Chris@148: /* -*- c-basic-offset: 4 indent-tabs-mode: nil -*-  vi:set ts=8 sts=4 sw=4: */
Chris@148: 
Chris@148: /*
Chris@148:     Sonic Visualiser
Chris@148:     An audio file viewer and annotation editor.
Chris@148:     Centre for Digital Music, Queen Mary, University of London.
Chris@148:     This file copyright 2006 Chris Cannam.
Chris@148:     
Chris@148:     This program is free software; you can redistribute it and/or
Chris@148:     modify it under the terms of the GNU General Public License as
Chris@148:     published by the Free Software Foundation; either version 2 of the
Chris@148:     License, or (at your option) any later version.  See the file
Chris@148:     COPYING included with this distribution for more information.
Chris@148: */
Chris@148: 
Chris@148: #include "CSVFileReader.h"
Chris@148: 
Chris@150: #include "model/Model.h"
Chris@148: #include "base/RealTime.h"
Chris@148: #include "model/SparseOneDimensionalModel.h"
Chris@148: #include "model/SparseTimeValueModel.h"
Chris@152: #include "model/EditableDenseThreeDimensionalModel.h"
Chris@308: #include "DataFileReaderFactory.h"
Chris@148: 
Chris@148: #include <QFile>
Chris@148: #include <QString>
Chris@148: #include <QRegExp>
Chris@148: #include <QStringList>
Chris@148: #include <QTextStream>
Chris@148: #include <QFrame>
Chris@148: #include <QGridLayout>
Chris@148: #include <QPushButton>
Chris@148: #include <QHBoxLayout>
Chris@148: #include <QVBoxLayout>
Chris@148: #include <QTableWidget>
Chris@148: #include <QComboBox>
Chris@148: #include <QLabel>
Chris@148: 
Chris@148: #include <iostream>
Chris@148: 
Chris@148: CSVFileReader::CSVFileReader(QString path, size_t mainModelSampleRate) :
Chris@148:     m_file(0),
Chris@148:     m_mainModelSampleRate(mainModelSampleRate)
Chris@148: {
Chris@148:     m_file = new QFile(path);
Chris@148:     bool good = false;
Chris@148:     
Chris@148:     if (!m_file->exists()) {
Chris@148: 	m_error = QFile::tr("File \"%1\" does not exist").arg(path);
Chris@148:     } else if (!m_file->open(QIODevice::ReadOnly | QIODevice::Text)) {
Chris@148: 	m_error = QFile::tr("Failed to open file \"%1\"").arg(path);
Chris@148:     } else {
Chris@148: 	good = true;
Chris@148:     }
Chris@148: 
Chris@148:     if (!good) {
Chris@148: 	delete m_file;
Chris@148: 	m_file = 0;
Chris@148:     }
Chris@148: }
Chris@148: 
Chris@148: CSVFileReader::~CSVFileReader()
Chris@148: {
Chris@148:     std::cerr << "CSVFileReader::~CSVFileReader: file is " << m_file << std::endl;
Chris@148: 
Chris@148:     if (m_file) {
Chris@148:         std::cerr << "CSVFileReader::CSVFileReader: Closing file" << std::endl;
Chris@148:         m_file->close();
Chris@148:     }
Chris@148:     delete m_file;
Chris@148: }
Chris@148: 
Chris@148: bool
Chris@148: CSVFileReader::isOK() const
Chris@148: {
Chris@148:     return (m_file != 0);
Chris@148: }
Chris@148: 
Chris@148: QString
Chris@148: CSVFileReader::getError() const
Chris@148: {
Chris@148:     return m_error;
Chris@148: }
Chris@148: 
Chris@148: Model *
Chris@148: CSVFileReader::load() const
Chris@148: {
Chris@148:     if (!m_file) return 0;
Chris@148: 
Chris@148:     CSVFormatDialog *dialog = new CSVFormatDialog
Chris@148: 	(0, m_file, m_mainModelSampleRate);
Chris@148: 
Chris@148:     if (dialog->exec() == QDialog::Rejected) {
Chris@148: 	delete dialog;
Chris@308:         throw DataFileReaderFactory::ImportCancelled;
Chris@148:     }
Chris@148: 
Chris@148:     CSVFormatDialog::ModelType   modelType = dialog->getModelType();
Chris@148:     CSVFormatDialog::TimingType timingType = dialog->getTimingType();
Chris@148:     CSVFormatDialog::TimeUnits   timeUnits = dialog->getTimeUnits();
Chris@148:     QString separator = dialog->getSeparator();
Chris@148:     size_t sampleRate = dialog->getSampleRate();
Chris@148:     size_t windowSize = dialog->getWindowSize();
Chris@148: 
Chris@148:     delete dialog;
Chris@148: 
Chris@148:     if (timingType == CSVFormatDialog::ExplicitTiming) {
Chris@148: 	windowSize = 1;
Chris@148: 	if (timeUnits == CSVFormatDialog::TimeSeconds) {
Chris@148: 	    sampleRate = m_mainModelSampleRate;
Chris@148: 	}
Chris@148:     }
Chris@148: 
Chris@148:     SparseOneDimensionalModel *model1 = 0;
Chris@148:     SparseTimeValueModel *model2 = 0;
Chris@152:     EditableDenseThreeDimensionalModel *model3 = 0;
Chris@148:     Model *model = 0;
Chris@148: 
Chris@148:     QTextStream in(m_file);
Chris@148:     in.seek(0);
Chris@148: 
Chris@148:     unsigned int warnings = 0, warnLimit = 10;
Chris@148:     unsigned int lineno = 0;
Chris@148: 
Chris@148:     float min = 0.0, max = 0.0;
Chris@148: 
Chris@148:     size_t frameNo = 0;
Chris@148: 
Chris@148:     while (!in.atEnd()) {
Chris@148: 
Chris@283:         // QTextStream's readLine doesn't cope with old-style Mac
Chris@283:         // CR-only line endings.  Why did they bother making the class
Chris@283:         // cope with more than one sort of line ending, if it still
Chris@283:         // can't be configured to cope with all the common sorts?
Chris@148: 
Chris@283:         // For the time being we'll deal with this case (which is
Chris@283:         // relatively uncommon for us, but still necessary to handle)
Chris@283:         // by reading the entire file using a single readLine, and
Chris@283:         // splitting it.  For CR and CR/LF line endings this will just
Chris@283:         // read a line at a time, and that's obviously OK.
Chris@148: 
Chris@283:         QString chunk = in.readLine();
Chris@283:         QStringList lines = chunk.split('\r', QString::SkipEmptyParts);
Chris@283:         
Chris@283:         for (size_t li = 0; li < lines.size(); ++li) {
Chris@148: 
Chris@283:             QString line = lines[li];
Chris@148: 
Chris@283:             if (line.startsWith("#")) continue;
Chris@283: 
Chris@283:             QStringList list = line.split(separator, QString::KeepEmptyParts);
Chris@283: 
Chris@283:             if (!model) {
Chris@283: 
Chris@283:                 switch (modelType) {
Chris@283: 
Chris@283:                 case CSVFormatDialog::OneDimensionalModel:
Chris@283:                     model1 = new SparseOneDimensionalModel(sampleRate, windowSize);
Chris@283:                     model = model1;
Chris@283:                     break;
Chris@148: 		
Chris@283:                 case CSVFormatDialog::TwoDimensionalModel:
Chris@283:                     model2 = new SparseTimeValueModel(sampleRate, windowSize, false);
Chris@283:                     model = model2;
Chris@283:                     break;
Chris@148: 		
Chris@283:                 case CSVFormatDialog::ThreeDimensionalModel:
Chris@283:                     model3 = new EditableDenseThreeDimensionalModel(sampleRate,
Chris@283:                                                                     windowSize,
Chris@283:                                                                     list.size());
Chris@283:                     model = model3;
Chris@283:                     break;
Chris@283:                 }
Chris@283:             }
Chris@148: 
Chris@283:             QStringList tidyList;
Chris@283:             QRegExp nonNumericRx("[^0-9.,+-]");
Chris@148: 
Chris@283:             for (int i = 0; i < list.size(); ++i) {
Chris@148: 	    
Chris@283:                 QString s(list[i].trimmed());
Chris@148: 
Chris@283:                 if (s.length() >= 2 && s.startsWith("\"") && s.endsWith("\"")) {
Chris@283:                     s = s.mid(1, s.length() - 2);
Chris@283:                 } else if (s.length() >= 2 && s.startsWith("'") && s.endsWith("'")) {
Chris@283:                     s = s.mid(1, s.length() - 2);
Chris@283:                 }
Chris@148: 
Chris@283:                 if (i == 0 && timingType == CSVFormatDialog::ExplicitTiming) {
Chris@148: 
Chris@283:                     bool ok = false;
Chris@283:                     QString numeric = s;
Chris@283:                     numeric.remove(nonNumericRx);
Chris@148: 
Chris@283:                     if (timeUnits == CSVFormatDialog::TimeSeconds) {
Chris@148: 
Chris@283:                         double time = numeric.toDouble(&ok);
Chris@283:                         frameNo = int(time * sampleRate + 0.00001);
Chris@148: 
Chris@283:                     } else {
Chris@148: 
Chris@283:                         frameNo = numeric.toInt(&ok);
Chris@148: 
Chris@283:                         if (timeUnits == CSVFormatDialog::TimeWindows) {
Chris@283:                             frameNo *= windowSize;
Chris@283:                         }
Chris@283:                     }
Chris@148: 			       
Chris@283:                     if (!ok) {
Chris@283:                         if (warnings < warnLimit) {
Chris@283:                             std::cerr << "WARNING: CSVFileReader::load: "
Chris@283:                                       << "Bad time format (\"" << s.toStdString()
Chris@283:                                       << "\") in data line "
Chris@283:                                       << lineno << ":" << std::endl;
Chris@283:                             std::cerr << line.toStdString() << std::endl;
Chris@283:                         } else if (warnings == warnLimit) {
Chris@283:                             std::cerr << "WARNING: Too many warnings" << std::endl;
Chris@283:                         }
Chris@283:                         ++warnings;
Chris@283:                     }
Chris@283:                 } else {
Chris@283:                     tidyList.push_back(s);
Chris@283:                 }
Chris@283:             }
Chris@148: 
Chris@283:             if (modelType == CSVFormatDialog::OneDimensionalModel) {
Chris@148: 	    
Chris@283:                 SparseOneDimensionalModel::Point point
Chris@283:                     (frameNo,
Chris@283:                      tidyList.size() > 0 ? tidyList[tidyList.size()-1] :
Chris@283:                      QString("%1").arg(lineno));
Chris@148: 
Chris@283:                 model1->addPoint(point);
Chris@148: 
Chris@283:             } else if (modelType == CSVFormatDialog::TwoDimensionalModel) {
Chris@148: 
Chris@283:                 SparseTimeValueModel::Point point
Chris@283:                     (frameNo,
Chris@283:                      tidyList.size() > 0 ? tidyList[0].toFloat() : 0.0,
Chris@283:                      tidyList.size() > 1 ? tidyList[1] : QString("%1").arg(lineno));
Chris@148: 
Chris@283:                 model2->addPoint(point);
Chris@148: 
Chris@283:             } else if (modelType == CSVFormatDialog::ThreeDimensionalModel) {
Chris@148: 
Chris@283:                 DenseThreeDimensionalModel::Column values;
Chris@148: 
Chris@283:                 for (int i = 0; i < tidyList.size(); ++i) {
Chris@148: 
Chris@283:                     bool ok = false;
Chris@283:                     float value = list[i].toFloat(&ok);
Chris@283:                     values.push_back(value);
Chris@148: 	    
Chris@283:                     if ((lineno == 0 && i == 0) || value < min) min = value;
Chris@283:                     if ((lineno == 0 && i == 0) || value > max) max = value;
Chris@148: 
Chris@283:                     if (!ok) {
Chris@283:                         if (warnings < warnLimit) {
Chris@283:                             std::cerr << "WARNING: CSVFileReader::load: "
Chris@283:                                       << "Non-numeric value in data line " << lineno
Chris@283:                                       << ":" << std::endl;
Chris@283:                             std::cerr << line.toStdString() << std::endl;
Chris@283:                             ++warnings;
Chris@283:                         } else if (warnings == warnLimit) {
Chris@283:                             std::cerr << "WARNING: Too many warnings" << std::endl;
Chris@283:                         }
Chris@283:                     }
Chris@283:                 }
Chris@148: 	
Chris@283:                 std::cerr << "Setting bin values for count " << lineno << ", frame "
Chris@283:                           << frameNo << ", time " << RealTime::frame2RealTime(frameNo, sampleRate) << std::endl;
Chris@148: 
Chris@283:                 model3->setColumn(frameNo / model3->getResolution(), values);
Chris@283:             }
Chris@148: 
Chris@283:             ++lineno;
Chris@283:             if (timingType == CSVFormatDialog::ImplicitTiming ||
Chris@283:                 list.size() == 0) {
Chris@283:                 frameNo += windowSize;
Chris@283:             }
Chris@283:         }
Chris@148:     }
Chris@148: 
Chris@148:     if (modelType == CSVFormatDialog::ThreeDimensionalModel) {
Chris@148: 	model3->setMinimumLevel(min);
Chris@148: 	model3->setMaximumLevel(max);
Chris@148:     }
Chris@148: 
Chris@148:     return model;
Chris@148: }
Chris@148: 
Chris@148: 
Chris@148: CSVFormatDialog::CSVFormatDialog(QWidget *parent, QFile *file,
Chris@148: 				 size_t defaultSampleRate) :
Chris@148:     QDialog(parent),
Chris@148:     m_modelType(OneDimensionalModel),
Chris@148:     m_timingType(ExplicitTiming),
Chris@148:     m_timeUnits(TimeAudioFrames),
Chris@148:     m_separator("")
Chris@148: {
Chris@148:     setModal(true);
Chris@148:     setWindowTitle(tr("Select Data Format"));
Chris@148: 
Chris@148:     (void)guessFormat(file);
Chris@148: 
Chris@148:     QGridLayout *layout = new QGridLayout;
Chris@148: 
Chris@308:     layout->addWidget(new QLabel(tr("<b>Select Data Format</b><p>Please select the correct data format for this file.")),
Chris@148: 		      0, 0, 1, 4);
Chris@148: 
Chris@148:     layout->addWidget(new QLabel(tr("Each row specifies:")), 1, 0);
Chris@148: 
Chris@148:     m_modelTypeCombo = new QComboBox;
Chris@148:     m_modelTypeCombo->addItem(tr("A point in time"));
Chris@148:     m_modelTypeCombo->addItem(tr("A value at a time"));
Chris@148:     m_modelTypeCombo->addItem(tr("A set of values"));
Chris@148:     layout->addWidget(m_modelTypeCombo, 1, 1, 1, 2);
Chris@148:     connect(m_modelTypeCombo, SIGNAL(activated(int)),
Chris@148: 	    this, SLOT(modelTypeChanged(int)));
Chris@148:     m_modelTypeCombo->setCurrentIndex(int(m_modelType));
Chris@148: 
Chris@148:     layout->addWidget(new QLabel(tr("The first column contains:")), 2, 0);
Chris@148:     
Chris@148:     m_timingTypeCombo = new QComboBox;
Chris@148:     m_timingTypeCombo->addItem(tr("Time, in seconds"));
Chris@148:     m_timingTypeCombo->addItem(tr("Time, in audio sample frames"));
Chris@148:     m_timingTypeCombo->addItem(tr("Data (rows are consecutive in time)"));
Chris@148:     layout->addWidget(m_timingTypeCombo, 2, 1, 1, 2);
Chris@148:     connect(m_timingTypeCombo, SIGNAL(activated(int)),
Chris@148: 	    this, SLOT(timingTypeChanged(int)));
Chris@148:     m_timingTypeCombo->setCurrentIndex(m_timingType == ExplicitTiming ?
Chris@148:                                        m_timeUnits == TimeSeconds ? 0 : 1 : 2);
Chris@148: 
Chris@148:     m_sampleRateLabel = new QLabel(tr("Audio sample rate (Hz):"));
Chris@148:     layout->addWidget(m_sampleRateLabel, 3, 0);
Chris@148:     
Chris@148:     size_t sampleRates[] = {
Chris@148: 	8000, 11025, 12000, 22050, 24000, 32000,
Chris@148: 	44100, 48000, 88200, 96000, 176400, 192000
Chris@148:     };
Chris@148: 
Chris@148:     m_sampleRateCombo = new QComboBox;
Chris@148:     m_sampleRate = defaultSampleRate;
Chris@148:     for (size_t i = 0; i < sizeof(sampleRates) / sizeof(sampleRates[0]); ++i) {
Chris@148: 	m_sampleRateCombo->addItem(QString("%1").arg(sampleRates[i]));
Chris@148: 	if (sampleRates[i] == m_sampleRate) m_sampleRateCombo->setCurrentIndex(i);
Chris@148:     }
Chris@148:     m_sampleRateCombo->setEditable(true);
Chris@148: 
Chris@148:     layout->addWidget(m_sampleRateCombo, 3, 1);
Chris@148:     connect(m_sampleRateCombo, SIGNAL(activated(QString)),
Chris@148: 	    this, SLOT(sampleRateChanged(QString)));
Chris@148:     connect(m_sampleRateCombo, SIGNAL(editTextChanged(QString)),
Chris@148: 	    this, SLOT(sampleRateChanged(QString)));
Chris@148: 
Chris@148:     m_windowSizeLabel = new QLabel(tr("Frame increment between rows:"));
Chris@148:     layout->addWidget(m_windowSizeLabel, 4, 0);
Chris@148: 
Chris@148:     m_windowSizeCombo = new QComboBox;
Chris@148:     m_windowSize = 1024;
Chris@148:     for (int i = 0; i <= 16; ++i) {
Chris@148: 	int value = 1 << i;
Chris@148: 	m_windowSizeCombo->addItem(QString("%1").arg(value));
Chris@259: 	if (value == int(m_windowSize)) m_windowSizeCombo->setCurrentIndex(i);
Chris@148:     }
Chris@148:     m_windowSizeCombo->setEditable(true);
Chris@148: 
Chris@148:     layout->addWidget(m_windowSizeCombo, 4, 1);
Chris@148:     connect(m_windowSizeCombo, SIGNAL(activated(QString)),
Chris@148: 	    this, SLOT(windowSizeChanged(QString)));
Chris@148:     connect(m_windowSizeCombo, SIGNAL(editTextChanged(QString)),
Chris@148: 	    this, SLOT(windowSizeChanged(QString)));
Chris@148: 
Chris@148:     layout->addWidget(new QLabel(tr("\nExample data from file:")), 5, 0, 1, 4);
Chris@148: 
Chris@148:     m_exampleWidget = new QTableWidget
Chris@148: 	(std::min(10, m_example.size()), m_maxExampleCols);
Chris@148: 
Chris@148:     layout->addWidget(m_exampleWidget, 6, 0, 1, 4);
Chris@148:     layout->setColumnStretch(3, 10);
Chris@148:     layout->setRowStretch(4, 10);
Chris@148: 
Chris@148:     QPushButton *ok = new QPushButton(tr("OK"));
Chris@148:     connect(ok, SIGNAL(clicked()), this, SLOT(accept()));
Chris@148:     ok->setDefault(true);
Chris@148: 
Chris@148:     QPushButton *cancel = new QPushButton(tr("Cancel"));
Chris@148:     connect(cancel, SIGNAL(clicked()), this, SLOT(reject()));
Chris@148: 
Chris@148:     QHBoxLayout *buttonLayout = new QHBoxLayout;
Chris@148:     buttonLayout->addStretch(1);
Chris@148:     buttonLayout->addWidget(ok);
Chris@148:     buttonLayout->addWidget(cancel);
Chris@148: 
Chris@148:     QVBoxLayout *mainLayout = new QVBoxLayout;
Chris@148:     mainLayout->addLayout(layout);
Chris@148:     mainLayout->addLayout(buttonLayout);
Chris@148: 
Chris@148:     setLayout(mainLayout);
Chris@148:     
Chris@148:     timingTypeChanged(m_timingTypeCombo->currentIndex());
Chris@148: }
Chris@148: 
Chris@148: CSVFormatDialog::~CSVFormatDialog()
Chris@148: {
Chris@148: }
Chris@148: 
Chris@148: void
Chris@148: CSVFormatDialog::populateExample()
Chris@148: {
Chris@148:     m_exampleWidget->setColumnCount
Chris@148: 	(m_timingType == ExplicitTiming ?
Chris@148: 	 m_maxExampleCols - 1 : m_maxExampleCols);
Chris@148: 
Chris@148:     m_exampleWidget->setHorizontalHeaderLabels(QStringList());
Chris@148: 
Chris@148:     for (int i = 0; i < m_example.size(); ++i) {
Chris@148: 	for (int j = 0; j < m_example[i].size(); ++j) {
Chris@148: 
Chris@148: 	    QTableWidgetItem *item = new QTableWidgetItem(m_example[i][j]);
Chris@148: 
Chris@148: 	    if (j == 0) {
Chris@148: 		if (m_timingType == ExplicitTiming) {
Chris@148: 		    m_exampleWidget->setVerticalHeaderItem(i, item);
Chris@148: 		    continue;
Chris@148: 		} else {
Chris@148: 		    QTableWidgetItem *header =
Chris@148: 			new QTableWidgetItem(QString("%1").arg(i));
Chris@148: 		    header->setFlags(Qt::ItemIsEnabled);
Chris@148: 		    m_exampleWidget->setVerticalHeaderItem(i, header);
Chris@148: 		}
Chris@148: 	    }
Chris@148: 	    int index = j;
Chris@148: 	    if (m_timingType == ExplicitTiming) --index;
Chris@148: 	    item->setFlags(Qt::ItemIsEnabled);
Chris@148: 	    m_exampleWidget->setItem(i, index, item);
Chris@148: 	}
Chris@148:     }
Chris@148: }
Chris@148: 
Chris@148: void
Chris@148: CSVFormatDialog::modelTypeChanged(int type)
Chris@148: {
Chris@148:     m_modelType = (ModelType)type;
Chris@148: 
Chris@148:     if (m_modelType == ThreeDimensionalModel) {
Chris@148:         // We can't load 3d models with explicit timing, because the 3d
Chris@148:         // model is dense so we need a fixed sample increment
Chris@148:         m_timingTypeCombo->setCurrentIndex(2);
Chris@148:         timingTypeChanged(2);
Chris@148:     }
Chris@148: }
Chris@148: 
Chris@148: void
Chris@148: CSVFormatDialog::timingTypeChanged(int type)
Chris@148: {
Chris@148:     switch (type) {
Chris@148: 
Chris@148:     case 0:
Chris@148: 	m_timingType = ExplicitTiming;
Chris@148: 	m_timeUnits = TimeSeconds;
Chris@148: 	m_sampleRateCombo->setEnabled(false);
Chris@148: 	m_sampleRateLabel->setEnabled(false);
Chris@148: 	m_windowSizeCombo->setEnabled(false);
Chris@148: 	m_windowSizeLabel->setEnabled(false);
Chris@148:         if (m_modelType == ThreeDimensionalModel) {
Chris@148:             m_modelTypeCombo->setCurrentIndex(1);
Chris@148:             modelTypeChanged(1);
Chris@148:         }
Chris@148: 	break;
Chris@148: 
Chris@148:     case 1:
Chris@148: 	m_timingType = ExplicitTiming;
Chris@148: 	m_timeUnits = TimeAudioFrames;
Chris@148: 	m_sampleRateCombo->setEnabled(true);
Chris@148: 	m_sampleRateLabel->setEnabled(true);
Chris@148: 	m_windowSizeCombo->setEnabled(false);
Chris@148: 	m_windowSizeLabel->setEnabled(false);
Chris@148:         if (m_modelType == ThreeDimensionalModel) {
Chris@148:             m_modelTypeCombo->setCurrentIndex(1);
Chris@148:             modelTypeChanged(1);
Chris@148:         }
Chris@148: 	break;
Chris@148: 
Chris@148:     case 2:
Chris@148: 	m_timingType = ImplicitTiming;
Chris@148: 	m_timeUnits = TimeWindows;
Chris@148: 	m_sampleRateCombo->setEnabled(true);
Chris@148: 	m_sampleRateLabel->setEnabled(true);
Chris@148: 	m_windowSizeCombo->setEnabled(true);
Chris@148: 	m_windowSizeLabel->setEnabled(true);
Chris@148: 	break;
Chris@148:     }
Chris@148: 
Chris@148:     populateExample();
Chris@148: }
Chris@148: 
Chris@148: void
Chris@148: CSVFormatDialog::sampleRateChanged(QString rateString)
Chris@148: {
Chris@148:     bool ok = false;
Chris@148:     int sampleRate = rateString.toInt(&ok);
Chris@148:     if (ok) m_sampleRate = sampleRate;
Chris@148: }
Chris@148: 
Chris@148: void
Chris@148: CSVFormatDialog::windowSizeChanged(QString sizeString)
Chris@148: {
Chris@148:     bool ok = false;
Chris@148:     int size = sizeString.toInt(&ok);
Chris@148:     if (ok) m_windowSize = size;
Chris@148: }
Chris@148: 
Chris@148: bool
Chris@148: CSVFormatDialog::guessFormat(QFile *file)
Chris@148: {
Chris@148:     QTextStream in(file);
Chris@148:     in.seek(0);
Chris@148: 
Chris@148:     unsigned int lineno = 0;
Chris@148: 
Chris@148:     bool nonIncreasingPrimaries = false;
Chris@148:     bool nonNumericPrimaries = false;
Chris@148:     bool floatPrimaries = false;
Chris@148:     bool variableItemCount = false;
Chris@148:     int itemCount = 1;
Chris@148:     int earliestNonNumericItem = -1;
Chris@148: 
Chris@148:     float prevPrimary = 0.0;
Chris@148: 
Chris@148:     m_maxExampleCols = 0;
Chris@148: 
Chris@148:     while (!in.atEnd()) {
Chris@148: 
Chris@283:         // See comment about line endings in load() above
Chris@148: 
Chris@283:         QString chunk = in.readLine();
Chris@283:         QStringList lines = chunk.split('\r', QString::SkipEmptyParts);
Chris@148: 
Chris@283:         for (size_t li = 0; li < lines.size(); ++li) {
Chris@283: 
Chris@283:             QString line = lines[li];
Chris@283: 
Chris@283:             if (line.startsWith("#")) continue;
Chris@283: 
Chris@283:             if (m_separator == "") {
Chris@283:                 //!!! to do: ask the user
Chris@283:                 if (line.split(",").size() >= 2) m_separator = ",";
Chris@283:                 else if (line.split("\t").size() >= 2) m_separator = "\t";
Chris@283:                 else if (line.split("|").size() >= 2) m_separator = "|";
Chris@283:                 else if (line.split("/").size() >= 2) m_separator = "/";
Chris@283:                 else if (line.split(":").size() >= 2) m_separator = ":";
Chris@283:                 else m_separator = " ";
Chris@283:             }
Chris@283: 
Chris@283:             QStringList list = line.split(m_separator);
Chris@283:             QStringList tidyList;
Chris@283: 
Chris@283:             for (int i = 0; i < list.size(); ++i) {
Chris@148: 	    
Chris@283:                 QString s(list[i]);
Chris@283:                 bool numeric = false;
Chris@148: 
Chris@283:                 if (s.length() >= 2 && s.startsWith("\"") && s.endsWith("\"")) {
Chris@283:                     s = s.mid(1, s.length() - 2);
Chris@283:                 } else if (s.length() >= 2 && s.startsWith("'") && s.endsWith("'")) {
Chris@283:                     s = s.mid(1, s.length() - 2);
Chris@283:                 } else {
Chris@283:                     (void)s.toFloat(&numeric);
Chris@283:                 }
Chris@148: 
Chris@283:                 tidyList.push_back(s);
Chris@148: 
Chris@283:                 if (lineno == 0 || (list.size() < itemCount)) {
Chris@283:                     itemCount = list.size();
Chris@283:                 } else {
Chris@283:                     if (itemCount != list.size()) {
Chris@283:                         variableItemCount = true;
Chris@283:                     }
Chris@283:                 }
Chris@148: 	    
Chris@283:                 if (i == 0) { // primary
Chris@148: 
Chris@283:                     if (numeric) {
Chris@148: 
Chris@283:                         float primary = s.toFloat();
Chris@148: 
Chris@283:                         if (lineno > 0 && primary <= prevPrimary) {
Chris@283:                             nonIncreasingPrimaries = true;
Chris@283:                         }
Chris@148: 
Chris@283:                         if (s.contains(".") || s.contains(",")) {
Chris@283:                             floatPrimaries = true;
Chris@283:                         }
Chris@148: 
Chris@283:                         prevPrimary = primary;
Chris@148: 
Chris@283:                     } else {
Chris@283:                         nonNumericPrimaries = true;
Chris@283:                     }
Chris@283:                 } else { // secondary
Chris@148: 
Chris@283:                     if (!numeric) {
Chris@283:                         if (earliestNonNumericItem < 0 ||
Chris@283:                             i < earliestNonNumericItem) {
Chris@283:                             earliestNonNumericItem = i;
Chris@283:                         }
Chris@283:                     }
Chris@283:                 }
Chris@283:             }
Chris@148: 
Chris@283:             if (lineno < 10) {
Chris@283:                 m_example.push_back(tidyList);
Chris@283:                 if (lineno == 0 || tidyList.size() > m_maxExampleCols) {
Chris@283:                     m_maxExampleCols = tidyList.size();
Chris@283:                 }
Chris@283:             }
Chris@148: 
Chris@283:             ++lineno;
Chris@148: 
Chris@283:             if (lineno == 50) break;
Chris@283:         }
Chris@148:     }
Chris@148: 
Chris@148:     if (nonNumericPrimaries || nonIncreasingPrimaries) {
Chris@148: 	
Chris@148: 	// Primaries are probably not a series of times
Chris@148: 
Chris@148: 	m_timingType = ImplicitTiming;
Chris@148: 	m_timeUnits = TimeWindows;
Chris@148: 	
Chris@148: 	if (nonNumericPrimaries) {
Chris@148: 	    m_modelType = OneDimensionalModel;
Chris@148: 	} else if (itemCount == 1 || variableItemCount ||
Chris@148: 		   (earliestNonNumericItem != -1)) {
Chris@148: 	    m_modelType = TwoDimensionalModel;
Chris@148: 	} else {
Chris@148: 	    m_modelType = ThreeDimensionalModel;
Chris@148: 	}
Chris@148: 
Chris@148:     } else {
Chris@148: 
Chris@148: 	// Increasing numeric primaries -- likely to be time
Chris@148: 
Chris@148: 	m_timingType = ExplicitTiming;
Chris@148: 
Chris@148: 	if (floatPrimaries) {
Chris@148: 	    m_timeUnits = TimeSeconds;
Chris@148: 	} else {
Chris@148: 	    m_timeUnits = TimeAudioFrames;
Chris@148: 	}
Chris@148: 
Chris@148: 	if (itemCount == 1) {
Chris@148: 	    m_modelType = OneDimensionalModel;
Chris@148: 	} else if (variableItemCount || (earliestNonNumericItem != -1)) {
Chris@148: 	    if (earliestNonNumericItem != -1 && earliestNonNumericItem < 2) {
Chris@148: 		m_modelType = OneDimensionalModel;
Chris@148: 	    } else {
Chris@148: 		m_modelType = TwoDimensionalModel;
Chris@148: 	    }
Chris@148: 	} else {
Chris@148: 	    m_modelType = ThreeDimensionalModel;
Chris@148: 	}
Chris@148:     }
Chris@148: 
Chris@148:     std::cerr << "Estimated model type: " << m_modelType << std::endl;
Chris@148:     std::cerr << "Estimated timing type: " << m_timingType << std::endl;
Chris@148:     std::cerr << "Estimated units: " << m_timeUnits << std::endl;
Chris@148: 
Chris@148:     in.seek(0);
Chris@148:     return true;
Chris@148: }