Chris@392: /* -*- c-basic-offset: 4 indent-tabs-mode: nil -*-  vi:set ts=8 sts=4 sw=4: */
Chris@392: 
Chris@392: /*
Chris@392:     Sonic Visualiser
Chris@392:     An audio file viewer and annotation editor.
Chris@392:     Centre for Digital Music, Queen Mary, University of London.
Chris@392:     This file copyright 2006 Chris Cannam.
Chris@392:     
Chris@392:     This program is free software; you can redistribute it and/or
Chris@392:     modify it under the terms of the GNU General Public License as
Chris@392:     published by the Free Software Foundation; either version 2 of the
Chris@392:     License, or (at your option) any later version.  See the file
Chris@392:     COPYING included with this distribution for more information.
Chris@392: */
Chris@392: 
Chris@392: #include "CSVFormat.h"
Chris@392: 
Chris@392: #include <QFile>
Chris@392: #include <QString>
Chris@392: #include <QRegExp>
Chris@392: #include <QStringList>
Chris@392: #include <QTextStream>
Chris@392: 
Chris@392: #include <iostream>
Chris@392: 
Chris@392: CSVFormat::CSVFormat(QString filename) :
Chris@392:     m_modelType(TwoDimensionalModel),
Chris@392:     m_timingType(ExplicitTiming),
Chris@628:     m_durationType(Durations),
Chris@392:     m_timeUnits(TimeSeconds),
Chris@392:     m_separator(","),
Chris@392:     m_sampleRate(44100),
Chris@392:     m_windowSize(1024),
Chris@392:     m_behaviour(QString::KeepEmptyParts),
Chris@392:     m_maxExampleCols(0)
Chris@392: {
Chris@392:     QFile file(filename);
Chris@392:     if (!file.exists()) return;
Chris@392:     if (!file.open(QIODevice::ReadOnly | QIODevice::Text)) return;
Chris@392: 
Chris@392:     QTextStream in(&file);
Chris@392:     in.seek(0);
Chris@392: 
Chris@392:     unsigned int lineno = 0;
Chris@392: 
Chris@392:     bool nonIncreasingPrimaries = false;
Chris@628:     bool nonIncreasingSecondaries = false;
Chris@392:     bool nonNumericPrimaries = false;
Chris@392:     bool floatPrimaries = false;
Chris@392:     bool variableItemCount = false;
Chris@392:     int itemCount = 1;
Chris@392:     int earliestNonNumericItem = -1;
Chris@392: 
Chris@392:     float prevPrimary = 0.0;
Chris@628:     float prevSecondary = 0.0;
Chris@392: 
Chris@392:     m_maxExampleCols = 0;
Chris@392:     m_separator = "";
Chris@392: 
Chris@392:     while (!in.atEnd()) {
Chris@392: 
Chris@392:         // See comment about line endings in CSVFileReader::load() 
Chris@392: 
Chris@392:         QString chunk = in.readLine();
Chris@392:         QStringList lines = chunk.split('\r', QString::SkipEmptyParts);
Chris@392: 
Chris@392:         for (size_t li = 0; li < lines.size(); ++li) {
Chris@392: 
Chris@392:             QString line = lines[li];
Chris@392: 
Chris@392:             if (line.startsWith("#")) continue;
Chris@392: 
Chris@392:             m_behaviour = QString::KeepEmptyParts;
Chris@392: 
Chris@392:             if (m_separator == "") {
Chris@392:                 //!!! to do: ask the user
Chris@392:                 if (line.split(",").size() >= 2) m_separator = ",";
Chris@392:                 else if (line.split("\t").size() >= 2) m_separator = "\t";
Chris@392:                 else if (line.split("|").size() >= 2) m_separator = "|";
Chris@392:                 else if (line.split("/").size() >= 2) m_separator = "/";
Chris@392:                 else if (line.split(":").size() >= 2) m_separator = ":";
Chris@392:                 else {
Chris@392:                     m_separator = " ";
Chris@392:                     m_behaviour = QString::SkipEmptyParts;
Chris@392:                 }
Chris@392:             }
Chris@392: 
Chris@406: //            std::cerr << "separator = \"" << m_separator.toStdString() << "\"" << std::endl;
Chris@392: 
Chris@392:             QStringList list = line.split(m_separator, m_behaviour);
Chris@392:             QStringList tidyList;
Chris@392: 
Chris@392:             for (int i = 0; i < list.size(); ++i) {
Chris@392: 	    
Chris@392:                 QString s(list[i]);
Chris@392:                 bool numeric = false;
Chris@392: 
Chris@392:                 if (s.length() >= 2 && s.startsWith("\"") && s.endsWith("\"")) {
Chris@392:                     s = s.mid(1, s.length() - 2);
Chris@392:                 } else if (s.length() >= 2 && s.startsWith("'") && s.endsWith("'")) {
Chris@392:                     s = s.mid(1, s.length() - 2);
Chris@392:                 } else {
Chris@392:                     float f = s.toFloat(&numeric);
Chris@406: //                    std::cerr << "converted \"" << s.toStdString() << "\" to float, got " << f << " and success = " << numeric << std::endl;
Chris@392:                 }
Chris@392: 
Chris@392:                 tidyList.push_back(s);
Chris@392: 
Chris@392:                 if (lineno == 0 || (list.size() < itemCount)) {
Chris@392:                     itemCount = list.size();
Chris@392:                 } else {
Chris@392:                     if (itemCount != list.size()) {
Chris@392:                         variableItemCount = true;
Chris@392:                     }
Chris@392:                 }
Chris@392: 	    
Chris@392:                 if (i == 0) { // primary
Chris@392: 
Chris@392:                     if (numeric) {
Chris@392: 
Chris@392:                         float primary = s.toFloat();
Chris@392: 
Chris@392:                         if (lineno > 0 && primary <= prevPrimary) {
Chris@392:                             nonIncreasingPrimaries = true;
Chris@392:                         }
Chris@392: 
Chris@392:                         if (s.contains(".") || s.contains(",")) {
Chris@392:                             floatPrimaries = true;
Chris@392:                         }
Chris@392: 
Chris@392:                         prevPrimary = primary;
Chris@392: 
Chris@392:                     } else {
Chris@392:                         nonNumericPrimaries = true;
Chris@392:                     }
Chris@392:                 } else { // secondary
Chris@392: 
Chris@392:                     if (!numeric) {
Chris@392:                         if (earliestNonNumericItem < 0 ||
Chris@392:                             i < earliestNonNumericItem) {
Chris@392:                             earliestNonNumericItem = i;
Chris@392:                         }
Chris@628:                     } else if (i == 1) {
Chris@628:                         float secondary = s.toFloat();
Chris@628:                         if (lineno > 0 && secondary <= prevSecondary) {
Chris@628:                             nonIncreasingSecondaries = true;
Chris@628:                         }
Chris@628:                         prevSecondary = secondary;
Chris@392:                     }
Chris@392:                 }
Chris@392:             }
Chris@392: 
Chris@392:             if (lineno < 10) {
Chris@392:                 m_example.push_back(tidyList);
Chris@392:                 if (lineno == 0 || tidyList.size() > m_maxExampleCols) {
Chris@392:                     m_maxExampleCols = tidyList.size();
Chris@392:                 }
Chris@392:             }
Chris@392: 
Chris@392:             ++lineno;
Chris@392: 
Chris@392:             if (lineno == 50) break;
Chris@392:         }
Chris@392:     }
Chris@392: 
Chris@392:     if (nonNumericPrimaries || nonIncreasingPrimaries) {
Chris@392: 	
Chris@392: 	// Primaries are probably not a series of times
Chris@392: 
Chris@392: 	m_timingType = CSVFormat::ImplicitTiming;
Chris@392: 	m_timeUnits = CSVFormat::TimeWindows;
Chris@392: 	
Chris@392: 	if (nonNumericPrimaries) {
Chris@392: 	    m_modelType = CSVFormat::OneDimensionalModel;
Chris@392: 	} else if (itemCount == 1 || variableItemCount ||
Chris@392: 		   (earliestNonNumericItem != -1)) {
Chris@392: 	    m_modelType = CSVFormat::TwoDimensionalModel;
Chris@392: 	} else {
Chris@392: 	    m_modelType = CSVFormat::ThreeDimensionalModel;
Chris@392: 	}
Chris@392: 
Chris@392:     } else {
Chris@392: 
Chris@392: 	// Increasing numeric primaries -- likely to be time
Chris@392: 
Chris@392: 	m_timingType = CSVFormat::ExplicitTiming;
Chris@392: 
Chris@392: 	if (floatPrimaries) {
Chris@392: 	    m_timeUnits = CSVFormat::TimeSeconds;
Chris@392: 	} else {
Chris@392: 	    m_timeUnits = CSVFormat::TimeAudioFrames;
Chris@392: 	}
Chris@392: 
Chris@392: 	if (itemCount == 1) {
Chris@392: 	    m_modelType = CSVFormat::OneDimensionalModel;
Chris@392: 	} else if (variableItemCount || (earliestNonNumericItem != -1)) {
Chris@392: 	    if (earliestNonNumericItem != -1 && earliestNonNumericItem < 2) {
Chris@392: 		m_modelType = CSVFormat::OneDimensionalModel;
Chris@392: 	    } else {
Chris@392: 		m_modelType = CSVFormat::TwoDimensionalModel;
Chris@392: 	    }
Chris@392: 	} else {
Chris@392: 	    m_modelType = CSVFormat::ThreeDimensionalModel;
Chris@392: 	}
Chris@628: 
Chris@628:         if (nonIncreasingSecondaries) {
Chris@628:             m_durationType = Durations;
Chris@628:         } else {
Chris@628:             m_durationType = EndTimes;
Chris@628:         }
Chris@392:     }
Chris@392: 
Chris@392:     std::cerr << "Estimated model type: " << m_modelType << std::endl;
Chris@392:     std::cerr << "Estimated timing type: " << m_timingType << std::endl;
Chris@628:     std::cerr << "Estimated duration type: " << m_durationType << std::endl;
Chris@392:     std::cerr << "Estimated units: " << m_timeUnits << std::endl;
Chris@392: }
Chris@392: