Merge branch 'master' into optdefload

pull/13171/head
Joshua Haberman 4 years ago
commit 15e20636c3
  1. 652
      BUILD
  2. 72
      bazel/build_defs.bzl
  3. 10
      bazel/lua.BUILD
  4. 6
      bazel/ragel.BUILD
  5. 31
      bazel/upb_proto_library.bzl
  6. 3
      bazel/workspace_deps.bzl
  7. 53
      benchmarks/BUILD
  8. 142
      benchmarks/benchmark.cc
  9. 10
      benchmarks/compare.py
  10. 909
      benchmarks/descriptor.proto
  11. 894
      benchmarks/descriptor_sv.proto
  12. 89
      cmake/BUILD
  13. 109
      cmake/CMakeLists.txt
  14. 23
      cmake/README.md
  15. 44
      cmake/build_defs.bzl
  16. 0
      cmake/google/protobuf/descriptor.upb.c
  17. 0
      cmake/google/protobuf/descriptor.upb.h
  18. 18
      cmake/make_cmakelists.py
  19. 2
      cmake/staleness_test.py
  20. 13
      cmake/staleness_test_lib.py
  21. 2
      cmake/upb/json/parser.c
  22. 3
      examples/bazel/BUILD
  23. 2
      examples/bazel/test_binary.c
  24. 6
      kokoro/ubuntu/build.sh
  25. 283
      tests/BUILD
  26. 62
      tests/bindings/lua/BUILD
  27. 2
      tests/bindings/lua/main.c
  28. 8
      tests/bindings/lua/test.proto
  29. 2
      tests/bindings/lua/test_upb.lua
  30. 26
      tests/pb/test_decoder.cc
  31. 57
      tests/pb/test_encoder.cc
  32. 126
      tests/pb/test_varint.c
  33. 25
      tests/test_cpp.cc
  34. 14
      tests/test_generated_code.c
  35. 40
      upb/bindings/lua/BUILD
  36. 12
      upb/bindings/lua/def.c
  37. 2
      upb/bindings/lua/lua_proto_library.bzl
  38. 2
      upb/bindings/lua/msg.c
  39. 69
      upb/bindings/stdc++/string.h
  40. 228
      upb/decode.c
  41. 2
      upb/json/parser.rl
  42. 30
      upb/json/printer.c
  43. 12
      upb/json_decode.c
  44. 10
      upb/json_encode.c
  45. 4
      upb/pb/textprinter.c
  46. 4
      upb/pb/varint.int.h
  47. 75
      upb/port_def.inc
  48. 7
      upb/port_undef.inc
  49. 2
      upb/text_encode.c
  50. 4
      upb/upb.c
  51. 8
      upb/upb.h
  52. 8
      upb/upb.hpp
  53. 35
      upbc/BUILD

652
BUILD

@ -1,13 +1,12 @@
load(
"//bazel:build_defs.bzl",
"generated_file_staleness_test",
"licenses", # copybara:strip_for_google3
"make_shell_script",
"upb_amalgamation",
"UPB_DEFAULT_COPTS",
"upb_amalgamation", # copybara:strip_for_google3
)
load(
"//bazel:upb_proto_library.bzl",
"upb_proto_library",
"upb_proto_library_copts",
"upb_proto_reflection_library",
)
@ -16,10 +15,7 @@ load(
"@rules_proto//proto:defs.bzl",
"proto_library",
)
load(
"//:upb/bindings/lua/lua_proto_library.bzl",
"lua_proto_library",
)
# copybara:strip_end
licenses(["notice"]) # BSD (Google-authored w/ possible external contributions)
@ -29,23 +25,6 @@ exports_files([
"build_defs",
])
CPPOPTS = [
# copybara:strip_for_google3_begin
"-Wextra",
# "-Wshorten-64-to-32", # not in GCC (and my Kokoro images doesn't have Clang)
"-Werror",
"-Wno-long-long",
# copybara:strip_end
]
COPTS = CPPOPTS + [
# copybara:strip_for_google3_begin
"-pedantic",
"-Werror=pedantic",
"-Wstrict-prototypes",
# copybara:strip_end
]
config_setting(
name = "darwin",
values = {"cpu": "darwin"},
@ -57,9 +36,10 @@ config_setting(
constraint_values = ["@bazel_tools//platforms:windows"],
)
config_setting(
name = "fuzz",
values = {"define": "fuzz=true"},
upb_proto_library_copts(
name = "upb_proto_library_copts__for_generated_code_only_do_not_use",
copts = UPB_DEFAULT_COPTS,
visibility = ["//visibility:public"],
)
# Public C/C++ libraries #######################################################
@ -69,10 +49,12 @@ cc_library(
srcs = [
"upb/port.c",
],
copts = UPB_DEFAULT_COPTS,
textual_hdrs = [
"upb/port_def.inc",
"upb/port_undef.inc",
],
visibility = ["//tests:__pkg__"],
)
cc_library(
@ -94,10 +76,7 @@ cc_library(
"upb/upb.h",
"upb/upb.hpp",
],
copts = select({
":windows": [],
"//conditions:default": COPTS,
}),
copts = UPB_DEFAULT_COPTS,
visibility = ["//visibility:public"],
deps = [":port"],
)
@ -115,10 +94,7 @@ cc_library(
"upb/port_def.inc",
"upb/port_undef.inc",
],
copts = select({
":windows": [],
"//conditions:default": COPTS,
}),
copts = UPB_DEFAULT_COPTS,
visibility = ["//visibility:public"],
deps = [
":table",
@ -150,10 +126,7 @@ cc_library(
"upb/def.hpp",
"upb/reflection.h",
],
copts = select({
":windows": [],
"//conditions:default": COPTS,
}),
copts = UPB_DEFAULT_COPTS,
visibility = ["//visibility:public"],
deps = [
":descriptor_upb_proto",
@ -171,10 +144,7 @@ cc_library(
hdrs = [
"upb/text_encode.h",
],
copts = select({
":windows": [],
"//conditions:default": COPTS,
}),
copts = UPB_DEFAULT_COPTS,
visibility = ["//visibility:public"],
deps = [
":port",
@ -192,10 +162,8 @@ cc_library(
"upb/json_decode.h",
"upb/json_encode.h",
],
copts = select({
":windows": [],
"//conditions:default": COPTS,
}),
copts = UPB_DEFAULT_COPTS,
visibility = ["//tests:__pkg__"],
deps = [
":port",
":reflection",
@ -208,6 +176,7 @@ cc_library(
cc_library(
name = "table",
hdrs = ["upb/table.int.h"],
visibility = ["//tests:__pkg__"],
deps = [
":port",
":upb",
@ -227,10 +196,8 @@ cc_library(
"upb/handlers.h",
"upb/sink.h",
],
copts = select({
":windows": [],
"//conditions:default": COPTS,
}),
copts = UPB_DEFAULT_COPTS,
visibility = ["//tests:__pkg__"],
deps = [
":port",
":reflection",
@ -255,10 +222,8 @@ cc_library(
"upb/pb/encoder.h",
"upb/pb/textprinter.h",
],
copts = select({
":windows": [],
"//conditions:default": COPTS,
}),
copts = UPB_DEFAULT_COPTS,
visibility = ["//tests:__pkg__"],
deps = [
":descriptor_upb_proto",
":handlers",
@ -280,401 +245,23 @@ cc_library(
"upb/json/parser.h",
"upb/json/printer.h",
],
copts = select({
":windows": [],
"//conditions:default": COPTS,
}),
deps = [
":upb",
":upb_pb",
],
)
# copybara:strip_end
cc_library(
name = "upb_cc_bindings",
hdrs = [
"upb/bindings/stdc++/string.h",
],
deps = [
":descriptor_upb_proto",
":handlers",
":port",
":upb",
],
)
# upb compiler #################################################################
cc_library(
name = "upbc_generator",
srcs = [
"upbc/generator.cc",
"upbc/message_layout.cc",
"upbc/message_layout.h",
],
hdrs = ["upbc/generator.h"],
copts = select({
":windows": [],
"//conditions:default": CPPOPTS,
}),
deps = [
"@com_google_absl//absl/base:core_headers",
"@com_google_absl//absl/container:flat_hash_map",
"@com_google_absl//absl/strings",
"@com_google_protobuf//:protobuf",
"@com_google_protobuf//:protoc_lib",
],
)
cc_binary(
name = "protoc-gen-upb",
srcs = ["upbc/main.cc"],
copts = select({
":windows": [],
"//conditions:default": CPPOPTS,
}),
visibility = ["//visibility:public"],
deps = [
":upbc_generator",
"@com_google_protobuf//:protoc_lib",
],
)
# We strip the tests and remaining rules from google3 until the upb_proto_library()
# and upb_proto_reflection_library() rules are fixed.
# C/C++ tests ##################################################################
upb_proto_reflection_library(
name = "descriptor_upbreflection",
deps = ["@com_google_protobuf//:descriptor_proto"],
)
cc_binary(
name = "benchmark",
testonly = 1,
srcs = ["tests/benchmark.cc"],
deps = [
":descriptor_upb_proto",
":descriptor_upbreflection",
"@com_github_google_benchmark//:benchmark_main",
"@com_google_protobuf//:protobuf",
],
)
cc_library(
name = "upb_test",
testonly = 1,
srcs = [
"tests/testmain.cc",
],
hdrs = [
"tests/test_util.h",
"tests/upb_test.h",
],
copts = select({
":windows": [],
"//conditions:default": CPPOPTS,
}),
deps = [
":handlers",
":port",
":upb",
],
)
cc_test(
name = "test_varint",
srcs = [
"tests/pb/test_varint.c",
"upb/pb/varint.int.h",
],
copts = select({
":windows": [],
"//conditions:default": COPTS,
}),
deps = [
":port",
":upb",
":upb_pb",
":upb_test",
],
)
proto_library(
name = "test_proto",
testonly = 1,
srcs = ["tests/test.proto"],
)
upb_proto_library(
name = "test_upb_proto",
testonly = 1,
deps = [":test_proto"],
)
cc_test(
name = "test_generated_code",
srcs = ["tests/test_generated_code.c"],
copts = UPB_DEFAULT_COPTS,
visibility = ["//tests:__pkg__"],
deps = [
":test_messages_proto3_proto_upb",
":empty_upbdefs_proto",
":test_upb_proto",
":upb_test",
],
)
proto_library(
name = "empty_proto",
srcs = ["tests/empty.proto"],
)
upb_proto_reflection_library(
name = "empty_upbdefs_proto",
testonly = 1,
deps = [":empty_proto"],
)
upb_proto_library(
name = "test_messages_proto3_proto_upb",
testonly = 1,
deps = ["@com_google_protobuf//:test_messages_proto3_proto"],
)
proto_library(
name = "test_decoder_proto",
srcs = [
"tests/pb/test_decoder.proto",
],
)
upb_proto_reflection_library(
name = "test_decoder_upb_proto",
deps = [":test_decoder_proto"],
)
cc_test(
name = "test_decoder",
srcs = [
"tests/pb/test_decoder.cc",
"upb/pb/varint.int.h",
],
copts = select({
":windows": [],
"//conditions:default": CPPOPTS,
}),
deps = [
":handlers",
":port",
":test_decoder_upb_proto",
":upb",
":upb_pb",
":upb_test",
],
)
proto_library(
name = "test_cpp_proto",
srcs = [
"tests/test_cpp.proto",
],
)
upb_proto_reflection_library(
name = "test_cpp_upb_proto",
deps = ["test_cpp_proto"],
)
cc_test(
name = "test_cpp",
srcs = ["tests/test_cpp.cc"],
copts = select({
":windows": [],
"//conditions:default": CPPOPTS,
}),
deps = [
":handlers",
":port",
":reflection",
":test_cpp_upb_proto",
":upb",
":upb_pb",
":upb_test",
],
)
cc_test(
name = "test_table",
srcs = ["tests/test_table.cc"],
copts = select({
":windows": [],
"//conditions:default": CPPOPTS,
}),
deps = [
":port",
":table",
":upb",
":upb_test",
],
)
# OSS-Fuzz test
cc_binary(
name = "file_descriptor_parsenew_fuzzer",
testonly = 1,
srcs = ["tests/file_descriptor_parsenew_fuzzer.cc"],
copts = select({
":windows": [],
"//conditions:default": CPPOPTS,
}) + select({
"//conditions:default": [],
":fuzz": ["-fsanitize=fuzzer,address"],
}),
defines = select({
"//conditions:default": [],
":fuzz": ["HAVE_FUZZER"],
}),
deps = [
":descriptor_upb_proto",
":upb",
],
)
# copybara:strip_for_google3_begin
cc_test(
name = "test_encoder",
srcs = ["tests/pb/test_encoder.cc"],
copts = select({
":windows": [],
"//conditions:default": CPPOPTS,
}),
deps = [
":descriptor_upb_proto",
":descriptor_upbreflection",
":upb",
":upb_cc_bindings",
":upb_pb",
":upb_test",
],
)
proto_library(
name = "test_json_enum_from_separate",
srcs = ["tests/json/enum_from_separate_file.proto"],
deps = [":test_json_proto"],
)
proto_library(
name = "test_json_proto",
srcs = ["tests/json/test.proto"],
)
upb_proto_reflection_library(
name = "test_json_upb_proto_reflection",
deps = ["test_json_proto"],
)
upb_proto_library(
name = "test_json_enum_from_separate_upb_proto",
deps = [":test_json_enum_from_separate"],
)
upb_proto_library(
name = "test_json_upb_proto",
deps = [":test_json_proto"],
)
cc_test(
name = "test_json",
srcs = [
"tests/json/test_json.cc",
],
copts = select({
":windows": [],
"//conditions:default": CPPOPTS,
}),
deps = [
":test_json_upb_proto",
":test_json_upb_proto_reflection",
":upb_json",
":upb_test",
],
)
# copybara:strip_end
upb_proto_library(
name = "conformance_proto_upb",
testonly = 1,
deps = ["@com_google_protobuf//:conformance_proto"],
)
upb_proto_reflection_library(
name = "conformance_proto_upbdefs",
testonly = 1,
deps = ["@com_google_protobuf//:conformance_proto"],
)
upb_proto_reflection_library(
name = "test_messages_proto2_upbdefs",
testonly = 1,
deps = ["@com_google_protobuf//:test_messages_proto2_proto"],
)
upb_proto_reflection_library(
name = "test_messages_proto3_upbdefs",
testonly = 1,
deps = ["@com_google_protobuf//:test_messages_proto3_proto"],
)
cc_binary(
name = "conformance_upb",
testonly = 1,
srcs = [
"tests/conformance_upb.c",
],
data = [
"tests/conformance_upb_failures.txt",
],
copts = select({
":windows": [],
"//conditions:default": COPTS,
}) + ["-Ibazel-out/k8-fastbuild/bin"],
deps = [
":port",
":conformance_proto_upb",
":conformance_proto_upbdefs",
":json",
":reflection",
":test_messages_proto2_upbdefs",
":test_messages_proto3_upbdefs",
":textformat",
":upb",
],
)
make_shell_script(
name = "gen_test_conformance_upb",
out = "test_conformance_upb.sh",
contents = "external/com_google_protobuf/conformance_test_runner " +
" --enforce_recommended " +
" --failure_list ./tests/conformance_upb_failures.txt" +
" ./conformance_upb",
)
sh_test(
name = "test_conformance_upb",
srcs = ["test_conformance_upb.sh"],
data = [
"tests/conformance_upb_failures.txt",
":conformance_upb",
"@com_google_protobuf//:conformance_test_runner",
],
deps = ["@bazel_tools//tools/bash/runfiles"],
genrule(
name = "generate_json_ragel",
srcs = ["//:upb/json/parser.rl"],
outs = ["upb/json/parser.c"],
cmd = "$(location @ragel//:ragelc) -C -o upb/json/parser.c $< && mv upb/json/parser.c $@",
tools = ["@ragel//:ragelc"],
visibility = ["//cmake:__pkg__"],
)
# copybara:strip_for_google3_begin
# Amalgamation #################################################################
py_binary(
@ -704,15 +291,11 @@ cc_library(
name = "amalgamation",
srcs = ["upb.c"],
hdrs = ["upb.h"],
copts = select({
":windows": [],
"//conditions:default": COPTS,
}),
copts = UPB_DEFAULT_COPTS,
)
upb_amalgamation(
name = "gen_php_amalgamation",
prefix = "php-",
outs = [
"php-upb.c",
"php-upb.h",
@ -726,21 +309,18 @@ upb_amalgamation(
":port",
":json",
],
prefix = "php-",
)
cc_library(
name = "php_amalgamation",
srcs = ["php-upb.c"],
hdrs = ["php-upb.h"],
copts = select({
":windows": [],
"//conditions:default": COPTS,
}),
copts = UPB_DEFAULT_COPTS,
)
upb_amalgamation(
name = "gen_ruby_amalgamation",
prefix = "ruby-",
outs = [
"ruby-upb.c",
"ruby-upb.h",
@ -753,102 +333,37 @@ upb_amalgamation(
":port",
":json",
],
prefix = "ruby-",
)
cc_library(
name = "ruby_amalgamation",
srcs = ["ruby-upb.c"],
hdrs = ["ruby-upb.h"],
copts = select({
":windows": [],
"//conditions:default": COPTS,
}),
copts = UPB_DEFAULT_COPTS,
)
# Lua ##########################################################################
cc_library(
name = "lupb",
srcs = [
"upb/bindings/lua/def.c",
"upb/bindings/lua/msg.c",
"upb/bindings/lua/upb.c",
],
hdrs = [
"upb/bindings/lua/upb.h",
],
deps = [
":reflection",
":textformat",
":upb",
"@lua//:liblua",
exports_files(
[
"upb/json/parser.rl",
"BUILD",
"WORKSPACE",
],
visibility = ["//cmake:__pkg__"],
)
cc_test(
name = "test_lua",
srcs = ["tests/bindings/lua/main.c"],
data = [
"tests/bindings/lua/test_upb.lua",
exports_files(
[
"third_party/lunit/console.lua",
"third_party/lunit/lunit.lua",
"upb/bindings/lua/upb.lua",
":descriptor_proto_lua",
":test_messages_proto3_proto_lua",
":test_messages_proto2_proto_lua",
":test_proto_lua",
"@com_google_protobuf//:conformance_proto",
"@com_google_protobuf//:descriptor_proto",
],
linkstatic = 1,
deps = [
":lupb",
"@lua//:liblua",
],
)
cc_binary(
name = "protoc-gen-lua",
srcs = ["upb/bindings/lua/upbc.cc"],
copts = select({
":windows": [],
"//conditions:default": CPPOPTS,
}),
visibility = ["//visibility:public"],
deps = [
"@com_google_absl//absl/strings",
"@com_google_protobuf//:protoc_lib",
],
visibility = ["//tests/bindings/lua:__pkg__"],
)
lua_proto_library(
name = "test_proto_lua",
testonly = 1,
deps = [":test_proto"],
)
lua_proto_library(
name = "descriptor_proto_lua",
deps = ["@com_google_protobuf//:descriptor_proto"],
)
lua_proto_library(
name = "test_messages_proto3_proto_lua",
testonly = 1,
deps = ["@com_google_protobuf//:test_messages_proto3_proto"],
)
lua_proto_library(
name = "test_messages_proto2_proto_lua",
testonly = 1,
deps = ["@com_google_protobuf//:test_messages_proto2_proto"],
)
# Test the CMake build #########################################################
filegroup(
name = "cmake_files",
srcs = glob([
"upb/json/parser.c",
"CMakeLists.txt",
"generated_for_cmake/**/*",
"google/**/*",
@ -857,82 +372,7 @@ filegroup(
"tests/**/*",
"third_party/**/*",
]),
)
make_shell_script(
name = "gen_run_cmake_build",
out = "run_cmake_build.sh",
contents = "find . && mkdir build && cd build && cmake .. && make -j8 && make test",
)
sh_test(
name = "cmake_build",
srcs = ["run_cmake_build.sh"],
data = [":cmake_files"],
deps = ["@bazel_tools//tools/bash/runfiles"],
)
# Generated files ##############################################################
exports_files(["tools/staleness_test.py"])
py_library(
name = "staleness_test_lib",
testonly = 1,
srcs = ["tools/staleness_test_lib.py"],
)
py_binary(
name = "make_cmakelists",
srcs = ["tools/make_cmakelists.py"],
)
genrule(
name = "gen_cmakelists",
srcs = [
"BUILD",
"WORKSPACE",
":cmake_files",
],
outs = ["generated-in/CMakeLists.txt"],
cmd = "$(location :make_cmakelists) $@",
tools = [":make_cmakelists"],
)
genrule(
name = "generate_json_ragel",
srcs = ["upb/json/parser.rl"],
outs = ["upb/json/parser.c"],
cmd = "$(location @ragel//:ragelc) -C -o upb/json/parser.c $< && mv upb/json/parser.c $@",
tools = ["@ragel//:ragelc"],
)
genrule(
name = "copy_json_ragel",
srcs = ["upb/json/parser.c"],
outs = ["generated-in/generated_for_cmake/upb/json/parser.c"],
cmd = "cp $< $@",
)
genrule(
name = "copy_protos",
srcs = [":descriptor_upb_proto"],
outs = [
"generated-in/generated_for_cmake/google/protobuf/descriptor.upb.c",
"generated-in/generated_for_cmake/google/protobuf/descriptor.upb.h",
],
cmd = "cp $(SRCS) $(@D)/generated-in/generated_for_cmake/google/protobuf",
)
generated_file_staleness_test(
name = "test_generated_files",
outs = [
"CMakeLists.txt",
"generated_for_cmake/google/protobuf/descriptor.upb.c",
"generated_for_cmake/google/protobuf/descriptor.upb.h",
"generated_for_cmake/upb/json/parser.c",
],
generated_pattern = "generated-in/%s",
visibility = ["//cmake:__pkg__"],
)
# copybara:strip_end

@ -2,6 +2,30 @@
load(":upb_proto_library.bzl", "GeneratedSrcsInfo")
UPB_DEFAULT_CPPOPTS = select({
"//:windows": [],
"//conditions:default": [
# copybara:strip_for_google3_begin
"-Wextra",
# "-Wshorten-64-to-32", # not in GCC (and my Kokoro images doesn't have Clang)
"-Werror",
"-Wno-long-long",
# copybara:strip_end
],
})
UPB_DEFAULT_COPTS = select({
"//:windows": [],
"//conditions:default": [
# copybara:strip_for_google3_begin
"-std=c99",
"-pedantic",
"-Werror=pedantic",
"-Wstrict-prototypes",
# copybara:strip_end
],
})
def _librule(name):
return name + "_lib"
@ -58,50 +82,6 @@ def make_shell_script(name, contents, out):
cmd = "(cat <<'HEREDOC'\n%s\nHEREDOC\n) > $@" % contents,
)
def generated_file_staleness_test(name, outs, generated_pattern):
"""Tests that checked-in file(s) match the contents of generated file(s).
The resulting test will verify that all output files exist and have the
correct contents. If the test fails, it can be invoked with --fix to
bring the checked-in files up to date.
Args:
name: Name of the rule.
outs: the checked-in files that are copied from generated files.
generated_pattern: the pattern for transforming each "out" file into a
generated file. For example, if generated_pattern="generated/%s" then
a file foo.txt will look for generated file generated/foo.txt.
"""
script_name = name + ".py"
script_src = "//:tools/staleness_test.py"
# Filter out non-existing rules so Blaze doesn't error out before we even
# run the test.
existing_outs = native.glob(include = outs)
# The file list contains a few extra bits of information at the end.
# These get unpacked by the Config class in staleness_test_lib.py.
file_list = outs + [generated_pattern, native.package_name() or ".", name]
native.genrule(
name = name + "_makescript",
outs = [script_name],
srcs = [script_src],
testonly = 1,
cmd = "cat $(location " + script_src + ") > $@; " +
"sed -i.bak -e 's|INSERT_FILE_LIST_HERE|" + "\\\n ".join(file_list) + "|' $@",
)
native.py_test(
name = name,
srcs = [script_name],
data = existing_outs + [generated_pattern % file for file in outs],
deps = [
"//:staleness_test_lib",
],
)
# upb_amalgamation() rule, with file_list aspect.
SrcList = provider(
@ -156,7 +136,3 @@ upb_amalgamation = rule(
},
implementation = _upb_amalgamation,
)
def licenses(*args):
# No-op (for Google-internal usage).
pass

@ -4,7 +4,6 @@ package(
cc_library(
name = "liblua_headers",
defines = ["LUA_USE_LINUX"],
hdrs = [
"src/lauxlib.h",
"src/lua.h",
@ -12,6 +11,7 @@ cc_library(
"src/luaconf.h",
"src/lualib.h",
],
defines = ["LUA_USE_LINUX"],
includes = ["src"],
)
@ -72,7 +72,6 @@ cc_library(
"src/lzio.c",
"src/lzio.h",
],
defines = ["LUA_USE_LINUX"],
hdrs = [
"src/lauxlib.h",
"src/lua.h",
@ -80,6 +79,7 @@ cc_library(
"src/luaconf.h",
"src/lualib.h",
],
defines = ["LUA_USE_LINUX"],
includes = ["src"],
linkopts = [
"-lm",
@ -92,11 +92,11 @@ cc_binary(
srcs = [
"src/lua.c",
],
deps = [
":liblua",
],
linkopts = [
"-lreadline",
"-rdynamic",
],
deps = [
":liblua",
],
)

@ -1,4 +1,3 @@
package(
default_visibility = ["//visibility:public"],
)
@ -158,7 +157,10 @@ cc_binary(
"aapl/avlimelkey.h",
"aapl/avltree.h",
],
includes = ["ragel", "aapl"],
includes = [
"aapl",
"ragel",
],
)
config_h_contents = """

@ -18,6 +18,7 @@ def _get_real_short_path(file):
if short_path.startswith("../"):
second_slash = short_path.index("/", 3)
short_path = short_path[second_slash + 1:]
# Sometimes it has another few prefixes like:
# _virtual_imports/any_proto/google/protobuf/any.proto
# We want just google/protobuf/any.proto.
@ -51,7 +52,7 @@ def _filter_none(elems):
out.append(elem)
return out
def _cc_library_func(ctx, name, hdrs, srcs, dep_ccinfos):
def _cc_library_func(ctx, name, hdrs, srcs, copts, dep_ccinfos):
"""Like cc_library(), but callable from rules.
Args:
@ -87,6 +88,7 @@ def _cc_library_func(ctx, name, hdrs, srcs, dep_ccinfos):
name = name,
srcs = srcs,
public_hdrs = hdrs,
user_compile_flags = copts,
compilation_contexts = compilation_contexts,
**blaze_only_args
)
@ -105,6 +107,22 @@ def _cc_library_func(ctx, name, hdrs, srcs, dep_ccinfos):
linking_context = linking_context,
)
# Dummy rule to expose select() copts to aspects ##############################
_UpbProtoLibraryCopts = provider(
fields = {
"copts": "copts for upb_proto_library()",
},
)
def upb_proto_library_copts_impl(ctx):
return _UpbProtoLibraryCopts(copts = ctx.attr.copts)
upb_proto_library_copts = rule(
implementation = upb_proto_library_copts_impl,
attrs = {"copts": attr.string_list(default = [])},
)
# upb_proto_library / upb_proto_reflection_library shared code #################
GeneratedSrcsInfo = provider(
@ -198,6 +216,7 @@ def _upb_proto_aspect_impl(target, ctx, cc_provider, file_provider):
name = ctx.rule.attr.name + ctx.attr._ext,
hdrs = files.hdrs,
srcs = files.srcs,
copts = ctx.attr._copts[_UpbProtoLibraryCopts].copts,
dep_ccinfos = dep_ccinfos,
)
return [cc_provider(cc_info = cc_info), file_provider(srcs = files)]
@ -221,10 +240,13 @@ def _maybe_add(d):
_upb_proto_library_aspect = aspect(
attrs = _maybe_add({
"_copts": attr.label(
default = "//:upb_proto_library_copts__for_generated_code_only_do_not_use",
),
"_upbc": attr.label(
executable = True,
cfg = "host",
default = "//:protoc-gen-upb",
default = "//upbc:protoc-gen-upb",
),
"_protoc": attr.label(
executable = True,
@ -266,10 +288,13 @@ upb_proto_library = rule(
_upb_proto_reflection_library_aspect = aspect(
attrs = _maybe_add({
"_copts": attr.label(
default = "//:upb_proto_library_copts__for_generated_code_only_do_not_use",
),
"_upbc": attr.label(
executable = True,
cfg = "host",
default = "//:protoc-gen-upb",
default = "//upbc:protoc-gen-upb",
),
"_protoc": attr.label(
executable = True,

@ -1,4 +1,3 @@
load("@bazel_tools//tools/build_defs/repo:http.bzl", "http_archive")
load("@bazel_tools//tools/build_defs/repo:git.bzl", "git_repository")
load("@bazel_tools//tools/build_defs/repo:utils.bzl", "maybe")
@ -9,7 +8,7 @@ def upb_deps():
name = "com_google_absl",
commit = "df3ea785d8c30a9503321a3d35ee7d35808f190d", # LTS 2020-02-25
remote = "https://github.com/abseil/abseil-cpp.git",
shallow_since = "1583355457 -0500"
shallow_since = "1583355457 -0500",
)
maybe(

@ -0,0 +1,53 @@
load(
"//bazel:upb_proto_library.bzl",
"upb_proto_library",
"upb_proto_reflection_library",
)
licenses(["notice"])
proto_library(
name = "benchmark_descriptor_proto",
srcs = ["descriptor.proto"],
)
upb_proto_library(
name = "benchmark_descriptor_upb_proto",
deps = [":benchmark_descriptor_proto"],
)
upb_proto_reflection_library(
name = "benchmark_descriptor_upb_proto_reflection",
deps = [":benchmark_descriptor_proto"],
)
cc_proto_library(
name = "benchmark_descriptor_cc_proto",
deps = [":benchmark_descriptor_proto"],
)
proto_library(
name = "benchmark_descriptor_sv_proto",
srcs = ["descriptor_sv.proto"],
)
cc_proto_library(
name = "benchmark_descriptor_sv_cc_proto",
deps = [":benchmark_descriptor_sv_proto"],
)
cc_binary(
name = "benchmark",
testonly = 1,
srcs = ["benchmark.cc"],
deps = [
":benchmark_descriptor_cc_proto",
":benchmark_descriptor_sv_cc_proto",
":benchmark_descriptor_upb_proto",
":benchmark_descriptor_upb_proto_reflection",
"//:descriptor_upb_proto",
"//:reflection",
"@com_github_google_benchmark//:benchmark_main",
"@com_google_protobuf//:protobuf",
],
)

@ -1,12 +1,20 @@
#include <string.h>
#include <benchmark/benchmark.h>
#include <string.h>
// For benchmarks of parsing speed.
#include "benchmarks/descriptor.pb.h"
#include "benchmarks/descriptor.upb.h"
#include "benchmarks/descriptor.upbdefs.h"
#include "benchmarks/descriptor_sv.pb.h"
// For for benchmarks of building descriptors.
#include "google/protobuf/descriptor.upb.h"
#include "google/protobuf/descriptor.upbdefs.h"
#include "google/protobuf/descriptor.pb.h"
#include "upb/def.hpp"
upb_strview descriptor = google_protobuf_descriptor_proto_upbdefinit.descriptor;
upb_strview descriptor = benchmarks_descriptor_proto_upbdefinit.descriptor;
namespace protobuf = ::google::protobuf;
/* A buffer big enough to parse descriptor.proto without going to heap. */
@ -93,12 +101,12 @@ static void BM_LoadDescriptor_Proto2(benchmark::State& state) {
}
BENCHMARK(BM_LoadDescriptor_Proto2);
static void BM_ParseDescriptor_Upb_LargeInitialBlock(benchmark::State& state) {
static void BM_Parse_Upb_FileDesc_WithArena(benchmark::State& state) {
size_t bytes = 0;
for (auto _ : state) {
upb_arena* arena = upb_arena_init(buf, sizeof(buf), NULL);
google_protobuf_FileDescriptorProto* set =
google_protobuf_FileDescriptorProto_parse(descriptor.data,
upb_arena* arena = upb_arena_new();
upb_benchmark_FileDescriptorProto* set =
upb_benchmark_FileDescriptorProto_parse(descriptor.data,
descriptor.size, arena);
if (!set) {
printf("Failed to parse.\n");
@ -109,14 +117,14 @@ static void BM_ParseDescriptor_Upb_LargeInitialBlock(benchmark::State& state) {
}
state.SetBytesProcessed(state.iterations() * descriptor.size);
}
BENCHMARK(BM_ParseDescriptor_Upb_LargeInitialBlock);
BENCHMARK(BM_Parse_Upb_FileDesc_WithArena);
static void BM_ParseDescriptor_Upb(benchmark::State& state) {
static void BM_Parse_Upb_FileDesc_WithInitialBlock(benchmark::State& state) {
size_t bytes = 0;
for (auto _ : state) {
upb_arena* arena = upb_arena_new();
google_protobuf_FileDescriptorProto* set =
google_protobuf_FileDescriptorProto_parse(descriptor.data,
upb_arena* arena = upb_arena_init(buf, sizeof(buf), NULL);
upb_benchmark_FileDescriptorProto* set =
upb_benchmark_FileDescriptorProto_parse(descriptor.data,
descriptor.size, arena);
if (!set) {
printf("Failed to parse.\n");
@ -127,55 +135,60 @@ static void BM_ParseDescriptor_Upb(benchmark::State& state) {
}
state.SetBytesProcessed(state.iterations() * descriptor.size);
}
BENCHMARK(BM_ParseDescriptor_Upb);
static void BM_ParseDescriptor_Proto2_NoArena(benchmark::State& state) {
size_t bytes = 0;
for (auto _ : state) {
protobuf::FileDescriptorProto proto;
protobuf::StringPiece input(descriptor.data,descriptor.size);
bool ok = proto.ParseFrom<protobuf::MessageLite::kMergePartial>(input);
if (!ok) {
printf("Failed to parse.\n");
exit(1);
}
bytes += descriptor.size;
BENCHMARK(BM_Parse_Upb_FileDesc_WithInitialBlock);
template <class P>
struct NoArena {
public:
P* GetProto() { return &proto_; }
private:
P proto_;
};
template <class P>
struct WithArena {
public:
P* GetProto() { return protobuf::Arena::CreateMessage<P>(&arena_); }
private:
protobuf::Arena arena_;
};
template <class P>
struct WithInitialBlock {
public:
WithInitialBlock() : arena_(GetOptions()) {}
P* GetProto() { return protobuf::Arena::CreateMessage<P>(&arena_); }
private:
protobuf::ArenaOptions GetOptions() {
protobuf::ArenaOptions opts;
opts.initial_block = buf;
opts.initial_block_size = sizeof(buf);
return opts;
}
state.SetBytesProcessed(state.iterations() * descriptor.size);
}
BENCHMARK(BM_ParseDescriptor_Proto2_NoArena);
static void BM_ParseDescriptor_Proto2_Arena(benchmark::State& state) {
size_t bytes = 0;
for (auto _ : state) {
protobuf::Arena arena;
protobuf::StringPiece input(descriptor.data,descriptor.size);
auto proto = protobuf::Arena::CreateMessage<protobuf::FileDescriptorProto>(
&arena);
bool ok = proto->ParseFrom<protobuf::MessageLite::kMergePartial>(input);
protobuf::Arena arena_;
};
if (!ok) {
printf("Failed to parse.\n");
exit(1);
}
bytes += descriptor.size;
}
state.SetBytesProcessed(state.iterations() * descriptor.size);
}
BENCHMARK(BM_ParseDescriptor_Proto2_Arena);
using FileDesc = ::upb_benchmark::FileDescriptorProto;
using FileDescSV = ::upb_benchmark::sv::FileDescriptorProto;
static void BM_ParseDescriptor_Proto2_Arena_LargeInitialBlock(benchmark::State& state) {
const protobuf::MessageLite::ParseFlags kMergePartial =
protobuf::MessageLite::ParseFlags::kMergePartial;
const protobuf::MessageLite::ParseFlags kAliasStrings =
protobuf::MessageLite::ParseFlags::kMergePartialWithAliasing;
template <class P, template <class> class Factory,
protobuf::MessageLite::ParseFlags kParseFlags = kMergePartial>
void BM_Parse_Proto2(benchmark::State& state) {
size_t bytes = 0;
protobuf::ArenaOptions opts;
opts.initial_block = buf;
opts.initial_block_size = sizeof(buf);
for (auto _ : state) {
protobuf::Arena arena(opts);
Factory<P> proto_factory;
auto proto = proto_factory.GetProto();
protobuf::StringPiece input(descriptor.data,descriptor.size);
auto proto = protobuf::Arena::CreateMessage<protobuf::FileDescriptorProto>(
&arena);
bool ok = proto->ParseFrom<protobuf::MessageLite::kMergePartial>(input);
bool ok = proto->template ParseFrom<kParseFlags>(input);
if (!ok) {
printf("Failed to parse.\n");
exit(1);
@ -184,11 +197,19 @@ static void BM_ParseDescriptor_Proto2_Arena_LargeInitialBlock(benchmark::State&
}
state.SetBytesProcessed(state.iterations() * descriptor.size);
}
BENCHMARK(BM_ParseDescriptor_Proto2_Arena_LargeInitialBlock);
BENCHMARK_TEMPLATE(BM_Parse_Proto2, FileDesc, NoArena);
BENCHMARK_TEMPLATE(BM_Parse_Proto2, FileDesc, WithArena);
BENCHMARK_TEMPLATE(BM_Parse_Proto2, FileDesc, WithInitialBlock);
//BENCHMARK_TEMPLATE(BM_Parse_Proto2, FileDescSV, NoArena);
//BENCHMARK_TEMPLATE(BM_Parse_Proto2, FileDescSV, WithArena);
BENCHMARK_TEMPLATE(BM_Parse_Proto2, FileDescSV, WithInitialBlock);
//BENCHMARK_TEMPLATE(BM_Parse_Proto2, FileDescSV, NoArena, kAliasStrings);
//BENCHMARK_TEMPLATE(BM_Parse_Proto2, FileDescSV, WithArena, kAliasStrings);
BENCHMARK_TEMPLATE(BM_Parse_Proto2, FileDescSV, WithInitialBlock, kAliasStrings);
static void BM_SerializeDescriptor_Proto2(benchmark::State& state) {
size_t bytes = 0;
protobuf::FileDescriptorProto proto;
upb_benchmark::FileDescriptorProto proto;
proto.ParseFromArray(descriptor.data, descriptor.size);
for (auto _ : state) {
proto.SerializePartialToArray(buf, sizeof(buf));
@ -201,9 +222,9 @@ BENCHMARK(BM_SerializeDescriptor_Proto2);
static void BM_SerializeDescriptor_Upb(benchmark::State& state) {
int64_t total = 0;
upb_arena* arena = upb_arena_new();
google_protobuf_FileDescriptorProto* set =
google_protobuf_FileDescriptorProto_parse(descriptor.data,
descriptor.size, arena);
upb_benchmark_FileDescriptorProto* set =
upb_benchmark_FileDescriptorProto_parse(descriptor.data, descriptor.size,
arena);
if (!set) {
printf("Failed to parse.\n");
exit(1);
@ -211,7 +232,8 @@ static void BM_SerializeDescriptor_Upb(benchmark::State& state) {
for (auto _ : state) {
upb_arena* enc_arena = upb_arena_init(buf, sizeof(buf), NULL);
size_t size;
char *data = google_protobuf_FileDescriptorProto_serialize(set, enc_arena, &size);
char* data =
upb_benchmark_FileDescriptorProto_serialize(set, enc_arena, &size);
if (!data) {
printf("Failed to serialize.\n");
exit(1);

@ -30,12 +30,12 @@ def Run(cmd):
def Benchmark(outbase, bench_cpu=True, runs=12):
tmpfile = "/tmp/bench-output.json"
Run("rm -rf {}".format(tmpfile))
Run("CC=clang bazel test :all")
Run("CC=clang bazel test ...")
if bench_cpu:
Run("CC=clang bazel build -c opt --copt=-march=native :benchmark")
Run("CC=clang bazel build -c opt --copt=-march=native benchmarks:benchmark")
Run("./bazel-bin/benchmark --benchmark_out_format=json --benchmark_out={} --benchmark_repetitions={}".format(tmpfile, runs))
Run("./bazel-bin/benchmarks/benchmark --benchmark_out_format=json --benchmark_out={} --benchmark_repetitions={}".format(tmpfile, runs))
with open(tmpfile) as f:
bench_json = json.load(f)
@ -48,8 +48,8 @@ def Benchmark(outbase, bench_cpu=True, runs=12):
values = (name, run["iterations"], run["cpu_time"])
print("{} {} {} ns/op".format(*values), file=f)
Run("CC=clang bazel build -c opt --copt=-g :conformance_upb")
Run("cp -f bazel-bin/conformance_upb {}.bin".format(outbase))
Run("CC=clang bazel build -c opt --copt=-g tests:conformance_upb")
Run("cp -f bazel-bin/tests/conformance_upb {}.bin".format(outbase))
baseline = "master"

@ -0,0 +1,909 @@
// 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.
// Author: kenton@google.com (Kenton Varda)
// Based on original Protocol Buffers design by
// Sanjay Ghemawat, Jeff Dean, and others.
//
// The messages in this file describe the definitions found in .proto files.
// A valid .proto file can be translated directly to a FileDescriptorProto
// without any other information (e.g. without reading its imports).
syntax = "proto2";
package upb_benchmark;
option go_package = "google.golang.org/protobuf/types/descriptorpb";
option java_package = "com.google.protobuf";
option java_outer_classname = "DescriptorProtos";
option csharp_namespace = "Google.Protobuf.Reflection";
option objc_class_prefix = "GPB";
option cc_enable_arenas = true;
// descriptor.proto must be optimized for speed because reflection-based
// algorithms don't work during bootstrapping.
option optimize_for = SPEED;
// The protocol compiler can output a FileDescriptorSet containing the .proto
// files it parses.
message FileDescriptorSet {
repeated FileDescriptorProto file = 1;
}
// Describes a complete .proto file.
message FileDescriptorProto {
optional string name = 1; // file name, relative to root of source tree
optional string package = 2; // e.g. "foo", "foo.bar", etc.
// Names of files imported by this file.
repeated string dependency = 3;
// Indexes of the public imported files in the dependency list above.
repeated int32 public_dependency = 10;
// Indexes of the weak imported files in the dependency list.
// For Google-internal migration only. Do not use.
repeated int32 weak_dependency = 11;
// All top-level definitions in this file.
repeated DescriptorProto message_type = 4;
repeated EnumDescriptorProto enum_type = 5;
repeated ServiceDescriptorProto service = 6;
repeated FieldDescriptorProto extension = 7;
optional FileOptions options = 8;
// This field contains optional information about the original source code.
// You may safely remove this entire field without harming runtime
// functionality of the descriptors -- the information is needed only by
// development tools.
optional SourceCodeInfo source_code_info = 9;
// The syntax of the proto file.
// The supported values are "proto2" and "proto3".
optional string syntax = 12;
}
// Describes a message type.
message DescriptorProto {
optional string name = 1;
repeated FieldDescriptorProto field = 2;
repeated FieldDescriptorProto extension = 6;
repeated DescriptorProto nested_type = 3;
repeated EnumDescriptorProto enum_type = 4;
message ExtensionRange {
optional int32 start = 1; // Inclusive.
optional int32 end = 2; // Exclusive.
optional ExtensionRangeOptions options = 3;
}
repeated ExtensionRange extension_range = 5;
repeated OneofDescriptorProto oneof_decl = 8;
optional MessageOptions options = 7;
// Range of reserved tag numbers. Reserved tag numbers may not be used by
// fields or extension ranges in the same message. Reserved ranges may
// not overlap.
message ReservedRange {
optional int32 start = 1; // Inclusive.
optional int32 end = 2; // Exclusive.
}
repeated ReservedRange reserved_range = 9;
// Reserved field names, which may not be used by fields in the same message.
// A given name may only be reserved once.
repeated string reserved_name = 10;
}
message ExtensionRangeOptions {
// The parser stores options it doesn't recognize here. See above.
repeated UninterpretedOption uninterpreted_option = 999;
// Clients can define custom options in extensions of this message. See above.
extensions 1000 to max;
}
// Describes a field within a message.
message FieldDescriptorProto {
enum Type {
// 0 is reserved for errors.
// Order is weird for historical reasons.
TYPE_DOUBLE = 1;
TYPE_FLOAT = 2;
// Not ZigZag encoded. Negative numbers take 10 bytes. Use TYPE_SINT64 if
// negative values are likely.
TYPE_INT64 = 3;
TYPE_UINT64 = 4;
// Not ZigZag encoded. Negative numbers take 10 bytes. Use TYPE_SINT32 if
// negative values are likely.
TYPE_INT32 = 5;
TYPE_FIXED64 = 6;
TYPE_FIXED32 = 7;
TYPE_BOOL = 8;
TYPE_STRING = 9;
// Tag-delimited aggregate.
// Group type is deprecated and not supported in proto3. However, Proto3
// implementations should still be able to parse the group wire format and
// treat group fields as unknown fields.
TYPE_GROUP = 10;
TYPE_MESSAGE = 11; // Length-delimited aggregate.
// New in version 2.
TYPE_BYTES = 12;
TYPE_UINT32 = 13;
TYPE_ENUM = 14;
TYPE_SFIXED32 = 15;
TYPE_SFIXED64 = 16;
TYPE_SINT32 = 17; // Uses ZigZag encoding.
TYPE_SINT64 = 18; // Uses ZigZag encoding.
}
enum Label {
// 0 is reserved for errors
LABEL_OPTIONAL = 1;
LABEL_REQUIRED = 2;
LABEL_REPEATED = 3;
}
optional string name = 1;
optional int32 number = 3;
optional Label label = 4;
// If type_name is set, this need not be set. If both this and type_name
// are set, this must be one of TYPE_ENUM, TYPE_MESSAGE or TYPE_GROUP.
optional Type type = 5;
// For message and enum types, this is the name of the type. If the name
// starts with a '.', it is fully-qualified. Otherwise, C++-like scoping
// rules are used to find the type (i.e. first the nested types within this
// message are searched, then within the parent, on up to the root
// namespace).
optional string type_name = 6;
// For extensions, this is the name of the type being extended. It is
// resolved in the same manner as type_name.
optional string extendee = 2;
// For numeric types, contains the original text representation of the value.
// For booleans, "true" or "false".
// For strings, contains the default text contents (not escaped in any way).
// For bytes, contains the C escaped value. All bytes >= 128 are escaped.
// TODO(kenton): Base-64 encode?
optional string default_value = 7;
// If set, gives the index of a oneof in the containing type's oneof_decl
// list. This field is a member of that oneof.
optional int32 oneof_index = 9;
// JSON name of this field. The value is set by protocol compiler. If the
// user has set a "json_name" option on this field, that option's value
// will be used. Otherwise, it's deduced from the field's name by converting
// it to camelCase.
optional string json_name = 10;
optional FieldOptions options = 8;
// If true, this is a proto3 "optional". When a proto3 field is optional, it
// tracks presence regardless of field type.
//
// When proto3_optional is true, this field must be belong to a oneof to
// signal to old proto3 clients that presence is tracked for this field. This
// oneof is known as a "synthetic" oneof, and this field must be its sole
// member (each proto3 optional field gets its own synthetic oneof). Synthetic
// oneofs exist in the descriptor only, and do not generate any API. Synthetic
// oneofs must be ordered after all "real" oneofs.
//
// For message fields, proto3_optional doesn't create any semantic change,
// since non-repeated message fields always track presence. However it still
// indicates the semantic detail of whether the user wrote "optional" or not.
// This can be useful for round-tripping the .proto file. For consistency we
// give message fields a synthetic oneof also, even though it is not required
// to track presence. This is especially important because the parser can't
// tell if a field is a message or an enum, so it must always create a
// synthetic oneof.
//
// Proto2 optional fields do not set this flag, because they already indicate
// optional with `LABEL_OPTIONAL`.
optional bool proto3_optional = 17;
}
// Describes a oneof.
message OneofDescriptorProto {
optional string name = 1;
optional OneofOptions options = 2;
}
// Describes an enum type.
message EnumDescriptorProto {
optional string name = 1;
repeated EnumValueDescriptorProto value = 2;
optional EnumOptions options = 3;
// Range of reserved numeric values. Reserved values may not be used by
// entries in the same enum. Reserved ranges may not overlap.
//
// Note that this is distinct from DescriptorProto.ReservedRange in that it
// is inclusive such that it can appropriately represent the entire int32
// domain.
message EnumReservedRange {
optional int32 start = 1; // Inclusive.
optional int32 end = 2; // Inclusive.
}
// Range of reserved numeric values. Reserved numeric values may not be used
// by enum values in the same enum declaration. Reserved ranges may not
// overlap.
repeated EnumReservedRange reserved_range = 4;
// Reserved enum value names, which may not be reused. A given name may only
// be reserved once.
repeated string reserved_name = 5;
}
// Describes a value within an enum.
message EnumValueDescriptorProto {
optional string name = 1;
optional int32 number = 2;
optional EnumValueOptions options = 3;
}
// Describes a service.
message ServiceDescriptorProto {
optional string name = 1;
repeated MethodDescriptorProto method = 2;
optional ServiceOptions options = 3;
}
// Describes a method of a service.
message MethodDescriptorProto {
optional string name = 1;
// Input and output type names. These are resolved in the same way as
// FieldDescriptorProto.type_name, but must refer to a message type.
optional string input_type = 2;
optional string output_type = 3;
optional MethodOptions options = 4;
// Identifies if client streams multiple client messages
optional bool client_streaming = 5 [default = false];
// Identifies if server streams multiple server messages
optional bool server_streaming = 6 [default = false];
}
// ===================================================================
// Options
// Each of the definitions above may have "options" attached. These are
// just annotations which may cause code to be generated slightly differently
// or may contain hints for code that manipulates protocol messages.
//
// Clients may define custom options as extensions of the *Options messages.
// These extensions may not yet be known at parsing time, so the parser cannot
// store the values in them. Instead it stores them in a field in the *Options
// message called uninterpreted_option. This field must have the same name
// across all *Options messages. We then use this field to populate the
// extensions when we build a descriptor, at which point all protos have been
// parsed and so all extensions are known.
//
// Extension numbers for custom options may be chosen as follows:
// * For options which will only be used within a single application or
// organization, or for experimental options, use field numbers 50000
// through 99999. It is up to you to ensure that you do not use the
// same number for multiple options.
// * For options which will be published and used publicly by multiple
// independent entities, e-mail protobuf-global-extension-registry@google.com
// to reserve extension numbers. Simply provide your project name (e.g.
// Objective-C plugin) and your project website (if available) -- there's no
// need to explain how you intend to use them. Usually you only need one
// extension number. You can declare multiple options with only one extension
// number by putting them in a sub-message. See the Custom Options section of
// the docs for examples:
// https://developers.google.com/protocol-buffers/docs/proto#options
// If this turns out to be popular, a web service will be set up
// to automatically assign option numbers.
message FileOptions {
// Sets the Java package where classes generated from this .proto will be
// placed. By default, the proto package is used, but this is often
// inappropriate because proto packages do not normally start with backwards
// domain names.
optional string java_package = 1;
// If set, all the classes from the .proto file are wrapped in a single
// outer class with the given name. This applies to both Proto1
// (equivalent to the old "--one_java_file" option) and Proto2 (where
// a .proto always translates to a single class, but you may want to
// explicitly choose the class name).
optional string java_outer_classname = 8;
// If set true, then the Java code generator will generate a separate .java
// file for each top-level message, enum, and service defined in the .proto
// file. Thus, these types will *not* be nested inside the outer class
// named by java_outer_classname. However, the outer class will still be
// generated to contain the file's getDescriptor() method as well as any
// top-level extensions defined in the file.
optional bool java_multiple_files = 10 [default = false];
// This option does nothing.
optional bool java_generate_equals_and_hash = 20 [deprecated=true];
// If set true, then the Java2 code generator will generate code that
// throws an exception whenever an attempt is made to assign a non-UTF-8
// byte sequence to a string field.
// Message reflection will do the same.
// However, an extension field still accepts non-UTF-8 byte sequences.
// This option has no effect on when used with the lite runtime.
optional bool java_string_check_utf8 = 27 [default = false];
// Generated classes can be optimized for speed or code size.
enum OptimizeMode {
SPEED = 1; // Generate complete code for parsing, serialization,
// etc.
CODE_SIZE = 2; // Use ReflectionOps to implement these methods.
LITE_RUNTIME = 3; // Generate code using MessageLite and the lite runtime.
}
optional OptimizeMode optimize_for = 9 [default = SPEED];
// Sets the Go package where structs generated from this .proto will be
// placed. If omitted, the Go package will be derived from the following:
// - The basename of the package import path, if provided.
// - Otherwise, the package statement in the .proto file, if present.
// - Otherwise, the basename of the .proto file, without extension.
optional string go_package = 11;
// Should generic services be generated in each language? "Generic" services
// are not specific to any particular RPC system. They are generated by the
// main code generators in each language (without additional plugins).
// Generic services were the only kind of service generation supported by
// early versions of google.protobuf.
//
// Generic services are now considered deprecated in favor of using plugins
// that generate code specific to your particular RPC system. Therefore,
// these default to false. Old code which depends on generic services should
// explicitly set them to true.
optional bool cc_generic_services = 16 [default = false];
optional bool java_generic_services = 17 [default = false];
optional bool py_generic_services = 18 [default = false];
optional bool php_generic_services = 42 [default = false];
// Is this file deprecated?
// Depending on the target platform, this can emit Deprecated annotations
// for everything in the file, or it will be completely ignored; in the very
// least, this is a formalization for deprecating files.
optional bool deprecated = 23 [default = false];
// Enables the use of arenas for the proto messages in this file. This applies
// only to generated classes for C++.
optional bool cc_enable_arenas = 31 [default = true];
// Sets the objective c class prefix which is prepended to all objective c
// generated classes from this .proto. There is no default.
optional string objc_class_prefix = 36;
// Namespace for generated classes; defaults to the package.
optional string csharp_namespace = 37;
// By default Swift generators will take the proto package and CamelCase it
// replacing '.' with underscore and use that to prefix the types/symbols
// defined. When this options is provided, they will use this value instead
// to prefix the types/symbols defined.
optional string swift_prefix = 39;
// Sets the php class prefix which is prepended to all php generated classes
// from this .proto. Default is empty.
optional string php_class_prefix = 40;
// Use this option to change the namespace of php generated classes. Default
// is empty. When this option is empty, the package name will be used for
// determining the namespace.
optional string php_namespace = 41;
// Use this option to change the namespace of php generated metadata classes.
// Default is empty. When this option is empty, the proto file name will be
// used for determining the namespace.
optional string php_metadata_namespace = 44;
// Use this option to change the package of ruby generated classes. Default
// is empty. When this option is not set, the package name will be used for
// determining the ruby package.
optional string ruby_package = 45;
// The parser stores options it doesn't recognize here.
// See the documentation for the "Options" section above.
repeated UninterpretedOption uninterpreted_option = 999;
// Clients can define custom options in extensions of this message.
// See the documentation for the "Options" section above.
extensions 1000 to max;
reserved 38;
}
message MessageOptions {
// Set true to use the old proto1 MessageSet wire format for extensions.
// This is provided for backwards-compatibility with the MessageSet wire
// format. You should not use this for any other reason: It's less
// efficient, has fewer features, and is more complicated.
//
// The message must be defined exactly as follows:
// message Foo {
// option message_set_wire_format = true;
// extensions 4 to max;
// }
// Note that the message cannot have any defined fields; MessageSets only
// have extensions.
//
// All extensions of your type must be singular messages; e.g. they cannot
// be int32s, enums, or repeated messages.
//
// Because this is an option, the above two restrictions are not enforced by
// the protocol compiler.
optional bool message_set_wire_format = 1 [default = false];
// Disables the generation of the standard "descriptor()" accessor, which can
// conflict with a field of the same name. This is meant to make migration
// from proto1 easier; new code should avoid fields named "descriptor".
optional bool no_standard_descriptor_accessor = 2 [default = false];
// Is this message deprecated?
// Depending on the target platform, this can emit Deprecated annotations
// for the message, or it will be completely ignored; in the very least,
// this is a formalization for deprecating messages.
optional bool deprecated = 3 [default = false];
// Whether the message is an automatically generated map entry type for the
// maps field.
//
// For maps fields:
// map<KeyType, ValueType> map_field = 1;
// The parsed descriptor looks like:
// message MapFieldEntry {
// option map_entry = true;
// optional KeyType key = 1;
// optional ValueType value = 2;
// }
// repeated MapFieldEntry map_field = 1;
//
// Implementations may choose not to generate the map_entry=true message, but
// use a native map in the target language to hold the keys and values.
// The reflection APIs in such implementations still need to work as
// if the field is a repeated message field.
//
// NOTE: Do not set the option in .proto files. Always use the maps syntax
// instead. The option should only be implicitly set by the proto compiler
// parser.
optional bool map_entry = 7;
reserved 8; // javalite_serializable
reserved 9; // javanano_as_lite
// The parser stores options it doesn't recognize here. See above.
repeated UninterpretedOption uninterpreted_option = 999;
// Clients can define custom options in extensions of this message. See above.
extensions 1000 to max;
}
message FieldOptions {
// The ctype option instructs the C++ code generator to use a different
// representation of the field than it normally would. See the specific
// options below. This option is not yet implemented in the open source
// release -- sorry, we'll try to include it in a future version!
optional CType ctype = 1 [default = STRING];
enum CType {
// Default mode.
STRING = 0;
CORD = 1;
STRING_PIECE = 2;
}
// The packed option can be enabled for repeated primitive fields to enable
// a more efficient representation on the wire. Rather than repeatedly
// writing the tag and type for each element, the entire array is encoded as
// a single length-delimited blob. In proto3, only explicit setting it to
// false will avoid using packed encoding.
optional bool packed = 2;
// The jstype option determines the JavaScript type used for values of the
// field. The option is permitted only for 64 bit integral and fixed types
// (int64, uint64, sint64, fixed64, sfixed64). A field with jstype JS_STRING
// is represented as JavaScript string, which avoids loss of precision that
// can happen when a large value is converted to a floating point JavaScript.
// Specifying JS_NUMBER for the jstype causes the generated JavaScript code to
// use the JavaScript "number" type. The behavior of the default option
// JS_NORMAL is implementation dependent.
//
// This option is an enum to permit additional types to be added, e.g.
// goog.math.Integer.
optional JSType jstype = 6 [default = JS_NORMAL];
enum JSType {
// Use the default type.
JS_NORMAL = 0;
// Use JavaScript strings.
JS_STRING = 1;
// Use JavaScript numbers.
JS_NUMBER = 2;
}
// Should this field be parsed lazily? Lazy applies only to message-type
// fields. It means that when the outer message is initially parsed, the
// inner message's contents will not be parsed but instead stored in encoded
// form. The inner message will actually be parsed when it is first accessed.
//
// This is only a hint. Implementations are free to choose whether to use
// eager or lazy parsing regardless of the value of this option. However,
// setting this option true suggests that the protocol author believes that
// using lazy parsing on this field is worth the additional bookkeeping
// overhead typically needed to implement it.
//
// This option does not affect the public interface of any generated code;
// all method signatures remain the same. Furthermore, thread-safety of the
// interface is not affected by this option; const methods remain safe to
// call from multiple threads concurrently, while non-const methods continue
// to require exclusive access.
//
//
// Note that implementations may choose not to check required fields within
// a lazy sub-message. That is, calling IsInitialized() on the outer message
// may return true even if the inner message has missing required fields.
// This is necessary because otherwise the inner message would have to be
// parsed in order to perform the check, defeating the purpose of lazy
// parsing. An implementation which chooses not to check required fields
// must be consistent about it. That is, for any particular sub-message, the
// implementation must either *always* check its required fields, or *never*
// check its required fields, regardless of whether or not the message has
// been parsed.
optional bool lazy = 5 [default = false];
// Is this field deprecated?
// Depending on the target platform, this can emit Deprecated annotations
// for accessors, or it will be completely ignored; in the very least, this
// is a formalization for deprecating fields.
optional bool deprecated = 3 [default = false];
// For Google-internal migration only. Do not use.
optional bool weak = 10 [default = false];
// The parser stores options it doesn't recognize here. See above.
repeated UninterpretedOption uninterpreted_option = 999;
// Clients can define custom options in extensions of this message. See above.
extensions 1000 to max;
reserved 4; // removed jtype
}
message OneofOptions {
// The parser stores options it doesn't recognize here. See above.
repeated UninterpretedOption uninterpreted_option = 999;
// Clients can define custom options in extensions of this message. See above.
extensions 1000 to max;
}
message EnumOptions {
// Set this option to true to allow mapping different tag names to the same
// value.
optional bool allow_alias = 2;
// Is this enum deprecated?
// Depending on the target platform, this can emit Deprecated annotations
// for the enum, or it will be completely ignored; in the very least, this
// is a formalization for deprecating enums.
optional bool deprecated = 3 [default = false];
reserved 5; // javanano_as_lite
// The parser stores options it doesn't recognize here. See above.
repeated UninterpretedOption uninterpreted_option = 999;
// Clients can define custom options in extensions of this message. See above.
extensions 1000 to max;
}
message EnumValueOptions {
// Is this enum value deprecated?
// Depending on the target platform, this can emit Deprecated annotations
// for the enum value, or it will be completely ignored; in the very least,
// this is a formalization for deprecating enum values.
optional bool deprecated = 1 [default = false];
// The parser stores options it doesn't recognize here. See above.
repeated UninterpretedOption uninterpreted_option = 999;
// Clients can define custom options in extensions of this message. See above.
extensions 1000 to max;
}
message ServiceOptions {
// Note: Field numbers 1 through 32 are reserved for Google's internal RPC
// framework. We apologize for hoarding these numbers to ourselves, but
// we were already using them long before we decided to release Protocol
// Buffers.
// Is this service deprecated?
// Depending on the target platform, this can emit Deprecated annotations
// for the service, or it will be completely ignored; in the very least,
// this is a formalization for deprecating services.
optional bool deprecated = 33 [default = false];
// The parser stores options it doesn't recognize here. See above.
repeated UninterpretedOption uninterpreted_option = 999;
// Clients can define custom options in extensions of this message. See above.
extensions 1000 to max;
}
message MethodOptions {
// Note: Field numbers 1 through 32 are reserved for Google's internal RPC
// framework. We apologize for hoarding these numbers to ourselves, but
// we were already using them long before we decided to release Protocol
// Buffers.
// Is this method deprecated?
// Depending on the target platform, this can emit Deprecated annotations
// for the method, or it will be completely ignored; in the very least,
// this is a formalization for deprecating methods.
optional bool deprecated = 33 [default = false];
// Is this method side-effect-free (or safe in HTTP parlance), or idempotent,
// or neither? HTTP based RPC implementation may choose GET verb for safe
// methods, and PUT verb for idempotent methods instead of the default POST.
enum IdempotencyLevel {
IDEMPOTENCY_UNKNOWN = 0;
NO_SIDE_EFFECTS = 1; // implies idempotent
IDEMPOTENT = 2; // idempotent, but may have side effects
}
optional IdempotencyLevel idempotency_level = 34
[default = IDEMPOTENCY_UNKNOWN];
// The parser stores options it doesn't recognize here. See above.
repeated UninterpretedOption uninterpreted_option = 999;
// Clients can define custom options in extensions of this message. See above.
extensions 1000 to max;
}
// A message representing a option the parser does not recognize. This only
// appears in options protos created by the compiler::Parser class.
// DescriptorPool resolves these when building Descriptor objects. Therefore,
// options protos in descriptor objects (e.g. returned by Descriptor::options(),
// or produced by Descriptor::CopyTo()) will never have UninterpretedOptions
// in them.
message UninterpretedOption {
// The name of the uninterpreted option. Each string represents a segment in
// a dot-separated name. is_extension is true iff a segment represents an
// extension (denoted with parentheses in options specs in .proto files).
// E.g.,{ ["foo", false], ["bar.baz", true], ["qux", false] } represents
// "foo.(bar.baz).qux".
message NamePart {
optional string name_part = 1;
optional bool is_extension = 2;
}
repeated NamePart name = 2;
// The value of the uninterpreted option, in whatever type the tokenizer
// identified it as during parsing. Exactly one of these should be set.
optional string identifier_value = 3;
optional uint64 positive_int_value = 4;
optional int64 negative_int_value = 5;
optional double double_value = 6;
optional bytes string_value = 7;
optional string aggregate_value = 8;
}
// ===================================================================
// Optional source code info
// Encapsulates information about the original source file from which a
// FileDescriptorProto was generated.
message SourceCodeInfo {
// A Location identifies a piece of source code in a .proto file which
// corresponds to a particular definition. This information is intended
// to be useful to IDEs, code indexers, documentation generators, and similar
// tools.
//
// For example, say we have a file like:
// message Foo {
// optional string foo = 1;
// }
// Let's look at just the field definition:
// optional string foo = 1;
// ^ ^^ ^^ ^ ^^^
// a bc de f ghi
// We have the following locations:
// span path represents
// [a,i) [ 4, 0, 2, 0 ] The whole field definition.
// [a,b) [ 4, 0, 2, 0, 4 ] The label (optional).
// [c,d) [ 4, 0, 2, 0, 5 ] The type (string).
// [e,f) [ 4, 0, 2, 0, 1 ] The name (foo).
// [g,h) [ 4, 0, 2, 0, 3 ] The number (1).
//
// Notes:
// - A location may refer to a repeated field itself (i.e. not to any
// particular index within it). This is used whenever a set of elements are
// logically enclosed in a single code segment. For example, an entire
// extend block (possibly containing multiple extension definitions) will
// have an outer location whose path refers to the "extensions" repeated
// field without an index.
// - Multiple locations may have the same path. This happens when a single
// logical declaration is spread out across multiple places. The most
// obvious example is the "extend" block again -- there may be multiple
// extend blocks in the same scope, each of which will have the same path.
// - A location's span is not always a subset of its parent's span. For
// example, the "extendee" of an extension declaration appears at the
// beginning of the "extend" block and is shared by all extensions within
// the block.
// - Just because a location's span is a subset of some other location's span
// does not mean that it is a descendant. For example, a "group" defines
// both a type and a field in a single declaration. Thus, the locations
// corresponding to the type and field and their components will overlap.
// - Code which tries to interpret locations should probably be designed to
// ignore those that it doesn't understand, as more types of locations could
// be recorded in the future.
repeated Location location = 1;
message Location {
// Identifies which part of the FileDescriptorProto was defined at this
// location.
//
// Each element is a field number or an index. They form a path from
// the root FileDescriptorProto to the place where the definition. For
// example, this path:
// [ 4, 3, 2, 7, 1 ]
// refers to:
// file.message_type(3) // 4, 3
// .field(7) // 2, 7
// .name() // 1
// This is because FileDescriptorProto.message_type has field number 4:
// repeated DescriptorProto message_type = 4;
// and DescriptorProto.field has field number 2:
// repeated FieldDescriptorProto field = 2;
// and FieldDescriptorProto.name has field number 1:
// optional string name = 1;
//
// Thus, the above path gives the location of a field name. If we removed
// the last element:
// [ 4, 3, 2, 7 ]
// this path refers to the whole field declaration (from the beginning
// of the label to the terminating semicolon).
repeated int32 path = 1 [packed = true];
// Always has exactly three or four elements: start line, start column,
// end line (optional, otherwise assumed same as start line), end column.
// These are packed into a single field for efficiency. Note that line
// and column numbers are zero-based -- typically you will want to add
// 1 to each before displaying to a user.
repeated int32 span = 2 [packed = true];
// If this SourceCodeInfo represents a complete declaration, these are any
// comments appearing before and after the declaration which appear to be
// attached to the declaration.
//
// A series of line comments appearing on consecutive lines, with no other
// tokens appearing on those lines, will be treated as a single comment.
//
// leading_detached_comments will keep paragraphs of comments that appear
// before (but not connected to) the current element. Each paragraph,
// separated by empty lines, will be one comment element in the repeated
// field.
//
// Only the comment content is provided; comment markers (e.g. //) are
// stripped out. For block comments, leading whitespace and an asterisk
// will be stripped from the beginning of each line other than the first.
// Newlines are included in the output.
//
// Examples:
//
// optional int32 foo = 1; // Comment attached to foo.
// // Comment attached to bar.
// optional int32 bar = 2;
//
// optional string baz = 3;
// // Comment attached to baz.
// // Another line attached to baz.
//
// // Comment attached to qux.
// //
// // Another line attached to qux.
// optional double qux = 4;
//
// // Detached comment for corge. This is not leading or trailing comments
// // to qux or corge because there are blank lines separating it from
// // both.
//
// // Detached comment for corge paragraph 2.
//
// optional string corge = 5;
// /* Block comment attached
// * to corge. Leading asterisks
// * will be removed. */
// /* Block comment attached to
// * grault. */
// optional int32 grault = 6;
//
// // ignored detached comments.
optional string leading_comments = 3;
optional string trailing_comments = 4;
repeated string leading_detached_comments = 6;
}
}
// Describes the relationship between generated code and its original source
// file. A GeneratedCodeInfo message is associated with only one generated
// source file, but may contain references to different source .proto files.
message GeneratedCodeInfo {
// An Annotation connects some span of text in generated code to an element
// of its generating .proto file.
repeated Annotation annotation = 1;
message Annotation {
// Identifies the element in the original source .proto file. This field
// is formatted the same as SourceCodeInfo.Location.path.
repeated int32 path = 1 [packed = true];
// Identifies the filesystem path to the original source .proto.
optional string source_file = 2;
// Identifies the starting offset in bytes in the generated code
// that relates to the identified object.
optional int32 begin = 3;
// Identifies the ending offset in bytes in the generated code that
// relates to the identified offset. The end offset should be one past
// the last relevant byte (so the length of the text = end - begin).
optional int32 end = 4;
}
}

@ -0,0 +1,894 @@
// 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.
// Author: kenton@google.com (Kenton Varda)
// Based on original Protocol Buffers design by
// Sanjay Ghemawat, Jeff Dean, and others.
//
// The messages in this file describe the definitions found in .proto files.
// A valid .proto file can be translated directly to a FileDescriptorProto
// without any other information (e.g. without reading its imports).
syntax = "proto2";
package upb_benchmark.sv;
option go_package = "google.golang.org/protobuf/types/descriptorpb";
option java_package = "com.google.protobuf";
option java_outer_classname = "DescriptorProtos";
option csharp_namespace = "Google.Protobuf.Reflection";
option objc_class_prefix = "GPB";
option cc_enable_arenas = true;
// descriptor.proto must be optimized for speed because reflection-based
// algorithms don't work during bootstrapping.
option optimize_for = SPEED;
// The protocol compiler can output a FileDescriptorSet containing the .proto
// files it parses.
message FileDescriptorSet {
repeated FileDescriptorProto file = 1;
}
// Describes a complete .proto file.
message FileDescriptorProto {
optional string name = 1
[ctype = STRING_PIECE]; // file name, relative to root of source tree
optional string package = 2
[ctype = STRING_PIECE]; // e.g. "foo", "foo.bar", etc.
// Names of files imported by this file.
repeated string dependency = 3 [ctype = STRING_PIECE];
// Indexes of the public imported files in the dependency list above.
repeated int32 public_dependency = 10;
// Indexes of the weak imported files in the dependency list.
// For Google-internal migration only. Do not use.
repeated int32 weak_dependency = 11;
// All top-level definitions in this file.
repeated DescriptorProto message_type = 4;
repeated EnumDescriptorProto enum_type = 5;
repeated ServiceDescriptorProto service = 6;
repeated FieldDescriptorProto extension = 7;
optional FileOptions options = 8;
// This field contains optional information about the original source code.
// You may safely remove this entire field without harming runtime
// functionality of the descriptors -- the information is needed only by
// development tools.
optional SourceCodeInfo source_code_info = 9;
// The syntax of the proto file.
// The supported values are "proto2" and "proto3".
optional string syntax = 12 [ctype = STRING_PIECE];
}
// Describes a message type.
message DescriptorProto {
optional string name = 1 [ctype = STRING_PIECE];
repeated FieldDescriptorProto field = 2;
repeated FieldDescriptorProto extension = 6;
repeated DescriptorProto nested_type = 3;
repeated EnumDescriptorProto enum_type = 4;
message ExtensionRange {
optional int32 start = 1; // Inclusive.
optional int32 end = 2; // Exclusive.
optional ExtensionRangeOptions options = 3;
}
repeated ExtensionRange extension_range = 5;
repeated OneofDescriptorProto oneof_decl = 8;
optional MessageOptions options = 7;
// Range of reserved tag numbers. Reserved tag numbers may not be used by
// fields or extension ranges in the same message. Reserved ranges may
// not overlap.
message ReservedRange {
optional int32 start = 1; // Inclusive.
optional int32 end = 2; // Exclusive.
}
repeated ReservedRange reserved_range = 9;
// Reserved field names, which may not be used by fields in the same message.
// A given name may only be reserved once.
repeated string reserved_name = 10 [ctype = STRING_PIECE];
}
message ExtensionRangeOptions {
// The parser stores options it doesn't recognize here. See above.
repeated UninterpretedOption uninterpreted_option = 999;
// Clients can define custom options in extensions of this message. See above.
extensions 1000 to max;
}
// Describes a field within a message.
message FieldDescriptorProto {
enum Type {
// 0 is reserved for errors.
// Order is weird for historical reasons.
TYPE_DOUBLE = 1;
TYPE_FLOAT = 2;
// Not ZigZag encoded. Negative numbers take 10 bytes. Use TYPE_SINT64 if
// negative values are likely.
TYPE_INT64 = 3;
TYPE_UINT64 = 4;
// Not ZigZag encoded. Negative numbers take 10 bytes. Use TYPE_SINT32 if
// negative values are likely.
TYPE_INT32 = 5;
TYPE_FIXED64 = 6;
TYPE_FIXED32 = 7;
TYPE_BOOL = 8;
TYPE_STRING = 9;
// Tag-delimited aggregate.
// Group type is deprecated and not supported in proto3. However, Proto3
// implementations should still be able to parse the group wire format and
// treat group fields as unknown fields.
TYPE_GROUP = 10;
TYPE_MESSAGE = 11; // Length-delimited aggregate.
// New in version 2.
TYPE_BYTES = 12;
TYPE_UINT32 = 13;
TYPE_ENUM = 14;
TYPE_SFIXED32 = 15;
TYPE_SFIXED64 = 16;
TYPE_SINT32 = 17; // Uses ZigZag encoding.
TYPE_SINT64 = 18; // Uses ZigZag encoding.
}
enum Label {
// 0 is reserved for errors
LABEL_OPTIONAL = 1;
LABEL_REQUIRED = 2;
LABEL_REPEATED = 3;
}
optional string name = 1 [ctype = STRING_PIECE];
optional int32 number = 3;
optional Label label = 4;
// If type_name is set, this need not be set. If both this and type_name
// are set, this must be one of TYPE_ENUM, TYPE_MESSAGE or TYPE_GROUP.
optional Type type = 5;
// For message and enum types, this is the name of the type. If the name
// starts with a '.', it is fully-qualified. Otherwise, C++-like scoping
// rules are used to find the type (i.e. first the nested types within this
// message are searched, then within the parent, on up to the root
// namespace).
optional string type_name = 6 [ctype = STRING_PIECE];
// For extensions, this is the name of the type being extended. It is
// resolved in the same manner as type_name.
optional string extendee = 2 [ctype = STRING_PIECE];
// For numeric types, contains the original text representation of the value.
// For booleans, "true" or "false".
// For strings, contains the default text contents (not escaped in any way).
// For bytes, contains the C escaped value. All bytes >= 128 are escaped.
// TODO(kenton): Base-64 encode?
optional string default_value = 7 [ctype = STRING_PIECE];
// If set, gives the index of a oneof in the containing type's oneof_decl
// list. This field is a member of that oneof.
optional int32 oneof_index = 9;
// JSON name of this field. The value is set by protocol compiler. If the
// user has set a "json_name" option on this field, that option's value
// will be used. Otherwise, it's deduced from the field's name by converting
// it to camelCase.
optional string json_name = 10 [ctype = STRING_PIECE];
optional FieldOptions options = 8;
// If true, this is a proto3 "optional". When a proto3 field is optional, it
// tracks presence regardless of field type.
//
// When proto3_optional is true, this field must be belong to a oneof to
// signal to old proto3 clients that presence is tracked for this field. This
// oneof is known as a "synthetic" oneof, and this field must be its sole
// member (each proto3 optional field gets its own synthetic oneof). Synthetic
// oneofs exist in the descriptor only, and do not generate any API. Synthetic
// oneofs must be ordered after all "real" oneofs.
//
// For message fields, proto3_optional doesn't create any semantic change,
// since non-repeated message fields always track presence. However it still
// indicates the semantic detail of whether the user wrote "optional" or not.
// This can be useful for round-tripping the .proto file. For consistency we
// give message fields a synthetic oneof also, even though it is not required
// to track presence. This is especially important because the parser can't
// tell if a field is a message or an enum, so it must always create a
// synthetic oneof.
//
// Proto2 optional fields do not set this flag, because they already indicate
// optional with `LABEL_OPTIONAL`.
optional bool proto3_optional = 17;
}
// Describes a oneof.
message OneofDescriptorProto {
optional string name = 1 [ctype = STRING_PIECE];
optional OneofOptions options = 2;
}
// Describes an enum type.
message EnumDescriptorProto {
optional string name = 1 [ctype = STRING_PIECE];
repeated EnumValueDescriptorProto value = 2;
optional EnumOptions options = 3;
// Range of reserved numeric values. Reserved values may not be used by
// entries in the same enum. Reserved ranges may not overlap.
//
// Note that this is distinct from DescriptorProto.ReservedRange in that it
// is inclusive such that it can appropriately represent the entire int32
// domain.
message EnumReservedRange {
optional int32 start = 1; // Inclusive.
optional int32 end = 2; // Inclusive.
}
// Range of reserved numeric values. Reserved numeric values may not be used
// by enum values in the same enum declaration. Reserved ranges may not
// overlap.
repeated EnumReservedRange reserved_range = 4;
// Reserved enum value names, which may not be reused. A given name may only
// be reserved once.
repeated string reserved_name = 5 [ctype = STRING_PIECE];
}
// Describes a value within an enum.
message EnumValueDescriptorProto {
optional string name = 1 [ctype = STRING_PIECE];
optional int32 number = 2;
optional EnumValueOptions options = 3;
}
// Describes a service.
message ServiceDescriptorProto {
optional string name = 1 [ctype = STRING_PIECE];
repeated MethodDescriptorProto method = 2;
optional ServiceOptions options = 3;
}
// Describes a method of a service.
message MethodDescriptorProto {
optional string name = 1 [ctype = STRING_PIECE];
// Input and output type names. These are resolved in the same way as
// FieldDescriptorProto.type_name, but must refer to a message type.
optional string input_type = 2 [ctype = STRING_PIECE];
optional string output_type = 3 [ctype = STRING_PIECE];
optional MethodOptions options = 4;
// Identifies if client streams multiple client messages
optional bool client_streaming = 5 [default = false];
// Identifies if server streams multiple server messages
optional bool server_streaming = 6 [default = false];
}
// ===================================================================
// Options
// Each of the definitions above may have "options" attached. These are
// just annotations which may cause code to be generated slightly differently
// or may contain hints for code that manipulates protocol messages.
//
// Clients may define custom options as extensions of the *Options messages.
// These extensions may not yet be known at parsing time, so the parser cannot
// store the values in them. Instead it stores them in a field in the *Options
// message called uninterpreted_option. This field must have the same name
// across all *Options messages. We then use this field to populate the
// extensions when we build a descriptor, at which point all protos have been
// parsed and so all extensions are known.
//
// Extension numbers for custom options may be chosen as follows:
// * For options which will only be used within a single application or
// organization, or for experimental options, use field numbers 50000
// through 99999. It is up to you to ensure that you do not use the
// same number for multiple options.
// * For options which will be published and used publicly by multiple
// independent entities, e-mail protobuf-global-extension-registry@google.com
// to reserve extension numbers. Simply provide your project name (e.g.
// Objective-C plugin) and your project website (if available) -- there's no
// need to explain how you intend to use them. Usually you only need one
// extension number. You can declare multiple options with only one extension
// number by putting them in a sub-message. See the Custom Options section of
// the docs for examples:
// https://developers.google.com/protocol-buffers/docs/proto#options
// If this turns out to be popular, a web service will be set up
// to automatically assign option numbers.
message FileOptions {
// Sets the Java package where classes generated from this .proto will be
// placed. By default, the proto package is used, but this is often
// inappropriate because proto packages do not normally start with backwards
// domain names.
optional string java_package = 1 [ctype = STRING_PIECE];
// If set, all the classes from the .proto file are wrapped in a single
// outer class with the given name. This applies to both Proto1
// (equivalent to the old "--one_java_file" option) and Proto2 (where
// a .proto always translates to a single class, but you may want to
// explicitly choose the class name).
optional string java_outer_classname = 8 [ctype = STRING_PIECE];
// If set true, then the Java code generator will generate a separate .java
// file for each top-level message, enum, and service defined in the .proto
// file. Thus, these types will *not* be nested inside the outer class
// named by java_outer_classname. However, the outer class will still be
// generated to contain the file's getDescriptor() method as well as any
// top-level extensions defined in the file.
optional bool java_multiple_files = 10 [default = false];
// This option does nothing.
optional bool java_generate_equals_and_hash = 20 [deprecated = true];
// If set true, then the Java2 code generator will generate code that
// throws an exception whenever an attempt is made to assign a non-UTF-8
// byte sequence to a string field.
// Message reflection will do the same.
// However, an extension field still accepts non-UTF-8 byte sequences.
// This option has no effect on when used with the lite runtime.
optional bool java_string_check_utf8 = 27 [default = false];
// Generated classes can be optimized for speed or code size.
enum OptimizeMode {
SPEED = 1; // Generate complete code for parsing, serialization,
// etc.
CODE_SIZE = 2; // Use ReflectionOps to implement these methods.
LITE_RUNTIME = 3; // Generate code using MessageLite and the lite runtime.
}
optional OptimizeMode optimize_for = 9 [default = SPEED];
// Sets the Go package where structs generated from this .proto will be
// placed. If omitted, the Go package will be derived from the following:
// - The basename of the package import path, if provided.
// - Otherwise, the package statement in the .proto file, if present.
// - Otherwise, the basename of the .proto file, without extension.
optional string go_package = 11 [ctype = STRING_PIECE];
// Should generic services be generated in each language? "Generic" services
// are not specific to any particular RPC system. They are generated by the
// main code generators in each language (without additional plugins).
// Generic services were the only kind of service generation supported by
// early versions of google.protobuf.
//
// Generic services are now considered deprecated in favor of using plugins
// that generate code specific to your particular RPC system. Therefore,
// these default to false. Old code which depends on generic services should
// explicitly set them to true.
optional bool cc_generic_services = 16 [default = false];
optional bool java_generic_services = 17 [default = false];
optional bool py_generic_services = 18 [default = false];
optional bool php_generic_services = 42 [default = false];
// Is this file deprecated?
// Depending on the target platform, this can emit Deprecated annotations
// for everything in the file, or it will be completely ignored; in the very
// least, this is a formalization for deprecating files.
optional bool deprecated = 23 [default = false];
// Enables the use of arenas for the proto messages in this file. This applies
// only to generated classes for C++.
optional bool cc_enable_arenas = 31 [default = true];
// Sets the objective c class prefix which is prepended to all objective c
// generated classes from this .proto. There is no default.
optional string objc_class_prefix = 36 [ctype = STRING_PIECE];
// Namespace for generated classes; defaults to the package.
optional string csharp_namespace = 37 [ctype = STRING_PIECE];
// By default Swift generators will take the proto package and CamelCase it
// replacing '.' with underscore and use that to prefix the types/symbols
// defined. When this options is provided, they will use this value instead
// to prefix the types/symbols defined.
optional string swift_prefix = 39 [ctype = STRING_PIECE];
// Sets the php class prefix which is prepended to all php generated classes
// from this .proto. Default is empty.
optional string php_class_prefix = 40 [ctype = STRING_PIECE];
// Use this option to change the namespace of php generated classes. Default
// is empty. When this option is empty, the package name will be used for
// determining the namespace.
optional string php_namespace = 41 [ctype = STRING_PIECE];
// Use this option to change the namespace of php generated metadata classes.
// Default is empty. When this option is empty, the proto file name will be
// used for determining the namespace.
optional string php_metadata_namespace = 44 [ctype = STRING_PIECE];
// Use this option to change the package of ruby generated classes. Default
// is empty. When this option is not set, the package name will be used for
// determining the ruby package.
optional string ruby_package = 45 [ctype = STRING_PIECE];
// The parser stores options it doesn't recognize here.
// See the documentation for the "Options" section above.
repeated UninterpretedOption uninterpreted_option = 999;
// Clients can define custom options in extensions of this message.
// See the documentation for the "Options" section above.
extensions 1000 to max;
reserved 38;
}
message MessageOptions {
// Set true to use the old proto1 MessageSet wire format for extensions.
// This is provided for backwards-compatibility with the MessageSet wire
// format. You should not use this for any other reason: It's less
// efficient, has fewer features, and is more complicated.
//
// The message must be defined exactly as follows:
// message Foo {
// option message_set_wire_format = true;
// extensions 4 to max;
// }
// Note that the message cannot have any defined fields; MessageSets only
// have extensions.
//
// All extensions of your type must be singular messages; e.g. they cannot
// be int32s, enums, or repeated messages.
//
// Because this is an option, the above two restrictions are not enforced by
// the protocol compiler.
optional bool message_set_wire_format = 1 [default = false];
// Disables the generation of the standard "descriptor()" accessor, which can
// conflict with a field of the same name. This is meant to make migration
// from proto1 easier; new code should avoid fields named "descriptor".
optional bool no_standard_descriptor_accessor = 2 [default = false];
// Is this message deprecated?
// Depending on the target platform, this can emit Deprecated annotations
// for the message, or it will be completely ignored; in the very least,
// this is a formalization for deprecating messages.
optional bool deprecated = 3 [default = false];
// Whether the message is an automatically generated map entry type for the
// maps field.
//
// For maps fields:
// map<KeyType, ValueType> map_field = 1;
// The parsed descriptor looks like:
// message MapFieldEntry {
// option map_entry = true;
// optional KeyType key = 1;
// optional ValueType value = 2;
// }
// repeated MapFieldEntry map_field = 1;
//
// Implementations may choose not to generate the map_entry=true message, but
// use a native map in the target language to hold the keys and values.
// The reflection APIs in such implementations still need to work as
// if the field is a repeated message field.
//
// NOTE: Do not set the option in .proto files. Always use the maps syntax
// instead. The option should only be implicitly set by the proto compiler
// parser.
optional bool map_entry = 7;
reserved 8; // javalite_serializable
reserved 9; // javanano_as_lite
// The parser stores options it doesn't recognize here. See above.
repeated UninterpretedOption uninterpreted_option = 999;
// Clients can define custom options in extensions of this message. See above.
extensions 1000 to max;
}
message FieldOptions {
// The ctype option instructs the C++ code generator to use a different
// representation of the field than it normally would. See the specific
// options below. This option is not yet implemented in the open source
// release -- sorry, we'll try to include it in a future version!
optional CType ctype = 1 [default = STRING];
enum CType {
// Default mode.
STRING = 0;
CORD = 1;
STRING_PIECE = 2;
}
// The packed option can be enabled for repeated primitive fields to enable
// a more efficient representation on the wire. Rather than repeatedly
// writing the tag and type for each element, the entire array is encoded as
// a single length-delimited blob. In proto3, only explicit setting it to
// false will avoid using packed encoding.
optional bool packed = 2;
// The jstype option determines the JavaScript type used for values of the
// field. The option is permitted only for 64 bit integral and fixed types
// (int64, uint64, sint64, fixed64, sfixed64). A field with jstype JS_STRING
// is represented as JavaScript string, which avoids loss of precision that
// can happen when a large value is converted to a floating point JavaScript.
// Specifying JS_NUMBER for the jstype causes the generated JavaScript code to
// use the JavaScript "number" type. The behavior of the default option
// JS_NORMAL is implementation dependent.
//
// This option is an enum to permit additional types to be added, e.g.
// goog.math.Integer.
optional JSType jstype = 6 [default = JS_NORMAL];
enum JSType {
// Use the default type.
JS_NORMAL = 0;
// Use JavaScript strings.
JS_STRING = 1;
// Use JavaScript numbers.
JS_NUMBER = 2;
}
// Should this field be parsed lazily? Lazy applies only to message-type
// fields. It means that when the outer message is initially parsed, the
// inner message's contents will not be parsed but instead stored in encoded
// form. The inner message will actually be parsed when it is first accessed.
//
// This is only a hint. Implementations are free to choose whether to use
// eager or lazy parsing regardless of the value of this option. However,
// setting this option true suggests that the protocol author believes that
// using lazy parsing on this field is worth the additional bookkeeping
// overhead typically needed to implement it.
//
// This option does not affect the public interface of any generated code;
// all method signatures remain the same. Furthermore, thread-safety of the
// interface is not affected by this option; const methods remain safe to
// call from multiple threads concurrently, while non-const methods continue
// to require exclusive access.
//
//
// Note that implementations may choose not to check required fields within
// a lazy sub-message. That is, calling IsInitialized() on the outer message
// may return true even if the inner message has missing required fields.
// This is necessary because otherwise the inner message would have to be
// parsed in order to perform the check, defeating the purpose of lazy
// parsing. An implementation which chooses not to check required fields
// must be consistent about it. That is, for any particular sub-message, the
// implementation must either *always* check its required fields, or *never*
// check its required fields, regardless of whether or not the message has
// been parsed.
optional bool lazy = 5 [default = false];
// Is this field deprecated?
// Depending on the target platform, this can emit Deprecated annotations
// for accessors, or it will be completely ignored; in the very least, this
// is a formalization for deprecating fields.
optional bool deprecated = 3 [default = false];
// For Google-internal migration only. Do not use.
optional bool weak = 10 [default = false];
// The parser stores options it doesn't recognize here. See above.
repeated UninterpretedOption uninterpreted_option = 999;
// Clients can define custom options in extensions of this message. See above.
extensions 1000 to max;
reserved 4; // removed jtype
}
message OneofOptions {
// The parser stores options it doesn't recognize here. See above.
repeated UninterpretedOption uninterpreted_option = 999;
// Clients can define custom options in extensions of this message. See above.
extensions 1000 to max;
}
message EnumOptions {
// Set this option to true to allow mapping different tag names to the same
// value.
optional bool allow_alias = 2;
// Is this enum deprecated?
// Depending on the target platform, this can emit Deprecated annotations
// for the enum, or it will be completely ignored; in the very least, this
// is a formalization for deprecating enums.
optional bool deprecated = 3 [default = false];
reserved 5; // javanano_as_lite
// The parser stores options it doesn't recognize here. See above.
repeated UninterpretedOption uninterpreted_option = 999;
// Clients can define custom options in extensions of this message. See above.
extensions 1000 to max;
}
message EnumValueOptions {
// Is this enum value deprecated?
// Depending on the target platform, this can emit Deprecated annotations
// for the enum value, or it will be completely ignored; in the very least,
// this is a formalization for deprecating enum values.
optional bool deprecated = 1 [default = false];
// The parser stores options it doesn't recognize here. See above.
repeated UninterpretedOption uninterpreted_option = 999;
// Clients can define custom options in extensions of this message. See above.
extensions 1000 to max;
}
message ServiceOptions {
// Note: Field numbers 1 through 32 are reserved for Google's internal RPC
// framework. We apologize for hoarding these numbers to ourselves, but
// we were already using them long before we decided to release Protocol
// Buffers.
// Is this service deprecated?
// Depending on the target platform, this can emit Deprecated annotations
// for the service, or it will be completely ignored; in the very least,
// this is a formalization for deprecating services.
optional bool deprecated = 33 [default = false];
// The parser stores options it doesn't recognize here. See above.
repeated UninterpretedOption uninterpreted_option = 999;
// Clients can define custom options in extensions of this message. See above.
extensions 1000 to max;
}
message MethodOptions {
// Note: Field numbers 1 through 32 are reserved for Google's internal RPC
// framework. We apologize for hoarding these numbers to ourselves, but
// we were already using them long before we decided to release Protocol
// Buffers.
// Is this method deprecated?
// Depending on the target platform, this can emit Deprecated annotations
// for the method, or it will be completely ignored; in the very least,
// this is a formalization for deprecating methods.
optional bool deprecated = 33 [default = false];
// Is this method side-effect-free (or safe in HTTP parlance), or idempotent,
// or neither? HTTP based RPC implementation may choose GET verb for safe
// methods, and PUT verb for idempotent methods instead of the default POST.
enum IdempotencyLevel {
IDEMPOTENCY_UNKNOWN = 0;
NO_SIDE_EFFECTS = 1; // implies idempotent
IDEMPOTENT = 2; // idempotent, but may have side effects
}
optional IdempotencyLevel idempotency_level = 34
[default = IDEMPOTENCY_UNKNOWN];
// The parser stores options it doesn't recognize here. See above.
repeated UninterpretedOption uninterpreted_option = 999;
// Clients can define custom options in extensions of this message. See above.
extensions 1000 to max;
}
// A message representing a option the parser does not recognize. This only
// appears in options protos created by the compiler::Parser class.
// DescriptorPool resolves these when building Descriptor objects. Therefore,
// options protos in descriptor objects (e.g. returned by Descriptor::options(),
// or produced by Descriptor::CopyTo()) will never have UninterpretedOptions
// in them.
message UninterpretedOption {
// The name of the uninterpreted option. Each string represents a segment in
// a dot-separated name. is_extension is true iff a segment represents an
// extension (denoted with parentheses in options specs in .proto files).
// E.g.,{ ["foo", false], ["bar.baz", true], ["qux", false] } represents
// "foo.(bar.baz).qux".
message NamePart {
optional string name_part = 1 [ctype = STRING_PIECE];
optional bool is_extension = 2;
}
repeated NamePart name = 2;
// The value of the uninterpreted option, in whatever type the tokenizer
// identified it as during parsing. Exactly one of these should be set.
optional string identifier_value = 3 [ctype = STRING_PIECE];
optional uint64 positive_int_value = 4;
optional int64 negative_int_value = 5;
optional double double_value = 6;
optional bytes string_value = 7;
optional string aggregate_value = 8 [ctype = STRING_PIECE];
}
// ===================================================================
// Optional source code info
// Encapsulates information about the original source file from which a
// FileDescriptorProto was generated.
message SourceCodeInfo {
// A Location identifies a piece of source code in a .proto file which
// corresponds to a particular definition. This information is intended
// to be useful to IDEs, code indexers, documentation generators, and similar
// tools.
//
// For example, say we have a file like:
// message Foo {
// optional string foo = 1 [ctype = STRING_PIECE];
// }
// Let's look at just the field definition:
// optional string foo = 1 [ctype = STRING_PIECE];
// ^ ^^ ^^ ^ ^^^
// a bc de f ghi
// We have the following locations:
// span path represents
// [a,i) [ 4, 0, 2, 0 ] The whole field definition.
// [a,b) [ 4, 0, 2, 0, 4 ] The label (optional).
// [c,d) [ 4, 0, 2, 0, 5 ] The type (string).
// [e,f) [ 4, 0, 2, 0, 1 ] The name (foo).
// [g,h) [ 4, 0, 2, 0, 3 ] The number (1).
//
// Notes:
// - A location may refer to a repeated field itself (i.e. not to any
// particular index within it). This is used whenever a set of elements are
// logically enclosed in a single code segment. For example, an entire
// extend block (possibly containing multiple extension definitions) will
// have an outer location whose path refers to the "extensions" repeated
// field without an index.
// - Multiple locations may have the same path. This happens when a single
// logical declaration is spread out across multiple places. The most
// obvious example is the "extend" block again -- there may be multiple
// extend blocks in the same scope, each of which will have the same path.
// - A location's span is not always a subset of its parent's span. For
// example, the "extendee" of an extension declaration appears at the
// beginning of the "extend" block and is shared by all extensions within
// the block.
// - Just because a location's span is a subset of some other location's span
// does not mean that it is a descendant. For example, a "group" defines
// both a type and a field in a single declaration. Thus, the locations
// corresponding to the type and field and their components will overlap.
// - Code which tries to interpret locations should probably be designed to
// ignore those that it doesn't understand, as more types of locations could
// be recorded in the future.
repeated Location location = 1;
message Location {
// Identifies which part of the FileDescriptorProto was defined at this
// location.
//
// Each element is a field number or an index. They form a path from
// the root FileDescriptorProto to the place where the definition. For
// example, this path:
// [ 4, 3, 2, 7, 1 ]
// refers to:
// file.message_type(3) // 4, 3
// .field(7) // 2, 7
// .name() // 1
// This is because FileDescriptorProto.message_type has field number 4:
// repeated DescriptorProto message_type = 4;
// and DescriptorProto.field has field number 2:
// repeated FieldDescriptorProto field = 2;
// and FieldDescriptorProto.name has field number 1:
// optional string name = 1 [ctype = STRING_PIECE];
//
// Thus, the above path gives the location of a field name. If we removed
// the last element:
// [ 4, 3, 2, 7 ]
// this path refers to the whole field declaration (from the beginning
// of the label to the terminating semicolon).
repeated int32 path = 1 [packed = true];
// Always has exactly three or four elements: start line, start column,
// end line (optional, otherwise assumed same as start line), end column.
// These are packed into a single field for efficiency. Note that line
// and column numbers are zero-based -- typically you will want to add
// 1 to each before displaying to a user.
repeated int32 span = 2 [packed = true];
// If this SourceCodeInfo represents a complete declaration, these are any
// comments appearing before and after the declaration which appear to be
// attached to the declaration.
//
// A series of line comments appearing on consecutive lines, with no other
// tokens appearing on those lines, will be treated as a single comment.
//
// leading_detached_comments will keep paragraphs of comments that appear
// before (but not connected to) the current element. Each paragraph,
// separated by empty lines, will be one comment element in the repeated
// field.
//
// Only the comment content is provided; comment markers (e.g. //) are
// stripped out. For block comments, leading whitespace and an asterisk
// will be stripped from the beginning of each line other than the first.
// Newlines are included in the output.
//
// Examples:
//
// optional int32 foo = 1; // Comment attached to foo.
// // Comment attached to bar.
// optional int32 bar = 2;
//
// optional string baz = 3 [ctype = STRING_PIECE];
// // Comment attached to baz.
// // Another line attached to baz.
//
// // Comment attached to qux.
// //
// // Another line attached to qux.
// optional double qux = 4;
//
// // Detached comment for corge. This is not leading or trailing comments
// // to qux or corge because there are blank lines separating it from
// // both.
//
// // Detached comment for corge paragraph 2.
//
// optional string corge = 5 [ctype = STRING_PIECE];
// /* Block comment attached
// * to corge. Leading asterisks
// * will be removed. */
// /* Block comment attached to
// * grault. */
// optional int32 grault = 6;
//
// // ignored detached comments.
optional string leading_comments = 3 [ctype = STRING_PIECE];
optional string trailing_comments = 4 [ctype = STRING_PIECE];
repeated string leading_detached_comments = 6 [ctype = STRING_PIECE];
}
}
// Describes the relationship between generated code and its original source
// file. A GeneratedCodeInfo message is associated with only one generated
// source file, but may contain references to different source .proto files.
message GeneratedCodeInfo {
// An Annotation connects some span of text in generated code to an element
// of its generating .proto file.
repeated Annotation annotation = 1;
message Annotation {
// Identifies the element in the original source .proto file. This field
// is formatted the same as SourceCodeInfo.Location.path.
repeated int32 path = 1 [packed = true];
// Identifies the filesystem path to the original source .proto.
optional string source_file = 2 [ctype = STRING_PIECE];
// Identifies the starting offset in bytes in the generated code
// that relates to the identified object.
optional int32 begin = 3;
// Identifies the ending offset in bytes in the generated code that
// relates to the identified offset. The end offset should be one past
// the last relevant byte (so the length of the text = end - begin).
optional int32 end = 4;
}
}

@ -0,0 +1,89 @@
load(
":build_defs.bzl",
"generated_file_staleness_test",
)
load(
"//bazel:build_defs.bzl",
"make_shell_script",
)
licenses(["notice"])
exports_files(["staleness_test.py"])
py_library(
name = "staleness_test_lib",
testonly = 1,
srcs = ["staleness_test_lib.py"],
)
py_binary(
name = "make_cmakelists",
srcs = ["make_cmakelists.py"],
)
genrule(
name = "gen_cmakelists",
srcs = [
"//:BUILD",
"//:WORKSPACE",
"//:cmake_files",
":cmake_files",
],
outs = ["generated-in/CMakeLists.txt"],
cmd = "$(location :make_cmakelists) $@",
tools = [":make_cmakelists"],
)
genrule(
name = "copy_json_ragel",
srcs = ["//:upb/json/parser.c"],
outs = ["generated-in/upb/json/parser.c"],
cmd = "cp $< $@",
)
genrule(
name = "copy_protos",
srcs = ["//:descriptor_upb_proto"],
outs = [
"generated-in/google/protobuf/descriptor.upb.c",
"generated-in/google/protobuf/descriptor.upb.h",
],
cmd = "cp $(SRCS) $(@D)/generated-in/google/protobuf",
)
generated_file_staleness_test(
name = "test_generated_files",
outs = [
"CMakeLists.txt",
"google/protobuf/descriptor.upb.c",
"google/protobuf/descriptor.upb.h",
"upb/json/parser.c",
],
generated_pattern = "generated-in/%s",
)
# Test the CMake build #########################################################
filegroup(
name = "cmake_files",
srcs = glob([
"**/*",
]),
)
make_shell_script(
name = "gen_run_cmake_build",
out = "run_cmake_build.sh",
contents = "find . && mkdir build && cd build && cmake ../cmake && make -j8 && make test",
)
sh_test(
name = "cmake_build",
srcs = ["run_cmake_build.sh"],
data = [
":cmake_files",
"//:cmake_files",
],
deps = ["@bazel_tools//tools/bash/runfiles"],
)

@ -12,6 +12,7 @@ cmake_minimum_required (VERSION 3.0)
cmake_policy(SET CMP0048 NEW)
project(upb)
set(CMAKE_C_STANDARD 99)
# Prevent CMake from setting -rdynamic on Linux (!!).
@ -48,8 +49,8 @@ if(UPB_ENABLE_UBSAN)
set(CMAKE_SHARED_LINKER_FLAGS "${CMAKE_SHARED_LINKER_FLAGS} -fsanitize=address")
endif()
include_directories(.)
include_directories(generated_for_cmake)
include_directories(..)
include_directories(../cmake)
include_directories(${CMAKE_CURRENT_BINARY_DIR})
if(APPLE)
@ -61,21 +62,21 @@ endif()
enable_testing()
add_library(port
upb/port.c)
../upb/port.c)
add_library(upb
upb/decode.c
upb/encode.c
upb/msg.c
upb/msg.h
upb/table.c
upb/table.int.h
upb/upb.c
upb/upb.int.h
third_party/wyhash/wyhash.h
upb/decode.h
upb/encode.h
upb/upb.h
upb/upb.hpp)
../upb/decode.c
../upb/encode.c
../upb/msg.c
../upb/msg.h
../upb/table.c
../upb/table.int.h
../upb/upb.c
../upb/upb.int.h
../third_party/wyhash/wyhash.h
../upb/decode.h
../upb/encode.h
../upb/upb.h
../upb/upb.hpp)
target_link_libraries(upb
port)
add_library(generated_code_support__only_for_generated_code_do_not_use__i_give_permission_to_break_me INTERFACE)
@ -83,28 +84,28 @@ target_link_libraries(generated_code_support__only_for_generated_code_do_not_use
table
upb)
add_library(reflection
upb/def.c
upb/msg.h
upb/reflection.c
upb/def.h
upb/def.hpp
upb/reflection.h)
../upb/def.c
../upb/msg.h
../upb/reflection.c
../upb/def.h
../upb/def.hpp
../upb/reflection.h)
target_link_libraries(reflection
descriptor_upb_proto
port
table
upb)
add_library(textformat
upb/text_encode.c
upb/text_encode.h)
../upb/text_encode.c
../upb/text_encode.h)
target_link_libraries(textformat
port
reflection)
add_library(json
upb/json_decode.c
upb/json_encode.c
upb/json_decode.h
upb/json_encode.h)
../upb/json_decode.c
../upb/json_encode.c
../upb/json_decode.h
../upb/json_encode.h)
target_link_libraries(json
port
reflection
@ -114,27 +115,27 @@ target_link_libraries(table INTERFACE
port
upb)
add_library(handlers
upb/handlers.c
upb/handlers-inl.h
upb/sink.c
upb/handlers.h
upb/sink.h)
../upb/handlers.c
../upb/handlers-inl.h
../upb/sink.c
../upb/handlers.h
../upb/sink.h)
target_link_libraries(handlers
port
reflection
table
upb)
add_library(upb_pb
upb/pb/compile_decoder.c
upb/pb/decoder.c
upb/pb/decoder.int.h
upb/pb/encoder.c
upb/pb/textprinter.c
upb/pb/varint.c
upb/pb/varint.int.h
upb/pb/decoder.h
upb/pb/encoder.h
upb/pb/textprinter.h)
../upb/pb/compile_decoder.c
../upb/pb/decoder.c
../upb/pb/decoder.int.h
../upb/pb/encoder.c
../upb/pb/textprinter.c
../upb/pb/varint.c
../upb/pb/varint.int.h
../upb/pb/decoder.h
../upb/pb/encoder.h
../upb/pb/textprinter.h)
target_link_libraries(upb_pb
descriptor_upb_proto
handlers
@ -143,26 +144,12 @@ target_link_libraries(upb_pb
table
upb)
add_library(upb_json
generated_for_cmake/upb/json/parser.c
upb/json/printer.c
upb/json/parser.h
upb/json/printer.h)
../cmake/upb/json/parser.c
../upb/json/printer.c
../upb/json/parser.h
../upb/json/printer.h)
target_link_libraries(upb_json
upb
upb_pb)
add_library(upb_cc_bindings INTERFACE)
target_link_libraries(upb_cc_bindings INTERFACE
descriptor_upb_proto
handlers
port
upb)
add_library(upb_test
tests/testmain.cc
tests/test_util.h
tests/upb_test.h)
target_link_libraries(upb_test
handlers
port
upb)

@ -0,0 +1,23 @@
# upb CMake build (EXPERIMENTAL)
upb's CMake support is experimental. The core library builds successfully
under CMake, and this is verified by the Bazel tests in this directory.
However there is no support for building the upb compiler or for generating
.upb.c/upb.h files. This means upb's CMake support is incomplete at best,
unless your application is intended to be purely reflective.
If you find this CMake setup useful in its current state, please consider
filing an issue so we know. If you have suggestions for how it could be
more useful (and particularly if you can contribute some code for it)
please feel free to file an issue for that too. Do keep in mind that upb
does not currently provide any ABI stability, so we want to avoid providing
a shared library.
The CMakeLists.txt is generated from the Bazel BUILD files using the Python
scripts in this directory. We want to avoid having two separate sources of
truth that both need to be updated when a file is added or removed.
This directory also contains some generated files that would be created
on the fly during a Bazel build. These are automaticaly kept in sync by
the Bazel test `//cmake:test_generated_files`.

@ -0,0 +1,44 @@
def generated_file_staleness_test(name, outs, generated_pattern):
"""Tests that checked-in file(s) match the contents of generated file(s).
The resulting test will verify that all output files exist and have the
correct contents. If the test fails, it can be invoked with --fix to
bring the checked-in files up to date.
Args:
name: Name of the rule.
outs: the checked-in files that are copied from generated files.
generated_pattern: the pattern for transforming each "out" file into a
generated file. For example, if generated_pattern="generated/%s" then
a file foo.txt will look for generated file generated/foo.txt.
"""
script_name = name + ".py"
script_src = ":staleness_test.py"
# Filter out non-existing rules so Blaze doesn't error out before we even
# run the test.
existing_outs = native.glob(include = outs)
# The file list contains a few extra bits of information at the end.
# These get unpacked by the Config class in staleness_test_lib.py.
file_list = outs + [generated_pattern, native.package_name() or ".", name]
native.genrule(
name = name + "_makescript",
outs = [script_name],
srcs = [script_src],
testonly = 1,
cmd = "cat $(location " + script_src + ") > $@; " +
"sed -i.bak -e 's|INSERT_FILE_LIST_HERE|" + "\\\n ".join(file_list) + "|' $@",
)
native.py_test(
name = name,
srcs = [script_name],
data = existing_outs + [generated_pattern % file for file in outs],
deps = [
":staleness_test_lib",
],
)

@ -46,9 +46,9 @@ class BuildFileFunctions(object):
found_files = []
for file in files:
if os.path.isfile(file):
found_files.append(file)
elif os.path.isfile("generated_for_cmake/" + file):
found_files.append("generated_for_cmake/" + file)
found_files.append("../" + file)
elif os.path.isfile("cmake/" + file):
found_files.append("../cmake/" + file)
else:
print("Warning: no such file: " + file)
@ -117,6 +117,9 @@ class BuildFileFunctions(object):
def proto_library(self, **kwargs):
pass
def cc_proto_library(self, **kwargs):
pass
def generated_file_staleness_test(self, **kwargs):
pass
@ -126,6 +129,9 @@ class BuildFileFunctions(object):
def upb_proto_library(self, **kwargs):
pass
def upb_proto_library_copts(self, **kwargs):
pass
def upb_proto_reflection_library(self, **kwargs):
pass
@ -163,6 +169,7 @@ class WorkspaceFileFunctions(object):
def workspace(self, **kwargs):
self.converter.prelude += "project(%s)\n" % (kwargs["name"])
self.converter.prelude += "set(CMAKE_C_STANDARD 99)\n"
def http_archive(self, **kwargs):
pass
@ -239,8 +246,8 @@ class Converter(object):
set(CMAKE_SHARED_LINKER_FLAGS "${CMAKE_SHARED_LINKER_FLAGS} -fsanitize=address")
endif()
include_directories(.)
include_directories(generated_for_cmake)
include_directories(..)
include_directories(../cmake)
include_directories(${CMAKE_CURRENT_BINARY_DIR})
if(APPLE)
@ -260,6 +267,7 @@ converter = Converter()
def GetDict(obj):
ret = {}
ret["UPB_DEFAULT_COPTS"] = [] # HACK
for k in dir(obj):
if not k.startswith("_"):
ret[k] = getattr(obj, k);

@ -6,7 +6,7 @@ with the actual list of files before we actually run the script.
from __future__ import absolute_import
from tools import staleness_test_lib
from cmake import staleness_test_lib
import unittest
import sys

@ -7,6 +7,7 @@ generated_file_staleness_test() rules.
from __future__ import absolute_import
from __future__ import print_function
import sys
import os
from shutil import copyfile
@ -47,13 +48,13 @@ def _GetFilePairs(config):
ret = []
has_bazel_genfiles = os.path.exists("bazel-genfiles")
has_bazel_genfiles = os.path.exists("bazel-bin")
for filename in config.file_list:
target = os.path.join(config.package_name, filename)
generated = os.path.join(config.package_name, config.pattern % filename)
if has_bazel_genfiles:
generated = os.path.join("bazel-genfiles", generated)
generated = os.path.join("bazel-bin", generated)
# Generated files should always exist. Blaze should guarantee this before
# we are run.
@ -61,6 +62,7 @@ def _GetFilePairs(config):
print("Generated file '%s' does not exist." % generated)
print("Please run this command to generate it:")
print(" bazel build %s:%s" % (config.package_name, config.target_name))
sys.exit(1)
ret.append(_FilePair(target, generated))
return ret
@ -87,10 +89,9 @@ def _GetMissingAndStaleFiles(file_pairs):
missing_files.append(pair)
continue
generated = open(pair.generated).read()
target = open(pair.target).read()
if generated != target:
stale_files.append(pair)
with open(pair.generated) as g, open(pair.target) as t:
if g.read() != t.read():
stale_files.append(pair)
return missing_files, stale_files

@ -953,7 +953,7 @@ static bool parse_number_from_buffer(upb_json_parser *p, const char *buf,
upb_fieldtype_t type = upb_fielddef_type(p->top->f);
double val;
double dummy;
double inf = UPB_INFINITY;
double inf = INFINITY;
errno = 0;

@ -1,6 +1,8 @@
load("@rules_proto//proto:defs.bzl", "proto_library")
load("@upb//bazel:upb_proto_library.bzl", "upb_proto_library")
licenses(["notice"])
proto_library(
name = "foo_proto",
srcs = ["foo.proto"],
@ -15,4 +17,5 @@ cc_binary(
name = "test_binary",
srcs = ["test_binary.c"],
deps = [":foo_upbproto"],
copts = ["-std=c99"],
)

@ -1,7 +1,7 @@
#include <time.h>
#include "foo.upb.h"
#include "examples/bazel/foo.upb.h"
int main() {
upb_arena *arena = upb_arena_new();

@ -19,12 +19,12 @@ which bazel
bazel version
cd $(dirname $0)/../..
bazel test --test_output=errors :all
bazel test --test_output=errors ...
if [[ $(uname) = "Linux" ]]; then
# Verify the ASAN build. Have to exclude test_conformance_upb as protobuf
# currently leaks memory in the conformance test runner.
bazel test --copt=-fsanitize=address --linkopt=-fsanitize=address --test_output=errors :all
bazel test --copt=-fsanitize=address --linkopt=-fsanitize=address --test_output=errors ...
# Verify the UBSan build. Have to exclude Lua as the version we are using
# fails some UBSan tests.
@ -35,5 +35,5 @@ if [[ $(uname) = "Linux" ]]; then
fi
if which valgrind; then
bazel test --run_under='valgrind --leak-check=full --error-exitcode=1' :all -- -:test_conformance_upb -:cmake_build
bazel test --run_under='valgrind --leak-check=full --error-exitcode=1' ... -- -tests:test_conformance_upb -cmake:cmake_build
fi

@ -0,0 +1,283 @@
load(
"//bazel:build_defs.bzl",
"UPB_DEFAULT_COPTS",
"UPB_DEFAULT_CPPOPTS",
"make_shell_script",
)
load(
"//bazel:upb_proto_library.bzl",
"upb_proto_library",
"upb_proto_reflection_library",
)
licenses(["notice"])
config_setting(
name = "fuzz",
values = {"define": "fuzz=true"},
)
cc_library(
name = "upb_test",
testonly = 1,
srcs = [
"testmain.cc",
],
hdrs = [
"test_util.h",
"upb_test.h",
],
copts = UPB_DEFAULT_CPPOPTS,
deps = [
"//:handlers",
"//:port",
"//:upb",
],
)
proto_library(
name = "test_proto",
testonly = 1,
srcs = ["test.proto"],
)
upb_proto_library(
name = "test_upb_proto",
testonly = 1,
deps = [":test_proto"],
)
cc_test(
name = "test_generated_code",
srcs = ["test_generated_code.c"],
copts = UPB_DEFAULT_COPTS,
deps = [
":empty_upbdefs_proto",
":test_messages_proto3_proto_upb",
":test_upb_proto",
":upb_test",
],
)
proto_library(
name = "empty_proto",
srcs = ["empty.proto"],
)
upb_proto_reflection_library(
name = "empty_upbdefs_proto",
testonly = 1,
deps = [":empty_proto"],
)
upb_proto_library(
name = "test_messages_proto3_proto_upb",
testonly = 1,
deps = ["@com_google_protobuf//:test_messages_proto3_proto"],
)
proto_library(
name = "test_decoder_proto",
srcs = [
"pb/test_decoder.proto",
],
)
upb_proto_reflection_library(
name = "test_decoder_upb_proto",
deps = [":test_decoder_proto"],
)
cc_test(
name = "test_decoder",
srcs = ["pb/test_decoder.cc"],
copts = UPB_DEFAULT_CPPOPTS,
deps = [
":test_decoder_upb_proto",
":upb_test",
"//:handlers",
"//:port",
"//:upb",
"//:upb_pb",
],
)
proto_library(
name = "test_cpp_proto",
srcs = [
"test_cpp.proto",
],
)
upb_proto_reflection_library(
name = "test_cpp_upb_proto",
deps = ["test_cpp_proto"],
)
cc_test(
name = "test_cpp",
srcs = ["test_cpp.cc"],
copts = UPB_DEFAULT_CPPOPTS,
deps = [
":test_cpp_upb_proto",
":upb_test",
"//:handlers",
"//:port",
"//:reflection",
"//:upb",
"//:upb_pb",
],
)
cc_test(
name = "test_table",
srcs = ["test_table.cc"],
copts = UPB_DEFAULT_CPPOPTS,
deps = [
":upb_test",
"//:port",
"//:table",
"//:upb",
],
)
# OSS-Fuzz test
cc_binary(
name = "file_descriptor_parsenew_fuzzer",
testonly = 1,
srcs = ["file_descriptor_parsenew_fuzzer.cc"],
copts = UPB_DEFAULT_CPPOPTS + select({
"//conditions:default": [],
":fuzz": ["-fsanitize=fuzzer,address"],
}),
defines = select({
"//conditions:default": [],
":fuzz": ["HAVE_FUZZER"],
}),
deps = [
"//:descriptor_upb_proto",
"//:upb",
],
)
# copybara:strip_for_google3_begin
cc_test(
name = "test_encoder",
srcs = ["pb/test_encoder.cc"],
copts = UPB_DEFAULT_CPPOPTS,
deps = [
":upb_test",
"//:descriptor_upb_proto",
"//:descriptor_upb_proto_reflection",
"//:upb",
"//:upb_pb",
],
)
proto_library(
name = "test_json_enum_from_separate",
srcs = ["json/enum_from_separate_file.proto"],
deps = [":test_json_proto"],
)
proto_library(
name = "test_json_proto",
srcs = ["json/test.proto"],
)
upb_proto_reflection_library(
name = "test_json_upb_proto_reflection",
deps = ["test_json_proto"],
)
upb_proto_library(
name = "test_json_enum_from_separate_upb_proto",
deps = [":test_json_enum_from_separate"],
)
upb_proto_library(
name = "test_json_upb_proto",
deps = [":test_json_proto"],
)
cc_test(
name = "test_json",
srcs = [
"json/test_json.cc",
],
copts = UPB_DEFAULT_CPPOPTS,
deps = [
":test_json_upb_proto",
":test_json_upb_proto_reflection",
":upb_test",
"//:upb_json",
],
)
# copybara:strip_end
upb_proto_library(
name = "conformance_proto_upb",
testonly = 1,
deps = ["@com_google_protobuf//:conformance_proto"],
)
upb_proto_reflection_library(
name = "conformance_proto_upbdefs",
testonly = 1,
deps = ["@com_google_protobuf//:conformance_proto"],
)
upb_proto_reflection_library(
name = "test_messages_proto2_upbdefs",
testonly = 1,
deps = ["@com_google_protobuf//:test_messages_proto2_proto"],
)
upb_proto_reflection_library(
name = "test_messages_proto3_upbdefs",
testonly = 1,
deps = ["@com_google_protobuf//:test_messages_proto3_proto"],
)
cc_binary(
name = "conformance_upb",
testonly = 1,
srcs = [
"conformance_upb.c",
],
copts = UPB_DEFAULT_COPTS,
data = [
"conformance_upb_failures.txt",
],
deps = [
":conformance_proto_upb",
":conformance_proto_upbdefs",
":test_messages_proto2_upbdefs",
":test_messages_proto3_upbdefs",
"//:json",
"//:port",
"//:reflection",
"//:textformat",
"//:upb",
],
)
make_shell_script(
name = "gen_test_conformance_upb",
out = "test_conformance_upb.sh",
contents = "external/com_google_protobuf/conformance_test_runner " +
" --enforce_recommended " +
" --failure_list ./tests/conformance_upb_failures.txt" +
" ./tests/conformance_upb",
)
sh_test(
name = "test_conformance_upb",
srcs = ["test_conformance_upb.sh"],
data = [
"conformance_upb_failures.txt",
":conformance_upb",
"@com_google_protobuf//:conformance_test_runner",
],
deps = ["@bazel_tools//tools/bash/runfiles"],
)

@ -0,0 +1,62 @@
load(
"//upb/bindings/lua:lua_proto_library.bzl",
"lua_proto_library",
)
load(
"//bazel:build_defs.bzl",
"UPB_DEFAULT_COPTS",
)
licenses(["notice"])
cc_test(
name = "test_lua",
srcs = ["main.c"],
data = [
"test_upb.lua",
":descriptor_proto_lua",
":test_messages_proto2_proto_lua",
":test_messages_proto3_proto_lua",
":test_proto_lua",
"//:third_party/lunit/console.lua",
"//:third_party/lunit/lunit.lua",
"//upb/bindings/lua:upb.lua",
"@com_google_protobuf//:conformance_proto",
"@com_google_protobuf//:descriptor_proto",
],
copts = UPB_DEFAULT_COPTS,
linkstatic = 1,
deps = [
"//upb/bindings/lua:lupb",
"@lua//:liblua",
],
)
proto_library(
name = "test_proto",
testonly = 1,
srcs = ["test.proto"],
)
lua_proto_library(
name = "test_proto_lua",
testonly = 1,
deps = [":test_proto"],
)
lua_proto_library(
name = "descriptor_proto_lua",
deps = ["@com_google_protobuf//:descriptor_proto"],
)
lua_proto_library(
name = "test_messages_proto3_proto_lua",
testonly = 1,
deps = ["@com_google_protobuf//:test_messages_proto3_proto"],
)
lua_proto_library(
name = "test_messages_proto2_proto_lua",
testonly = 1,
deps = ["@com_google_protobuf//:test_messages_proto2_proto"],
)

@ -34,7 +34,7 @@ const char *init =
"upb/bindings/lua/?.lua"
"'";
int main() {
int main(int argc, char **argv) {
int ret = 0;
L = luaL_newstate();
luaL_openlibs(L);

@ -0,0 +1,8 @@
syntax = "proto2";
package upb_test;
message MapTest {
map<string, double> map_string_double = 1;
}

@ -1,7 +1,7 @@
local upb = require "lupb"
local lunit = require "lunit"
local upb_test = require "tests.test_pb"
local upb_test = require "tests.bindings.lua.test_pb"
local test_messages_proto3 = require "google.protobuf.test_messages_proto3_pb"
local test_messages_proto2 = require "google.protobuf.test_messages_proto2_pb"
local descriptor = require "google.protobuf.descriptor_pb"

@ -43,7 +43,6 @@
#else // AMALGAMATED
#include "upb/handlers.h"
#include "upb/pb/decoder.h"
#include "upb/pb/varint.int.h"
#include "upb/upb.h"
#endif // !AMALGAMATED
@ -103,7 +102,7 @@ using std::string;
void vappendf(string* str, const char *format, va_list args) {
va_list copy;
_upb_va_copy(copy, args);
va_copy(copy, args);
int count = vsnprintf(NULL, 0, format, args);
if (count >= 0)
@ -136,6 +135,29 @@ void PrintBinary(const string& str) {
}
}
#define UPB_PB_VARINT_MAX_LEN 10
static size_t upb_vencode64(uint64_t val, char *buf) {
size_t i;
if (val == 0) { buf[0] = 0; return 1; }
i = 0;
while (val) {
uint8_t byte = val & 0x7fU;
val >>= 7;
if (val) byte |= 0x80U;
buf[i++] = byte;
}
return i;
}
static uint32_t upb_zzenc_32(int32_t n) {
return ((uint32_t)n << 1) ^ (n >> 31);
}
static uint64_t upb_zzenc_64(int64_t n) {
return ((uint64_t)n << 1) ^ (n >> 63);
}
/* Routines for building arbitrary protos *************************************/
const string empty;

@ -5,12 +5,65 @@
#include "google/protobuf/descriptor.upbdefs.h"
#include "tests/test_util.h"
#include "tests/upb_test.h"
#include "upb/bindings/stdc++/string.h"
#include "upb/pb/decoder.h"
#include "upb/pb/encoder.h"
#include "upb/port_def.inc"
#include "upb/upb.hpp"
template <class T>
class FillStringHandler {
public:
static void SetHandler(upb_byteshandler* handler) {
upb_byteshandler_setstartstr(handler, &FillStringHandler::StartString,
NULL);
upb_byteshandler_setstring(handler, &FillStringHandler::StringBuf, NULL);
}
private:
// TODO(haberman): add UpbBind/UpbMakeHandler support to BytesHandler so these
// can be prettier callbacks.
static void* StartString(void *c, const void *hd, size_t size) {
UPB_UNUSED(hd);
UPB_UNUSED(size);
T* str = static_cast<T*>(c);
str->clear();
return c;
}
static size_t StringBuf(void* c, const void* hd, const char* buf, size_t n,
const upb_bufhandle* h) {
UPB_UNUSED(hd);
UPB_UNUSED(h);
T* str = static_cast<T*>(c);
try {
str->append(buf, n);
return n;
} catch (const std::exception&) {
return 0;
}
}
};
class StringSink {
public:
template <class T>
explicit StringSink(T* target) {
// TODO(haberman): we need to avoid rebuilding a new handler every time,
// but with class globals disallowed for google3 C++ this is tricky.
upb_byteshandler_init(&handler_);
FillStringHandler<T>::SetHandler(&handler_);
input_.Reset(&handler_, target);
}
upb::BytesSink input() { return input_; }
private:
upb_byteshandler handler_;
upb::BytesSink input_;
};
void test_pb_roundtrip() {
std::string input(
google_protobuf_descriptor_proto_upbdefinit.descriptor.data,
@ -29,7 +82,7 @@ void test_pb_roundtrip() {
const upb::pb::DecoderMethodPtr method = decoder_cache.Get(md);
std::string output;
upb::StringSink string_sink(&output);
StringSink string_sink(&output);
upb::pb::EncoderPtr encoder =
upb::pb::EncoderPtr::Create(&arena, encoder_handlers, string_sink.input());
upb::pb::DecoderPtr decoder =

@ -1,126 +0,0 @@
#include <stdio.h>
#include "upb/pb/varint.int.h"
#include "tests/upb_test.h"
#include "upb/port_def.inc"
/* Test that we can round-trip from int->varint->int. */
static void test_varint_for_num(upb_decoderet (*decoder)(const char*),
uint64_t num) {
char buf[16];
size_t bytes;
upb_decoderet r;
memset(buf, 0xff, sizeof(buf));
bytes = upb_vencode64(num, buf);
if (num <= UINT32_MAX) {
uint64_t encoded = upb_vencode32((uint32_t)num);
char buf2[16];
upb_decoderet r;
memset(buf2, 0, sizeof(buf2));
memcpy(&buf2, &encoded, 8);
#ifdef UPB_BIG_ENDIAN
char swap[8];
swap[0] = buf2[7];
swap[1] = buf2[6];
swap[2] = buf2[5];
swap[3] = buf2[4];
swap[4] = buf2[3];
swap[5] = buf2[2];
swap[6] = buf2[1];
swap[7] = buf2[0];
buf2[0] = swap[0];
buf2[1] = swap[1];
buf2[2] = swap[2];
buf2[3] = swap[3];
buf2[4] = swap[4];
buf2[5] = swap[5];
buf2[6] = swap[6];
buf2[7] = swap[7];
#endif
r = decoder(buf2);
ASSERT(r.val == num);
ASSERT(r.p == buf2 + upb_value_size(encoded));
ASSERT(upb_zzenc_32(upb_zzdec_32((uint32_t)num)) == num);
}
r = decoder(buf);
ASSERT(r.val == num);
ASSERT(r.p == buf + bytes);
ASSERT(upb_zzenc_64(upb_zzdec_64(num)) == num);
}
/* Making up for the lack of 64-bit constants in C89. */
static uint64_t make_u64(uint32_t high, uint32_t low) {
uint64_t ret = high;
ret = (ret << 32) | low;
return ret;
}
static void test_varint_decoder(upb_decoderet (*decoder)(const char*)) {
#define TEST(bytes, expected_val) {\
size_t n = sizeof(bytes) - 1; /* for NULL */ \
char buf[UPB_PB_VARINT_MAX_LEN]; \
upb_decoderet r; \
memset(buf, 0xff, sizeof(buf)); \
memcpy(buf, bytes, n); \
r = decoder(buf); \
ASSERT(r.val == expected_val); \
ASSERT(r.p == buf + n); \
}
uint64_t num;
char twelvebyte[16] = {-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 1, 1};
const char *twelvebyte_buf = twelvebyte;
/* A varint that terminates before hitting the end of the provided buffer,
* but in too many bytes (11 instead of 10). */
upb_decoderet r = decoder(twelvebyte_buf);
ASSERT(r.p == NULL);
TEST("\x00", 0UL);
TEST("\x01", 1UL);
TEST("\x81\x14", 0xa01UL);
TEST("\x81\x03", 0x181UL);
TEST("\x81\x83\x07", 0x1c181UL);
TEST("\x81\x83\x87\x0f", 0x1e1c181UL);
TEST("\x81\x83\x87\x8f\x1f", make_u64(0x1, 0xf1e1c181UL));
TEST("\x81\x83\x87\x8f\x9f\x3f", make_u64(0x1f9, 0xf1e1c181UL));
TEST("\x81\x83\x87\x8f\x9f\xbf\x7f", make_u64(0x1fdf9, 0xf1e1c181UL));
TEST("\x81\x83\x87\x8f\x9f\xbf\xff\x01", make_u64(0x3fdf9, 0xf1e1c181UL));
TEST("\x81\x83\x87\x8f\x9f\xbf\xff\x81\x03",
make_u64(0x303fdf9, 0xf1e1c181UL));
TEST("\x81\x83\x87\x8f\x9f\xbf\xff\x81\x83\x07",
make_u64(0x8303fdf9, 0xf1e1c181UL));
#undef TEST
for (num = 5; num * 1.5 < UINT64_MAX; num *= 1.5) {
test_varint_for_num(decoder, num);
}
test_varint_for_num(decoder, 0);
}
#define TEST_VARINT_DECODER(decoder) \
/* Create non-inline versions for convenient inspection of assembly language \
* output. */ \
upb_decoderet _upb_vdecode_ ## decoder(const char *p) { \
return upb_vdecode_ ## decoder(p); \
} \
void test_ ## decoder(void) { \
test_varint_decoder(&_upb_vdecode_ ## decoder); \
} \
TEST_VARINT_DECODER(check2_branch32)
TEST_VARINT_DECODER(check2_branch64)
int run_tests(int argc, char *argv[]) {
UPB_UNUSED(argc);
UPB_UNUSED(argv);
test_check2_branch32();
test_check2_branch64();
return 0;
}

@ -952,6 +952,31 @@ void TestArena() {
}
}
void TestInlinedArena() {
int n = 100000;
struct Decrementer {
Decrementer(int* _p) : p(_p) {}
~Decrementer() { (*p)--; }
int* p;
};
{
upb::InlinedArena<1024> arena;
for (int i = 0; i < n; i++) {
arena.Own(new Decrementer(&n));
// Intersperse allocation and ensure we can write to it.
int* val = static_cast<int*>(upb_arena_malloc(arena.ptr(), sizeof(int)));
*val = i;
}
// Test a large allocation.
upb_arena_malloc(arena.ptr(), 1000000);
}
ASSERT(n == 0);
}
extern "C" {
int run_tests() {

@ -24,7 +24,7 @@ const int32_t test_int32_2 = -20;
const int32_t test_int32_3 = 30;
const int32_t test_int32_4 = -40;
static void test_scalars() {
static void test_scalars(void) {
upb_arena *arena = upb_arena_new();
protobuf_test_messages_proto3_TestAllTypesProto3 *msg =
protobuf_test_messages_proto3_TestAllTypesProto3_new(arena);
@ -117,7 +117,7 @@ static void check_string_map_one_entry(
ASSERT(!const_ent);
}
static void test_string_double_map() {
static void test_string_double_map(void) {
upb_arena *arena = upb_arena_new();
upb_strview serialized;
upb_test_MapTest *msg = upb_test_MapTest_new(arena);
@ -141,7 +141,7 @@ static void test_string_double_map() {
upb_arena_free(arena);
}
static void test_string_map() {
static void test_string_map(void) {
upb_arena *arena = upb_arena_new();
protobuf_test_messages_proto3_TestAllTypesProto3 *msg =
protobuf_test_messages_proto3_TestAllTypesProto3_new(arena);
@ -259,7 +259,7 @@ static void check_int32_map_one_entry(
ASSERT(!const_ent);
}
static void test_int32_map() {
static void test_int32_map(void) {
upb_arena *arena = upb_arena_new();
protobuf_test_messages_proto3_TestAllTypesProto3 *msg =
protobuf_test_messages_proto3_TestAllTypesProto3_new(arena);
@ -328,7 +328,7 @@ static void test_int32_map() {
upb_arena_free(arena);
}
void test_repeated() {
void test_repeated(void) {
upb_arena *arena = upb_arena_new();
protobuf_test_messages_proto3_TestAllTypesProto3 *msg =
protobuf_test_messages_proto3_TestAllTypesProto3_new(arena);
@ -347,7 +347,7 @@ void test_repeated() {
upb_arena_free(arena);
}
void test_null_decode_buf() {
void test_null_decode_buf(void) {
upb_arena *arena = upb_arena_new();
protobuf_test_messages_proto3_TestAllTypesProto3 *msg =
protobuf_test_messages_proto3_TestAllTypesProto3_parse(NULL, 0, arena);
@ -359,7 +359,7 @@ void test_null_decode_buf() {
upb_arena_free(arena);
}
void test_status_truncation() {
void test_status_truncation(void) {
int i, j;
upb_status status;
upb_status status2;

@ -0,0 +1,40 @@
load(
"//bazel:build_defs.bzl",
"UPB_DEFAULT_COPTS",
"UPB_DEFAULT_CPPOPTS",
)
licenses(["notice"])
cc_library(
name = "lupb",
srcs = [
"def.c",
"msg.c",
"upb.c",
],
hdrs = [
"upb.h",
],
copts = UPB_DEFAULT_COPTS,
visibility = ["//visibility:public"],
deps = [
"//:reflection",
"//:textformat",
"//:upb",
"@lua//:liblua",
],
)
cc_binary(
name = "protoc-gen-lua",
srcs = ["upbc.cc"],
copts = UPB_DEFAULT_CPPOPTS,
visibility = ["//visibility:public"],
deps = [
"@com_google_absl//absl/strings",
"@com_google_protobuf//:protoc_lib",
],
)
exports_files(["upb.lua"])

@ -56,8 +56,8 @@ static void lupb_wrapper_pushwrapper(lua_State *L, int narg, const void *def,
* wrapper for field |f| of this msgdef.
*/
void lupb_msgdef_pushsubmsgdef(lua_State *L, const upb_fielddef *f) {
assert(luaL_testudata(L, -1, LUPB_MSGDEF));
const upb_msgdef *m = upb_fielddef_msgsubdef(f);
assert(m);
assert(upb_fielddef_containingtype(f) == lupb_msgdef_check(L, -1));
lupb_wrapper_pushwrapper(L, -1, m, LUPB_MSGDEF);
lua_replace(L, -2); /* Replace msgdef with submsgdef. */
@ -251,8 +251,8 @@ static int lupb_oneofiter_next(lua_State *L) {
}
static int lupb_oneofdef_fields(lua_State *L) {
lupb_oneofdef_check(L, 1);
int *index = lua_newuserdata(L, sizeof(int));
lupb_oneofdef_check(L, 1);
*index = 0;
/* Closure upvalues are: oneofdef, index. */
@ -408,8 +408,8 @@ static int lupb_msgfielditer_next(lua_State *L) {
}
static int lupb_msgdef_fields(lua_State *L) {
lupb_msgdef_check(L, 1);
int *index = lua_newuserdata(L, sizeof(int));
lupb_msgdef_check(L, 1);
*index = 0;
/* Closure upvalues are: msgdef, index. */
@ -441,8 +441,8 @@ static int lupb_msgoneofiter_next(lua_State *L) {
}
static int lupb_msgdef_oneofs(lua_State *L) {
lupb_msgdef_check(L, 1);
int *index = lua_newuserdata(L, sizeof(int));
lupb_msgdef_check(L, 1);
*index = 0;
/* Closure upvalues are: msgdef, index. */
@ -750,7 +750,7 @@ static int lupb_symtab_addfile(lua_State *L) {
size_t len;
upb_symtab *s = lupb_symtab_check(L, 1);
const char *str = luaL_checklstring(L, 2, &len);
upb_arena *arena = lupb_arena_pushnew(L);;
upb_arena *arena = lupb_arena_pushnew(L);
const google_protobuf_FileDescriptorProto *file;
const upb_filedef *file_def;
upb_status status;
@ -776,7 +776,7 @@ static int lupb_symtab_addset(lua_State *L) {
google_protobuf_FileDescriptorSet *set;
upb_symtab *s = lupb_symtab_check(L, 1);
const char *str = luaL_checklstring(L, 2, &len);
upb_arena *arena = lupb_arena_pushnew(L);;
upb_arena *arena = lupb_arena_pushnew(L);
upb_status status;
upb_status_clear(&status);

@ -82,7 +82,7 @@ _lua_proto_library_aspect = aspect(
"_upbc": attr.label(
executable = True,
cfg = "host",
default = "//:protoc-gen-lua",
default = "//upb/bindings/lua:protoc-gen-lua",
),
"_protoc": attr.label(
executable = True,

@ -564,8 +564,8 @@ static int lupb_mapiter_next(lua_State *L) {
* pairs(map)
*/
static int lupb_map_pairs(lua_State *L) {
lupb_map_check(L, 1);
size_t *iter = lua_newuserdata(L, sizeof(*iter));
lupb_map_check(L, 1);
*iter = UPB_MAP_BEGIN;
lua_pushvalue(L, 1);

@ -1,69 +0,0 @@
#ifndef UPB_STDCPP_H_
#define UPB_STDCPP_H_
#include "upb/sink.h"
#include "upb/port_def.inc"
namespace upb {
template <class T>
class FillStringHandler {
public:
static void SetHandler(upb_byteshandler* handler) {
upb_byteshandler_setstartstr(handler, &FillStringHandler::StartString,
NULL);
upb_byteshandler_setstring(handler, &FillStringHandler::StringBuf, NULL);
}
private:
// TODO(haberman): add UpbBind/UpbMakeHandler support to BytesHandler so these
// can be prettier callbacks.
static void* StartString(void *c, const void *hd, size_t size) {
UPB_UNUSED(hd);
UPB_UNUSED(size);
T* str = static_cast<T*>(c);
str->clear();
return c;
}
static size_t StringBuf(void* c, const void* hd, const char* buf, size_t n,
const upb_bufhandle* h) {
UPB_UNUSED(hd);
UPB_UNUSED(h);
T* str = static_cast<T*>(c);
try {
str->append(buf, n);
return n;
} catch (const std::exception&) {
return 0;
}
}
};
class StringSink {
public:
template <class T>
explicit StringSink(T* target) {
// TODO(haberman): we need to avoid rebuilding a new handler every time,
// but with class globals disallowed for google3 C++ this is tricky.
upb_byteshandler_init(&handler_);
FillStringHandler<T>::SetHandler(&handler_);
input_.Reset(&handler_, target);
}
BytesSink input() { return input_; }
private:
upb_byteshandler handler_;
BytesSink input_;
};
} // namespace upb
#include "upb/port_undef.inc"
#endif // UPB_STDCPP_H_

@ -139,10 +139,14 @@ static const int8_t delim_ops[37] = {
/* Data pertaining to the parse. */
typedef struct {
const char *limit; /* End of delimited region or end of buffer. */
upb_arena arena;
const char *end; /* Can read up to 16 bytes slop beyond this. */
const char *limit_ptr; /* = end + UPB_MIN(limit, 0) */
int limit; /* Submessage limit relative to end. */
int depth;
uint32_t end_group; /* Set to field number of END_GROUP tag, if any. */
bool alias;
char patch[32];
upb_arena arena;
jmp_buf err;
} upb_decstate;
@ -150,7 +154,7 @@ typedef union {
bool bool_val;
uint32_t uint32_val;
uint64_t uint64_val;
upb_strview str_val;
uint32_t size;
} wireval;
static const char *decode_msg(upb_decstate *d, const char *ptr, upb_msg *msg,
@ -200,41 +204,48 @@ static bool decode_reserve(upb_decstate *d, upb_array *arr, size_t elem) {
return need_realloc;
}
typedef struct {
const char *ptr;
uint64_t val;
} decode_vret;
UPB_NOINLINE
static const char *decode_longvarint64(upb_decstate *d, const char *ptr,
const char *limit, uint64_t *val) {
uint8_t byte;
int bitpos = 0;
uint64_t out = 0;
do {
if (bitpos >= 70 || ptr == limit) decode_err(d);
byte = *ptr;
out |= (uint64_t)(byte & 0x7F) << bitpos;
ptr++;
bitpos += 7;
} while (byte & 0x80);
*val = out;
return ptr;
static decode_vret decode_longvarint64(const char *ptr, uint64_t val) {
decode_vret ret = {NULL, 0};
uint64_t byte;
int i;
for (i = 1; i < 10; i++) {
byte = (uint8_t)ptr[i];
val += (byte - 1) << (i * 7);
if (!(byte & 0x80)) {
ret.ptr = ptr + i + 1;
ret.val = val;
return ret;
}
}
return ret;
}
UPB_FORCEINLINE
static const char *decode_varint64(upb_decstate *d, const char *ptr,
const char *limit, uint64_t *val) {
if (UPB_LIKELY(ptr < limit && (*ptr & 0x80) == 0)) {
*val = (uint8_t)*ptr;
uint64_t *val) {
uint64_t byte = (uint8_t)*ptr;
if (UPB_LIKELY((byte & 0x80) == 0)) {
*val = byte;
return ptr + 1;
} else {
return decode_longvarint64(d, ptr, limit, val);
decode_vret res = decode_longvarint64(ptr, byte);
if (!res.ptr) decode_err(d);
*val = res.val;
return res.ptr;
}
}
UPB_FORCEINLINE
static const char *decode_varint32(upb_decstate *d, const char *ptr,
const char *limit, uint32_t *val) {
uint32_t *val) {
uint64_t u64;
ptr = decode_varint64(d, ptr, limit, &u64);
ptr = decode_varint64(d, ptr, &u64);
if (u64 > UINT32_MAX) decode_err(d);
*val = (uint32_t)u64;
return ptr;
@ -287,17 +298,82 @@ static upb_msg *decode_newsubmsg(upb_decstate *d, const upb_msglayout *layout,
return _upb_msg_new_inl(subl, &d->arena);
}
static void decode_tosubmsg(upb_decstate *d, upb_msg *submsg,
const upb_msglayout *layout,
const upb_msglayout_field *field, upb_strview val) {
static int decode_pushlimit(upb_decstate *d, const char *ptr, int size) {
int limit = size + (int)(ptr - d->end);
int delta = d->limit - limit;
d->limit = limit;
d->limit_ptr = d->end + UPB_MIN(0, limit);
return delta;
}
static void decode_poplimit(upb_decstate *d, int saved_delta) {
d->limit += saved_delta;
d->limit_ptr = d->end + UPB_MIN(0, d->limit);
}
typedef struct {
bool ok;
const char *ptr;
} decode_doneret;
UPB_NOINLINE
static const char *decode_isdonefallback(upb_decstate *d, const char *ptr,
int overrun) {
if (overrun < d->limit) {
/* Need to copy remaining data into patch buffer. */
UPB_ASSERT(overrun < 16);
memset(d->patch + 16, 0, 16);
memcpy(d->patch, d->end, 16);
ptr = &d->patch[0] + overrun;
d->end = &d->patch[16];
d->limit -= 16;
d->limit_ptr = d->end + d->limit;
d->alias = false;
UPB_ASSERT(ptr < d->limit_ptr);
return ptr;
} else {
decode_err(d);
}
}
UPB_FORCEINLINE
static bool decode_isdone(upb_decstate *d, const char **ptr) {
int overrun = *ptr - d->end;
if (UPB_LIKELY(*ptr < d->limit_ptr)) {
return false;
} else if (UPB_LIKELY(overrun == d->limit)) {
return true;
} else {
*ptr = decode_isdonefallback(d, *ptr, overrun);
return false;
}
}
static const char *decode_readstr(upb_decstate *d, const char *ptr, int size,
upb_strview *str) {
if (d->alias) {
str->data = ptr;
} else {
char *data = upb_arena_malloc(&d->arena, size);
if (!data) decode_err(d);
memcpy(data, ptr, size);
str->data = data;
}
str->size = size;
return ptr + size;
}
static const char *decode_tosubmsg(upb_decstate *d, const char *ptr,
upb_msg *submsg, const upb_msglayout *layout,
const upb_msglayout_field *field, int size) {
const upb_msglayout *subl = layout->submsgs[field->submsg_index];
const char *saved_limit = d->limit;
int saved_delta = decode_pushlimit(d, ptr, size);
if (--d->depth < 0) decode_err(d);
d->limit = val.data + val.size;
decode_msg(d, val.data, submsg, subl);
d->limit = saved_limit;
ptr = decode_msg(d, ptr, submsg, subl);
decode_poplimit(d, saved_delta);
if (d->end_group != 0) decode_err(d);
d->depth++;
return ptr;
}
static const char *decode_group(upb_decstate *d, const char *ptr,
@ -345,15 +421,14 @@ static const char *decode_toarray(upb_decstate *d, const char *ptr,
memcpy(mem, &val, 1 << op);
return ptr;
case OP_STRING:
decode_verifyutf8(d, val.str_val.data, val.str_val.size);
decode_verifyutf8(d, ptr, val.size);
/* Fallthrough. */
case OP_BYTES:
case OP_BYTES: {
/* Append bytes. */
mem =
UPB_PTR_AT(_upb_array_ptr(arr), arr->len * sizeof(upb_strview), void);
upb_strview *str = (upb_strview*)_upb_array_ptr(arr) + arr->len;
arr->len++;
memcpy(mem, &val, sizeof(upb_strview));
return ptr;
return decode_readstr(d, ptr, val.size, str);
}
case OP_SUBMSG: {
/* Append submessage / group. */
upb_msg *submsg = decode_newsubmsg(d, layout, field);
@ -361,26 +436,25 @@ static const char *decode_toarray(upb_decstate *d, const char *ptr,
submsg;
arr->len++;
if (UPB_UNLIKELY(field->descriptortype == UPB_DTYPE_GROUP)) {
ptr = decode_togroup(d, ptr, submsg, layout, field);
return decode_togroup(d, ptr, submsg, layout, field);
} else {
decode_tosubmsg(d, submsg, layout, field, val.str_val);
return decode_tosubmsg(d, ptr, submsg, layout, field, val.size);
}
return ptr;
}
case OP_FIXPCK_LG2(2):
case OP_FIXPCK_LG2(3): {
/* Fixed packed. */
int lg2 = op - OP_FIXPCK_LG2(0);
int mask = (1 << lg2) - 1;
size_t count = val.str_val.size >> lg2;
if ((val.str_val.size & mask) != 0) {
size_t count = val.size >> lg2;
if ((val.size & mask) != 0) {
decode_err(d); /* Length isn't a round multiple of elem size. */
}
decode_reserve(d, arr, count);
mem = UPB_PTR_AT(_upb_array_ptr(arr), arr->len << lg2, void);
arr->len += count;
memcpy(mem, val.str_val.data, val.str_val.size);
return ptr;
memcpy(mem, ptr, val.size); /* XXX: ptr boundary. */
return ptr + val.size;
}
case OP_VARPCK_LG2(0):
case OP_VARPCK_LG2(2):
@ -388,12 +462,11 @@ static const char *decode_toarray(upb_decstate *d, const char *ptr,
/* Varint packed. */
int lg2 = op - OP_VARPCK_LG2(0);
int scale = 1 << lg2;
const char *ptr = val.str_val.data;
const char *end = ptr + val.str_val.size;
int saved_limit = decode_pushlimit(d, ptr, val.size);
char *out = UPB_PTR_AT(_upb_array_ptr(arr), arr->len << lg2, void);
while (ptr < end) {
while (!decode_isdone(d, &ptr)) {
wireval elem;
ptr = decode_varint64(d, ptr, end, &elem.uint64_val);
ptr = decode_varint64(d, ptr, &elem.uint64_val);
decode_munge(field->descriptortype, &elem);
if (decode_reserve(d, arr, 1)) {
out = UPB_PTR_AT(_upb_array_ptr(arr), arr->len << lg2, void);
@ -402,7 +475,7 @@ static const char *decode_toarray(upb_decstate *d, const char *ptr,
memcpy(out, &elem, scale);
out += scale;
}
if (ptr != end) decode_err(d);
decode_poplimit(d, saved_limit);
return ptr;
}
default:
@ -410,9 +483,9 @@ static const char *decode_toarray(upb_decstate *d, const char *ptr,
}
}
static void decode_tomap(upb_decstate *d, upb_msg *msg,
const upb_msglayout *layout,
const upb_msglayout_field *field, wireval val) {
static const char *decode_tomap(upb_decstate *d, const char *ptr, upb_msg *msg,
const upb_msglayout *layout,
const upb_msglayout_field *field, wireval val) {
upb_map **map_p = UPB_PTR_AT(msg, field->offset, upb_map *);
upb_map *map = *map_p;
upb_map_entry ent;
@ -440,10 +513,9 @@ static void decode_tomap(upb_decstate *d, upb_msg *msg,
ent.v.val = upb_value_ptr(_upb_msg_new(entry->submsgs[0], &d->arena));
}
decode_tosubmsg(d, &ent.k, layout, field, val.str_val);
/* Insert into map. */
ptr = decode_tosubmsg(d, ptr, &ent.k, layout, field, val.size);
_upb_map_set(map, &ent.k, map->key_size, &ent.v, map->val_size, &d->arena);
return ptr;
}
static const char *decode_tomsg(upb_decstate *d, const char *ptr, upb_msg *msg,
@ -477,16 +549,15 @@ static const char *decode_tomsg(upb_decstate *d, const char *ptr, upb_msg *msg,
if (UPB_UNLIKELY(type == UPB_DTYPE_GROUP)) {
ptr = decode_togroup(d, ptr, submsg, layout, field);
} else {
decode_tosubmsg(d, submsg, layout, field, val.str_val);
ptr = decode_tosubmsg(d, ptr, submsg, layout, field, val.size);
}
break;
}
case OP_STRING:
decode_verifyutf8(d, val.str_val.data, val.str_val.size);
decode_verifyutf8(d, ptr, val.size);
/* Fallthrough. */
case OP_BYTES:
memcpy(mem, &val, sizeof(upb_strview));
break;
return decode_readstr(d, ptr, val.size, mem);
case OP_SCALAR_LG2(3):
memcpy(mem, &val, 8);
break;
@ -505,7 +576,7 @@ static const char *decode_tomsg(upb_decstate *d, const char *ptr, upb_msg *msg,
static const char *decode_msg(upb_decstate *d, const char *ptr, upb_msg *msg,
const upb_msglayout *layout) {
while (ptr < d->limit) {
while (!decode_isdone(d, &ptr)) {
uint32_t tag;
const upb_msglayout_field *field;
int field_number;
@ -514,7 +585,7 @@ static const char *decode_msg(upb_decstate *d, const char *ptr, upb_msg *msg,
wireval val;
int op;
ptr = decode_varint32(d, ptr, d->limit, &tag);
ptr = decode_varint32(d, ptr, &tag);
field_number = tag >> 3;
wire_type = tag & 7;
@ -522,12 +593,11 @@ static const char *decode_msg(upb_decstate *d, const char *ptr, upb_msg *msg,
switch (wire_type) {
case UPB_WIRE_TYPE_VARINT:
ptr = decode_varint64(d, ptr, d->limit, &val.uint64_val);
ptr = decode_varint64(d, ptr, &val.uint64_val);
op = varint_ops[field->descriptortype];
decode_munge(field->descriptortype, &val);
break;
case UPB_WIRE_TYPE_32BIT:
if (d->limit - ptr < 4) decode_err(d);
memcpy(&val.uint32_val, ptr, 4);
val.uint32_val = _upb_be_swap32(val.uint32_val);
ptr += 4;
@ -535,7 +605,6 @@ static const char *decode_msg(upb_decstate *d, const char *ptr, upb_msg *msg,
if (((1 << field->descriptortype) & fixed32_ok) == 0) goto unknown;
break;
case UPB_WIRE_TYPE_64BIT:
if (d->limit - ptr < 8) decode_err(d);
memcpy(&val.uint64_val, ptr, 8);
val.uint64_val = _upb_be_swap64(val.uint64_val);
ptr += 8;
@ -543,16 +612,12 @@ static const char *decode_msg(upb_decstate *d, const char *ptr, upb_msg *msg,
if (((1 << field->descriptortype) & fixed64_ok) == 0) goto unknown;
break;
case UPB_WIRE_TYPE_DELIMITED: {
uint32_t size;
int ndx = field->descriptortype;
if (_upb_isrepeated(field)) ndx += 18;
ptr = decode_varint32(d, ptr, d->limit, &size);
if (size >= INT32_MAX || (size_t)(d->limit - ptr) < size) {
ptr = decode_varint32(d, ptr, &val.size);
if (val.size >= INT32_MAX || ptr - d->end + val.size > d->limit) {
decode_err(d); /* Length overflow. */
}
val.str_val.data = ptr;
val.str_val.size = size;
ptr += size;
op = delim_ops[ndx];
break;
}
@ -576,7 +641,7 @@ static const char *decode_msg(upb_decstate *d, const char *ptr, upb_msg *msg,
ptr = decode_toarray(d, ptr, msg, layout, field, val, op);
break;
case _UPB_LABEL_MAP:
decode_tomap(d, msg, layout, field, val);
ptr = decode_tomap(d, ptr, msg, layout, field, val);
break;
default:
ptr = decode_tomsg(d, ptr, msg, layout, field, val, op);
@ -590,6 +655,7 @@ static const char *decode_msg(upb_decstate *d, const char *ptr, upb_msg *msg,
ptr = decode_group(d, ptr, NULL, NULL, field_number);
}
if (msg) {
if (wire_type == UPB_WIRE_TYPE_DELIMITED) ptr += val.size;
if (!_upb_msg_addunknown(msg, field_start, ptr - field_start,
&d->arena)) {
decode_err(d);
@ -598,7 +664,6 @@ static const char *decode_msg(upb_decstate *d, const char *ptr, upb_msg *msg,
}
}
if (ptr != d->limit) decode_err(d);
return ptr;
}
@ -607,9 +672,22 @@ bool upb_decode(const char *buf, size_t size, void *msg, const upb_msglayout *l,
bool ok;
upb_decstate state;
if (size == 0) return true;
if (size == 0) {
return true;
} else if (size < 16) {
memset(&state.patch, 0, 32);
memcpy(&state.patch, buf, size);
buf = state.patch;
state.end = buf + size;
state.limit = 0;
state.alias = false;
} else {
state.end = buf + size - 16;
state.limit = 16;
state.alias = true;
}
state.limit = buf + size;
state.limit_ptr = state.end;
state.depth = 64;
state.end_group = 0;
state.arena.head = arena->head;

@ -951,7 +951,7 @@ static bool parse_number_from_buffer(upb_json_parser *p, const char *buf,
upb_fieldtype_t type = upb_fielddef_type(p->top->f);
double val;
double dummy;
double inf = UPB_INFINITY;
double inf = INFINITY;
errno = 0;

@ -7,7 +7,9 @@
#include <ctype.h>
#include <inttypes.h>
#include <math.h>
#include <stdint.h>
#include <stdio.h>
#include <string.h>
#include <time.h>
@ -139,7 +141,7 @@ static void putstring(upb_json_printer *p, const char *buf, size_t len) {
char escape_buf[8];
if (!escape) {
unsigned char byte = (unsigned char)c;
_upb_snprintf(escape_buf, sizeof(escape_buf), "\\u%04x", (int)byte);
snprintf(escape_buf, sizeof(escape_buf), "\\u%04x", (int)byte);
escape = escape_buf;
}
@ -178,53 +180,53 @@ const char neginf[] = "\"-Infinity\"";
const char inf[] = "\"Infinity\"";
static size_t fmt_double(double val, char* buf, size_t length) {
if (val == UPB_INFINITY) {
if (val == INFINITY) {
CHKLENGTH(length >= strlen(inf));
strcpy(buf, inf);
return strlen(inf);
} else if (val == -UPB_INFINITY) {
} else if (val == -INFINITY) {
CHKLENGTH(length >= strlen(neginf));
strcpy(buf, neginf);
return strlen(neginf);
} else {
size_t n = _upb_snprintf(buf, length, "%.17g", val);
size_t n = snprintf(buf, length, "%.17g", val);
CHKLENGTH(n > 0 && n < length);
return n;
}
}
static size_t fmt_float(float val, char* buf, size_t length) {
size_t n = _upb_snprintf(buf, length, "%.8g", val);
size_t n = snprintf(buf, length, "%.8g", val);
CHKLENGTH(n > 0 && n < length);
return n;
}
static size_t fmt_bool(bool val, char* buf, size_t length) {
size_t n = _upb_snprintf(buf, length, "%s", (val ? "true" : "false"));
size_t n = snprintf(buf, length, "%s", (val ? "true" : "false"));
CHKLENGTH(n > 0 && n < length);
return n;
}
static size_t fmt_int64_as_number(int64_t val, char* buf, size_t length) {
size_t n = _upb_snprintf(buf, length, "%" PRId64, val);
size_t n = snprintf(buf, length, "%" PRId64, val);
CHKLENGTH(n > 0 && n < length);
return n;
}
static size_t fmt_uint64_as_number(uint64_t val, char* buf, size_t length) {
size_t n = _upb_snprintf(buf, length, "%" PRIu64, val);
size_t n = snprintf(buf, length, "%" PRIu64, val);
CHKLENGTH(n > 0 && n < length);
return n;
}
static size_t fmt_int64_as_string(int64_t val, char* buf, size_t length) {
size_t n = _upb_snprintf(buf, length, "\"%" PRId64 "\"", val);
size_t n = snprintf(buf, length, "\"%" PRId64 "\"", val);
CHKLENGTH(n > 0 && n < length);
return n;
}
static size_t fmt_uint64_as_string(uint64_t val, char* buf, size_t length) {
size_t n = _upb_snprintf(buf, length, "\"%" PRIu64 "\"", val);
size_t n = snprintf(buf, length, "\"%" PRIu64 "\"", val);
CHKLENGTH(n > 0 && n < length);
return n;
}
@ -870,12 +872,12 @@ static bool printer_enddurationmsg(void *closure, const void *handler_data,
return false;
}
_upb_snprintf(buffer, sizeof(buffer), "%ld", (long)p->seconds);
snprintf(buffer, sizeof(buffer), "%ld", (long)p->seconds);
base_len = strlen(buffer);
if (p->nanos != 0) {
char nanos_buffer[UPB_DURATION_MAX_NANO_LEN + 3];
_upb_snprintf(nanos_buffer, sizeof(nanos_buffer), "%.9f",
snprintf(nanos_buffer, sizeof(nanos_buffer), "%.9f",
p->nanos / 1000000000.0);
/* Remove trailing 0. */
for (i = UPB_DURATION_MAX_NANO_LEN + 2;
@ -949,8 +951,8 @@ static bool printer_endtimestampmsg(void *closure, const void *handler_data,
"%Y-%m-%dT%H:%M:%S", gmtime(&time));
if (p->nanos != 0) {
char nanos_buffer[UPB_TIMESTAMP_MAX_NANO_LEN + 3];
_upb_snprintf(nanos_buffer, sizeof(nanos_buffer), "%.9f",
p->nanos / 1000000000.0);
snprintf(nanos_buffer, sizeof(nanos_buffer), "%.9f",
p->nanos / 1000000000.0);
/* Remove trailing 0. */
for (i = UPB_TIMESTAMP_MAX_NANO_LEN + 2;
nanos_buffer[i] == '0'; i--) {

@ -5,6 +5,7 @@
#include <float.h>
#include <inttypes.h>
#include <limits.h>
#include <math.h>
#include <setjmp.h>
#include <stdlib.h>
#include <string.h>
@ -747,11 +748,11 @@ static upb_msgval jsondec_double(jsondec *d, const upb_fielddef *f) {
case JD_STRING:
str = jsondec_string(d);
if (jsondec_streql(str, "NaN")) {
val.double_val = UPB_NAN;
val.double_val = NAN;
} else if (jsondec_streql(str, "Infinity")) {
val.double_val = UPB_INFINITY;
val.double_val = INFINITY;
} else if (jsondec_streql(str, "-Infinity")) {
val.double_val = -UPB_INFINITY;
val.double_val = -INFINITY;
} else {
val.double_val = strtod(str.data, NULL);
}
@ -761,7 +762,7 @@ static upb_msgval jsondec_double(jsondec *d, const upb_fielddef *f) {
}
if (upb_fielddef_type(f) == UPB_TYPE_FLOAT) {
if (val.double_val != UPB_INFINITY && val.double_val != -UPB_INFINITY &&
if (val.double_val != INFINITY && val.double_val != -INFINITY &&
(val.double_val > FLT_MAX || val.double_val < -FLT_MAX)) {
jsondec_err(d, "Float out of range");
}
@ -1096,6 +1097,7 @@ static void jsondec_duration(jsondec *d, upb_msg *msg, const upb_msgdef *m) {
upb_strview str = jsondec_string(d);
const char *ptr = str.data;
const char *end = ptr + str.size;
const int64_t max = (uint64_t)3652500 * 86400;
/* "3.000000001s", "3s", etc. */
ptr = jsondec_buftoint64(d, ptr, end, &seconds.int64_val);
@ -1105,7 +1107,7 @@ static void jsondec_duration(jsondec *d, upb_msg *msg, const upb_msgdef *m) {
jsondec_err(d, "Malformed duration");
}
if (seconds.int64_val < -315576000000LL || seconds.int64_val > 315576000000LL) {
if (seconds.int64_val < -max || seconds.int64_val > max) {
jsondec_err(d, "Duration out of range");
}

@ -4,14 +4,16 @@
#include <ctype.h>
#include <float.h>
#include <inttypes.h>
#include <math.h>
#include <setjmp.h>
#include <stdarg.h>
#include <stdio.h>
#include <string.h>
#include <setjmp.h>
#include "upb/decode.h"
#include "upb/reflection.h"
/* Must be last. */
#include "upb/port_def.inc"
typedef struct {
@ -76,7 +78,7 @@ static void jsonenc_printf(jsonenc *e, const char *fmt, ...) {
va_list args;
va_start(args, fmt);
n = _upb_vsnprintf(e->ptr, have, fmt, args);
n = vsnprintf(e->ptr, have, fmt, args);
va_end(args);
if (UPB_LIKELY(have > n)) {
@ -268,9 +270,9 @@ static void jsonenc_string(jsonenc *e, upb_strview str) {
}
static void jsonenc_double(jsonenc *e, const char *fmt, double val) {
if (val == UPB_INFINITY) {
if (val == INFINITY) {
jsonenc_putstr(e, "\"Infinity\"");
} else if (val == -UPB_INFINITY) {
} else if (val == -INFINITY) {
jsonenc_putstr(e, "\"-Infinity\"");
} else if (val != val) {
jsonenc_putstr(e, "\"NaN\"");

@ -105,8 +105,8 @@ bool putf(upb_textprinter *p, const char *fmt, ...) {
va_start(args, fmt);
/* Run once to get the length of the string. */
_upb_va_copy(args_copy, args);
len = _upb_vsnprintf(NULL, 0, fmt, args_copy);
va_copy(args_copy, args);
len = vsnprintf(NULL, 0, fmt, args_copy);
va_end(args_copy);
/* + 1 for NULL terminator (vsprintf() requires it even if we don't). */

@ -150,9 +150,7 @@ UPB_INLINE uint64_t upb_vencode32(uint32_t val) {
uint64_t ret = 0;
UPB_ASSERT(bytes <= 5);
memcpy(&ret, buf, bytes);
#ifdef UPB_BIG_ENDIAN
ret = byteswap64(ret);
#endif
ret = _upb_be_swap64(ret);
UPB_ASSERT(ret <= 0xffffffffffU);
return ret;
}

@ -20,6 +20,15 @@
*
* This file is private and must not be included by users!
*/
#if !(__STDC_VERSION__ >= 199901L || __cplusplus >= 201103L)
#error upb requires C99 or C++11
#endif
#if (defined(_MSC_VER) && _MSC_VER < 1900)
#error upb requires MSVC >= 2015.
#endif
#include <stdint.h>
#include <stddef.h>
@ -68,12 +77,6 @@
#define UPB_UNLIKELY(x) (x)
#endif
/* Define UPB_BIG_ENDIAN manually if you're on big endian and your compiler
* doesn't provide these preprocessor symbols. */
#if defined(__BYTE_ORDER__) && (__BYTE_ORDER__ == __ORDER_BIG_ENDIAN__)
#define UPB_BIG_ENDIAN
#endif
/* Macros for function attributes on compilers that support them. */
#ifdef __GNUC__
#define UPB_FORCEINLINE __inline__ __attribute__((always_inline))
@ -89,49 +92,6 @@
#define UPB_NORETURN
#endif
#if __STDC_VERSION__ >= 199901L || __cplusplus >= 201103L
/* C99/C++11 versions. */
#include <stdio.h>
#define _upb_snprintf snprintf
#define _upb_vsnprintf vsnprintf
#define _upb_va_copy(a, b) va_copy(a, b)
#elif defined(_MSC_VER)
/* Microsoft C/C++ versions. */
#include <stdarg.h>
#include <stdio.h>
#if _MSC_VER < 1900
int msvc_snprintf(char* s, size_t n, const char* format, ...);
int msvc_vsnprintf(char* s, size_t n, const char* format, va_list arg);
#define UPB_MSVC_VSNPRINTF
#define _upb_snprintf msvc_snprintf
#define _upb_vsnprintf msvc_vsnprintf
#else
#define _upb_snprintf snprintf
#define _upb_vsnprintf vsnprintf
#endif
#define _upb_va_copy(a, b) va_copy(a, b)
#elif defined __GNUC__
/* A few hacky workarounds for functions not in C89.
* For internal use only!
* TODO(haberman): fix these by including our own implementations, or finding
* another workaround.
*/
#define _upb_snprintf __builtin_snprintf
#define _upb_vsnprintf __builtin_vsnprintf
#define _upb_va_copy(a, b) __va_copy(a, b)
#else
#error Need implementations of [v]snprintf and va_copy
#endif
#ifdef __cplusplus
#if __cplusplus >= 201103L || defined(__GXX_EXPERIMENTAL_CXX0X__) || \
(defined(_MSC_VER) && _MSC_VER >= 1900)
/* C++11 is present */
#else
#error upb requires C++11 for C++ support
#endif
#endif
#define UPB_MAX(x, y) ((x) > (y) ? (x) : (y))
#define UPB_MIN(x, y) ((x) < (y) ? (x) : (y))
@ -159,29 +119,12 @@ int msvc_vsnprintf(char* s, size_t n, const char* format, va_list arg);
#define UPB_ASSERT(expr) assert(expr)
#endif
/* UPB_ASSERT_DEBUGVAR(): assert that uses functions or variables that only
* exist in debug mode. This turns into regular assert. */
#define UPB_ASSERT_DEBUGVAR(expr) assert(expr)
#if defined(__GNUC__) || defined(__clang__)
#define UPB_UNREACHABLE() do { assert(0); __builtin_unreachable(); } while(0)
#else
#define UPB_UNREACHABLE() do { assert(0); } while(0)
#endif
/* UPB_INFINITY representing floating-point positive infinity. */
#include <math.h>
#ifdef INFINITY
#define UPB_INFINITY INFINITY
#else
#define UPB_INFINITY (1.0 / 0.0)
#endif
#ifdef NAN
#define UPB_NAN NAN
#else
#define UPB_NAN (0.0 / 0.0)
#endif
#if defined(__SANITIZE_ADDRESS__)
#define UPB_ASAN 1
#ifdef __cplusplus

@ -18,14 +18,7 @@
#undef UPB_UNUSED
#undef UPB_ASSUME
#undef UPB_ASSERT
#undef UPB_ASSERT_DEBUGVAR
#undef UPB_UNREACHABLE
#undef UPB_INFINITY
#undef UPB_NAN
#undef UPB_MSVC_VSNPRINTF
#undef _upb_snprintf
#undef _upb_vsnprintf
#undef _upb_va_copy
#undef UPB_POISON_MEMORY_REGION
#undef UPB_UNPOISON_MEMORY_REGION
#undef UPB_ASAN

@ -43,7 +43,7 @@ static void txtenc_printf(txtenc *e, const char *fmt, ...) {
va_list args;
va_start(args, fmt);
n = _upb_vsnprintf(e->ptr, have, fmt, args);
n = vsnprintf(e->ptr, have, fmt, args);
va_end(args);
if (UPB_LIKELY(have > n)) {

@ -40,7 +40,7 @@ void upb_status_seterrf(upb_status *status, const char *fmt, ...) {
void upb_status_vseterrf(upb_status *status, const char *fmt, va_list args) {
if (!status) return;
status->ok = false;
_upb_vsnprintf(status->msg, sizeof(status->msg), fmt, args);
vsnprintf(status->msg, sizeof(status->msg), fmt, args);
status->msg[UPB_STATUS_MAX_MESSAGE - 1] = '\0';
}
@ -49,7 +49,7 @@ void upb_status_vappenderrf(upb_status *status, const char *fmt, va_list args) {
if (!status) return;
status->ok = false;
len = strlen(status->msg);
_upb_vsnprintf(status->msg + len, sizeof(status->msg) - len, fmt, args);
vsnprintf(status->msg + len, sizeof(status->msg) - len, fmt, args);
status->msg[UPB_STATUS_MAX_MESSAGE - 1] = '\0';
}

@ -301,7 +301,7 @@ UPB_INLINE uint32_t _upb_be_swap32(uint32_t val) {
return val;
} else {
return ((val & 0xff) << 24) | ((val & 0xff00) << 8) |
((val & 0xff0000ULL) >> 8) | ((val & 0xff000000ULL) >> 24);
((val & 0xff0000) >> 8) | ((val & 0xff000000) >> 24);
}
}
@ -309,11 +309,7 @@ UPB_INLINE uint64_t _upb_be_swap64(uint64_t val) {
if (_upb_isle()) {
return val;
} else {
return ((val & 0xff) << 56) | ((val & 0xff00) << 40) |
((val & 0xff0000) << 24) | ((val & 0xff000000) << 8) |
((val & 0xff00000000ULL) >> 8) | ((val & 0xff0000000000ULL) >> 24) |
((val & 0xff000000000000ULL) >> 40) |
((val & 0xff00000000000000ULL) >> 56);
return ((uint64_t)_upb_be_swap32(val) << 32) | _upb_be_swap32(val >> 32);
}
}

@ -41,6 +41,9 @@ class Arena {
public:
// A simple arena with no initial memory block and the default allocator.
Arena() : ptr_(upb_arena_new(), upb_arena_free) {}
Arena(char *initial_block, size_t size)
: ptr_(upb_arena_init(initial_block, size, &upb_alloc_global),
upb_arena_free) {}
upb_arena* ptr() { return ptr_.get(); }
@ -71,15 +74,12 @@ class Arena {
template <int N>
class InlinedArena : public Arena {
public:
InlinedArena() : ptr_(upb_arena_new(&initial_block_, N, &upb_alloc_global)) {}
upb_arena* ptr() { return ptr_.get(); }
InlinedArena() : Arena(initial_block_, N) {}
private:
InlinedArena(const InlinedArena*) = delete;
InlinedArena& operator=(const InlinedArena*) = delete;
std::unique_ptr<upb_arena, decltype(&upb_arena_free)> ptr_;
char initial_block_[N];
};

@ -0,0 +1,35 @@
load(
"//bazel:build_defs.bzl",
"UPB_DEFAULT_CPPOPTS",
)
licenses(["notice"])
cc_library(
name = "upbc_generator",
srcs = [
"generator.cc",
"message_layout.cc",
"message_layout.h",
],
hdrs = ["generator.h"],
copts = UPB_DEFAULT_CPPOPTS,
deps = [
"@com_google_absl//absl/base:core_headers",
"@com_google_absl//absl/container:flat_hash_map",
"@com_google_absl//absl/strings",
"@com_google_protobuf//:protobuf",
"@com_google_protobuf//:protoc_lib",
],
)
cc_binary(
name = "protoc-gen-upb",
srcs = ["main.cc"],
copts = UPB_DEFAULT_CPPOPTS,
visibility = ["//visibility:public"],
deps = [
":upbc_generator",
"@com_google_protobuf//:protoc_lib",
],
)
Loading…
Cancel
Save