path: root/src/tz_info.cc

#include "common.hh"

#include "io.hh"
#include "logger.hh"
#include "tz_info.hh"
#include "tz_str.hh"

#include <byteswap.h>
#include <errno.h>
#include <string.h>
#include <vector>
#include <unistd.h>

namespace {

struct Header {
  char magic[4];
  char ver;
  char unused[15];
  uint32_t isutcnt;
  uint32_t isstdcnt;
  uint32_t leapcnt;
  uint32_t timecnt;
  uint32_t typecnt;
  uint32_t charcnt;
};

#if !HAVE_ATTRIBUTE_PACKED
#pragma pack(push)
#endif
struct
#if HAVE_ATTRIBUTE_PACKED
__attribute__((packed))
#endif
LocalTimeType {
  int32_t utoff;
  uint8_t dst;
  uint8_t idx;
};

struct
#if HAVE_ATTRIBUTE_PACKED
__attribute__((packed))
#endif
LeapSecond {
  int64_t occur;
  int32_t corr;
};
#if !HAVE_ATTRIBUTE_PACKED
#pragma pack(pop)
#endif

struct Data {
  std::vector<int64_t> transition_times;
  std::vector<uint8_t> transition_types;
  std::vector<LocalTimeType> local_time_type_records;
  std::vector<char> time_zone_designations;
  std::vector<LeapSecond> leap_second_records;
  std::vector<uint8_t> standard_or_wall_indicators;
  std::vector<uint8_t> ut_local_indicators;
};

inline uint32_t ntoh(uint32_t value) {
#ifdef WORDS_BIGENDIAN
  return value;
#else
  return bswap_32(value);
#endif
}

inline int32_t ntoh(int32_t value) {
#ifdef WORDS_BIGENDIAN
  return value;
#else
  return bswap_32(value);
#endif
}

inline int64_t ntoh(int64_t value) {
#ifdef WORDS_BIGENDIAN
  return value;
#else
  return bswap_64(value);
#endif
}

class TzInfoImpl : public TzInfo {
public:
  TzInfoImpl(std::shared_ptr<Logger> logger,
             std::filesystem::path tzinfo_dir)
    : logger_(std::move(logger)), base_(std::move(tzinfo_dir)) {
    static_assert(sizeof(Header) == 44, "Header must be packed");
  }

  std::optional<time_t> get_local_time(std::string_view tzname,
                                       time_t utc_time) const override {
    if (!base_.empty()) {
      std::filesystem::path zone = base_ / tzname;
      auto fd = io::open(zone, io::open_flags::rdonly);
      if (fd) {
        return read(zone, fd.get(), utc_time);
      } else {
        logger_->warn("Unable to open %s for reading: %s.", zone.c_str(),
                      strerror(errno));
      }
    }
    return std::nullopt;
  }

private:
  std::optional<time_t> read(std::filesystem::path const& zone, int fd,
                             time_t utc_time) const {
    Header header;
    if (!io::read_all(fd, &header, sizeof(header))) {
      logger_->warn("%s: Error reading: %s.", zone.c_str(),
                    strerror(errno));
      return std::nullopt;
    }
    if (!check_and_fix_header(header)) {
      logger_->warn("%s: Not a TZinfo file.", zone.c_str());
      return std::nullopt;
    }
    // Skip V1, 32bit time stamps are just so 90's.
    auto ret = lseek(fd, data_size(header, '\0'), SEEK_CUR);
    if (ret == -1) {
      logger_->warn("%s: Error seeking: %s.", zone.c_str(), strerror(errno));
      return std::nullopt;
    }
    if (!io::read_all(fd, &header, sizeof(header))) {
      logger_->warn("%s: Error reading: %s.", zone.c_str(),
                    strerror(errno));
      return std::nullopt;
    }
    if (header.ver < '2' || !check_and_fix_header(header)) {
      logger_->warn("%s: Not a TZinfo V2+ file.", zone.c_str());
      return std::nullopt;
    }
    Data data;
    if (!read_data(fd, header, data)) {
      logger_->warn("%s: Error reading: %s.", zone.c_str(),
                    strerror(errno));
      return std::nullopt;
    }
    std::string footer;
    // Note that read_footer might read past the footer, which is currently
    // fine (as it is a footer) but good to know for future developers.
    if (!read_footer(fd, footer)) {
      logger_->warn("%s: Error reading (4): %s.", zone.c_str(),
                    strerror(errno));
      return std::nullopt;
    }

    return utc_to_local(data, footer, utc_time);
  }

  static std::optional<time_t> utc_to_local(Data const& data,
                                            std::string const& footer,
                                            time_t utc) {
    if (data.transition_times.empty())
      return utc_to_local(footer, utc);
    if (utc < data.transition_times.front())
      return std::nullopt;
    size_t i = 0;
    while (i < data.transition_times.size() && utc > data.transition_times[i])
      ++i;

    if (i == 0)
      return std::nullopt;
    if (i == data.transition_times.size() && !footer.empty())
      return utc_to_local(footer, utc);
    --i;

    return utc_to_local(data.local_time_type_records[
                            data.transition_types[i]], utc);
  }

  static std::optional<time_t> utc_to_local(LocalTimeType const& local_time,
                                     time_t utc) {
    return utc += local_time.utoff;
  }

  static std::optional<time_t> utc_to_local(std::string const& tz, time_t utc) {
    return tz::get_local_time(tz, utc);
  }

  static size_t data_size(Header const& header, char version) {
    auto time_size = version >= '2' ? 8 : 4;
    return header.timecnt * time_size + header.timecnt + header.typecnt * 6 +
      header.charcnt + header.leapcnt * (time_size + 4) + header.isstdcnt +
      header.isutcnt;
  }

  static bool read_data(int fd, Header const& header, Data& data) {
    assert(header.ver >= '2');
    data.transition_times.resize(header.timecnt);
    data.transition_types.resize(header.timecnt);
    static_assert(sizeof(LocalTimeType) == 6, "LocalTimeType must be packed");
    data.local_time_type_records.resize(header.typecnt);
    data.time_zone_designations.resize(header.charcnt);
    static_assert(sizeof(LeapSecond) == 12, "LeapSecond must be packed");
    data.leap_second_records.resize(header.leapcnt);
    data.standard_or_wall_indicators.resize(header.isstdcnt);
    data.ut_local_indicators.resize(header.isutcnt);
    if (!io::read_all(fd, data.transition_times.data(),
                      header.timecnt * sizeof(int64_t)) ||
        !io::read_all(fd, data.transition_types.data(), header.timecnt) ||
        !io::read_all(fd, data.local_time_type_records.data(),
                      header.typecnt * sizeof(LocalTimeType)) ||
        !io::read_all(fd, data.time_zone_designations.data(), header.charcnt) ||
        !io::read_all(fd, data.leap_second_records.data(),
                      header.leapcnt * sizeof(LeapSecond)) ||
        !io::read_all(fd, data.standard_or_wall_indicators.data(),
                      header.isstdcnt) ||
        !io::read_all(fd, data.ut_local_indicators.data(), header.isutcnt))
      return false;

    {
      int64_t last = std::numeric_limits<int64_t>::min();
      for (auto& time : data.transition_times) {
        time = ntoh(time);
        if (time <= last)
          return false;
        last = time;
      }
    }

    for (auto const& type : data.transition_types)
      if (type >= header.typecnt)
        return false;

    for (auto& local : data.local_time_type_records) {
      local.utoff = ntoh(local.utoff);
      if (local.utoff == -2147483648)
        return false;
      if (local.dst != 0 && local.dst != 1)
        return false;
      if (local.idx >= header.charcnt)
        return false;
    }

    if (data.time_zone_designations.empty() ||
        data.time_zone_designations.back() != '\0')
      return false;

    {
      LeapSecond last;
      last.occur = -1;
      last.corr = 0;
      for (auto& leap : data.leap_second_records) {
        leap.occur = ntoh(leap.occur);
        leap.corr = ntoh(leap.corr);
        if (last.occur == -1) {
          if (leap.occur < 0)
            return false;
        } else {
          if (leap.occur <= last.occur)
            return false;
        }
        auto diff = last.corr - leap.corr;
        if (diff != 1 && diff != -1)
          return false;
        last.occur = leap.occur + 2419199;
        last.corr = leap.corr;
      }
    }

    for (auto const& standard_or_wall : data.standard_or_wall_indicators)
      if (standard_or_wall != 1 && standard_or_wall != 0)
        return false;

    for (auto const& ut_or_local : data.ut_local_indicators)
      if (ut_or_local != 1 && ut_or_local != 0)
        return false;

    return true;
  }

  static bool read_footer(int fd, std::string& footer) {
    size_t offset = 0;
    footer.resize(128);
    while (true) {
      auto got = io::read(fd, footer.data() + offset, footer.size() - offset);
      if (got <= 0)
        return false;
      if (footer.front() != '\n')
        return false;
      auto end = footer.substr(0, got).find('\n', std::max(1lu, offset));
      if (end != std::string::npos) {
        footer = footer.substr(1, end - 1);
        return true;
      }
      offset += got;
      footer.resize(offset + 128);
    }
  }

  static bool check_and_fix_header(Header& header) {
    if (memcmp(header.magic, "TZif", 4))
      return false;
    header.isutcnt = ntoh(header.isutcnt);
    header.isstdcnt = ntoh(header.isstdcnt);
    header.leapcnt = ntoh(header.leapcnt);
    header.timecnt = ntoh(header.timecnt);
    header.typecnt = ntoh(header.typecnt);
    header.charcnt = ntoh(header.charcnt);

    if (header.isutcnt != 0 && header.isutcnt != header.typecnt)
      return false;
    if (header.isstdcnt != 0 && header.isstdcnt != header.typecnt)
      return false;
    if (header.typecnt == 0)
      return false;
    if (header.charcnt == 0)
      return false;

    return true;
  }

  std::shared_ptr<Logger> logger_;
  std::filesystem::path base_;
};

}  // namespace

std::unique_ptr<TzInfo> TzInfo::create(std::shared_ptr<Logger> logger,
                                       std::filesystem::path tzinfo_dir) {
  return std::make_unique<TzInfoImpl>(std::move(logger), std::move(tzinfo_dir));
}