Method load

src/openms/source/FORMAT/MSPFile.cpp:52–323

  MSPFile::~MSPFile() = default;

  void MSPFile::load(const String & filename, vector<PeptideIdentification> & ids, PeakMap & exp)
  {
    if (!File::exists(filename))
    {
      throw Exception::FileNotFound(__FILE__, __LINE__, OPENMS_PRETTY_FUNCTION, filename);
    }
    if (!File::readable(filename))
    {
      throw Exception::FileNotReadable(__FILE__, __LINE__, OPENMS_PRETTY_FUNCTION, filename);
    }

    // groups everything inside the shortest pair of parentheses
    const std::regex rex(R"(\((.*?)\))");
    // matches 2+ whitespaces or tabs or returns "   ", "\t", "\r"
    // Note: this is a hack because one of the encountered formats has single whitespaces in quotes.
    // TODO choose a format during construction of the class. If we actually knew how to call and define them.
    const std::regex ws_rex(R"(\s{2,}|\t|\r)");

    exp.reset();

    //set DocumentIdentifier
    exp.setLoadedFileType(filename);
    exp.setLoadedFilePath(filename);

    String line;
    ifstream is(filename.c_str());

    std::map<String, String> modname_to_unimod;
    modname_to_unimod["Pyro-glu"] = "Gln->pyro-Glu";
    modname_to_unimod["CAM"] = "Carbamidomethyl";
    modname_to_unimod["AB_old_ICATd8"] = "ICAT-D:2H(8)";
    modname_to_unimod["AB_old_ICATd0"] = "ICAT-D";

    bool parse_headers(param_.getValue("parse_headers").toBool());
    bool parse_peakinfo(param_.getValue("parse_peakinfo").toBool());
    bool parse_firstpeakinfo_only(param_.getValue("parse_firstpeakinfo_only").toBool());
    std::string instrument((std::string)param_.getValue("instrument"));
    bool inst_type_correct(true);
    [[maybe_unused]] bool spectrast_format(false); // TODO: implement usage
    Size spectrum_number = 0;

    PeakSpectrum spec;

    // line number counter
    Size line_number = 0;

    while (getline(is, line))
    {
      ++line_number;

      if (line.hasPrefix("Name:"))
      {
        vector<String> split, split2;
        line.split(' ', split);
        split[1].split('/', split2);
        String peptide = split2[0];
        // in newer NIST versions, the charge is followed by the modification(s) e.g. "_1(0,A,Acetyl)"
        vector<String> split3;