197440f459544938f019c5ea15893b0416e46ded6Primiano Tucci/*
297440f459544938f019c5ea15893b0416e46ded6Primiano Tucci * Copyright (C) 2017 The Android Open Source Project
397440f459544938f019c5ea15893b0416e46ded6Primiano Tucci *
497440f459544938f019c5ea15893b0416e46ded6Primiano Tucci * Licensed under the Apache License, Version 2.0 (the "License");
597440f459544938f019c5ea15893b0416e46ded6Primiano Tucci * you may not use this file except in compliance with the License.
697440f459544938f019c5ea15893b0416e46ded6Primiano Tucci * You may obtain a copy of the License at
797440f459544938f019c5ea15893b0416e46ded6Primiano Tucci *
897440f459544938f019c5ea15893b0416e46ded6Primiano Tucci *      http://www.apache.org/licenses/LICENSE-2.0
997440f459544938f019c5ea15893b0416e46ded6Primiano Tucci *
1097440f459544938f019c5ea15893b0416e46ded6Primiano Tucci * Unless required by applicable law or agreed to in writing, software
1197440f459544938f019c5ea15893b0416e46ded6Primiano Tucci * distributed under the License is distributed on an "AS IS" BASIS,
1297440f459544938f019c5ea15893b0416e46ded6Primiano Tucci * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
1397440f459544938f019c5ea15893b0416e46ded6Primiano Tucci * See the License for the specific language governing permissions and
1497440f459544938f019c5ea15893b0416e46ded6Primiano Tucci * limitations under the License.
1597440f459544938f019c5ea15893b0416e46ded6Primiano Tucci */
1697440f459544938f019c5ea15893b0416e46ded6Primiano Tucci
174f9b6d7b7b2ad7b86e1577a18dc6e2df91b98164Primiano Tucci#ifndef INCLUDE_PERFETTO_PROTOZERO_PROTO_UTILS_H_
184f9b6d7b7b2ad7b86e1577a18dc6e2df91b98164Primiano Tucci#define INCLUDE_PERFETTO_PROTOZERO_PROTO_UTILS_H_
1997440f459544938f019c5ea15893b0416e46ded6Primiano Tucci
2097440f459544938f019c5ea15893b0416e46ded6Primiano Tucci#include <inttypes.h>
2197440f459544938f019c5ea15893b0416e46ded6Primiano Tucci#include <stddef.h>
2297440f459544938f019c5ea15893b0416e46ded6Primiano Tucci
2397440f459544938f019c5ea15893b0416e46ded6Primiano Tucci#include <type_traits>
2497440f459544938f019c5ea15893b0416e46ded6Primiano Tucci
2597440f459544938f019c5ea15893b0416e46ded6Primiano Tuccinamespace protozero {
2697440f459544938f019c5ea15893b0416e46ded6Primiano Tuccinamespace proto_utils {
2797440f459544938f019c5ea15893b0416e46ded6Primiano Tucci
2897440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// See https://developers.google.com/protocol-buffers/docs/encoding wire types.
2997440f459544938f019c5ea15893b0416e46ded6Primiano Tucci
3097440f459544938f019c5ea15893b0416e46ded6Primiano Tuccienum FieldType : uint32_t {
3197440f459544938f019c5ea15893b0416e46ded6Primiano Tucci  kFieldTypeVarInt = 0,
3297440f459544938f019c5ea15893b0416e46ded6Primiano Tucci  kFieldTypeFixed64 = 1,
3397440f459544938f019c5ea15893b0416e46ded6Primiano Tucci  kFieldTypeLengthDelimited = 2,
3497440f459544938f019c5ea15893b0416e46ded6Primiano Tucci  kFieldTypeFixed32 = 5,
3597440f459544938f019c5ea15893b0416e46ded6Primiano Tucci};
3697440f459544938f019c5ea15893b0416e46ded6Primiano Tucci
3797440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// Maximum message size supported: 256 MiB (4 x 7-bit due to varint encoding).
3897440f459544938f019c5ea15893b0416e46ded6Primiano Tucciconstexpr size_t kMessageLengthFieldSize = 4;
3997440f459544938f019c5ea15893b0416e46ded6Primiano Tucciconstexpr size_t kMaxMessageLength = (1u << (kMessageLengthFieldSize * 7)) - 1;
4097440f459544938f019c5ea15893b0416e46ded6Primiano Tucci
4197440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// Field tag is encoded as 32-bit varint (5 bytes at most).
4297440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// Largest value of simple (not length-delimited) field is 64-bit varint
4397440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// (10 bytes at most). 15 bytes buffer is enough to store a simple field.
4497440f459544938f019c5ea15893b0416e46ded6Primiano Tucciconstexpr size_t kMaxTagEncodedSize = 5;
4597440f459544938f019c5ea15893b0416e46ded6Primiano Tucciconstexpr size_t kMaxSimpleFieldEncodedSize = kMaxTagEncodedSize + 10;
4697440f459544938f019c5ea15893b0416e46ded6Primiano Tucci
4797440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// Proto types: (int|uint|sint)(32|64), bool, enum.
4897440f459544938f019c5ea15893b0416e46ded6Primiano Tucciconstexpr uint32_t MakeTagVarInt(uint32_t field_id) {
4997440f459544938f019c5ea15893b0416e46ded6Primiano Tucci  return (field_id << 3) | kFieldTypeVarInt;
5097440f459544938f019c5ea15893b0416e46ded6Primiano Tucci}
5197440f459544938f019c5ea15893b0416e46ded6Primiano Tucci
5297440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// Proto types: fixed64, sfixed64, fixed32, sfixed32, double, float.
5397440f459544938f019c5ea15893b0416e46ded6Primiano Tuccitemplate <typename T>
5497440f459544938f019c5ea15893b0416e46ded6Primiano Tucciconstexpr uint32_t MakeTagFixed(uint32_t field_id) {
5597440f459544938f019c5ea15893b0416e46ded6Primiano Tucci  static_assert(sizeof(T) == 8 || sizeof(T) == 4, "Value must be 4 or 8 bytes");
5697440f459544938f019c5ea15893b0416e46ded6Primiano Tucci  return (field_id << 3) |
5797440f459544938f019c5ea15893b0416e46ded6Primiano Tucci         (sizeof(T) == 8 ? kFieldTypeFixed64 : kFieldTypeFixed32);
5897440f459544938f019c5ea15893b0416e46ded6Primiano Tucci}
5997440f459544938f019c5ea15893b0416e46ded6Primiano Tucci
6097440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// Proto types: string, bytes, embedded messages.
6197440f459544938f019c5ea15893b0416e46ded6Primiano Tucciconstexpr uint32_t MakeTagLengthDelimited(uint32_t field_id) {
6297440f459544938f019c5ea15893b0416e46ded6Primiano Tucci  return (field_id << 3) | kFieldTypeLengthDelimited;
6397440f459544938f019c5ea15893b0416e46ded6Primiano Tucci}
6497440f459544938f019c5ea15893b0416e46ded6Primiano Tucci
6597440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// Proto types: sint64, sint32.
6697440f459544938f019c5ea15893b0416e46ded6Primiano Tuccitemplate <typename T>
6797440f459544938f019c5ea15893b0416e46ded6Primiano Tucciinline typename std::make_unsigned<T>::type ZigZagEncode(T value) {
6897440f459544938f019c5ea15893b0416e46ded6Primiano Tucci  return static_cast<typename std::make_unsigned<T>::type>(
6997440f459544938f019c5ea15893b0416e46ded6Primiano Tucci      (value << 1) ^ (value >> (sizeof(T) * 8 - 1)));
7097440f459544938f019c5ea15893b0416e46ded6Primiano Tucci}
7197440f459544938f019c5ea15893b0416e46ded6Primiano Tucci
7297440f459544938f019c5ea15893b0416e46ded6Primiano Tuccitemplate <typename T>
7397440f459544938f019c5ea15893b0416e46ded6Primiano Tucciinline uint8_t* WriteVarInt(T value, uint8_t* target) {
7497440f459544938f019c5ea15893b0416e46ded6Primiano Tucci  // Avoid arithmetic (sign expanding) shifts.
7597440f459544938f019c5ea15893b0416e46ded6Primiano Tucci  using UnsignedType = typename std::make_unsigned<T>::type;
7697440f459544938f019c5ea15893b0416e46ded6Primiano Tucci  UnsignedType unsigned_value = static_cast<UnsignedType>(value);
7797440f459544938f019c5ea15893b0416e46ded6Primiano Tucci
7897440f459544938f019c5ea15893b0416e46ded6Primiano Tucci  while (unsigned_value >= 0x80) {
7997440f459544938f019c5ea15893b0416e46ded6Primiano Tucci    *target++ = static_cast<uint8_t>(unsigned_value) | 0x80;
8097440f459544938f019c5ea15893b0416e46ded6Primiano Tucci    unsigned_value >>= 7;
8197440f459544938f019c5ea15893b0416e46ded6Primiano Tucci  }
8297440f459544938f019c5ea15893b0416e46ded6Primiano Tucci  *target = static_cast<uint8_t>(unsigned_value);
8397440f459544938f019c5ea15893b0416e46ded6Primiano Tucci  return target + 1;
8497440f459544938f019c5ea15893b0416e46ded6Primiano Tucci}
8597440f459544938f019c5ea15893b0416e46ded6Primiano Tucci
8697440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// Writes a fixed-size redundant encoding of the given |value|. This is
8797440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// used to backfill fixed-size reservations for the length field using a
8897440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// non-canonical varint encoding (e.g. \x81\x80\x80\x00 instead of \x01).
8997440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// See https://github.com/google/protobuf/issues/1530.
9097440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// In particular, this is used for nested messages. The size of a nested message
9197440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// is not known until all its field have been written. |kMessageLengthFieldSize|
9297440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// bytes are reserved to encode the size field and backfilled at the end.
9397440f459544938f019c5ea15893b0416e46ded6Primiano Tucciinline void WriteRedundantVarInt(uint32_t value, uint8_t* buf) {
9497440f459544938f019c5ea15893b0416e46ded6Primiano Tucci  for (size_t i = 0; i < kMessageLengthFieldSize; ++i) {
9597440f459544938f019c5ea15893b0416e46ded6Primiano Tucci    const uint8_t msb = (i < kMessageLengthFieldSize - 1) ? 0x80 : 0;
9697440f459544938f019c5ea15893b0416e46ded6Primiano Tucci    buf[i] = static_cast<uint8_t>(value) | msb;
9797440f459544938f019c5ea15893b0416e46ded6Primiano Tucci    value >>= 7;
9897440f459544938f019c5ea15893b0416e46ded6Primiano Tucci  }
9997440f459544938f019c5ea15893b0416e46ded6Primiano Tucci}
10097440f459544938f019c5ea15893b0416e46ded6Primiano Tucci
10197440f459544938f019c5ea15893b0416e46ded6Primiano Tuccitemplate <uint32_t field_id>
10297440f459544938f019c5ea15893b0416e46ded6Primiano Tuccivoid StaticAssertSingleBytePreamble() {
10397440f459544938f019c5ea15893b0416e46ded6Primiano Tucci  static_assert(field_id < 16,
10497440f459544938f019c5ea15893b0416e46ded6Primiano Tucci                "Proto field id too big to fit in a single byte preamble");
10597440f459544938f019c5ea15893b0416e46ded6Primiano Tucci};
10697440f459544938f019c5ea15893b0416e46ded6Primiano Tucci
10797440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// Parses a VarInt from the encoded buffer [start, end). |end| is STL-style and
10897440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// points one byte past the end of buffer.
10997440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// The parsed int value is stored in the output arg |value|. Returns a pointer
11097440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// to the next unconsumed byte (so start < retval <= end).
11197440f459544938f019c5ea15893b0416e46ded6Primiano Tucciconst uint8_t* ParseVarInt(const uint8_t* start,
11297440f459544938f019c5ea15893b0416e46ded6Primiano Tucci                           const uint8_t* end,
11397440f459544938f019c5ea15893b0416e46ded6Primiano Tucci                           uint64_t* value);
11497440f459544938f019c5ea15893b0416e46ded6Primiano Tucci
11597440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// Parses a protobuf field and computes its id, type and value.
11697440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// Returns a pointer to the next unconsumed byte (|start| < retval <= end) that
11797440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// is either the beginning of the next field or the end of the parent message.
11897440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// In the case of a kFieldTypeLengthDelimited field, |field_intvalue| will
11997440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// store the length of the payload (either a string or a nested message). In
12097440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// this case, the start of the payload will be at (return value) -
12197440f459544938f019c5ea15893b0416e46ded6Primiano Tucci// |field_intvalue|.
12297440f459544938f019c5ea15893b0416e46ded6Primiano Tucciconst uint8_t* ParseField(const uint8_t* start,
12397440f459544938f019c5ea15893b0416e46ded6Primiano Tucci                          const uint8_t* end,
12497440f459544938f019c5ea15893b0416e46ded6Primiano Tucci                          uint32_t* field_id,
12597440f459544938f019c5ea15893b0416e46ded6Primiano Tucci                          FieldType* field_type,
12697440f459544938f019c5ea15893b0416e46ded6Primiano Tucci                          uint64_t* field_intvalue);
12797440f459544938f019c5ea15893b0416e46ded6Primiano Tucci
12897440f459544938f019c5ea15893b0416e46ded6Primiano Tucci}  // namespace proto_utils
12997440f459544938f019c5ea15893b0416e46ded6Primiano Tucci}  // namespace protozero
13097440f459544938f019c5ea15893b0416e46ded6Primiano Tucci
1314f9b6d7b7b2ad7b86e1577a18dc6e2df91b98164Primiano Tucci#endif  // INCLUDE_PERFETTO_PROTOZERO_PROTO_UTILS_H_
132