path: root/gdbremote_parser.cpp

/* SPDX-License-Identifier: Unlicense
 */

#include "gdbremote_parser.hpp"

#include <cctype>
#include <cassert>
#include <cstring>

using namespace GDBRemote;

enum class TokenType: uint8_t {
    kUnknown = 0,
    kSeparatorComma,
    kSeparatorColon,
    kSeparatorSemicolon,
    kNumeric,
    kHexNumeric,
    kArbitrary,
};

static inline bool isseparator(uint8_t byte)
{
    return ':' == byte || ';' == byte || ',' == byte;
}

static inline bool IsTokenTypeSeparator(TokenType token_type)
{
    return token_type == TokenType::kSeparatorColon ||
        token_type == TokenType::kSeparatorComma ||
        token_type == TokenType::kSeparatorSemicolon;
}

static inline TokenType separatorTypeFromByte(uint8_t byte)
{
    if (':' == byte) return TokenType::kSeparatorColon;
    if (';' == byte) return TokenType::kSeparatorSemicolon;
    return TokenType::kSeparatorComma;
}

struct Token {
    TokenType type{};
    std::string data{};

    static std::vector<Token> Tokenize(std::string packet_data)
    {
        if (packet_data.length() == 0) return std::vector<Token>{};
        std::vector<Token> tokens{};
        TokenType type{};
        size_t i{}, offset{};
        for (; i < packet_data.length(); i++) {
            assert(type != TokenType::kSeparatorColon);
            assert(type != TokenType::kSeparatorSemicolon);
            assert(type != TokenType::kSeparatorComma);
            const uint8_t byte = packet_data[i];
            if (isseparator(byte)) {
                if (i > offset) {
                    assert(type != TokenType::kUnknown);
                    tokens.push_back(Token{type, packet_data.substr(offset, i - offset)});
                    offset = i;
                }
                tokens.push_back(Token{
                        separatorTypeFromByte(byte),
                        packet_data.substr(offset, 1),
                        });
                offset++;
                type = TokenType::kUnknown;
            } else if (isdigit(byte)) {
                if (type == TokenType::kUnknown) {
                    type = TokenType::kNumeric;
                }
            } else if (isxdigit(byte)) {
                if (type != TokenType::kArbitrary) {
                    type = TokenType::kHexNumeric;
                }
            } else {
                type = TokenType::kArbitrary;
            }
        }
        if (i > offset) {
            assert(type != TokenType::kUnknown);
            tokens.push_back(Token{type, packet_data.substr(offset, i - offset)});
        }
        return tokens;
    }
};

struct Command {
    std::string name;
    Packet (*parse)(const std::vector<Token>&&);
    static Packet parseNone(const std::vector<Token>&&)
    {
        return Packet::None();
    }
    static Packet parseQuerySupported(const std::vector<Token>&&)
    {
        // TODO arguments
        return Packet{PacketType::kQuerySupported};
    }
    static Packet parseQueryHaltReason(const std::vector<Token>&&)
    {
        return Packet{PacketType::kQueryHaltReason};
    }
    static Packet parseQueryC(const std::vector<Token>&&)
    {
        // TODO arguments
        return Packet{PacketType::kQueryC};
    }
    static Packet parseQueryFThreadInfo(const std::vector<Token>&&)
    {
        return Packet{PacketType::kQueryFThreadInfo};
    }
    static Packet parseQuerySThreadInfo(const std::vector<Token>&&)
    {
        return Packet{PacketType::kQuerySThreadInfo};
    }
    static Packet parseQueryRTOSThreadInfo(const std::vector<Token>&&)
    {
        // TODO arguments
        return Packet{PacketType::kQueryRTOSThreadInfo};
    }
    static Packet parseQueryAttached(const std::vector<Token>&&)
    {
        // TODO arguments
        return Packet{PacketType::kQueryAttached};
    }
    static Packet parseQueryTraceStatus(const std::vector<Token>&&)
    {
        return Packet{PacketType::kQueryTraceStatus};
    }
    static Packet parseSetThreadForCont(const std::vector<Token>&&)
    {
        // TODO arguments
        return Packet{PacketType::kSetThreadForCont};
    }
    static Packet parseSetThreadForOps(const std::vector<Token>&&)
    {
        // TODO arguments
        return Packet{PacketType::kSetThreadForOps};
    }
    static Packet parseMustReplyEmpty(const std::vector<Token>&&)
    {
        return Packet{PacketType::kVMustReplyEmpty};
    }
    static Packet parseEnableExtendedMode(const std::vector<Token>&&)
    {
        return Packet{PacketType::kEnableExtendedMode};
    }
    static Packet parseInterrupt(const std::vector<Token>&&)
    {
        return Packet{PacketType::kInterrupt};
    }
    static Packet parseContinueVCont(const std::vector<Token>&& tokens)
    {
        const std::string first = tokens[0].data;
        constexpr size_t command_len = strlen("vCont");
        // Check if it is "vCont?" command
        if (first.length() > command_len && first[command_len] == '?') {
            return Packet{PacketType::kContinueAskSupported};
        }
        // TODO arguments
        return Packet{PacketType::kContinue};
    }
    static Packet parseContinue(const std::vector<Token>&&)
    {
        return Packet{PacketType::kContinue};
    }
    static Packet parseReadMemory(const std::vector<Token>&& tokens)
    {
        if (tokens.size() < 3 ||
                tokens[0].type != TokenType::kArbitrary ||
                !IsTokenTypeSeparator(tokens[1].type) ||
                tokens[2].type != TokenType::kNumeric)
        {
            return Packet::None();
        }
        constexpr size_t command_len = strlen("m");
        const std::string first = tokens[0].data;
        const std::string second = tokens[2].data;
        const uint32_t offset = strtol(
                    first.substr(command_len, first.length()-command_len).c_str(),
                    nullptr,
                    16);
        const uint32_t lenght = strtol(second.c_str(), nullptr, 16);
        return Packet{
            PacketType::kReadMemory,
            std::make_unique<const PacketDataReadMemory>(offset, lenght),
        };
    }
    static Packet parseReadGeneralRegisters(const std::vector<Token>&&)
    {
        return Packet{PacketType::kReadGeneralRegisters};
    }
    static Packet parseStep(const std::vector<Token>&&)
    {
        return Packet{PacketType::kStep};
    }
};

static const Command commands[] = {
    { "qSupported",         Command::parseQuerySupported },
    { "?",                  Command::parseQueryHaltReason },
    { "qC",                 Command::parseQueryC },
    { "qfThreadInfo",       Command::parseQueryFThreadInfo },
    { "qsThreadInfo",       Command::parseQuerySThreadInfo },
    { "qL",                 Command::parseQueryRTOSThreadInfo },
    { "qAttached",          Command::parseQueryAttached },
    { "qTStatus",           Command::parseQueryTraceStatus },
    { "Hc",                 Command::parseSetThreadForCont },
    { "Hg",                 Command::parseSetThreadForOps },
    { "vMustReplyEmpty",    Command::parseMustReplyEmpty },
    { "!",                  Command::parseEnableExtendedMode },
    { "\x03",               Command::parseInterrupt },
    { "vCtrlC",             Command::parseInterrupt },
    { "vCont",              Command::parseContinueVCont },
    { "c",                  Command::parseContinue },
    { "m",                  Command::parseReadMemory },
    { "g",                  Command::parseReadGeneralRegisters },
    { "s",                  Command::parseStep },
};

Packet Packet::Parse(std::string packet_data)
{
    const auto tokens = Token::Tokenize(packet_data);
    if (tokens.size() == 0) {
        printf("Warning: no tokens to parse\n");
        return Packet::None();
    }
    const auto first_token = tokens[0];
    for (const auto& command: commands) {
        const auto cmdlen = command.name.length();
        const auto toklen = first_token.data.length();
        // Make sure first token fully includes the command from the beginning
        // and not necessary fully equals to it.
        // TODO tokenize as if the command is always a separate token
        if (cmdlen <= toklen && 0 == memcmp(
                    &command.name[0], &first_token.data[0], cmdlen))
        {
            return command.parse(std::move(tokens));
        }
    }
    return Packet::None();
}

static inline uint8_t parseChecksumFromHexChars(uint8_t first, uint8_t second)
{
    // Assume, that given bytes are valid hex digits in ASCII
    first = ((first & 0x40) ? (first + 9) : first) & 0x0f;
    second = ((second & 0x40) ? (second + 9) : second) & 0x0f;
    return ((first << 4) | second) & 0xff;
}

std::unique_ptr<ExchangeResult> ExchangeContext::Consume(uint8_t byte)
{
    _last_packet += byte;
    switch (_parsing_state) {
    case ParsingState::kIdle:
        if (byte == '$') {
            transit(ParsingState::kPacketData);
        } else if (byte == 0x03) {
            _last_packet = std::string(1, 0x03);
            return ExchangeResult::Ack(_last_packet);
        }
        break;
    case ParsingState::kPacketData:
        if (byte == '#') {
            transit(ParsingState::kChecksum1);
        } else {
            _checksum = (_checksum + byte) & 0xff;
            _packet_data += byte;
        }
        break;
    case ParsingState::kPacketDataBinSecond:
        // TODO escaped and binary data
        break;
    case ParsingState::kChecksum1:
        if (isxdigit(byte)) {
            transit(ParsingState::kChecksum2);
            _given_checksum_first_byte = byte;
        } else {
            transit(ParsingState::kIdle);
            return ExchangeResult::Nak();
        }
        break;
    case ParsingState::kChecksum2:
        const uint8_t checksum = _checksum;
        std::string packet_data(std::move(_packet_data));
        transit(ParsingState::kIdle);
        if (isxdigit(byte)) {
            const uint8_t given_checksum =
                parseChecksumFromHexChars(_given_checksum_first_byte, byte);
            if (given_checksum == checksum) {
                return ExchangeResult::Ack(packet_data);
            } else {
                printf(
                        "Checksum mismatch: received=%02x, expected=%02x\n",
                        given_checksum, checksum);
                printf("Packet data: \"%s\"\n", packet_data.c_str());
                return ExchangeResult::Nak();
            }
        } else {
            return ExchangeResult::Nak();
        }
        break;
    }
    return nullptr;
}

static inline uint8_t CalcPacketDataChecksum(std::string packet) {
    uint8_t checksum{};
    for (const auto c: packet) {
        checksum = (checksum + c) & 0xff;
    }
    return checksum;
}

std::string ExchangeContext::WrapDataToSend(std::string packet)
{
    const uint8_t checksum = CalcPacketDataChecksum(packet);
    const uint8_t c1 = (checksum >> 4) & 0x0f;
    const uint8_t c2 = checksum& 0x0f;
    const char checksum_str[2] = {
        static_cast<char>(c1 < 0xa ? c1 + '0' : c1 + ('a' - 0xa)),
        static_cast<char>(c2 < 0xa ? c2 + '0' : c2 + ('a' - 0xa)),
    };
    return "$" + packet + "#" + std::string(checksum_str, 2);
}

void ExchangeContext::transit(ParsingState new_state)
{
    if (0) {
        printf(
                "GDBRemote ParsingState: %s -> %s\n",
                parsingStateToString(_parsing_state),
                parsingStateToString(new_state));
    }
    switch (new_state) {
    case ParsingState::kIdle:
        _checksum = 0;
        _packet_data.clear();
        break;
    case ParsingState::kPacketData:
        _last_packet = "$";
        break;
    case ParsingState::kPacketDataBinSecond:
        break;
    case ParsingState::kChecksum1:
        break;
    case ParsingState::kChecksum2:
        break;
    }
    _parsing_state = new_state;
}

const char* ExchangeContext::parsingStateToString(ParsingState state)
{
    switch (state) {
    case ParsingState::kIdle:
        return "Idle";
    case ParsingState::kPacketData:
        return "PacketData";
    case ParsingState::kPacketDataBinSecond:
        return "PacketDataBinSecond";
    case ParsingState::kChecksum1:
        return "Checksum1";
    case ParsingState::kChecksum2:
        return "Checksum2";
    }
    return "<unknown>";
}