Method NewFromFile

src/DartsDict.cpp:216–324

}

DartsDictPtr DartsDict::NewFromFile(FILE* fp) {
  DartsDictPtr dict(new DartsDict());
  auto internal = dict->internal;

  size_t headerLen = strlen(OCDHEADER);
  std::vector<char> headerBuffer(headerLen);
  size_t bytesRead = fread(headerBuffer.data(), sizeof(char), headerLen, fp);
  if (bytesRead != headerLen || memcmp(headerBuffer.data(), OCDHEADER, headerLen) != 0) {
    throw InvalidFormat("Invalid OpenCC dictionary header");
  }

  // Measure remaining bytes for bounds checking
  long currentOffset = ftell(fp);
  fseek(fp, 0L, SEEK_END);
  long fileEnd = ftell(fp);
  fseek(fp, currentOffset, SEEK_SET);
  size_t remainingSize =
      (fileEnd > currentOffset)
          ? static_cast<size_t>(fileEnd - currentOffset)
          : 0;

  // Detect 32-bit vs 64-bit unit size by reading 8 bytes and checking
  // whether bytes [4..7] are all zero.
  //   - Old 64-bit build: dartsSize field is uint64_t (8 bytes); high 32 bits
  //     are zero for any realistic file size → bytes [4..7] == 0.
  //   - 32-bit build (new or old): dartsSize field is uint32_t (4 bytes);
  //     bytes [4..7] are the first 4 bytes of the darts array (non-zero for
  //     any valid array whose root unit has a non-zero offset).
  uint8_t probe[8];
  if (fread(probe, 1, 8, fp) != 8) {
    throw InvalidFormat("Invalid OpenCC dictionary header (dartsSize)");
  }
  bool is64bit =
      (probe[4] == 0 && probe[5] == 0 && probe[6] == 0 && probe[7] == 0);

  if (is64bit) {
    uint64_t dartsSize64;
    memcpy(&dartsSize64, probe, 8);
    size_t dartsSize = static_cast<size_t>(dartsSize64);
    if (dartsSize > remainingSize) {
      throw InvalidFormat(
          "Invalid OpenCC dictionary (dartsSize exceeds file size)");
    }
    if (dartsSize % sizeof(LegacyUnit64) != 0) {
      throw InvalidFormat("Invalid legacy OCD dictionary unit alignment");
    }
    std::unique_ptr<void, decltype(&free)> buffer(malloc(dartsSize), free);
    if (!buffer) {
      throw std::bad_alloc();
    }
    bytesRead = fread(buffer.get(), 1, dartsSize, fp);
    if (bytesRead != dartsSize) {
      throw InvalidFormat("Invalid legacy OCD dictionary size mismatch");
    }

    BinaryDictPtr binary = BinaryDict::NewFromFile(fp);

    internal->buffer = buffer.release();