C++ Points::InitializeFromFile方法代码示例

/**
 * binary to txt
 */
int Format0(int argc, char** argv)
{
  assert(argv[1][0] == '0' && argv[1][1] == 0);
  if (argc != 6) {
    cout
        << "-- For type 0, read txt file and output binary file, [size] and [dimension] is essential"
        << endl;
    cout << "   " << argv[0] << " 0 txt_file_in binary_file_out size dimension"
        << endl;
    return -1;
  }

  int size = atoi(argv[4]);
  int dimension = atoi(argv[5]);

  Points<DefaultDataTypes> points;
  points.InitializeFromFile(argv[2], size, dimension);
  points.SavePoints(argv[3]);

  return 0;
}

int main(int argc, char** argv)
{
  typedef typename DefaultDataTypes::Value ValueType;
  typedef typename DefaultDataTypes::Dist DistType;
  typedef typename DefaultDataTypes::Index IndexType;
  typedef typename DefaultDataTypes::Dim DimType;
  const size_t BitLength = DefaultHashTypes::BitLength;

  if (argc < 2) {
    cout << "Usage: " << argv[0]
        << " config_file_name [config_key=config_value ...]" << endl;
    return -1;
  }
  //load config from config file
  Config config(argv[1]);

  //load config from argv, note that this may override the key_value of the original
  for (int i = 2; i < argc; i++) {
    string k_v(argv[i]);
    size_t pos = k_v.find("=");
    if (pos == string::npos) {
      cout << "Unrecognized arg:" << k_v << endl;
      cout << "Usage: " << argv[0]
          << " config_file_name [config_key=config_value ...]" << endl;
      return -1;
    } else {
      string key = k_v.substr(0, pos);
      string value = k_v.substr(pos + 1);
      config.Add(key, value);
    }
  }

  if (config.Read<bool>(kShowConfigKey)) {
    cout << "==============config content=============" << endl;
    cout << config;
    cout << "=========================================" << endl;
  }

#ifdef NNPLUS_DEBUG
  std::cout << "DEBUG mode on" << std::endl;
#else
  std::cout << "DEBUG mode off" << std::endl;
#endif

#ifdef USE_PARALLELIZATION
  //control threads used in the whole program
  int open_MPI_threads = config.Read<int>(kOpenMPIThreadsNumKey);
  if (open_MPI_threads != -1 && open_MPI_threads > 0) {
    omp_set_num_threads(open_MPI_threads);
    cout << "omp_set_num_threads:" << open_MPI_threads << endl;
  } else {
    cout << "use default omp_threads" << endl;
  }
#endif

  unsigned int random_seed = config.Read<unsigned int>(kRandomSeedKey);
  srand(random_seed);

  Stopwatch timer("");
  timer.Reset();
  timer.Start();

  bool data_format_binary_flag = config.Read<bool>(kDataFormatBinaryKey);
  Points<DefaultDataTypes> dps;
  string input_data_file_name = config.Read<string>(kDataFileNameKey);
  if (data_format_binary_flag) {
    dps.LoadPoints(input_data_file_name.c_str());
  } else {
    dps.InitializeFromFile(input_data_file_name.c_str(),
        config.Read<IndexType>(kTextDataSizeKey),
        config.Read<DimType>(kTextDataDimKey));
  }
  Points<DefaultDataTypes> qps;
  string input_query_file_name = config.Read<string>(kQueryFileNameKey);
  if (data_format_binary_flag) {
    qps.LoadPoints(input_query_file_name.c_str());
  } else {
    qps.InitializeFromFile(input_query_file_name.c_str(),
        config.Read<IndexType>(kTextQuerySizeKey),
        config.Read<DimType>(kTextQueryDimKey));
  }
  assert(dps.dim_ == qps.dim_);
  cout << "- Reading Data Finished (" << timer.GetTime() << " seconds)" << endl;

  // check expansion and save kg_load time
  SearchEngine<DefaultHashTypes, DefaultDataTypes>::BuildParams bp;
  string search_algorithm = config.Read<string>(kSearchAlgorithm);
  if (search_algorithm == "HashHKM") {
    bp.algorithm = SearchEngine<DefaultHashTypes, DefaultDataTypes>::HashHKM;
    bp.forest_size = config.Read<size_t>(kForestTreeCountKey);
    bp.tree_height = config.Read<size_t>(kTreeHeightKey);
    bp.branches = config.ReadVector<size_t>(kTreeBranchesKey);
  } else if (search_algorithm == "HGNNS") {
    bp.algorithm = SearchEngine<DefaultHashTypes, DefaultDataTypes>::HGNNS;
    bp.k_neighbor = config.Read<size_t>(kKnnGraphNeighborKey);
    bp.forest_size = config.Read<size_t>(kForestTreeCountKey);
    bp.tree_height = config.Read<size_t>(kTreeHeightKey);
    bp.branches = config.ReadVector<size_t>(kTreeBranchesKey);
    bp.keep_branches = config.ReadVector<size_t>(kKeepBranchesKey);
    bp.hamming_keep_branches = config.ReadVector<size_t>(
//.........这里部分代码省略.........

int main(int argc, char** argv)
{
  typedef typename DefaultDataTypes::Value ValueType;
  typedef typename DefaultDataTypes::Dist DistType;
  typedef typename DefaultDataTypes::Index IndexType;
  typedef typename DefaultDataTypes::Dim DimType;

  if (argc < 2) {
    cout << "Usage: " << argv[0]
        << " config_file_name [config_key=config_value ...]" << endl;
    return -1;
  }
  //load config from config file
  Config config(argv[1]);

  //load config from argv, note that this may override the key_value of the original
  for (int i = 2; i < argc; i++) {
    string k_v(argv[i]);
    size_t pos = k_v.find("=");
    if (pos == string::npos) {
      cout << "Unrecognized arg:" << k_v << endl;
      cout << "Usage: " << argv[0]
          << " config_file_name [config_key=config_value ...]" << endl;
      return -1;
    } else {
      string key = k_v.substr(0, pos);
      string value = k_v.substr(pos + 1);
      config.Add(key, value);
    }
  }

  if (config.Read<bool>(kShowConfigKey)) {
    cout << "==============config content=============" << endl;
    cout << config;
    cout << "=========================================" << endl;
  }

  unsigned int random_seed = config.Read<unsigned int>(kRandomSeedKey);
  srand(random_seed);

  Stopwatch timer("");
  timer.Reset();
  timer.Start();

  Points<DefaultDataTypes> dps;
  string input_data_file_name = config.Read<string>(kDataFileNameKey);
  bool data_format_binary_flag = config.Read<bool>(kDataFormatBinaryKey);
  if (data_format_binary_flag) {
    dps.LoadPoints(input_data_file_name.c_str());
  } else {
    dps.InitializeFromFile(input_data_file_name.c_str(),
        config.Read<IndexType>(kTextDataSizeKey),
        config.Read<DimType>(kTextDataDimKey));
  }

  Points<DefaultDataTypes> qps;
  string input_query_file_name = config.Read<string>(kQueryFileNameKey);
  if (data_format_binary_flag) {
    qps.LoadPoints(input_query_file_name.c_str());
  } else {
    qps.InitializeFromFile(input_query_file_name.c_str(),
        config.Read<IndexType>(kTextQuerySizeKey),
        config.Read<DimType>(kTextQueryDimKey));
  }
  assert(dps.dim_ == qps.dim_);

  cout << "- Reading Data Finished (" << timer.GetTime() << " seconds)" << endl;
  if (config.Read<bool>(kSavePointsKey)) {
    cout << "Saving data points to "
        << config.Read<string>(kSaveDataPointsFileName) << endl;
    dps.SavePoints(config.Read<string>(kSaveDataPointsFileName).c_str());
    cout << "Saving query points to "
        << config.Read<string>(kSaveQueryPointsFileName) << endl;
    qps.SavePoints(config.Read<string>(kSaveQueryPointsFileName).c_str());
  }

  size_t knn = config.Read<size_t>(kNearKey);

  timer.Reset();
  timer.Start();
  // load ground truth
  Groundtruth<DefaultDataTypes> groundtruth;
  groundtruth.Initialize(dps, qps, knn,
      &ComputeEuclideanDistance<ValueType, DistType>);

  string gt_file_name_prefix = config.Read<string>(
      kGroundtruthFileNamePrefixKey);
  const size_t kMaxFileNameLength = 256;
  char gt_file_name[kMaxFileNameLength];
  sprintf(gt_file_name, "%s_d%d_q%d_k%d", gt_file_name_prefix.c_str(),
      (int) dps.size_, (int) qps.size_, (int) knn);
  FILE *gt_file = fopen(gt_file_name, "rb");
  if (gt_file != NULL) {
    std::cout << "-- Groundtruth file exists, " << gt_file_name << std::endl;
    std::cout << "-- Loading Groundtruth ..." << std::endl;
    groundtruth.Load(gt_file);
    fclose(gt_file);
  } else {
    std::cout << "-- Groundtruth file not exists, " << gt_file_name
        << std::endl;
//.........这里部分代码省略.........