From 7ff8413a0317566d67b5b4abcb6ee2b72e9d465b Mon Sep 17 00:00:00 2001 From: Joshua Haberman Date: Thu, 13 Dec 2018 05:40:33 -0800 Subject: Ported some tests. --- tests/pb/test_decoder.proto | 86 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 86 insertions(+) create mode 100644 tests/pb/test_decoder.proto (limited to 'tests/pb/test_decoder.proto') diff --git a/tests/pb/test_decoder.proto b/tests/pb/test_decoder.proto new file mode 100644 index 0000000..8197dea --- /dev/null +++ b/tests/pb/test_decoder.proto @@ -0,0 +1,86 @@ + +syntax = "proto2"; + +enum TestEnum { + FOO = 1; +} + +message DecoderTest { + optional double f_double = 1; + optional float f_float = 2; + optional int64 f_int64 = 3; + optional uint64 f_uint64 = 4; + optional int32 f_int32 = 5; + optional fixed64 f_fixed64 = 6; + optional fixed32 f_fixed32 = 7; + optional bool f_bool = 8; + optional string f_string = 9; + optional DecoderTest f_message = 11; + optional bytes f_bytes = 12; + optional uint32 f_uint32 = 13; + optional TestEnum f_enum = 14; + optional sfixed32 f_sfixed32 = 15; + optional sfixed64 f_sfixed64 = 16; + optional sint32 f_sint32 = 17; + optional sint64 f_sint64 = 18; + + optional string nop_field = 40; + + repeated double r_double = 536869912; + repeated float r_float = 536869913; + repeated int64 r_int64 = 536869914; + repeated uint64 r_uint64 = 536869915; + repeated int32 r_int32 = 536869916; + repeated fixed64 r_fixed64 = 536869917; + repeated fixed32 r_fixed32 = 536869918; + repeated bool r_bool = 536869919; + repeated string r_string = 536869920; + repeated DecoderTest r_message = 536869922; + repeated bytes r_bytes = 536869923; + repeated uint32 r_uint32 = 536869924; + repeated TestEnum r_enum = 536869925; + repeated sfixed32 r_sfixed32 = 536869926; + repeated sfixed64 r_sfixed64 = 536869927; + repeated sint32 r_sint32 = 536869928; + repeated sint64 r_sint64 = 536869929; + + optional group F_group = 10 { + optional double f_double = 1; + optional float f_float = 2; + optional int64 f_int64 = 3; + optional uint64 f_uint64 = 4; + optional int32 f_int32 = 5; + optional fixed64 f_fixed64 = 6; + optional fixed32 f_fixed32 = 7; + optional bool f_bool = 8; + optional string f_string = 9; + optional DecoderTest f_message = 11; + optional bytes f_bytes = 12; + optional uint32 f_uint32 = 13; + optional TestEnum f_enum = 14; + optional sfixed32 f_sfixed32 = 15; + optional sfixed64 f_sfixed64 = 16; + optional sint32 f_sint32 = 17; + optional sint64 f_sint64 = 18; + } + + optional group R_group = 536869921 { + optional double f_double = 1; + optional float f_float = 2; + optional int64 f_int64 = 3; + optional uint64 f_uint64 = 4; + optional int32 f_int32 = 5; + optional fixed64 f_fixed64 = 6; + optional fixed32 f_fixed32 = 7; + optional bool f_bool = 8; + optional string f_string = 9; + optional DecoderTest f_message = 11; + optional bytes f_bytes = 12; + optional uint32 f_uint32 = 13; + optional TestEnum f_enum = 14; + optional sfixed32 f_sfixed32 = 15; + optional sfixed64 f_sfixed64 = 16; + optional sint32 f_sint32 = 17; + optional sint64 f_sint64 = 18; + } +} -- cgit v1.2.3 From 377871f10403c7b4e1cc6f769b9443b5197aecc8 Mon Sep 17 00:00:00 2001 From: Joshua Haberman Date: Sun, 16 Dec 2018 14:32:14 -0800 Subject: Got test_decoder working! --- BUILD | 15 + CMakeLists.txt | 1 + build_defs.bzl | 51 ++- tests/pb/test_decoder.cc | 91 +++--- tests/pb/test_decoder.proto | 42 +++ tools/dump_cinit.lua | 749 -------------------------------------------- tools/make_cmakelists.py | 3 + tools/upbc.lua | 91 ------ upb/def.c | 51 ++- upb/def.h | 10 +- upb/upb.h | 1 - upbc/generator.cc | 104 ++++++ 12 files changed, 317 insertions(+), 892 deletions(-) delete mode 100644 tools/dump_cinit.lua delete mode 100644 tools/upbc.lua (limited to 'tests/pb/test_decoder.proto') diff --git a/BUILD b/BUILD index 9b8513d..b3f397a 100644 --- a/BUILD +++ b/BUILD @@ -8,6 +8,7 @@ load( "make_shell_script", "upb_amalgamation", "upb_proto_library", + "upb_proto_reflection_library", ) # C/C++ rules ################################################################## @@ -167,10 +168,24 @@ cc_test( ], ) +proto_library( + name = "test_decoder_proto", + srcs = [ + "tests/pb/test_decoder.proto" + ] +) + +upb_proto_reflection_library( + name = "test_decoder_upbproto", + deps = ["test_decoder_proto"], + upbc = ":protoc-gen-upb", +) + cc_test( name = "test_decoder", srcs = ["tests/pb/test_decoder.cc"], deps = [ + ":test_decoder_upbproto", ":upb_pb", ":upb_test", ], diff --git a/CMakeLists.txt b/CMakeLists.txt index 6133c16..3439aac 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -131,6 +131,7 @@ add_executable(test_decoder tests/pb/test_decoder.cc) add_test(NAME test_decoder COMMAND test_decoder) target_link_libraries(test_decoder + test_decoder_upbproto upb_pb upb_test) add_executable(test_encoder diff --git a/build_defs.bzl b/build_defs.bzl index 8251014..3867976 100644 --- a/build_defs.bzl +++ b/build_defs.bzl @@ -217,7 +217,7 @@ def _remove_up(string): return _remove_suffix(string, ".proto") -def _upb_proto_library_srcs_impl(ctx): +def _upb_proto_srcs_impl(ctx, suffix): sources = [] outs = [] include_dirs = {} @@ -225,14 +225,19 @@ def _upb_proto_library_srcs_impl(ctx): if hasattr(dep, 'proto'): for src in dep.proto.transitive_sources: sources.append(src) - include_dirs[_remove_suffix(src.path, _remove_up(src.short_path) + "." + src.extension)] = True - outs.append(ctx.actions.declare_file(_remove_up(src.short_path) + ".upb.h")) - outs.append(ctx.actions.declare_file(_remove_up(src.short_path) + ".upb.c")) - outdir = _remove_suffix(outs[-1].path, _remove_up(src.short_path) + ".upb.c") + include_dir = _remove_suffix(src.path, _remove_up(src.short_path) + "." + src.extension) + if include_dir: + include_dirs[include_dir] = True + outs.append(ctx.actions.declare_file(_remove_up(src.short_path) + suffix + ".h")) + outs.append(ctx.actions.declare_file(_remove_up(src.short_path) + suffix + ".c")) + outdir = _remove_suffix(outs[-1].path, _remove_up(src.short_path) + suffix + ".c") source_paths = [d.path for d in sources] include_args = ["-I" + root for root in include_dirs.keys()] + print(source_paths) + print(include_args) + ctx.actions.run( inputs = [ctx.executable.upbc] + sources, outputs = outs, @@ -243,6 +248,12 @@ def _upb_proto_library_srcs_impl(ctx): return [DefaultInfo(files = depset(outs))] +def _upb_proto_library_srcs_impl(ctx): + return _upb_proto_srcs_impl(ctx, ".upb") + +def _upb_proto_reflection_library_srcs_impl(ctx): + return _upb_proto_srcs_impl(ctx, ".upbdefs") + _upb_proto_library_srcs = rule( implementation = _upb_proto_library_srcs_impl, attrs = { @@ -272,3 +283,33 @@ def upb_proto_library(name, deps, upbc): deps = [":upb"], copts = ["-Ibazel-out/k8-fastbuild/bin"], ) + +_upb_proto_reflection_library_srcs = rule( + implementation = _upb_proto_reflection_library_srcs_impl, + attrs = { + "upbc": attr.label( + executable = True, + cfg = "host", + ), + "protoc": attr.label( + executable = True, + cfg = "host", + default = "@com_google_protobuf//:protoc", + ), + "deps": attr.label_list(), + } +) + +def upb_proto_reflection_library(name, deps, upbc): + srcs_rule = name + "_defsrcs.cc" + _upb_proto_reflection_library_srcs( + name = srcs_rule, + upbc = upbc, + deps = deps, + ) + native.cc_library( + name = name, + srcs = [":" + srcs_rule], + deps = [":upb"], + copts = ["-Ibazel-out/k8-fastbuild/bin"], + ) diff --git a/tests/pb/test_decoder.cc b/tests/pb/test_decoder.cc index a931779..d0e3fa3 100644 --- a/tests/pb/test_decoder.cc +++ b/tests/pb/test_decoder.cc @@ -36,6 +36,7 @@ #include "tests/test_util.h" #include "tests/upb_test.h" +#include "tests/pb/test_decoder.upbdefs.h" #ifdef AMALGAMATED #include "upb.h" @@ -387,7 +388,6 @@ void reg_subm(upb_handlers *h, uint32_t num) { ASSERT( h->SetStartSubMessageHandler(f, UpbBind(startsubmsg, new uint32_t(num)))); ASSERT(h->SetEndSubMessageHandler(f, UpbBind(endsubmsg, new uint32_t(num)))); - ASSERT(upb_handlers_setsubhandlers(h, f, h)); } void reg_str(upb_handlers *h, uint32_t num) { @@ -399,52 +399,60 @@ void reg_str(upb_handlers *h, uint32_t num) { ASSERT(h->SetStringHandler(f, UpbBind(value_string, new uint32_t(num)))); } -upb::reffed_ptr NewHandlers(TestMode mode) { - - upb::reffed_ptr h(upb::Handlers::New(NewMessageDef().get())); +struct HandlerRegisterData { + TestMode mode; +}; - if (mode == ALL_HANDLERS) { +void callback(const void *closure, upb_handlers *h) { + const HandlerRegisterData* data = + static_cast(closure); + if (data->mode == ALL_HANDLERS) { h->SetStartMessageHandler(UpbMakeHandler(startmsg)); h->SetEndMessageHandler(UpbMakeHandler(endmsg)); // Register handlers for each type. - reg(h.get(), UPB_DESCRIPTOR_TYPE_DOUBLE); - reg (h.get(), UPB_DESCRIPTOR_TYPE_FLOAT); - reg (h.get(), UPB_DESCRIPTOR_TYPE_INT64); - reg(h.get(), UPB_DESCRIPTOR_TYPE_UINT64); - reg (h.get(), UPB_DESCRIPTOR_TYPE_INT32); - reg(h.get(), UPB_DESCRIPTOR_TYPE_FIXED64); - reg(h.get(), UPB_DESCRIPTOR_TYPE_FIXED32); - reg (h.get(), UPB_DESCRIPTOR_TYPE_BOOL); - reg(h.get(), UPB_DESCRIPTOR_TYPE_UINT32); - reg (h.get(), UPB_DESCRIPTOR_TYPE_ENUM); - reg (h.get(), UPB_DESCRIPTOR_TYPE_SFIXED32); - reg (h.get(), UPB_DESCRIPTOR_TYPE_SFIXED64); - reg (h.get(), UPB_DESCRIPTOR_TYPE_SINT32); - reg (h.get(), UPB_DESCRIPTOR_TYPE_SINT64); - - reg_str(h.get(), UPB_DESCRIPTOR_TYPE_STRING); - reg_str(h.get(), UPB_DESCRIPTOR_TYPE_BYTES); - reg_str(h.get(), rep_fn(UPB_DESCRIPTOR_TYPE_STRING)); - reg_str(h.get(), rep_fn(UPB_DESCRIPTOR_TYPE_BYTES)); + reg(h, UPB_DESCRIPTOR_TYPE_DOUBLE); + reg (h, UPB_DESCRIPTOR_TYPE_FLOAT); + reg (h, UPB_DESCRIPTOR_TYPE_INT64); + reg(h, UPB_DESCRIPTOR_TYPE_UINT64); + reg (h, UPB_DESCRIPTOR_TYPE_INT32); + reg(h, UPB_DESCRIPTOR_TYPE_FIXED64); + reg(h, UPB_DESCRIPTOR_TYPE_FIXED32); + reg (h, UPB_DESCRIPTOR_TYPE_BOOL); + reg(h, UPB_DESCRIPTOR_TYPE_UINT32); + reg (h, UPB_DESCRIPTOR_TYPE_ENUM); + reg (h, UPB_DESCRIPTOR_TYPE_SFIXED32); + reg (h, UPB_DESCRIPTOR_TYPE_SFIXED64); + reg (h, UPB_DESCRIPTOR_TYPE_SINT32); + reg (h, UPB_DESCRIPTOR_TYPE_SINT64); + + reg_str(h, UPB_DESCRIPTOR_TYPE_STRING); + reg_str(h, UPB_DESCRIPTOR_TYPE_BYTES); + reg_str(h, rep_fn(UPB_DESCRIPTOR_TYPE_STRING)); + reg_str(h, rep_fn(UPB_DESCRIPTOR_TYPE_BYTES)); // Register submessage/group handlers that are self-recursive // to this type, eg: message M { optional M m = 1; } - reg_subm(h.get(), UPB_DESCRIPTOR_TYPE_MESSAGE); - reg_subm(h.get(), rep_fn(UPB_DESCRIPTOR_TYPE_MESSAGE)); - reg_subm(h.get(), UPB_DESCRIPTOR_TYPE_GROUP); - reg_subm(h.get(), rep_fn(UPB_DESCRIPTOR_TYPE_GROUP)); + reg_subm(h, UPB_DESCRIPTOR_TYPE_MESSAGE); + reg_subm(h, rep_fn(UPB_DESCRIPTOR_TYPE_MESSAGE)); + + if (h->message_def()->full_name() == std::string("DecoderTest")) { + reg_subm(h, UPB_DESCRIPTOR_TYPE_GROUP); + reg_subm(h, rep_fn(UPB_DESCRIPTOR_TYPE_GROUP)); + } // For NOP_FIELD we register no handlers, so we can pad a proto freely without // changing the output. } - - bool ok = h->Freeze(NULL); - ASSERT(ok); - - return h; } +upb::reffed_ptr NewHandlers(upb::SymbolTable* symtab, + TestMode mode) { + HandlerRegisterData handlerdata; + handlerdata.mode = mode; + return upb::Handlers::NewFrozen(DecoderTest_getmsgdef(symtab), callback, + &handlerdata); +} /* Running of test cases ******************************************************/ @@ -1132,14 +1140,11 @@ upb::reffed_ptr NewMethod( return cache.GetDecoderMethod(upb::pb::DecoderMethodOptions(dest_handlers)); } -void test_emptyhandlers(bool allowjit) { +void test_emptyhandlers(upb::SymbolTable* symtab, bool allowjit) { // Create an empty handlers to make sure that the decoder can handle empty // messages. - upb::reffed_ptr md = upb::MessageDef::New(); - ASSERT(md->set_full_name("Empty", NULL)); - ASSERT(md->Freeze(NULL)); - - upb::reffed_ptr h(upb::Handlers::New(md.get())); + const upb::MessageDef* md = Empty_getmsgdef(symtab); + upb::reffed_ptr h(upb::Handlers::New(md)); bool ok = h->Freeze(NULL); ASSERT(ok); upb::reffed_ptr method = @@ -1178,9 +1183,9 @@ upb::reffed_ptr method = void run_tests(bool use_jit) { upb::reffed_ptr method; upb::reffed_ptr handlers; - upb::SymbolTable symtab; + upb::SymbolTable* symtab = upb::SymbolTable::New(); - handlers = NewHandlers(test_mode); + handlers = NewHandlers(symtab, test_mode); global_handlers = handlers.get(); method = NewMethod(handlers.get(), use_jit); @@ -1191,7 +1196,9 @@ void run_tests(bool use_jit) { test_invalid(); test_valid(); - test_emptyhandlers(use_jit); + test_emptyhandlers(symtab, use_jit); + + upb::SymbolTable::Free(symtab); } void run_test_suite() { diff --git a/tests/pb/test_decoder.proto b/tests/pb/test_decoder.proto index 8197dea..e9fa6ad 100644 --- a/tests/pb/test_decoder.proto +++ b/tests/pb/test_decoder.proto @@ -5,6 +5,8 @@ enum TestEnum { FOO = 1; } +message Empty {} + message DecoderTest { optional double f_double = 1; optional float f_float = 2; @@ -62,6 +64,26 @@ message DecoderTest { optional sfixed64 f_sfixed64 = 16; optional sint32 f_sint32 = 17; optional sint64 f_sint64 = 18; + + optional string nop_field = 40; + + repeated double r_double = 536869912; + repeated float r_float = 536869913; + repeated int64 r_int64 = 536869914; + repeated uint64 r_uint64 = 536869915; + repeated int32 r_int32 = 536869916; + repeated fixed64 r_fixed64 = 536869917; + repeated fixed32 r_fixed32 = 536869918; + repeated bool r_bool = 536869919; + repeated string r_string = 536869920; + repeated DecoderTest r_message = 536869922; + repeated bytes r_bytes = 536869923; + repeated uint32 r_uint32 = 536869924; + repeated TestEnum r_enum = 536869925; + repeated sfixed32 r_sfixed32 = 536869926; + repeated sfixed64 r_sfixed64 = 536869927; + repeated sint32 r_sint32 = 536869928; + repeated sint64 r_sint64 = 536869929; } optional group R_group = 536869921 { @@ -82,5 +104,25 @@ message DecoderTest { optional sfixed64 f_sfixed64 = 16; optional sint32 f_sint32 = 17; optional sint64 f_sint64 = 18; + + optional string nop_field = 40; + + repeated double r_double = 536869912; + repeated float r_float = 536869913; + repeated int64 r_int64 = 536869914; + repeated uint64 r_uint64 = 536869915; + repeated int32 r_int32 = 536869916; + repeated fixed64 r_fixed64 = 536869917; + repeated fixed32 r_fixed32 = 536869918; + repeated bool r_bool = 536869919; + repeated string r_string = 536869920; + repeated DecoderTest r_message = 536869922; + repeated bytes r_bytes = 536869923; + repeated uint32 r_uint32 = 536869924; + repeated TestEnum r_enum = 536869925; + repeated sfixed32 r_sfixed32 = 536869926; + repeated sfixed64 r_sfixed64 = 536869927; + repeated sint32 r_sint32 = 536869928; + repeated sint64 r_sint64 = 536869929; } } diff --git a/tools/dump_cinit.lua b/tools/dump_cinit.lua deleted file mode 100644 index 93ee12e..0000000 --- a/tools/dump_cinit.lua +++ /dev/null @@ -1,749 +0,0 @@ ---[[ - - Routines for dumping internal data structures into C initializers - that can be compiled into a .o file. - ---]] - -local upbtable = require "upb.table" -local upb = require "upb" -local export = {} - --- A tiny little abstraction that decouples the dump_* functions from --- what they're writing to (appending to a string, writing to file I/O, etc). --- This could possibly matter since naive string building is O(n^2) in the --- number of appends. -function export.str_appender() - local str = "" - local function append(fmt, ...) - str = str .. string.format(fmt, ...) - end - local function get() - return str - end - return append, get -end - -function export.file_appender(file) - local f = file - local function append(fmt, ...) - f:write(string.format(fmt, ...)) - end - return append -end - -function handler_types(base) - local ret = {} - for k, _ in pairs(base) do - if string.find(k, "^" .. "HANDLER_") then - ret[#ret + 1] = k - end - end - return ret -end - -function octchar(num) - assert(num < 8) - local idx = num + 1 -- 1-based index - return string.sub("01234567", idx, idx) -end - -function c_escape(num) - assert(num < 256) - return string.format("\\%s%s%s", - octchar(math.floor(num / 64)), - octchar(math.floor(num / 8) % 8), - octchar(num % 8)); -end - --- const(f, label) -> UPB_LABEL_REPEATED, where f:label() == upb.LABEL_REPEATED -function const(obj, name, base) - local val = obj[name] - base = base or upb - - -- Support both f:label() and f.label. - if type(val) == "function" then - val = val(obj) - end - - for k, v in pairs(base) do - if v == val and string.find(k, "^" .. string.upper(name)) then - return "UPB_" .. k - end - end - assert(false, "Couldn't find UPB_" .. string.upper(name) .. - " constant for value: " .. val) -end - -function sortedkeys(tab) - arr = {} - for key in pairs(tab) do - arr[#arr + 1] = key - end - table.sort(arr) - return arr -end - -function sorted_defs(defs) - local sorted = {} - - for def in defs do - if def.type == deftype then - sorted[#sorted + 1] = def - end - end - - table.sort(sorted, - function(a, b) return a:full_name() < b:full_name() end) - - return sorted -end - -function constlist(pattern) - local ret = {} - for k, v in pairs(upb) do - if string.find(k, "^" .. pattern) then - ret[k] = v - end - end - return ret -end - -function boolstr(val) - if val == true then - return "true" - elseif val == false then - return "false" - else - assert(false, "Bad bool value: " .. tostring(val)) - end -end - ---[[ - - LinkTable: an object that tracks all linkable objects and their offsets to - facilitate linking. - ---]] - -local LinkTable = {} -function LinkTable:new(types) - local linktab = { - types = types, - table = {}, -- ptr -> {type, 0-based offset} - obj_arrays = {} -- Establishes the ordering for each object type - } - for type, _ in pairs(types) do - linktab.obj_arrays[type] = {} - end - setmetatable(linktab, {__index = LinkTable}) -- Inheritance - return linktab -end - --- Adds a new object to the sequence of objects of this type. -function LinkTable:add(objtype, ptr, obj) - obj = obj or ptr - assert(self.table[obj] == nil) - assert(self.types[objtype]) - local arr = self.obj_arrays[objtype] - self.table[ptr] = {objtype, #arr} - arr[#arr + 1] = obj -end - --- Returns a C symbol name for the given objtype and offset. -function LinkTable:csym(objtype, offset) - local typestr = assert(self.types[objtype]) - return string.format("%s[%d]", typestr, offset) -end - --- Returns the address of the given C object. -function LinkTable:addr(obj) - if obj == upbtable.NULL then - return "NULL" - else - local tabent = assert(self.table[obj], "unknown object: " .. tostring(obj)) - return "&" .. self:csym(tabent[1], tabent[2]) - end -end - --- Returns an array declarator indicating how many objects have been added. -function LinkTable:cdecl(objtype) - return self:csym(objtype, #self.obj_arrays[objtype]) -end - -function LinkTable:objs(objtype) - -- Return iterator function, allowing use as: - -- for obj in linktable:objs(type) do - -- -- ... - -- done - local array = self.obj_arrays[objtype] - local i = 0 - return function() - i = i + 1 - if array[i] then return array[i] end - end -end - -function LinkTable:empty(objtype) - return #self.obj_arrays[objtype] == 0 -end - ---[[ - - Dumper: an object that can dump C initializers for several constructs. - Uses a LinkTable to resolve references when necessary. - ---]] - -local Dumper = {} -function Dumper:new(linktab) - local obj = {linktab = linktab} - setmetatable(obj, {__index = Dumper}) -- Inheritance - return obj -end - --- Dumps a upb_tabval, eg: --- UPB_TABVALUE_INIT(5) -function Dumper:_value(val, upbtype) - if type(val) == "nil" then - return "UPB_TABVALUE_EMPTY_INIT" - elseif type(val) == "number" then - -- Use upbtype to disambiguate what kind of number it is. - if upbtype == upbtable.CTYPE_INT32 then - return string.format("UPB_TABVALUE_INT_INIT(%d)", val) - else - -- TODO(haberman): add support for these so we can properly support - -- default values. - error("Unsupported number type " .. upbtype) - end - elseif type(val) == "string" then - return string.format('UPB_TABVALUE_PTR_INIT("%s")', val) - else - -- We take this as an object reference that has an entry in the link table. - return string.format("UPB_TABVALUE_PTR_INIT(%s)", self.linktab:addr(val)) - end -end - --- Dumps a table key. -function Dumper:tabkey(key) - if type(key) == "nil" then - return "UPB_TABKEY_NONE" - elseif type(key) == "string" then - local len = #key - local len1 = c_escape(len % 256) - local len2 = c_escape(math.floor(len / 256) % 256) - local len3 = c_escape(math.floor(len / (256 * 256)) % 256) - local len4 = c_escape(math.floor(len / (256 * 256 * 256)) % 256) - return string.format('UPB_TABKEY_STR("%s", "%s", "%s", "%s", "%s")', - len1, len2, len3, len4, key) - else - return string.format("UPB_TABKEY_NUM(%d)", key) - end -end - --- Dumps a table entry. -function Dumper:tabent(ent) - local key = self:tabkey(ent.key) - local val = self:_value(ent.value, ent.valtype) - local next = self.linktab:addr(ent.next) - return string.format(' {%s, %s, %s},\n', key, val, next) -end - --- Dumps an inttable array entry. This is almost the same as value() above, --- except that nil values have a special value to indicate "empty". -function Dumper:arrayval(val) - if val.val then - return string.format(" %s,\n", self:_value(val.val, val.valtype)) - else - return " UPB_TABVALUE_EMPTY_INIT,\n" - end -end - --- Dumps an initializer for the given strtable/inttable (respectively). Its --- entries must have previously been added to the linktable. -function Dumper:strtable(t) - -- UPB_STRTABLE_INIT(count, mask, type, size_lg2, entries) - return string.format( - "UPB_STRTABLE_INIT(%d, %d, %s, %d, %s)", - t.count, t.mask, const(t, "ctype", upbtable) , t.size_lg2, - self.linktab:addr(t.entries[1].ptr)) -end - -function Dumper:inttable(t) - local lt = assert(self.linktab) - -- UPB_INTTABLE_INIT(count, mask, type, size_lg2, ent, a, asize, acount) - local entries = "NULL" - if #t.entries > 0 then - entries = lt:addr(t.entries[1].ptr) - end - return string.format( - "UPB_INTTABLE_INIT(%d, %d, %s, %d, %s, %s, %d, %d)", - t.count, t.mask, const(t, "ctype", upbtable), t.size_lg2, entries, - lt:addr(t.array[1].ptr), t.array_size, t.array_count) -end - --- A visitor for visiting all tables of a def. Used first to count entries --- and later to dump them. -local function gettables(def) - if def:def_type() == upb.DEF_MSG then - return {int = upbtable.msgdef_itof(def), str = upbtable.msgdef_ntof(def)} - elseif def:def_type() == upb.DEF_ENUM then - return {int = upbtable.enumdef_iton(def), str = upbtable.enumdef_ntoi(def)} - end -end - -local function emit_file_warning(filedef, append) - append('/* This file was generated by upbc (the upb compiler) from the input\n') - append(' * file:\n') - append(' *\n') - append(' * %s\n', filedef:name()) - append(' *\n') - append(' * Do not edit -- your changes will be discarded when the file is\n') - append(' * regenerated. */\n\n') -end - -local function join(...) - return table.concat({...}, ".") -end - -local function split(str) - local ret = {} - for word in string.gmatch(str, "%w+") do - table.insert(ret, word) - end - return ret -end - -local function to_cident(...) - return string.gsub(join(...), "[%./]", "_") -end - -local function to_preproc(...) - return string.upper(to_cident(...)) -end - --- Strips away last path element, ie: --- foo.Bar.Baz -> foo.Bar -local function remove_name(name) - local package_end = 0 - for i=1,string.len(name) do - if string.byte(name, i) == string.byte(".", 1) then - package_end = i - 1 - end - end - return string.sub(name, 1, package_end) -end - -local function start_namespace(package, append) - local package_components = split(package) - for _, component in ipairs(package_components) do - append("namespace %s {\n", component) - end -end - -local function end_namespace(package, append) - local package_components = split(package) - for i=#package_components,1,-1 do - append("} /* namespace %s */\n", package_components[i]) - end -end - -local function well_known_type(m) - local type_map = {} - type_map["google.protobuf.Duration"] = "UPB_WELLKNOWN_DURATION" - type_map["google.protobuf.Timestamp"] = "UPB_WELLKNOWN_TIMESTAMP" - type_map["google.protobuf.Value"] = "UPB_WELLKNOWN_VALUE" - type_map["google.protobuf.ListValue"] = "UPB_WELLKNOWN_LISTVALUE" - type_map["google.protobuf.Struct"] = "UPB_WELLKNOWN_STRUCT" - type_map["google.protobuf.DoubleValue"] = "UPB_WELLKNOWN_DOUBLEVALUE" - type_map["google.protobuf.FloatValue"] = "UPB_WELLKNOWN_FLOATVALUE" - type_map["google.protobuf.Int64Value"] = "UPB_WELLKNOWN_INT64VALUE" - type_map["google.protobuf.UInt64Value"] = "UPB_WELLKNOWN_UINT64VALUE" - type_map["google.protobuf.Int32Value"] = "UPB_WELLKNOWN_INT32VALUE" - type_map["google.protobuf.UInt32Value"] = "UPB_WELLKNOWN_UINT32VALUE" - type_map["google.protobuf.BoolValue"] = "UPB_WELLKNOWN_BOOLVALUE" - type_map["google.protobuf.StringValue"] = "UPB_WELLKNOWN_STRINGVALUE" - type_map["google.protobuf.BytesValue"] = "UPB_WELLKNOWN_BYTESVALUE" - local t = type_map[m:full_name()] - if (t == nil) then - t = "UPB_WELLKNOWN_UNSPECIFIED" - end - return t -end - ---[[ - - Top-level, exported dumper functions - ---]] - -local function dump_defs_c(filedef, append) - local defs = {} - for def in filedef:defs(upb.DEF_ANY) do - defs[#defs + 1] = def - if (def:def_type() == upb.DEF_MSG) then - for field in def:fields() do - defs[#defs + 1] = field - end - end - end - - -- Sort all defs by (type, name). - -- This gives us a linear ordering that we can use to create offsets into - -- shared arrays like REFTABLES, hash table entries, and arrays. - table.sort(defs, function(a, b) - if a:def_type() ~= b:def_type() then - return a:def_type() < b:def_type() - else - return a:full_name() < b:full_name() end - end - ) - - -- Perform pre-pass to build the link table. - local linktab = LinkTable:new{ - [upb.DEF_MSG] = "msgs", - [upb.DEF_FIELD] = "fields", - [upb.DEF_ENUM] = "enums", - intentries = "intentries", - strentries = "strentries", - arrays = "arrays", - } - local reftable_count = 0 - - for _, def in ipairs(defs) do - assert(def:is_frozen(), "can only dump frozen defs.") - linktab:add(def:def_type(), def) - reftable_count = reftable_count + 2 - local tables = gettables(def) - if tables then - for _, e in ipairs(tables.str.entries) do - linktab:add("strentries", e.ptr, e) - end - for _, e in ipairs(tables.int.entries) do - linktab:add("intentries", e.ptr, e) - end - for _, e in ipairs(tables.int.array) do - linktab:add("arrays", e.ptr, e) - end - end - end - - -- Emit forward declarations. - emit_file_warning(filedef, append) - append('#include "upb/def.h"\n') - append('#include "upb/structdefs.int.h"\n\n') - append("static const upb_msgdef %s;\n", linktab:cdecl(upb.DEF_MSG)) - append("static const upb_fielddef %s;\n", linktab:cdecl(upb.DEF_FIELD)) - if not linktab:empty(upb.DEF_ENUM) then - append("static const upb_enumdef %s;\n", linktab:cdecl(upb.DEF_ENUM)) - end - append("static const upb_tabent %s;\n", linktab:cdecl("strentries")) - if not linktab:empty("intentries") then - append("static const upb_tabent %s;\n", linktab:cdecl("intentries")) - end - append("static const upb_tabval %s;\n", linktab:cdecl("arrays")) - append("\n") - append("#ifdef UPB_DEBUG_REFS\n") - append("static upb_inttable reftables[%d];\n", reftable_count) - append("#endif\n") - append("\n") - - -- Emit defs. - local dumper = Dumper:new(linktab) - - local reftable = 0 - - append("static const upb_msgdef %s = {\n", linktab:cdecl(upb.DEF_MSG)) - for m in linktab:objs(upb.DEF_MSG) do - local tables = gettables(m) - -- UPB_MSGDEF_INIT(name, selector_count, submsg_field_count, itof, ntof, - -- refs, ref2s) - append(' UPB_MSGDEF_INIT("%s", %d, %d, %s, %s, %s, %s, %s,' .. - ' &reftables[%d], &reftables[%d]),\n', - m:full_name(), - upbtable.msgdef_selector_count(m), - upbtable.msgdef_submsg_field_count(m), - dumper:inttable(tables.int), - dumper:strtable(tables.str), - boolstr(m:_map_entry()), - const(m, "syntax"), - well_known_type(m), - reftable, reftable + 1) - reftable = reftable + 2 - end - append("};\n\n") - - append("static const upb_fielddef %s = {\n", linktab:cdecl(upb.DEF_FIELD)) - for f in linktab:objs(upb.DEF_FIELD) do - local subdef = "NULL" - if f:has_subdef() then - subdef = string.format("(const upb_def*)(%s)", linktab:addr(f:subdef())) - end - local intfmt - if f:type() == upb.TYPE_UINT32 or - f:type() == upb.TYPE_INT32 or - f:type() == upb.TYPE_UINT64 or - f:type() == upb.TYPE_INT64 then - intfmt = const(f, "intfmt") - else - intfmt = "0" - end - -- UPB_FIELDDEF_INIT(label, type, intfmt, tagdelim, is_extension, lazy, - -- packed, name, num, msgdef, subdef, selector_base, - -- index, -- default_value) - append(' UPB_FIELDDEF_INIT(%s, %s, %s, %s, %s, %s, %s, "%s", %d, %s, ' .. - '%s, %d, %d, {0},' .. -- TODO: support default value - '&reftables[%d], &reftables[%d]),\n', - const(f, "label"), const(f, "type"), intfmt, - boolstr(f:istagdelim()), boolstr(f:is_extension()), - boolstr(f:lazy()), boolstr(f:packed()), f:name(), f:number(), - linktab:addr(f:containing_type()), subdef, - upbtable.fielddef_selector_base(f), f:index(), - reftable, reftable + 1 - ) - reftable = reftable + 2 - end - append("};\n\n") - - if not linktab:empty(upb.DEF_ENUM) then - append("static const upb_enumdef %s = {\n", linktab:cdecl(upb.DEF_ENUM)) - for e in linktab:objs(upb.DEF_ENUM) do - local tables = gettables(e) - -- UPB_ENUMDEF_INIT(name, ntoi, iton, defaultval) - append(' UPB_ENUMDEF_INIT("%s", %s, %s, %d, ' .. - '&reftables[%d], &reftables[%d]),\n', - e:full_name(), - dumper:strtable(tables.str), - dumper:inttable(tables.int), - --e:default()) - 0, - reftable, reftable + 1) - reftable = reftable + 2 - end - append("};\n\n") - end - - append("static const upb_tabent %s = {\n", linktab:cdecl("strentries")) - for ent in linktab:objs("strentries") do - append(dumper:tabent(ent)) - end - append("};\n\n"); - - if not linktab:empty("intentries") then - append("static const upb_tabent %s = {\n", linktab:cdecl("intentries")) - for ent in linktab:objs("intentries") do - append(dumper:tabent(ent)) - end - append("};\n\n"); - end - - append("static const upb_tabval %s = {\n", linktab:cdecl("arrays")) - for ent in linktab:objs("arrays") do - append(dumper:arrayval(ent)) - end - append("};\n\n"); - - append("#ifdef UPB_DEBUG_REFS\n") - append("static upb_inttable reftables[%d] = {\n", reftable_count) - for i = 1,reftable_count do - append(" UPB_EMPTY_INTTABLE_INIT(UPB_CTYPE_PTR),\n") - end - append("};\n") - append("#endif\n\n") - - append("static const upb_msgdef *refm(const upb_msgdef *m, const void *owner) {\n") - append(" upb_msgdef_ref(m, owner);\n") - append(" return m;\n") - append("}\n\n") - append("static const upb_enumdef *refe(const upb_enumdef *e, const void *owner) {\n") - append(" upb_enumdef_ref(e, owner);\n") - append(" return e;\n") - append("}\n\n") - - append("/* Public API. */\n") - - for m in linktab:objs(upb.DEF_MSG) do - append("const upb_msgdef *upbdefs_%s_get(const void *owner)" .. - " { return refm(%s, owner); }\n", - to_cident(m:full_name()), linktab:addr(m)) - end - - append("\n") - - for e in linktab:objs(upb.DEF_ENUM) do - append("const upb_enumdef *upbdefs_%s_get(const void *owner)" .. - " { return refe(%s, owner); }\n", - to_cident(e:full_name()), linktab:addr(e)) - end - - return linktab -end - -local function dump_defs_for_type(format, defs, append) - local sorted = sorted_defs(defs) - for _, def in ipairs(sorted) do - append(format, to_cident(def:full_name()), def:full_name()) - end - - append("\n") -end - -local function make_children_map(file) - -- Maps file:package() or msg:full_name() -> children. - local map = {} - for def in file:defs(upb.DEF_ANY) do - local container = remove_name(def:full_name()) - if not map[container] then - map[container] = {} - end - table.insert(map[container], def) - end - - -- Sort all the lists for a consistent ordering. - for name, children in pairs(map) do - table.sort(children, function(a, b) return a:name() < b:name() end) - end - - return map -end - -local print_classes - -local function print_message(def, map, indent, append) - append("\n") - append("%sclass %s : public ::upb::reffed_ptr {\n", - indent, def:name()) - append("%s public:\n", indent) - append("%s %s(const ::upb::MessageDef* m, const void *ref_donor = NULL)\n", - indent, def:name()) - append("%s : reffed_ptr(m, ref_donor) {\n", indent) - append("%s UPB_ASSERT(upbdefs_%s_is(m));\n", indent, to_cident(def:full_name())) - append("%s }\n", indent) - append("\n") - append("%s static %s get() {\n", indent, def:name()) - append("%s const ::upb::MessageDef* m = upbdefs_%s_get(&m);\n", indent, to_cident(def:full_name())) - append("%s return %s(m, &m);\n", indent, def:name()) - append("%s }\n", indent) - -- TODO(haberman): add fields - print_classes(def:full_name(), map, indent .. " ", append) - append("%s};\n", indent) -end - -local function print_enum(def, indent, append) - append("\n") - append("%sclass %s : public ::upb::reffed_ptr {\n", - indent, def:name()) - append("%s public:\n", indent) - append("%s %s(const ::upb::EnumDef* e, const void *ref_donor = NULL)\n", - indent, def:name()) - append("%s : reffed_ptr(e, ref_donor) {\n", indent) - append("%s UPB_ASSERT(upbdefs_%s_is(e));\n", indent, to_cident(def:full_name())) - append("%s }\n", indent) - append("%s static %s get() {\n", indent, def:name()) - append("%s const ::upb::EnumDef* e = upbdefs_%s_get(&e);\n", indent, to_cident(def:full_name())) - append("%s return %s(e, &e);\n", indent, def:name()) - append("%s }\n", indent) - append("%s};\n", indent) -end - -function print_classes(name, map, indent, append) - if not map[name] then - return - end - - for _, def in ipairs(map[name]) do - if def:def_type() == upb.DEF_MSG then - print_message(def, map, indent, append) - elseif def:def_type() == upb.DEF_ENUM then - print_enum(def, indent, append) - else - error("Unknown def type for " .. def:full_name()) - end - end -end - -local function dump_defs_h(file, append, linktab) - local basename_preproc = to_preproc(file:name()) - append("/* This file contains accessors for a set of compiled-in defs.\n") - append(" * Note that unlike Google's protobuf, it does *not* define\n") - append(" * generated classes or any other kind of data structure for\n") - append(" * actually storing protobufs. It only contains *defs* which\n") - append(" * let you reflect over a protobuf *schema*.\n") - append(" */\n") - emit_file_warning(file, append) - append('#ifndef %s_UPB_H_\n', basename_preproc) - append('#define %s_UPB_H_\n\n', basename_preproc) - append('#include "upb/def.h"\n\n') - append('UPB_BEGIN_EXTERN_C\n\n') - - -- Dump C enums for proto enums. - - append("/* MessageDefs: call these functions to get a ref to a msgdef. */\n") - dump_defs_for_type( - "const upb_msgdef *upbdefs_%s_get(const void *owner);\n", - file:defs(upb.DEF_MSG), append) - - append("/* EnumDefs: call these functions to get a ref to an enumdef. */\n") - dump_defs_for_type( - "const upb_enumdef *upbdefs_%s_get(const void *owner);\n", - file:defs(upb.DEF_ENUM), append) - - append("/* Functions to test whether this message is of a certain type. */\n") - dump_defs_for_type( - "UPB_INLINE bool upbdefs_%s_is(const upb_msgdef *m) {\n" .. - " return strcmp(upb_msgdef_fullname(m), \"%s\") == 0;\n}\n", - file:defs(upb.DEF_MSG), append) - - append("/* Functions to test whether this enum is of a certain type. */\n") - dump_defs_for_type( - "UPB_INLINE bool upbdefs_%s_is(const upb_enumdef *e) {\n" .. - " return strcmp(upb_enumdef_fullname(e), \"%s\") == 0;\n}\n", - file:defs(upb.DEF_ENUM), append) - - append("\n") - - -- fields - local fields = {} - - for f in linktab:objs(upb.DEF_FIELD) do - local symname = f:containing_type():full_name() .. "." .. f:name() - fields[#fields + 1] = {to_cident(symname), f} - end - - table.sort(fields, function(a, b) return a[1] < b[1] end) - - append("/* Functions to get a fielddef from a msgdef reference. */\n") - for _, field in ipairs(fields) do - local f = field[2] - local msg_cident = to_cident(f:containing_type():full_name()) - local field_cident = to_cident(f:name()) - append("UPB_INLINE const upb_fielddef *upbdefs_%s_f_%s(const upb_msgdef *m) {" .. - " UPB_ASSERT(upbdefs_%s_is(m));" .. - " return upb_msgdef_itof(m, %d); }\n", - msg_cident, field_cident, msg_cident, f:number()) - end - - append('\nUPB_END_EXTERN_C\n\n') - - -- C++ wrappers. - local children_map = make_children_map(file) - - append("#ifdef __cplusplus\n\n") - append("namespace upbdefs {\n") - start_namespace(file:package(), append) - print_classes(file:package(), children_map, "", append) - append("\n") - end_namespace(file:package(), append) - append("} /* namespace upbdefs */\n\n") - append("#endif /* __cplusplus */\n") - - append("\n") - append('#endif /* %s_UPB_H_ */\n', basename_preproc) -end - -function export.dump_defs(filedef, append_h, append_c) - local linktab = dump_defs_c(filedef, append_c) - dump_defs_h(filedef, append_h, linktab) -end - -return export diff --git a/tools/make_cmakelists.py b/tools/make_cmakelists.py index b8f46b9..2b426b5 100755 --- a/tools/make_cmakelists.py +++ b/tools/make_cmakelists.py @@ -120,6 +120,9 @@ class BuildFileFunctions(object): def upb_proto_library(self, **kwargs): pass + def upb_proto_reflection_library(self, **kwargs): + pass + def genrule(self, **kwargs): pass diff --git a/tools/upbc.lua b/tools/upbc.lua deleted file mode 100644 index 80d2886..0000000 --- a/tools/upbc.lua +++ /dev/null @@ -1,91 +0,0 @@ ---[[ - - The upb compiler. It can write two different kinds of output - files: - - - generated code for a C API (foo.upb.h, foo.upb.c) - - (obsolete): definitions of upb defs. (foo.upbdefs.h, foo.upbdefs.c) - ---]] - -local dump_cinit = require "dump_cinit" -local upb = require "upb" - -local generate_upbdefs = false -local outdir = "." - -i = 1 -while i <= #arg do - argument = arg[i] - if argument.sub(argument, 1, 2) == "--" then - if argument == "--generate-upbdefs" then - generate_upbdefs = true - elseif argument == "--outdir" then - i = i + 1 - outdir = arg[i] - else - print("Unknown flag: " .. argument) - return 1 - end - else - if src then - print("upbc can only handle one input file at a time.") - return 1 - end - src = argument - end - i = i + 1 -end - -if not src then - print("Usage: upbc [--generate-upbdefs] ") - return 1 -end - -function strip_proto(filename) - return string.gsub(filename, '%.proto$','') -end - -local function open(filename) - local full_name = outdir .. "/" .. filename - return assert(io.open(full_name, "w"), "couldn't open " .. full_name) -end - --- Open input/output files. -local f = assert(io.open(src, "r"), "couldn't open input file " .. src) -local descriptor = f:read("*all") -local files = upb.load_descriptor(descriptor) -local symtab = upb.SymbolTable() - -for _, file in ipairs(files) do - symtab:add_file(file) - local outbase = strip_proto(file:name()) - - -- Write upbdefs. - - local hfilename = outbase .. ".upbdefs.h" - local cfilename = outbase .. ".upbdefs.c" - - if os.getenv("UPBC_VERBOSE") then - print("upbc:") - print(string.format(" source file=%s", src)) - print(string.format(" output file base=%s", outbase)) - print(string.format(" hfilename=%s", hfilename)) - print(string.format(" cfilename=%s", cfilename)) - end - - os.execute(string.format("mkdir -p `dirname %s`", outbase)) - - assert(generate_upbdefs) - -- Legacy generated defs. - local hfile = open(hfilename) - local cfile = open(cfilename) - - local happend = dump_cinit.file_appender(hfile) - local cappend = dump_cinit.file_appender(cfile) - - dump_cinit.dump_defs(file, happend, cappend) - - hfile:close() - cfile:close() -end diff --git a/upb/def.c b/upb/def.c index ba6de50..047684e 100644 --- a/upb/def.c +++ b/upb/def.c @@ -643,7 +643,7 @@ uint32_t upb_msgdef_submsgfieldcount(const upb_msgdef *m) { const upb_fielddef *upb_msgdef_itof(const upb_msgdef *m, uint32_t i) { upb_value val; return upb_inttable_lookup32(&m->itof, i, &val) ? - upb_value_getptr(val) : NULL; + upb_value_getconstptr(val) : NULL; } const upb_fielddef *upb_msgdef_ntof(const upb_msgdef *m, const char *name, @@ -1128,12 +1128,14 @@ static bool create_fielddef( if (m) { /* direct message field. */ + upb_value v, packed_v; + f = (upb_fielddef*)&m->fields[m->field_count++]; f->msgdef = m; f->is_extension_ = false; - upb_value packed_v = pack_def(f, UPB_DEFTYPE_FIELD); - upb_value v = upb_value_constptr(f); + packed_v = pack_def(f, UPB_DEFTYPE_FIELD); + v = upb_value_constptr(f); if (!upb_strtable_insert3(&m->ntof, name.data, name.size, packed_v, alloc)) { upb_status_seterrf(ctx->status, "duplicate field name (%s)", shortname); @@ -1580,5 +1582,48 @@ bool upb_symtab_addfile(upb_symtab *s, return ok; } +/* Include here since we want most of this file to be stdio-free. */ +#include + +bool _upb_symtab_loaddefinit(upb_symtab *s, const upb_def_init *init) { + /* Since this function should never fail (it would indicate a bug in upb) we + * print errors to stderr instead of returning error status to the user. */ + upb_def_init **deps = init->deps; + google_protobuf_FileDescriptorProto *file; + upb_arena arena; + upb_status status = UPB_STATUS_INIT; + + if (upb_strtable_lookup(&s->files, init->filename, NULL)) { + return true; + } + + for (; *deps; deps++) { + if (!_upb_symtab_loaddefinit(s, *deps)) goto err; + } + + upb_arena_init(&arena); + file = google_protobuf_FileDescriptorProto_parsenew(init->descriptor, &arena); + + if (!file) { + upb_status_seterrf( + &status, + "Failed to parse compiled-in descriptor for file '%s'. This should " + "never happen.", + init->filename); + goto err; + } + + if (!upb_symtab_addfile(s, file, &status)) goto err; + + upb_arena_uninit(&arena); + return true; + +err: + fprintf(stderr, "Error loading compiled-in descriptor: %s\n", + upb_status_errmsg(&status)); + upb_arena_uninit(&arena); + return false; +} + #undef CHK #undef CHK_OOM diff --git a/upb/def.h b/upb/def.h index 4fe6d04..e6fdf21 100644 --- a/upb/def.h +++ b/upb/def.h @@ -659,7 +659,6 @@ const upb_filedef *upb_filedef_dep(const upb_filedef *f, int i); const upb_msgdef *upb_filedef_msg(const upb_filedef *f, int i); const upb_enumdef *upb_filedef_enum(const upb_filedef *f, int i); - UPB_END_EXTERN_C #ifdef __cplusplus @@ -707,6 +706,15 @@ bool upb_symtab_addfile(upb_symtab *s, const google_protobuf_FileDescriptorProto* file, upb_status *status); +/* For generated code only: loads a generated descriptor. */ +typedef struct upb_def_init { + struct upb_def_init **deps; + const char *filename; + upb_stringview descriptor; +} upb_def_init; + +bool _upb_symtab_loaddefinit(upb_symtab *s, const upb_def_init *init); + UPB_END_EXTERN_C #ifdef __cplusplus diff --git a/upb/upb.h b/upb/upb.h index 020022b..2fb7a88 100644 --- a/upb/upb.h +++ b/upb/upb.h @@ -463,7 +463,6 @@ struct upb_alloc { UPB_INLINE void *upb_malloc(upb_alloc *alloc, size_t size) { UPB_ASSERT(alloc); - UPB_ASSERT(size < 65535); return alloc->func(alloc, NULL, 0, size); } diff --git a/upbc/generator.cc b/upbc/generator.cc index 68996a9..3b4f6ac 100644 --- a/upbc/generator.cc +++ b/upbc/generator.cc @@ -8,6 +8,7 @@ #include "absl/strings/substitute.h" #include "google/protobuf/compiler/code_generator.h" #include "google/protobuf/descriptor.h" +#include "google/protobuf/descriptor.pb.h" #include "google/protobuf/io/zero_copy_stream.h" #include "upbc/generator.h" @@ -32,6 +33,14 @@ static std::string SourceFilename(std::string proto_filename) { return StripExtension(proto_filename) + ".upb.c"; } +static std::string DefHeaderFilename(std::string proto_filename) { + return StripExtension(proto_filename) + ".upbdefs.h"; +} + +static std::string DefSourceFilename(std::string proto_filename) { + return StripExtension(proto_filename) + ".upbdefs.c"; +} + class Output { public: Output(protobuf::io::ZeroCopyOutputStream* stream) : stream_(stream) {} @@ -165,6 +174,10 @@ std::string ToCIdent(absl::string_view str) { return absl::StrReplaceAll(str, {{".", "_"}, {"/", "_"}}); } +std::string DefInitSymbol(const protobuf::FileDescriptor *file) { + return ToCIdent(file->name()) + "_upbdefinit"; +} + std::string ToPreproc(absl::string_view str) { return absl::AsciiStrToUpper(ToCIdent(str)); } @@ -558,6 +571,91 @@ void WriteSource(const protobuf::FileDescriptor* file, Output& output) { output("\n"); } +void GenerateMessageDefAccessor(const protobuf::Descriptor* d, Output& output) { + output("UPB_INLINE const upb_msgdef *$0_getmsgdef(upb_symtab *s) {\n", + ToCIdent(d->full_name())); + output(" _upb_symtab_loaddefinit(s, &$0);\n", DefInitSymbol(d->file())); + output(" return upb_symtab_lookupmsg(s, \"$0\");\n", d->full_name()); + output("}\n"); + output("\n"); + + for (int i = 0; i < d->nested_type_count(); i++) { + GenerateMessageDefAccessor(d->nested_type(i), output); + } +} + +void WriteDefHeader(const protobuf::FileDescriptor* file, Output& output) { + EmitFileWarning(file, output); + + output("extern upb_def_init $0;\n", DefInitSymbol(file)); + + for (int i = 0; i < file->message_type_count(); i++) { + GenerateMessageDefAccessor(file->message_type(i), output); + } +} + +// Escape C++ trigraphs by escaping question marks to \? +std::string EscapeTrigraphs(absl::string_view to_escape) { + return absl::StrReplaceAll(to_escape, {{"?", "\\?"}}); +} + +void WriteDefSource(const protobuf::FileDescriptor* file, Output& output) { + EmitFileWarning(file, output); + + output("#include \"upb/def.h\"\n"); + output("\n"); + + for (int i = 0; i < file->dependency_count(); i++) { + output("extern upb_def_init $0;\n", DefInitSymbol(file->dependency(i))); + } + + protobuf::FileDescriptorProto file_proto; + file->CopyTo(&file_proto); + std::string file_data; + file_proto.SerializeToString(&file_data); + + output("static const char descriptor[$0] =\n", file_data.size()); + + { + if (file_data.size() > 65535) { + // Workaround for MSVC: "Error C1091: compiler limit: string exceeds + // 65535 bytes in length". Declare a static array of chars rather than + // use a string literal. Only write 25 bytes per line. + static const int kBytesPerLine = 25; + output("{ "); + for (int i = 0; i < file_data.size();) { + for (int j = 0; j < kBytesPerLine && i < file_data.size(); ++i, ++j) { + output("'$0', ", absl::CEscape(file_data.substr(i, 1))); + } + output("\n"); + } + output("'\\0' }"); // null-terminate + } else { + // Only write 40 bytes per line. + static const int kBytesPerLine = 40; + for (int i = 0; i < file_data.size(); i += kBytesPerLine) { + output( + "\"$0\"\n", + EscapeTrigraphs(absl::CEscape(file_data.substr(i, kBytesPerLine)))); + } + } + output(";\n"); + } + + output("static upb_def_init *deps[$0] = {\n", file->dependency_count() + 1); + for (int i = 0; i < file->dependency_count(); i++) { + output(" $0,\n", DefInitSymbol(file->dependency(i))); + } + output(" NULL\n"); + output("};\n"); + + output("upb_def_init $0 = {\n", DefInitSymbol(file)); + output(" deps,\n"); + output(" \"$0\",\n", file->name()); + output(" UPB_STRINGVIEW_INIT(descriptor, $0)\n", file_data.size()); + output("};\n"); +} + bool Generator::Generate(const protobuf::FileDescriptor* file, const std::string& parameter, protoc::GeneratorContext* context, @@ -568,6 +666,12 @@ bool Generator::Generate(const protobuf::FileDescriptor* file, Output c_output(context->Open(SourceFilename(file->name()))); WriteSource(file, c_output); + Output h_def_output(context->Open(DefHeaderFilename(file->name()))); + WriteDefHeader(file, h_def_output); + + Output c_def_output(context->Open(DefSourceFilename(file->name()))); + WriteDefSource(file, c_def_output); + return true; } -- cgit v1.2.3