bool ParallelSampler::read_PPR_from_binary_file()

in para_graph_sampler/graph_engine/backend/ParallelSampler.cpp [145:231]

• 87 lines of code
• 22 McCabe index (conditional complexity)

bool ParallelSampler::read_PPR_from_binary_file(
      std::string name_in_neighs, 
      std::string name_in_scores,
      int k, 
      float alpha, 
      float epsilon
) {
  if (name_in_neighs.length() == 0 || name_in_scores.length() == 0) {
    return false;
  }
  std::ifstream fin_neighs(name_in_neighs, std::ios::in | std::ios::binary);
  std::ifstream fin_scores(name_in_scores, std::ios::in | std::ios::binary);
  if (fin_neighs.good() && fin_scores.good()) {
    if (fin_neighs.is_open()) {
      float alpha_ = -1.;
      float epsilon_ = -1.;
      int k_ = -1;
      unsigned int root_cnt = -1;
      fin_neighs.read((char*)&alpha_, sizeof(alpha_));
      fin_neighs.read((char*)&epsilon_, sizeof(epsilon_));
      fin_neighs.read((char*)&k_, sizeof(k_));
      if (alpha_ != alpha || epsilon_ > 1.1*epsilon || epsilon_ < 0.9*epsilon || k_ < k) {
        fin_neighs.close();
        return false;
      }
      std::cout << "meta data matches" << std::endl;
      fin_neighs.read((char*)&root_cnt, sizeof(root_cnt));
      for (unsigned int i = 0; i < root_cnt; i++) {
        assert(static_cast<NodeType>(root_cnt) == graph_full.num_nodes);
        unsigned int deg_;      // NOTE: if change dtype here, need to change the write function too. 
        fin_neighs.read((char*)&deg_, sizeof(deg_));
        auto deg_clip = std::min(static_cast<NodeType>(deg_), static_cast<NodeType>(k));
        for (NodeType j = 0; j < static_cast<NodeType>(deg_); j++) {
          NodeType nidx;
          fin_neighs.read((char*)&nidx, sizeof(nidx));
          if (j < deg_clip) {
            top_ppr_neighs[i].push_back(nidx);
          }
        }
      }
    } else {
      fin_neighs.close();
      return false;
    }
    fin_neighs.close();
    if (fin_scores.is_open()) {
      float alpha_ = -1.;
      float epsilon_ = -1.;
      int k_ = -1;
      unsigned int root_cnt = -1;
      fin_scores.read((char*)&alpha_, sizeof(alpha_));
      fin_scores.read((char*)&epsilon_, sizeof(epsilon_));
      fin_scores.read((char*)&k_, sizeof(k_));
      if (alpha_ != alpha || epsilon_ > 1.1*epsilon || epsilon_ < 0.9*epsilon || k_ < k) {
        fin_scores.close();
        std::fill(top_ppr_neighs.begin(), top_ppr_neighs.end(), std::vector<NodeType>());
        return false;
      }
      std::cout << "meta data matches" << std::endl;
      fin_scores.read((char*)&root_cnt, sizeof(root_cnt));
      for (unsigned int i = 0; i < root_cnt; i++) {
        assert(static_cast<NodeType>(root_cnt) == graph_full.num_nodes);
        unsigned int deg_;
        fin_scores.read((char*)&deg_, sizeof(deg_));
        auto deg_clip = std::min(deg_, static_cast<unsigned int>(k));
        for (unsigned int j = 0; j < deg_; j++) {
          PPRType nscore;
          fin_scores.read((char*)&nscore, sizeof(nscore));
          if (j < deg_clip) {
            top_ppr_scores[i].push_back(nscore);
            if (j > 0) {
              assert(top_ppr_scores[i][j-1] + 1e-25 >= top_ppr_scores[i][j]);
            }
          }
        }
      }
    } else {
      fin_scores.close();
      std::fill(top_ppr_scores.begin(), top_ppr_scores.end(), std::vector<PPRType>());
      return false;
    }
    fin_scores.close();
    return true;
  } else {
    return false;
  }
}