mixed with * other insert/remove calls. */ -bool upb_inttable_push(upb_inttable *t, upb_value val); +bool upb_inttable_push2(upb_inttable *t, upb_value val, upb_alloc *a); upb_value upb_inttable_pop(upb_inttable *t); +UPB_INLINE bool upb_inttable_push(upb_inttable *t, upb_value val) { + return upb_inttable_push2(t, val, &upb_alloc_global); +} + /* Convenience routines for inttables with pointer keys. */ -bool upb_inttable_insertptr(upb_inttable *t, const void *key, upb_value val); +bool upb_inttable_insertptr2(upb_inttable *t, const void *key, upb_value val, + upb_alloc *a); bool upb_inttable_removeptr(upb_inttable *t, const void *key, upb_value *val); bool upb_inttable_lookupptr( const upb_inttable *t, const void *key, upb_value *val); +UPB_INLINE bool upb_inttable_insertptr(upb_inttable *t, const void *key, + upb_value val) { + return upb_inttable_insertptr2(t, key, val, &upb_alloc_global); +} + /* Optimizes the table for the current set of entries, for both memory use and * lookup time. Client should call this after all entries have been inserted; * inserting more entries is legal, but will likely require a table resize. */ -void upb_inttable_compact(upb_inttable *t); +void upb_inttable_compact2(upb_inttable *t, upb_alloc *a); + +UPB_INLINE void upb_inttable_compact(upb_inttable *t) { + upb_inttable_compact2(t, &upb_alloc_global); +} /* A special-case inlinable version of the lookup routine for 32-bit * integers. */ @@ -957,7 +1230,7 @@ UPB_INLINE bool upb_inttable_lookup32(const upb_inttable *t, uint32_t key, } /* Exposed for testing only. */ -bool upb_strtable_resize(upb_strtable *t, size_t size_lg2); +bool upb_strtable_resize(upb_strtable *t, size_t size_lg2, upb_alloc *a); /* Iterators ******************************************************************/ @@ -1002,8 +1275,8 @@ typedef struct { void upb_strtable_begin(upb_strtable_iter *i, const upb_strtable *t); void upb_strtable_next(upb_strtable_iter *i); bool upb_strtable_done(const upb_strtable_iter *i); -const char *upb_strtable_iter_key(upb_strtable_iter *i); -size_t upb_strtable_iter_keylength(upb_strtable_iter *i); +const char *upb_strtable_iter_key(const upb_strtable_iter *i); +size_t upb_strtable_iter_keylength(const upb_strtable_iter *i); upb_value upb_strtable_iter_value(const upb_strtable_iter *i); void upb_strtable_iter_setdone(upb_strtable_iter *i); bool upb_strtable_iter_isequal(const upb_strtable_iter *i1, @@ -1056,7 +1329,10 @@ bool upb_inttable_iter_isequal(const upb_inttable_iter *i1, /* #define UPB_DEBUG_REFS */ #ifdef __cplusplus -namespace upb { class RefCounted; } +namespace upb { +class RefCounted; +template class reffed_ptr; +} #endif UPB_DECLARE_TYPE(upb::RefCounted, upb_refcounted) @@ -1124,10 +1400,12 @@ struct upb_refcounted { }; #ifdef UPB_DEBUG_REFS -#define UPB_REFCOUNT_INIT(refs, ref2s) \ - {&static_refcount, NULL, NULL, 0, true, refs, ref2s} +extern upb_alloc upb_alloc_debugrefs; +#define UPB_REFCOUNT_INIT(vtbl, refs, ref2s) \ + {&static_refcount, NULL, vtbl, 0, true, refs, ref2s} #else -#define UPB_REFCOUNT_INIT(refs, ref2s) {&static_refcount, NULL, NULL, 0, true} +#define UPB_REFCOUNT_INIT(vtbl, refs, ref2s) \ + {&static_refcount, NULL, vtbl, 0, true} #endif UPB_BEGIN_EXTERN_C @@ -1201,64 +1479,169 @@ bool upb_refcounted_init(upb_refcounted *r, const struct upb_refcounted_vtbl *vtbl, const void *owner); -/* Adds a ref from one refcounted object to another ("from" must not already - * own a ref). These refs may be circular; cycles will be collected correctly - * (if conservatively). These refs do not need to be freed in from's free() - * function. */ -void upb_refcounted_ref2(const upb_refcounted *r, upb_refcounted *from); +/* Adds a ref from one refcounted object to another ("from" must not already + * own a ref). These refs may be circular; cycles will be collected correctly + * (if conservatively). These refs do not need to be freed in from's free() + * function. */ +void upb_refcounted_ref2(const upb_refcounted *r, upb_refcounted *from); + +/* Removes a ref that was acquired from upb_refcounted_ref2(), and collects any + * object it can. This is only necessary when "from" no longer points to "r", + * and not from from's "free" function. */ +void upb_refcounted_unref2(const upb_refcounted *r, upb_refcounted *from); + +#define upb_ref2(r, from) \ + upb_refcounted_ref2((const upb_refcounted*)r, (upb_refcounted*)from) +#define upb_unref2(r, from) \ + upb_refcounted_unref2((const upb_refcounted*)r, (upb_refcounted*)from) + +/* Freezes all mutable object reachable by ref2() refs from the given roots. + * This will split refcounting groups into precise SCC groups, so that + * refcounting of frozen objects can be more aggressive. If memory allocation + * fails, or if more than 2**31 mutable objects are reachable from "roots", or + * if the maximum depth of the graph exceeds "maxdepth", false is returned and + * the objects are unchanged. + * + * After this operation succeeds, the objects are frozen/const, and may not be + * used through non-const pointers. In particular, they may not be passed as + * the second parameter of upb_refcounted_{ref,unref}2(). On the upside, all + * operations on frozen refcounteds are threadsafe, and objects will be freed + * at the precise moment that they become unreachable. + * + * Caller must own refs on each object in the "roots" list. */ +bool upb_refcounted_freeze(upb_refcounted *const*roots, int n, upb_status *s, + int maxdepth); + +/* Shared by all compiled-in refcounted objects. */ +extern uint32_t static_refcount; + +UPB_END_EXTERN_C + +#ifdef __cplusplus +/* C++ Wrappers. */ +namespace upb { +inline bool RefCounted::IsFrozen() const { + return upb_refcounted_isfrozen(this); +} +inline void RefCounted::Ref(const void *owner) const { + upb_refcounted_ref(this, owner); +} +inline void RefCounted::Unref(const void *owner) const { + upb_refcounted_unref(this, owner); +} +inline void RefCounted::DonateRef(const void *from, const void *to) const { + upb_refcounted_donateref(this, from, to); +} +inline void RefCounted::CheckRef(const void *owner) const { + upb_refcounted_checkref(this, owner); +} +} /* namespace upb */ +#endif + + +/* upb::reffed_ptr ************************************************************/ + +#ifdef __cplusplus + +#include /* For std::swap(). */ + +/* Provides RAII semantics for upb refcounted objects. Each reffed_ptr owns a + * ref on whatever object it points to (if any). */ +template class upb::reffed_ptr { + public: + reffed_ptr() : ptr_(NULL) {} + + /* If ref_donor is NULL, takes a new ref, otherwise adopts from ref_donor. */ + template + reffed_ptr(U* val, const void* ref_donor = NULL) + : ptr_(upb::upcast(val)) { + if (ref_donor) { + assert(ptr_); + ptr_->DonateRef(ref_donor, this); + } else if (ptr_) { + ptr_->Ref(this); + } + } + + template + reffed_ptr(const reffed_ptr& other) + : ptr_(upb::upcast(other.get())) { + if (ptr_) ptr_->Ref(this); + } + + reffed_ptr(const reffed_ptr& other) + : ptr_(upb::upcast(other.get())) { + if (ptr_) ptr_->Ref(this); + } + + ~reffed_ptr() { if (ptr_) ptr_->Unref(this); } + + template + reffed_ptr& operator=(const reffed_ptr& other) { + reset(other.get()); + return *this; + } + + reffed_ptr& operator=(const reffed_ptr& other) { + reset(other.get()); + return *this; + } + + /* TODO(haberman): add C++11 move construction/assignment for greater + * efficiency. */ + + void swap(reffed_ptr& other) { + if (ptr_ == other.ptr_) { + return; + } + + if (ptr_) ptr_->DonateRef(this, &other); + if (other.ptr_) other.ptr_->DonateRef(&other, this); + std::swap(ptr_, other.ptr_); + } + + T& operator*() const { + assert(ptr_); + return *ptr_; + } + + T* operator->() const { + assert(ptr_); + return ptr_; + } + + T* get() const { return ptr_; } -/* Removes a ref that was acquired from upb_refcounted_ref2(), and collects any - * object it can. This is only necessary when "from" no longer points to "r", - * and not from from's "free" function. */ -void upb_refcounted_unref2(const upb_refcounted *r, upb_refcounted *from); + /* If ref_donor is NULL, takes a new ref, otherwise adopts from ref_donor. */ + template + void reset(U* ptr = NULL, const void* ref_donor = NULL) { + reffed_ptr(ptr, ref_donor).swap(*this); + } -#define upb_ref2(r, from) \ - upb_refcounted_ref2((const upb_refcounted*)r, (upb_refcounted*)from) -#define upb_unref2(r, from) \ - upb_refcounted_unref2((const upb_refcounted*)r, (upb_refcounted*)from) + template + reffed_ptr down_cast() { + return reffed_ptr(upb::down_cast(get())); + } -/* Freezes all mutable object reachable by ref2() refs from the given roots. - * This will split refcounting groups into precise SCC groups, so that - * refcounting of frozen objects can be more aggressive. If memory allocation - * fails, or if more than 2**31 mutable objects are reachable from "roots", or - * if the maximum depth of the graph exceeds "maxdepth", false is returned and - * the objects are unchanged. - * - * After this operation succeeds, the objects are frozen/const, and may not be - * used through non-const pointers. In particular, they may not be passed as - * the second parameter of upb_refcounted_{ref,unref}2(). On the upside, all - * operations on frozen refcounteds are threadsafe, and objects will be freed - * at the precise moment that they become unreachable. - * - * Caller must own refs on each object in the "roots" list. */ -bool upb_refcounted_freeze(upb_refcounted *const*roots, int n, upb_status *s, - int maxdepth); + template + reffed_ptr dyn_cast() { + return reffed_ptr(upb::dyn_cast(get())); + } -/* Shared by all compiled-in refcounted objects. */ -extern uint32_t static_refcount; + /* Plain release() is unsafe; if we were the only owner, it would leak the + * object. Instead we provide this: */ + T* ReleaseTo(const void* new_owner) { + T* ret = NULL; + ptr_->DonateRef(this, new_owner); + std::swap(ret, ptr_); + return ret; + } -UPB_END_EXTERN_C + private: + T* ptr_; +}; -#ifdef __cplusplus -/* C++ Wrappers. */ -namespace upb { -inline bool RefCounted::IsFrozen() const { - return upb_refcounted_isfrozen(this); -} -inline void RefCounted::Ref(const void *owner) const { - upb_refcounted_ref(this, owner); -} -inline void RefCounted::Unref(const void *owner) const { - upb_refcounted_unref(this, owner); -} -inline void RefCounted::DonateRef(const void *from, const void *to) const { - upb_refcounted_donateref(this, from, to); -} -inline void RefCounted::CheckRef(const void *owner) const { - upb_refcounted_checkref(this, owner); -} -} /* namespace upb */ -#endif +#endif /* __cplusplus */ #endif /* UPB_REFCOUNT_H_ */ @@ -1511,6 +1894,11 @@ typedef enum { UPB_DESCRIPTOR_TYPE_SINT64 = 18 } upb_descriptortype_t; +typedef enum { + UPB_SYNTAX_PROTO2 = 2, + UPB_SYNTAX_PROTO3 = 3 +} upb_syntax_t; + /* Maximum field number allowed for FieldDefs. This is an inherent limit of the * protobuf wire format. */ #define UPB_MAX_FIELDNUMBER ((1 << 29) - 1) @@ -1902,6 +2290,10 @@ UPB_END_EXTERN_C typedef upb_inttable_iter upb_msg_field_iter; typedef upb_strtable_iter upb_msg_oneof_iter; +/* Well-known field tag numbers for map-entry messages. */ +#define UPB_MAPENTRY_KEY 1 +#define UPB_MAPENTRY_VALUE 2 + #ifdef __cplusplus /* Structure that describes a single .proto message type. @@ -1960,6 +2352,11 @@ class upb::MessageDef { bool AddOneof(OneofDef* o, Status* s); bool AddOneof(const reffed_ptr& o, Status* s); + upb_syntax_t syntax() const; + + /* Returns false if we don't support this syntax value. */ + bool set_syntax(upb_syntax_t syntax); + /* Set this to false to indicate that primitive fields should not have * explicit presence information associated with them. This will affect all * fields added to this message. Defaults to true. */ @@ -2150,15 +2547,20 @@ UPB_REFCOUNTED_CMETHODS(upb_msgdef, upb_msgdef_upcast2) bool upb_msgdef_freeze(upb_msgdef *m, upb_status *status); +upb_msgdef *upb_msgdef_dup(const upb_msgdef *m, const void *owner); const char *upb_msgdef_fullname(const upb_msgdef *m); const char *upb_msgdef_name(const upb_msgdef *m); -bool upb_msgdef_setfullname(upb_msgdef *m, const char *fullname, upb_status *s); +int upb_msgdef_numoneofs(const upb_msgdef *m); +upb_syntax_t upb_msgdef_syntax(const upb_msgdef *m); -upb_msgdef *upb_msgdef_dup(const upb_msgdef *m, const void *owner); bool upb_msgdef_addfield(upb_msgdef *m, upb_fielddef *f, const void *ref_donor, upb_status *s); bool upb_msgdef_addoneof(upb_msgdef *m, upb_oneofdef *o, const void *ref_donor, upb_status *s); +bool upb_msgdef_setfullname(upb_msgdef *m, const char *fullname, upb_status *s); +void upb_msgdef_setmapentry(upb_msgdef *m, bool map_entry); +bool upb_msgdef_mapentry(const upb_msgdef *m); +bool upb_msgdef_setsyntax(upb_msgdef *m, upb_syntax_t syntax); /* Field lookup in a couple of different variations: * - itof = int to field @@ -2200,18 +2602,21 @@ UPB_INLINE upb_oneofdef *upb_msgdef_ntoo_mutable(upb_msgdef *m, return (upb_oneofdef *)upb_msgdef_ntoo(m, name, len); } -void upb_msgdef_setmapentry(upb_msgdef *m, bool map_entry); -bool upb_msgdef_mapentry(const upb_msgdef *m); - -/* Well-known field tag numbers for map-entry messages. */ -#define UPB_MAPENTRY_KEY 1 -#define UPB_MAPENTRY_VALUE 2 +/* Lookup of either field or oneof by name. Returns whether either was found. + * If the return is true, then the found def will be set, and the non-found + * one set to NULL. */ +bool upb_msgdef_lookupname(const upb_msgdef *m, const char *name, size_t len, + const upb_fielddef **f, const upb_oneofdef **o); -const upb_oneofdef *upb_msgdef_findoneof(const upb_msgdef *m, - const char *name); -int upb_msgdef_numoneofs(const upb_msgdef *m); +UPB_INLINE bool upb_msgdef_lookupnamez(const upb_msgdef *m, const char *name, + const upb_fielddef **f, + const upb_oneofdef **o) { + return upb_msgdef_lookupname(m, name, strlen(name), f, o); +} -/* upb_msg_field_iter i; +/* Iteration over fields and oneofs. For example: + * + * upb_msg_field_iter i; * for(upb_msg_field_begin(&i, m); * !upb_msg_field_done(&i); * upb_msg_field_next(&i)) { @@ -2523,11 +2928,6 @@ UPB_END_EXTERN_C /* upb::FileDef ***************************************************************/ -typedef enum { - UPB_SYNTAX_PROTO2 = 2, - UPB_SYNTAX_PROTO3 = 3 -} upb_syntax_t; - #ifdef __cplusplus /* Class that represents a .proto file with some things defined in it. @@ -2890,12 +3290,18 @@ inline const char *MessageDef::full_name() const { inline const char *MessageDef::name() const { return upb_msgdef_name(this); } +inline upb_syntax_t MessageDef::syntax() const { + return upb_msgdef_syntax(this); +} inline bool MessageDef::set_full_name(const char* fullname, Status* s) { return upb_msgdef_setfullname(this, fullname, s); } inline bool MessageDef::set_full_name(const std::string& fullname, Status* s) { return upb_msgdef_setfullname(this, upb_safecstr(fullname), s); } +inline bool MessageDef::set_syntax(upb_syntax_t syntax) { + return upb_msgdef_setsyntax(this, syntax); +} inline bool MessageDef::Freeze(Status* status) { return upb_msgdef_freeze(this, status); } @@ -3309,8 +3715,8 @@ struct upb_def { bool came_from_user; }; -#define UPB_DEF_INIT(name, type, refs, ref2s) \ - { UPB_REFCOUNT_INIT(refs, ref2s), name, NULL, type, false } +#define UPB_DEF_INIT(name, type, vtbl, refs, ref2s) \ + { UPB_REFCOUNT_INIT(vtbl, refs, ref2s), name, NULL, type, false } /* upb_fielddef ***************************************************************/ @@ -3350,12 +3756,14 @@ struct upb_fielddef { uint32_t index_; }; +extern const struct upb_refcounted_vtbl upb_fielddef_vtbl; + #define UPB_FIELDDEF_INIT(label, type, intfmt, tagdelim, is_extension, lazy, \ packed, name, num, msgdef, subdef, selector_base, \ index, defaultval, refs, ref2s) \ { \ - UPB_DEF_INIT(name, UPB_DEF_FIELD, refs, ref2s), defaultval, {msgdef}, \ - {subdef}, NULL, false, false, \ + UPB_DEF_INIT(name, UPB_DEF_FIELD, &upb_fielddef_vtbl, refs, ref2s), \ + defaultval, {msgdef}, {subdef}, NULL, false, false, \ type == UPB_TYPE_STRING || type == UPB_TYPE_BYTES, true, is_extension, \ lazy, packed, intfmt, tagdelim, type, label, num, selector_base, index \ } @@ -3371,32 +3779,26 @@ struct upb_msgdef { /* Tables for looking up fields by number and name. */ upb_inttable itof; /* int to field */ - upb_strtable ntof; /* name to field */ + upb_strtable ntof; /* name to field/oneof */ - /* Tables for looking up oneofs by name. */ - upb_strtable ntoo; /* name to oneof */ - - /* Is this a map-entry message? - * TODO: set this flag properly for static descriptors; regenerate - * descriptor.upb.c. */ + /* Is this a map-entry message? */ bool map_entry; - /* Whether this message has proto2 or proto3 semantics. - * TODO: set this flag properly for static descriptors; regenerate - * descriptor.upb.c. */ + /* Whether this message has proto2 or proto3 semantics. */ upb_syntax_t syntax; /* TODO(haberman): proper extension ranges (there can be multiple). */ }; +extern const struct upb_refcounted_vtbl upb_msgdef_vtbl; + /* TODO: also support static initialization of the oneofs table. This will be * needed if we compile in descriptors that contain oneofs. */ #define UPB_MSGDEF_INIT(name, selector_count, submsg_field_count, itof, ntof, \ - refs, ref2s) \ + map_entry, syntax, refs, ref2s) \ { \ - UPB_DEF_INIT(name, UPB_DEF_MSG, refs, ref2s), selector_count, \ - submsg_field_count, itof, ntof, \ - UPB_EMPTY_STRTABLE_INIT(UPB_CTYPE_PTR), false, true \ + UPB_DEF_INIT(name, UPB_DEF_MSG, &upb_fielddef_vtbl, refs, ref2s), \ + selector_count, submsg_field_count, itof, ntof, map_entry, syntax \ } @@ -3410,8 +3812,11 @@ struct upb_enumdef { int32_t defaultval; }; +extern const struct upb_refcounted_vtbl upb_enumdef_vtbl; + #define UPB_ENUMDEF_INIT(name, ntoi, iton, defaultval, refs, ref2s) \ - { UPB_DEF_INIT(name, UPB_DEF_ENUM, refs, ref2s), ntoi, iton, defaultval } + { UPB_DEF_INIT(name, UPB_DEF_ENUM, &upb_enumdef_vtbl, refs, ref2s), ntoi, \ + iton, defaultval } /* upb_oneofdef ***************************************************************/ @@ -3425,8 +3830,10 @@ struct upb_oneofdef { const upb_msgdef *parent; }; +extern const struct upb_refcounted_vtbl upb_oneofdef_vtbl; + #define UPB_ONEOFDEF_INIT(name, ntof, itof, refs, ref2s) \ - { UPB_REFCOUNT_INIT(refs, ref2s), name, ntof, itof } + { UPB_REFCOUNT_INIT(&upb_oneofdef_vtbl, refs, ref2s), name, ntof, itof } /* upb_symtab *****************************************************************/ @@ -3437,9 +3844,6 @@ struct upb_symtab { upb_strtable symtab; }; -#define UPB_SYMTAB_INIT(symtab, refs, ref2s) \ - { UPB_REFCOUNT_INIT(refs, ref2s), symtab } - struct upb_filedef { upb_refcounted base; @@ -3451,6 +3855,8 @@ struct upb_filedef { upb_inttable deps; }; +extern const struct upb_refcounted_vtbl upb_filedef_vtbl; + #endif /* UPB_STATICINIT_H_ */ /* ** upb::Handlers (upb_handlers) @@ -5392,267 +5798,6 @@ inline BytesHandler::~BytesHandler() {} #endif /* UPB_HANDLERS_H */ /* -** upb::Environment (upb_env) -** -** A upb::Environment provides a means for injecting malloc and an -** error-reporting callback into encoders/decoders. This allows them to be -** independent of nearly all assumptions about their actual environment. -** -** It is also a container for allocating the encoders/decoders themselves that -** insulates clients from knowing their actual size. This provides ABI -** compatibility even if the size of the objects change. And this allows the -** structure definitions to be in the .c files instead of the .h files, making -** the .h files smaller and more readable. -*/ - - -#ifndef UPB_ENV_H_ -#define UPB_ENV_H_ - -#ifdef __cplusplus -namespace upb { -class Environment; -class SeededAllocator; -} -#endif - -UPB_DECLARE_TYPE(upb::Environment, upb_env) -UPB_DECLARE_TYPE(upb::SeededAllocator, upb_seededalloc) - -typedef void *upb_alloc_func(void *ud, void *ptr, size_t oldsize, size_t size); -typedef void upb_cleanup_func(void *ud); -typedef bool upb_error_func(void *ud, const upb_status *status); - -#ifdef __cplusplus - -/* An environment is *not* thread-safe. */ -class upb::Environment { - public: - Environment(); - ~Environment(); - - /* Set a custom memory allocation function for the environment. May ONLY - * be called before any calls to Malloc()/Realloc()/AddCleanup() below. - * If this is not called, the system realloc() function will be used. - * The given user pointer "ud" will be passed to the allocation function. - * - * The allocation function will not receive corresponding "free" calls. it - * must ensure that the memory is valid for the lifetime of the Environment, - * but it may be reclaimed any time thereafter. The likely usage is that - * "ud" points to a stateful allocator, and that the allocator frees all - * memory, arena-style, when it is destroyed. In this case the allocator must - * outlive the Environment. Another possibility is that the allocation - * function returns GC-able memory that is guaranteed to be GC-rooted for the - * life of the Environment. */ - void SetAllocationFunction(upb_alloc_func* alloc, void* ud); - - template - void SetAllocator(T* allocator) { - SetAllocationFunction(allocator->GetAllocationFunction(), allocator); - } - - /* Set a custom error reporting function. */ - void SetErrorFunction(upb_error_func* func, void* ud); - - /* Set the error reporting function to simply copy the status to the given - * status and abort. */ - void ReportErrorsTo(Status* status); - - /* Returns true if all allocations and AddCleanup() calls have succeeded, - * and no errors were reported with ReportError() (except ones that recovered - * successfully). */ - bool ok() const; - - /* Functions for use by encoders/decoders. **********************************/ - - /* Reports an error to this environment's callback, returning true if - * the caller should try to recover. */ - bool ReportError(const Status* status); - - /* Allocate memory. Uses the environment's allocation function. - * - * There is no need to free(). All memory will be freed automatically, but is - * guaranteed to outlive the Environment. */ - void* Malloc(size_t size); - - /* Reallocate memory. Preserves "oldsize" bytes from the existing buffer - * Requires: oldsize <= existing_size. - * - * TODO(haberman): should we also enforce that oldsize <= size? */ - void* Realloc(void* ptr, size_t oldsize, size_t size); - - /* Add a cleanup function to run when the environment is destroyed. - * Returns false on out-of-memory. - * - * The first call to AddCleanup() after SetAllocationFunction() is guaranteed - * to return true -- this makes it possible to robustly set a cleanup handler - * for a custom allocation function. */ - bool AddCleanup(upb_cleanup_func* func, void* ud); - - /* Total number of bytes that have been allocated. It is undefined what - * Realloc() does to this counter. */ - size_t BytesAllocated() const; - - private: - UPB_DISALLOW_COPY_AND_ASSIGN(Environment) - -#else -struct upb_env { -#endif /* __cplusplus */ - - bool ok_; - size_t bytes_allocated; - - /* Alloc function. */ - upb_alloc_func *alloc; - void *alloc_ud; - - /* Error-reporting function. */ - upb_error_func *err; - void *err_ud; - - /* Userdata for default alloc func. */ - void *default_alloc_ud; - - /* Cleanup entries. Pointer to a cleanup_ent, defined in env.c */ - void *cleanup_head; - - /* For future expansion, since the size of this struct is exposed to users. */ - void *future1; - void *future2; -}; - -UPB_BEGIN_EXTERN_C - -void upb_env_init(upb_env *e); -void upb_env_uninit(upb_env *e); -void upb_env_setallocfunc(upb_env *e, upb_alloc_func *func, void *ud); -void upb_env_seterrorfunc(upb_env *e, upb_error_func *func, void *ud); -void upb_env_reporterrorsto(upb_env *e, upb_status *status); -bool upb_env_ok(const upb_env *e); -bool upb_env_reporterror(upb_env *e, const upb_status *status); -void *upb_env_malloc(upb_env *e, size_t size); -void *upb_env_realloc(upb_env *e, void *ptr, size_t oldsize, size_t size); -bool upb_env_addcleanup(upb_env *e, upb_cleanup_func *func, void *ud); -size_t upb_env_bytesallocated(const upb_env *e); - -UPB_END_EXTERN_C - -#ifdef __cplusplus - -/* An allocator that allocates from an initial memory region (likely the stack) - * before falling back to another allocator. */ -class upb::SeededAllocator { - public: - SeededAllocator(void *mem, size_t len); - ~SeededAllocator(); - - /* Set a custom fallback memory allocation function for the allocator, to use - * once the initial region runs out. - * - * May ONLY be called before GetAllocationFunction(). If this is not - * called, the system realloc() will be the fallback allocator. */ - void SetFallbackAllocator(upb_alloc_func *alloc, void *ud); - - /* Gets the allocation function for this allocator. */ - upb_alloc_func* GetAllocationFunction(); - - private: - UPB_DISALLOW_COPY_AND_ASSIGN(SeededAllocator) - -#else -struct upb_seededalloc { -#endif /* __cplusplus */ - - /* Fallback alloc function. */ - upb_alloc_func *alloc; - upb_cleanup_func *alloc_cleanup; - void *alloc_ud; - bool need_cleanup; - bool returned_allocfunc; - - /* Userdata for default alloc func. */ - void *default_alloc_ud; - - /* Pointers for the initial memory region. */ - char *mem_base; - char *mem_ptr; - char *mem_limit; - - /* For future expansion, since the size of this struct is exposed to users. */ - void *future1; - void *future2; -}; - -UPB_BEGIN_EXTERN_C - -void upb_seededalloc_init(upb_seededalloc *a, void *mem, size_t len); -void upb_seededalloc_uninit(upb_seededalloc *a); -void upb_seededalloc_setfallbackalloc(upb_seededalloc *a, upb_alloc_func *func, - void *ud); -upb_alloc_func *upb_seededalloc_getallocfunc(upb_seededalloc *a); - -UPB_END_EXTERN_C - -#ifdef __cplusplus - -namespace upb { - -inline Environment::Environment() { - upb_env_init(this); -} -inline Environment::~Environment() { - upb_env_uninit(this); -} -inline void Environment::SetAllocationFunction(upb_alloc_func *alloc, - void *ud) { - upb_env_setallocfunc(this, alloc, ud); -} -inline void Environment::SetErrorFunction(upb_error_func *func, void *ud) { - upb_env_seterrorfunc(this, func, ud); -} -inline void Environment::ReportErrorsTo(Status* status) { - upb_env_reporterrorsto(this, status); -} -inline bool Environment::ok() const { - return upb_env_ok(this); -} -inline bool Environment::ReportError(const Status* status) { - return upb_env_reporterror(this, status); -} -inline void *Environment::Malloc(size_t size) { - return upb_env_malloc(this, size); -} -inline void *Environment::Realloc(void *ptr, size_t oldsize, size_t size) { - return upb_env_realloc(this, ptr, oldsize, size); -} -inline bool Environment::AddCleanup(upb_cleanup_func *func, void *ud) { - return upb_env_addcleanup(this, func, ud); -} -inline size_t Environment::BytesAllocated() const { - return upb_env_bytesallocated(this); -} - -inline SeededAllocator::SeededAllocator(void *mem, size_t len) { - upb_seededalloc_init(this, mem, len); -} -inline SeededAllocator::~SeededAllocator() { - upb_seededalloc_uninit(this); -} -inline void SeededAllocator::SetFallbackAllocator(upb_alloc_func *alloc, - void *ud) { - upb_seededalloc_setfallbackalloc(this, alloc, ud); -} -inline upb_alloc_func *SeededAllocator::GetAllocationFunction() { - return upb_seededalloc_getallocfunc(this); -} - -} /* namespace upb */ - -#endif /* __cplusplus */ - -#endif /* UPB_ENV_H_ */ -/* ** upb::Sink (upb_sink) ** upb::BytesSink (upb_bytessink) ** @@ -6529,7 +6674,11 @@ inline FileDef* Reader::file(size_t i) const { * actually storing protobufs. It only contains *defs* which * let you reflect over a protobuf *schema*. */ -/* This file was generated by upbc (the upb compiler). +/* This file was generated by upbc (the upb compiler) from the input + * file: + * + * upb/descriptor/descriptor.proto + * * Do not edit -- your changes will be discarded when the file is * regenerated. */ @@ -6818,347 +6967,347 @@ namespace upbdefs { namespace google { namespace protobuf { -class DescriptorProto : public upb::reffed_ptr { +class DescriptorProto : public ::upb::reffed_ptr { public: - DescriptorProto(const upb::MessageDef* m, const void *ref_donor = NULL) + DescriptorProto(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_DescriptorProto_is(m)); } static DescriptorProto get() { - const upb::MessageDef* m = upbdefs_google_protobuf_DescriptorProto_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_DescriptorProto_get(&m); return DescriptorProto(m, &m); } - class ExtensionRange : public upb::reffed_ptr { + class ExtensionRange : public ::upb::reffed_ptr { public: - ExtensionRange(const upb::MessageDef* m, const void *ref_donor = NULL) + ExtensionRange(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_DescriptorProto_ExtensionRange_is(m)); } static ExtensionRange get() { - const upb::MessageDef* m = upbdefs_google_protobuf_DescriptorProto_ExtensionRange_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_DescriptorProto_ExtensionRange_get(&m); return ExtensionRange(m, &m); } }; - class ReservedRange : public upb::reffed_ptr { + class ReservedRange : public ::upb::reffed_ptr { public: - ReservedRange(const upb::MessageDef* m, const void *ref_donor = NULL) + ReservedRange(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_DescriptorProto_ReservedRange_is(m)); } static ReservedRange get() { - const upb::MessageDef* m = upbdefs_google_protobuf_DescriptorProto_ReservedRange_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_DescriptorProto_ReservedRange_get(&m); return ReservedRange(m, &m); } }; }; -class EnumDescriptorProto : public upb::reffed_ptr { +class EnumDescriptorProto : public ::upb::reffed_ptr { public: - EnumDescriptorProto(const upb::MessageDef* m, const void *ref_donor = NULL) + EnumDescriptorProto(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_EnumDescriptorProto_is(m)); } static EnumDescriptorProto get() { - const upb::MessageDef* m = upbdefs_google_protobuf_EnumDescriptorProto_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_EnumDescriptorProto_get(&m); return EnumDescriptorProto(m, &m); } }; -class EnumOptions : public upb::reffed_ptr { +class EnumOptions : public ::upb::reffed_ptr { public: - EnumOptions(const upb::MessageDef* m, const void *ref_donor = NULL) + EnumOptions(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_EnumOptions_is(m)); } static EnumOptions get() { - const upb::MessageDef* m = upbdefs_google_protobuf_EnumOptions_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_EnumOptions_get(&m); return EnumOptions(m, &m); } }; -class EnumValueDescriptorProto : public upb::reffed_ptr { +class EnumValueDescriptorProto : public ::upb::reffed_ptr { public: - EnumValueDescriptorProto(const upb::MessageDef* m, const void *ref_donor = NULL) + EnumValueDescriptorProto(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_EnumValueDescriptorProto_is(m)); } static EnumValueDescriptorProto get() { - const upb::MessageDef* m = upbdefs_google_protobuf_EnumValueDescriptorProto_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_EnumValueDescriptorProto_get(&m); return EnumValueDescriptorProto(m, &m); } }; -class EnumValueOptions : public upb::reffed_ptr { +class EnumValueOptions : public ::upb::reffed_ptr { public: - EnumValueOptions(const upb::MessageDef* m, const void *ref_donor = NULL) + EnumValueOptions(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_EnumValueOptions_is(m)); } static EnumValueOptions get() { - const upb::MessageDef* m = upbdefs_google_protobuf_EnumValueOptions_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_EnumValueOptions_get(&m); return EnumValueOptions(m, &m); } }; -class FieldDescriptorProto : public upb::reffed_ptr { +class FieldDescriptorProto : public ::upb::reffed_ptr { public: - FieldDescriptorProto(const upb::MessageDef* m, const void *ref_donor = NULL) + FieldDescriptorProto(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_FieldDescriptorProto_is(m)); } static FieldDescriptorProto get() { - const upb::MessageDef* m = upbdefs_google_protobuf_FieldDescriptorProto_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_FieldDescriptorProto_get(&m); return FieldDescriptorProto(m, &m); } - class Label : public upb::reffed_ptr { + class Label : public ::upb::reffed_ptr { public: - Label(const upb::EnumDef* e, const void *ref_donor = NULL) + Label(const ::upb::EnumDef* e, const void *ref_donor = NULL) : reffed_ptr(e, ref_donor) { assert(upbdefs_google_protobuf_FieldDescriptorProto_Label_is(e)); } static Label get() { - const upb::EnumDef* e = upbdefs_google_protobuf_FieldDescriptorProto_Label_get(&e); + const ::upb::EnumDef* e = upbdefs_google_protobuf_FieldDescriptorProto_Label_get(&e); return Label(e, &e); } }; - class Type : public upb::reffed_ptr { + class Type : public ::upb::reffed_ptr { public: - Type(const upb::EnumDef* e, const void *ref_donor = NULL) + Type(const ::upb::EnumDef* e, const void *ref_donor = NULL) : reffed_ptr(e, ref_donor) { assert(upbdefs_google_protobuf_FieldDescriptorProto_Type_is(e)); } static Type get() { - const upb::EnumDef* e = upbdefs_google_protobuf_FieldDescriptorProto_Type_get(&e); + const ::upb::EnumDef* e = upbdefs_google_protobuf_FieldDescriptorProto_Type_get(&e); return Type(e, &e); } }; }; -class FieldOptions : public upb::reffed_ptr { +class FieldOptions : public ::upb::reffed_ptr { public: - FieldOptions(const upb::MessageDef* m, const void *ref_donor = NULL) + FieldOptions(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_FieldOptions_is(m)); } static FieldOptions get() { - const upb::MessageDef* m = upbdefs_google_protobuf_FieldOptions_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_FieldOptions_get(&m); return FieldOptions(m, &m); } - class CType : public upb::reffed_ptr { + class CType : public ::upb::reffed_ptr { public: - CType(const upb::EnumDef* e, const void *ref_donor = NULL) + CType(const ::upb::EnumDef* e, const void *ref_donor = NULL) : reffed_ptr(e, ref_donor) { assert(upbdefs_google_protobuf_FieldOptions_CType_is(e)); } static CType get() { - const upb::EnumDef* e = upbdefs_google_protobuf_FieldOptions_CType_get(&e); + const ::upb::EnumDef* e = upbdefs_google_protobuf_FieldOptions_CType_get(&e); return CType(e, &e); } }; - class JSType : public upb::reffed_ptr { + class JSType : public ::upb::reffed_ptr { public: - JSType(const upb::EnumDef* e, const void *ref_donor = NULL) + JSType(const ::upb::EnumDef* e, const void *ref_donor = NULL) : reffed_ptr(e, ref_donor) { assert(upbdefs_google_protobuf_FieldOptions_JSType_is(e)); } static JSType get() { - const upb::EnumDef* e = upbdefs_google_protobuf_FieldOptions_JSType_get(&e); + const ::upb::EnumDef* e = upbdefs_google_protobuf_FieldOptions_JSType_get(&e); return JSType(e, &e); } }; }; -class FileDescriptorProto : public upb::reffed_ptr { +class FileDescriptorProto : public ::upb::reffed_ptr { public: - FileDescriptorProto(const upb::MessageDef* m, const void *ref_donor = NULL) + FileDescriptorProto(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_FileDescriptorProto_is(m)); } static FileDescriptorProto get() { - const upb::MessageDef* m = upbdefs_google_protobuf_FileDescriptorProto_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_FileDescriptorProto_get(&m); return FileDescriptorProto(m, &m); } }; -class FileDescriptorSet : public upb::reffed_ptr { +class FileDescriptorSet : public ::upb::reffed_ptr { public: - FileDescriptorSet(const upb::MessageDef* m, const void *ref_donor = NULL) + FileDescriptorSet(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_FileDescriptorSet_is(m)); } static FileDescriptorSet get() { - const upb::MessageDef* m = upbdefs_google_protobuf_FileDescriptorSet_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_FileDescriptorSet_get(&m); return FileDescriptorSet(m, &m); } }; -class FileOptions : public upb::reffed_ptr { +class FileOptions : public ::upb::reffed_ptr { public: - FileOptions(const upb::MessageDef* m, const void *ref_donor = NULL) + FileOptions(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_FileOptions_is(m)); } static FileOptions get() { - const upb::MessageDef* m = upbdefs_google_protobuf_FileOptions_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_FileOptions_get(&m); return FileOptions(m, &m); } - class OptimizeMode : public upb::reffed_ptr { + class OptimizeMode : public ::upb::reffed_ptr { public: - OptimizeMode(const upb::EnumDef* e, const void *ref_donor = NULL) + OptimizeMode(const ::upb::EnumDef* e, const void *ref_donor = NULL) : reffed_ptr(e, ref_donor) { assert(upbdefs_google_protobuf_FileOptions_OptimizeMode_is(e)); } static OptimizeMode get() { - const upb::EnumDef* e = upbdefs_google_protobuf_FileOptions_OptimizeMode_get(&e); + const ::upb::EnumDef* e = upbdefs_google_protobuf_FileOptions_OptimizeMode_get(&e); return OptimizeMode(e, &e); } }; }; -class MessageOptions : public upb::reffed_ptr { +class MessageOptions : public ::upb::reffed_ptr { public: - MessageOptions(const upb::MessageDef* m, const void *ref_donor = NULL) + MessageOptions(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_MessageOptions_is(m)); } static MessageOptions get() { - const upb::MessageDef* m = upbdefs_google_protobuf_MessageOptions_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_MessageOptions_get(&m); return MessageOptions(m, &m); } }; -class MethodDescriptorProto : public upb::reffed_ptr { +class MethodDescriptorProto : public ::upb::reffed_ptr { public: - MethodDescriptorProto(const upb::MessageDef* m, const void *ref_donor = NULL) + MethodDescriptorProto(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_MethodDescriptorProto_is(m)); } static MethodDescriptorProto get() { - const upb::MessageDef* m = upbdefs_google_protobuf_MethodDescriptorProto_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_MethodDescriptorProto_get(&m); return MethodDescriptorProto(m, &m); } }; -class MethodOptions : public upb::reffed_ptr { +class MethodOptions : public ::upb::reffed_ptr { public: - MethodOptions(const upb::MessageDef* m, const void *ref_donor = NULL) + MethodOptions(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_MethodOptions_is(m)); } static MethodOptions get() { - const upb::MessageDef* m = upbdefs_google_protobuf_MethodOptions_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_MethodOptions_get(&m); return MethodOptions(m, &m); } }; -class OneofDescriptorProto : public upb::reffed_ptr { +class OneofDescriptorProto : public ::upb::reffed_ptr { public: - OneofDescriptorProto(const upb::MessageDef* m, const void *ref_donor = NULL) + OneofDescriptorProto(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_OneofDescriptorProto_is(m)); } static OneofDescriptorProto get() { - const upb::MessageDef* m = upbdefs_google_protobuf_OneofDescriptorProto_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_OneofDescriptorProto_get(&m); return OneofDescriptorProto(m, &m); } }; -class ServiceDescriptorProto : public upb::reffed_ptr { +class ServiceDescriptorProto : public ::upb::reffed_ptr { public: - ServiceDescriptorProto(const upb::MessageDef* m, const void *ref_donor = NULL) + ServiceDescriptorProto(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_ServiceDescriptorProto_is(m)); } static ServiceDescriptorProto get() { - const upb::MessageDef* m = upbdefs_google_protobuf_ServiceDescriptorProto_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_ServiceDescriptorProto_get(&m); return ServiceDescriptorProto(m, &m); } }; -class ServiceOptions : public upb::reffed_ptr { +class ServiceOptions : public ::upb::reffed_ptr { public: - ServiceOptions(const upb::MessageDef* m, const void *ref_donor = NULL) + ServiceOptions(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_ServiceOptions_is(m)); } static ServiceOptions get() { - const upb::MessageDef* m = upbdefs_google_protobuf_ServiceOptions_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_ServiceOptions_get(&m); return ServiceOptions(m, &m); } }; -class SourceCodeInfo : public upb::reffed_ptr { +class SourceCodeInfo : public ::upb::reffed_ptr { public: - SourceCodeInfo(const upb::MessageDef* m, const void *ref_donor = NULL) + SourceCodeInfo(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_SourceCodeInfo_is(m)); } static SourceCodeInfo get() { - const upb::MessageDef* m = upbdefs_google_protobuf_SourceCodeInfo_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_SourceCodeInfo_get(&m); return SourceCodeInfo(m, &m); } - class Location : public upb::reffed_ptr { + class Location : public ::upb::reffed_ptr { public: - Location(const upb::MessageDef* m, const void *ref_donor = NULL) + Location(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_SourceCodeInfo_Location_is(m)); } static Location get() { - const upb::MessageDef* m = upbdefs_google_protobuf_SourceCodeInfo_Location_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_SourceCodeInfo_Location_get(&m); return Location(m, &m); } }; }; -class UninterpretedOption : public upb::reffed_ptr { +class UninterpretedOption : public ::upb::reffed_ptr { public: - UninterpretedOption(const upb::MessageDef* m, const void *ref_donor = NULL) + UninterpretedOption(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_UninterpretedOption_is(m)); } static UninterpretedOption get() { - const upb::MessageDef* m = upbdefs_google_protobuf_UninterpretedOption_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_UninterpretedOption_get(&m); return UninterpretedOption(m, &m); } - class NamePart : public upb::reffed_ptr { + class NamePart : public ::upb::reffed_ptr { public: - NamePart(const upb::MessageDef* m, const void *ref_donor = NULL) + NamePart(const ::upb::MessageDef* m, const void *ref_donor = NULL) : reffed_ptr(m, ref_donor) { assert(upbdefs_google_protobuf_UninterpretedOption_NamePart_is(m)); } static NamePart get() { - const upb::MessageDef* m = upbdefs_google_protobuf_UninterpretedOption_NamePart_get(&m); + const ::upb::MessageDef* m = upbdefs_google_protobuf_UninterpretedOption_NamePart_get(&m); return NamePart(m, &m); } }; @@ -7178,7 +7327,6 @@ class UninterpretedOption : public upb::reffed_ptr { #ifndef UPB_DECODER_INT_H_ #define UPB_DECODER_INT_H_ -#include /* ** upb::pb::Decoder ** @@ -7278,7 +7426,7 @@ class upb::pb::DecoderMethod { * constructed. This hint may be an overestimate for some build configurations. * But if the decoder library is upgraded without recompiling the application, * it may be an underestimate. */ -#define UPB_PB_DECODER_SIZE 4408 +#define UPB_PB_DECODER_SIZE 4416 #ifdef __cplusplus @@ -8097,7 +8245,8 @@ extern "C" { #endif /* Loads a binary descriptor and returns a NULL-terminated array of unfrozen - * filedefs. The caller owns the returned array. */ + * filedefs. The caller owns the returned array, which must be freed with + * upb_gfree(). */ upb_filedef **upb_loaddescriptor(const char *buf, size_t n, const void *owner, upb_status *status); @@ -8244,7 +8393,7 @@ UPB_DECLARE_DERIVED_TYPE(upb::json::ParserMethod, upb::RefCounted, * constructed. This hint may be an overestimate for some build configurations. * But if the parser library is upgraded without recompiling the application, * it may be an underestimate. */ -#define UPB_JSON_PARSER_SIZE 4104 +#define UPB_JSON_PARSER_SIZE 4112 #ifdef __cplusplus @@ -8357,7 +8506,7 @@ UPB_DECLARE_TYPE(upb::json::Printer, upb_json_printer) /* upb::json::Printer *********************************************************/ -#define UPB_JSON_PRINTER_SIZE 168 +#define UPB_JSON_PRINTER_SIZE 176 #ifdef __cplusplus From 18b6a321b5fd8f53e1029c436c2ad5dd2d3122df Mon Sep 17 00:00:00 2001 From: Thomas Van Lenten Date: Tue, 26 Apr 2016 14:40:11 -0400 Subject: [PATCH 6/8] Proper checking of enum with non zero default proto2 syntax allows the first enum to have a non zero value. This means any field using that default has a non zero default without having an explicit default being set. So when deciding what runtime info is needed, don't rely on an explicit default, always check that the values aren't zero. Fixes https://github.com/google/protobuf/issues/1453 --- objectivec/Tests/GPBMessageTests.m | 7 +++++++ objectivec/Tests/unittest_objc.proto | 10 ++++++++++ .../protobuf/compiler/objectivec/objectivec_helpers.cc | 9 +++++---- 3 files changed, 22 insertions(+), 4 deletions(-) diff --git a/objectivec/Tests/GPBMessageTests.m b/objectivec/Tests/GPBMessageTests.m index 43546156c1..0779c5ae76 100644 --- a/objectivec/Tests/GPBMessageTests.m +++ b/objectivec/Tests/GPBMessageTests.m @@ -1947,4 +1947,11 @@ EnumTestMsg_MyEnum_NegTwo); } +- (void)testOneBasedEnumHolder { + // Test case for https://github.com/google/protobuf/issues/1453 + // Message with no explicit defaults, but a non zero default for an enum. + MessageWithOneBasedEnum *enumMsg = [MessageWithOneBasedEnum message]; + XCTAssertEqual(enumMsg.enumField, MessageWithOneBasedEnum_OneBasedEnum_One); +} + @end diff --git a/objectivec/Tests/unittest_objc.proto b/objectivec/Tests/unittest_objc.proto index 9483cb1d7f..6bcfbf705a 100644 --- a/objectivec/Tests/unittest_objc.proto +++ b/objectivec/Tests/unittest_objc.proto @@ -401,3 +401,13 @@ message EnumTestMsg { repeated MyEnum mumble = 4; } + +// Test case for https://github.com/google/protobuf/issues/1453 +// Message with no explicit defaults, but a non zero default for an enum. +message MessageWithOneBasedEnum { + enum OneBasedEnum { + ONE = 1; + TWO = 2; + } + optional OneBasedEnum enum_field = 1; +} diff --git a/src/google/protobuf/compiler/objectivec/objectivec_helpers.cc b/src/google/protobuf/compiler/objectivec/objectivec_helpers.cc index fda5180712..196b39dd93 100644 --- a/src/google/protobuf/compiler/objectivec/objectivec_helpers.cc +++ b/src/google/protobuf/compiler/objectivec/objectivec_helpers.cc @@ -757,10 +757,11 @@ bool HasNonZeroDefaultValue(const FieldDescriptor* field) { return false; } - if (!field->has_default_value()) { - // No custom default set in the proto file. - return false; - } + // As much as checking field->has_default_value() seems useful, it isn't + // because of enums. proto2 syntax allows the first item in an enum (the + // default) to be non zero. So checking field->has_default_value() would + // result in missing this non zero default. See MessageWithOneBasedEnum in + // objectivec/Tests/unittest_objc.proto for a test Message to confirm this. // Some proto file set the default to the zero value, so make sure the value // isn't the zero case. From bbb68fe63d0c77d19cb7ecccbfb77c8bdccf751c Mon Sep 17 00:00:00 2001 From: Adam Cozzette Date: Wed, 27 Apr 2016 10:40:54 -0700 Subject: [PATCH 7/8] Added dig and bsearch_index to RepeatedField methods forwarded to array This fixes the test_acts_likes_an_array test in RepeatedFieldTest, which checks that repeated fields respond to the same methods as regular Ruby arrays. The bsearch_index and dig array methods seem to be new in Ruby 2.3 and so we should support those. --- ruby/lib/google/protobuf/repeated_field.rb | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/ruby/lib/google/protobuf/repeated_field.rb b/ruby/lib/google/protobuf/repeated_field.rb index 16c843c07f..0ad2060d43 100644 --- a/ruby/lib/google/protobuf/repeated_field.rb +++ b/ruby/lib/google/protobuf/repeated_field.rb @@ -69,8 +69,8 @@ module Google # relationship explicit instead of implicit def_delegators :to_ary, :&, :*, :-, :'<=>', - :assoc, :bsearch, :combination, :compact, :count, :cycle, - :drop, :drop_while, :eql?, :fetch, :find_index, :flatten, + :assoc, :bsearch, :bsearch_index, :combination, :compact, :count, + :cycle, :dig, :drop, :drop_while, :eql?, :fetch, :find_index, :flatten, :include?, :index, :inspect, :join, :pack, :permutation, :product, :pretty_print, :pretty_print_cycle, :rassoc, :repeated_combination, :repeated_permutation, :reverse, From 30646288ad60fff7a3ce9c864a9bf481e6c4045a Mon Sep 17 00:00:00 2001 From: Thomas Van Lenten Date: Wed, 27 Apr 2016 13:11:16 -0400 Subject: [PATCH 8/8] Fix up -hash/-isEqual: for bool storage. Both methods weren't checking the has_bits (where the bools are stored), so it resulted in invalid results. Add a test that should shake out something like this in the future also. --- objectivec/GPBMessage.m | 18 +++++--- objectivec/Tests/GPBMessageTests.m | 65 ++++++++++++++++++++++++++++ objectivec/Tests/unittest_objc.proto | 36 +++++++++++++++ 3 files changed, 114 insertions(+), 5 deletions(-) diff --git a/objectivec/GPBMessage.m b/objectivec/GPBMessage.m index 0e1599dcc2..8134e25962 100644 --- a/objectivec/GPBMessage.m +++ b/objectivec/GPBMessage.m @@ -2603,9 +2603,13 @@ static void MergeRepeatedNotPackedFieldFromCodedInputStream( size_t fieldOffset = field->description_->offset; switch (fieldDataType) { case GPBDataTypeBool: { - BOOL *selfValPtr = (BOOL *)&selfStorage[fieldOffset]; - BOOL *otherValPtr = (BOOL *)&otherStorage[fieldOffset]; - if (*selfValPtr != *otherValPtr) { + // Bools are stored in has_bits to avoid needing explicit space in + // the storage structure. + // (the field number passed to the HasIvar helper doesn't really + // matter since the offset is never negative) + BOOL selfValue = GPBGetHasIvar(self, (int32_t)(fieldOffset), 0); + BOOL otherValue = GPBGetHasIvar(other, (int32_t)(fieldOffset), 0); + if (selfValue != otherValue) { return NO; } break; @@ -2714,8 +2718,12 @@ static void MergeRepeatedNotPackedFieldFromCodedInputStream( size_t fieldOffset = field->description_->offset; switch (fieldDataType) { case GPBDataTypeBool: { - BOOL *valPtr = (BOOL *)&storage[fieldOffset]; - result = prime * result + *valPtr; + // Bools are stored in has_bits to avoid needing explicit space in + // the storage structure. + // (the field number passed to the HasIvar helper doesn't really + // matter since the offset is never negative) + BOOL value = GPBGetHasIvar(self, (int32_t)(fieldOffset), 0); + result = prime * result + value; break; } case GPBDataTypeSFixed32: diff --git a/objectivec/Tests/GPBMessageTests.m b/objectivec/Tests/GPBMessageTests.m index 0779c5ae76..89d1fce2c3 100644 --- a/objectivec/Tests/GPBMessageTests.m +++ b/objectivec/Tests/GPBMessageTests.m @@ -1954,4 +1954,69 @@ XCTAssertEqual(enumMsg.enumField, MessageWithOneBasedEnum_OneBasedEnum_One); } +- (void)testBoolOffsetUsage { + // Bools use storage within has_bits; this test ensures that this is honored + // in all places where things should crash or fail based on reading out of + // field storage instead. + BoolOnlyMessage *msg1 = [BoolOnlyMessage message]; + BoolOnlyMessage *msg2 = [BoolOnlyMessage message]; + + msg1.boolField1 = YES; + msg2.boolField1 = YES; + msg1.boolField3 = YES; + msg2.boolField3 = YES; + msg1.boolField5 = YES; + msg2.boolField5 = YES; + msg1.boolField7 = YES; + msg2.boolField7 = YES; + msg1.boolField9 = YES; + msg2.boolField9 = YES; + msg1.boolField11 = YES; + msg2.boolField11 = YES; + msg1.boolField13 = YES; + msg2.boolField13 = YES; + msg1.boolField15 = YES; + msg2.boolField15 = YES; + msg1.boolField17 = YES; + msg2.boolField17 = YES; + msg1.boolField19 = YES; + msg2.boolField19 = YES; + msg1.boolField21 = YES; + msg2.boolField21 = YES; + msg1.boolField23 = YES; + msg2.boolField23 = YES; + msg1.boolField25 = YES; + msg2.boolField25 = YES; + msg1.boolField27 = YES; + msg2.boolField27 = YES; + msg1.boolField29 = YES; + msg2.boolField29 = YES; + msg1.boolField31 = YES; + msg2.boolField31 = YES; + + msg1.boolField32 = YES; + msg2.boolField32 = YES; + + XCTAssertTrue(msg1 != msg2); // Different pointers. + XCTAssertEqual([msg1 hash], [msg2 hash]); + XCTAssertEqualObjects(msg1, msg2); + + BoolOnlyMessage *msg1Prime = [[msg1 copy] autorelease]; + XCTAssertTrue(msg1Prime != msg1); // Different pointers. + XCTAssertEqual([msg1 hash], [msg1Prime hash]); + XCTAssertEqualObjects(msg1, msg1Prime); + + // Field set in one, but not the other means they don't match (even if + // set to default value). + msg1Prime.boolField2 = NO; + XCTAssertNotEqualObjects(msg1Prime, msg1); + // And when set to different values. + msg1.boolField2 = YES; + XCTAssertNotEqualObjects(msg1Prime, msg1); + // And then they match again. + msg1.boolField2 = NO; + XCTAssertEqualObjects(msg1Prime, msg1); + XCTAssertEqual([msg1 hash], [msg1Prime hash]); +} + @end diff --git a/objectivec/Tests/unittest_objc.proto b/objectivec/Tests/unittest_objc.proto index 6bcfbf705a..f6ab6a24ca 100644 --- a/objectivec/Tests/unittest_objc.proto +++ b/objectivec/Tests/unittest_objc.proto @@ -411,3 +411,39 @@ message MessageWithOneBasedEnum { } optional OneBasedEnum enum_field = 1; } + +// Message with all bools for testing things related to bool storage. +message BoolOnlyMessage { + optional bool bool_field_1 = 1; + optional bool bool_field_2 = 2; + optional bool bool_field_3 = 3; + optional bool bool_field_4 = 4; + optional bool bool_field_5 = 5; + optional bool bool_field_6 = 6; + optional bool bool_field_7 = 7; + optional bool bool_field_8 = 8; + optional bool bool_field_9 = 9; + optional bool bool_field_10 = 10; + optional bool bool_field_11 = 11; + optional bool bool_field_12 = 12; + optional bool bool_field_13 = 13; + optional bool bool_field_14 = 14; + optional bool bool_field_15 = 15; + optional bool bool_field_16 = 16; + optional bool bool_field_17 = 17; + optional bool bool_field_18 = 18; + optional bool bool_field_19 = 19; + optional bool bool_field_20 = 20; + optional bool bool_field_21 = 21; + optional bool bool_field_22 = 22; + optional bool bool_field_23 = 23; + optional bool bool_field_24 = 24; + optional bool bool_field_25 = 25; + optional bool bool_field_26 = 26; + optional bool bool_field_27 = 27; + optional bool bool_field_28 = 28; + optional bool bool_field_29 = 29; + optional bool bool_field_30 = 30; + optional bool bool_field_31 = 31; + optional bool bool_field_32 = 32; +}