Method load

src/openms/source/FORMAT/PercolatorInfile.cpp:59–265

  }

  vector<PeptideIdentification> PercolatorInfile::load(
    const String& pin_file,
    bool higher_score_better,
    const String& score_name,
    const StringList& extra_scores,
    StringList& filenames,
    String decoy_prefix)
  {
    CsvFile csv(pin_file, '\t');
    StringList header;
    //TODO DANGEROUS! Our CSV reader does not support comment lines!!
    csv.getRow(0, header);

    unordered_map<String, size_t> to_idx; // map column name to column index
    {
      int idx{};
      for (const auto& h : header) { to_idx[h] = idx++; }
    }

    int file_name_column_index{-1};
    if (auto it = std::find(header.begin(), header.end(), "FileName"); it != header.end())
    {
      file_name_column_index = it - header.begin();
    }

    // get column indices of extra scores
    std::set<String> found_extra_scores; // additional (non-main) scores that should be stored in the PeptideHit, order important for comparable idXML
    for (const String& s : extra_scores)
    {
      if (auto it = std::find(header.begin(), header.end(), s); it != header.end())
      {
        found_extra_scores.insert(s);
      }
      else
      {
        OPENMS_LOG_WARN << "Extra score: " << s << " not found in Percolator input file." << endl;
      }
    }

    // charge columns are not standardized, so we check for the format and create hash to lookup column name to charge mapping
    std::regex charge_one_hot_pattern("^charge\\d+$");
    std::regex sage_one_hot_pattern("^z=\\d+$");
    String charge_prefix;
    unordered_map<String, int> col_name_to_charge;

    // Special handling for sage: sage produces an additional column for PSMs outside of the "suggested" charge search range (e.g., charge 2-5).
    // The reason is that sage searches always for the charge annotated in the spectrum raw file. Only if the annotation is missing it will search
    // the suggested charge range.
    bool found_sage_otherz_charge_column{false}; 
    for (const String& c : header)
    {
      if (std::regex_match(c, charge_one_hot_pattern))
      {
        col_name_to_charge[c] = c.substr(6).toInt();
        charge_prefix = "charge";
      }
      else if (std::regex_match(c, sage_one_hot_pattern))
      {