xymsfilereader.cpp

Go to the documentation of this file.

/////////////////////// StdLib includes
#include <iostream>
#include <iomanip>
 
 
/////////////////////// Qt includes
#include <QDebug>
#include <QFile>
#include <QFileInfo>
 
 
/////////////////////// libpwiz includes
#include <pwiz/data/msdata/DefaultReaderList.hpp>
 
 
/////////////////////// Local includes
#include "xymsfilereader.h"
#include "pappsomspp/core/exception/exceptionnotfound.h"
#include "pappsomspp/core/utils.h"
#include "pappsomspp/core/types.h"
#include "pappsomspp/core/msrun/msrunid.h"
 
 
namespace pappso
{
 
 
XyMsFileReader::XyMsFileReader(const QString &file_name) : MsFileReader{file_name}
{
}
 
 
XyMsFileReader::~XyMsFileReader()
{
}
 
 
bool
XyMsFileReader::initialize(std::size_t &line_count)
{
  // Here we just test all the lines of the file to check that they comply with
  // the xy format.
 
  line_count = 0;
 
  QFile file(m_fileName);
 
  if(!file.open(QFile::ReadOnly | QFile::Text))
    {
      qDebug() << "Failed to open file" << m_fileName;
 
      return 0;
    }
 
  QRegularExpressionMatch regExpMatch;
 
  QString line;
  bool file_reading_failed = false;
 
  while(!file.atEnd())
    {
      line = file.readLine();
 
      // We only read a given number of lines from the file, that would be
      // enough to check if that file has the right syntax or not.
      // if(linesRead >= 2000)
      // return true;
 
      if(line.startsWith('#') || line.isEmpty() || Utils::endOfLineRegExp.match(line).hasMatch())
        continue;
 
      qDebug() << "Current xy format line:" << line;
 
      if(Utils::xyMassDataFormatRegExp.match(line).hasMatch())
        {
          ++line_count;
          continue;
        }
      else
        {
          // the first line of the text file may include column titles
          if(line_count > 0)
            {
              file_reading_failed = true;
              break;
            }
        }
    }
 
  if(!file_reading_failed && line_count >= 1)
    {
      m_fileFormat = Enums::MsDataFormat::xy;
      return true;
    }
 
  m_fileFormat = Enums::MsDataFormat::unknown;
 
  // qDebug() << "m_fileFormat: " << static_cast<int>(m_fileFormat);
 
  return false;
}
 
 
Enums::MsDataFormat
XyMsFileReader::getFileFormat()
{
  return m_fileFormat;
}
 
 
std::vector<MsRunIdCstSPtr>
XyMsFileReader::getMsRunIds(const QString &run_prefix)
{
  std::vector<MsRunIdCstSPtr> ms_run_ids;
 
  std::size_t line_count = 0;
 
  if(!initialize(line_count))
    return ms_run_ids;
 
  // Finally create the MsRunId with the file name.
  MsRunId ms_run_id(m_fileName);
  ms_run_id.setMsDataFormat(m_fileFormat);
 
  // We need to set the unambiguous xmlId string.
  ms_run_id.setXmlId(QString("%1%2").arg(run_prefix).arg(Utils::getLexicalOrderedString(0)));
 
  // Craft a meaningful sample name because otherwise all the files loaded from
  // text files will have the same sample name and it will be difficult to
  // differentiate them.
  // Orig version:
  // ms_run_id.setRunId("Single spectrum");
  // Now the sample name is nothing but the file name without the path.
 
  QFileInfo file_info(m_fileName);
 
  // qDebug() << "file name:" << m_fileName;
 
  QString sample_name = file_info.fileName();
 
  // qDebug() << "sample name:" << sample_name;
 
  ms_run_id.setRunId(sample_name);
 
  // Now set the sample name to the run id:
 
  ms_run_id.setSampleName(ms_run_id.getRunId());
 
  // Now set the sample name to the run id:
 
  ms_run_id.setSampleName(ms_run_id.getRunId());
 
  // qDebug() << __FILE__ << "@" << __LINE__ << __FUNCTION__ << "()"
  //<< "Current ms_run_id:" << ms_run_id.toString();
 
  // Finally make a shared pointer out of it and append it to the vector.
  ms_run_ids.push_back(std::make_shared<MsRunId>(ms_run_id));
 
  return ms_run_ids;
}
 
 
} // namespace pappso