annotate data/fileio/CSVFileReader.cpp @ 469:a8a7b8f698c8

* Revert revisions 1244 and 1245 on FileSource -- backing out its built-in persistent cache support, as CachedFile now does the same thing and will do so more properly
author Chris Cannam
date Tue, 28 Oct 2008 11:07:08 +0000
parents 183ee2a55fc7
children 6f8ee19984ad
rev   line source
Chris@148 1 /* -*- c-basic-offset: 4 indent-tabs-mode: nil -*- vi:set ts=8 sts=4 sw=4: */
Chris@148 2
Chris@148 3 /*
Chris@148 4 Sonic Visualiser
Chris@148 5 An audio file viewer and annotation editor.
Chris@148 6 Centre for Digital Music, Queen Mary, University of London.
Chris@148 7 This file copyright 2006 Chris Cannam.
Chris@148 8
Chris@148 9 This program is free software; you can redistribute it and/or
Chris@148 10 modify it under the terms of the GNU General Public License as
Chris@148 11 published by the Free Software Foundation; either version 2 of the
Chris@148 12 License, or (at your option) any later version. See the file
Chris@148 13 COPYING included with this distribution for more information.
Chris@148 14 */
Chris@148 15
Chris@148 16 #include "CSVFileReader.h"
Chris@148 17
Chris@150 18 #include "model/Model.h"
Chris@148 19 #include "base/RealTime.h"
Chris@148 20 #include "model/SparseOneDimensionalModel.h"
Chris@148 21 #include "model/SparseTimeValueModel.h"
Chris@152 22 #include "model/EditableDenseThreeDimensionalModel.h"
Chris@308 23 #include "DataFileReaderFactory.h"
Chris@148 24
Chris@148 25 #include <QFile>
Chris@148 26 #include <QString>
Chris@148 27 #include <QRegExp>
Chris@148 28 #include <QStringList>
Chris@148 29 #include <QTextStream>
Chris@148 30
Chris@148 31 #include <iostream>
Chris@148 32
Chris@392 33 CSVFileReader::CSVFileReader(QString path, CSVFormat format,
Chris@392 34 size_t mainModelSampleRate) :
Chris@392 35 m_format(format),
Chris@148 36 m_file(0),
Chris@148 37 m_mainModelSampleRate(mainModelSampleRate)
Chris@148 38 {
Chris@148 39 m_file = new QFile(path);
Chris@148 40 bool good = false;
Chris@148 41
Chris@148 42 if (!m_file->exists()) {
Chris@148 43 m_error = QFile::tr("File \"%1\" does not exist").arg(path);
Chris@148 44 } else if (!m_file->open(QIODevice::ReadOnly | QIODevice::Text)) {
Chris@148 45 m_error = QFile::tr("Failed to open file \"%1\"").arg(path);
Chris@148 46 } else {
Chris@148 47 good = true;
Chris@148 48 }
Chris@148 49
Chris@148 50 if (!good) {
Chris@148 51 delete m_file;
Chris@148 52 m_file = 0;
Chris@148 53 }
Chris@148 54 }
Chris@148 55
Chris@148 56 CSVFileReader::~CSVFileReader()
Chris@148 57 {
Chris@148 58 std::cerr << "CSVFileReader::~CSVFileReader: file is " << m_file << std::endl;
Chris@148 59
Chris@148 60 if (m_file) {
Chris@148 61 std::cerr << "CSVFileReader::CSVFileReader: Closing file" << std::endl;
Chris@148 62 m_file->close();
Chris@148 63 }
Chris@148 64 delete m_file;
Chris@148 65 }
Chris@148 66
Chris@148 67 bool
Chris@148 68 CSVFileReader::isOK() const
Chris@148 69 {
Chris@148 70 return (m_file != 0);
Chris@148 71 }
Chris@148 72
Chris@148 73 QString
Chris@148 74 CSVFileReader::getError() const
Chris@148 75 {
Chris@148 76 return m_error;
Chris@148 77 }
Chris@148 78
Chris@148 79 Model *
Chris@148 80 CSVFileReader::load() const
Chris@148 81 {
Chris@148 82 if (!m_file) return 0;
Chris@392 83 /*!!!
Chris@148 84 CSVFormatDialog *dialog = new CSVFormatDialog
Chris@148 85 (0, m_file, m_mainModelSampleRate);
Chris@148 86
Chris@148 87 if (dialog->exec() == QDialog::Rejected) {
Chris@148 88 delete dialog;
Chris@308 89 throw DataFileReaderFactory::ImportCancelled;
Chris@148 90 }
Chris@392 91 */
Chris@148 92
Chris@392 93 CSVFormat::ModelType modelType = m_format.getModelType();
Chris@392 94 CSVFormat::TimingType timingType = m_format.getTimingType();
Chris@392 95 CSVFormat::TimeUnits timeUnits = m_format.getTimeUnits();
Chris@392 96 QString separator = m_format.getSeparator();
Chris@392 97 QString::SplitBehavior behaviour = m_format.getSplitBehaviour();
Chris@392 98 size_t sampleRate = m_format.getSampleRate();
Chris@392 99 size_t windowSize = m_format.getWindowSize();
Chris@148 100
Chris@392 101 if (timingType == CSVFormat::ExplicitTiming) {
Chris@148 102 windowSize = 1;
Chris@392 103 if (timeUnits == CSVFormat::TimeSeconds) {
Chris@148 104 sampleRate = m_mainModelSampleRate;
Chris@148 105 }
Chris@148 106 }
Chris@148 107
Chris@148 108 SparseOneDimensionalModel *model1 = 0;
Chris@148 109 SparseTimeValueModel *model2 = 0;
Chris@152 110 EditableDenseThreeDimensionalModel *model3 = 0;
Chris@148 111 Model *model = 0;
Chris@148 112
Chris@148 113 QTextStream in(m_file);
Chris@148 114 in.seek(0);
Chris@148 115
Chris@148 116 unsigned int warnings = 0, warnLimit = 10;
Chris@148 117 unsigned int lineno = 0;
Chris@148 118
Chris@148 119 float min = 0.0, max = 0.0;
Chris@148 120
Chris@148 121 size_t frameNo = 0;
Chris@148 122
Chris@148 123 while (!in.atEnd()) {
Chris@148 124
Chris@283 125 // QTextStream's readLine doesn't cope with old-style Mac
Chris@283 126 // CR-only line endings. Why did they bother making the class
Chris@283 127 // cope with more than one sort of line ending, if it still
Chris@283 128 // can't be configured to cope with all the common sorts?
Chris@148 129
Chris@283 130 // For the time being we'll deal with this case (which is
Chris@283 131 // relatively uncommon for us, but still necessary to handle)
Chris@283 132 // by reading the entire file using a single readLine, and
Chris@283 133 // splitting it. For CR and CR/LF line endings this will just
Chris@283 134 // read a line at a time, and that's obviously OK.
Chris@148 135
Chris@283 136 QString chunk = in.readLine();
Chris@283 137 QStringList lines = chunk.split('\r', QString::SkipEmptyParts);
Chris@283 138
Chris@283 139 for (size_t li = 0; li < lines.size(); ++li) {
Chris@148 140
Chris@283 141 QString line = lines[li];
Chris@148 142
Chris@283 143 if (line.startsWith("#")) continue;
Chris@283 144
Chris@390 145 QStringList list = line.split(separator, behaviour);
Chris@283 146
Chris@283 147 if (!model) {
Chris@283 148
Chris@283 149 switch (modelType) {
Chris@283 150
Chris@392 151 case CSVFormat::OneDimensionalModel:
Chris@283 152 model1 = new SparseOneDimensionalModel(sampleRate, windowSize);
Chris@283 153 model = model1;
Chris@283 154 break;
Chris@148 155
Chris@392 156 case CSVFormat::TwoDimensionalModel:
Chris@283 157 model2 = new SparseTimeValueModel(sampleRate, windowSize, false);
Chris@283 158 model = model2;
Chris@283 159 break;
Chris@148 160
Chris@392 161 case CSVFormat::ThreeDimensionalModel:
Chris@283 162 model3 = new EditableDenseThreeDimensionalModel(sampleRate,
Chris@283 163 windowSize,
Chris@283 164 list.size());
Chris@283 165 model = model3;
Chris@283 166 break;
Chris@283 167 }
Chris@283 168 }
Chris@148 169
Chris@283 170 QStringList tidyList;
Chris@390 171 QRegExp nonNumericRx("[^0-9eE.,+-]");
Chris@148 172
Chris@283 173 for (int i = 0; i < list.size(); ++i) {
Chris@148 174
Chris@283 175 QString s(list[i].trimmed());
Chris@148 176
Chris@283 177 if (s.length() >= 2 && s.startsWith("\"") && s.endsWith("\"")) {
Chris@283 178 s = s.mid(1, s.length() - 2);
Chris@283 179 } else if (s.length() >= 2 && s.startsWith("'") && s.endsWith("'")) {
Chris@283 180 s = s.mid(1, s.length() - 2);
Chris@283 181 }
Chris@148 182
Chris@392 183 if (i == 0 && timingType == CSVFormat::ExplicitTiming) {
Chris@148 184
Chris@283 185 bool ok = false;
Chris@283 186 QString numeric = s;
Chris@283 187 numeric.remove(nonNumericRx);
Chris@148 188
Chris@392 189 if (timeUnits == CSVFormat::TimeSeconds) {
Chris@148 190
Chris@283 191 double time = numeric.toDouble(&ok);
Chris@283 192 frameNo = int(time * sampleRate + 0.00001);
Chris@148 193
Chris@283 194 } else {
Chris@148 195
Chris@283 196 frameNo = numeric.toInt(&ok);
Chris@148 197
Chris@392 198 if (timeUnits == CSVFormat::TimeWindows) {
Chris@283 199 frameNo *= windowSize;
Chris@283 200 }
Chris@283 201 }
Chris@148 202
Chris@283 203 if (!ok) {
Chris@283 204 if (warnings < warnLimit) {
Chris@283 205 std::cerr << "WARNING: CSVFileReader::load: "
Chris@283 206 << "Bad time format (\"" << s.toStdString()
Chris@283 207 << "\") in data line "
Chris@283 208 << lineno << ":" << std::endl;
Chris@283 209 std::cerr << line.toStdString() << std::endl;
Chris@283 210 } else if (warnings == warnLimit) {
Chris@283 211 std::cerr << "WARNING: Too many warnings" << std::endl;
Chris@283 212 }
Chris@283 213 ++warnings;
Chris@283 214 }
Chris@283 215 } else {
Chris@283 216 tidyList.push_back(s);
Chris@283 217 }
Chris@283 218 }
Chris@148 219
Chris@392 220 if (modelType == CSVFormat::OneDimensionalModel) {
Chris@148 221
Chris@283 222 SparseOneDimensionalModel::Point point
Chris@283 223 (frameNo,
Chris@283 224 tidyList.size() > 0 ? tidyList[tidyList.size()-1] :
Chris@283 225 QString("%1").arg(lineno));
Chris@148 226
Chris@283 227 model1->addPoint(point);
Chris@148 228
Chris@392 229 } else if (modelType == CSVFormat::TwoDimensionalModel) {
Chris@148 230
Chris@283 231 SparseTimeValueModel::Point point
Chris@283 232 (frameNo,
Chris@283 233 tidyList.size() > 0 ? tidyList[0].toFloat() : 0.0,
Chris@283 234 tidyList.size() > 1 ? tidyList[1] : QString("%1").arg(lineno));
Chris@148 235
Chris@283 236 model2->addPoint(point);
Chris@148 237
Chris@392 238 } else if (modelType == CSVFormat::ThreeDimensionalModel) {
Chris@148 239
Chris@283 240 DenseThreeDimensionalModel::Column values;
Chris@148 241
Chris@283 242 for (int i = 0; i < tidyList.size(); ++i) {
Chris@148 243
Chris@283 244 bool ok = false;
Chris@283 245 float value = list[i].toFloat(&ok);
Chris@283 246 values.push_back(value);
Chris@148 247
Chris@283 248 if ((lineno == 0 && i == 0) || value < min) min = value;
Chris@283 249 if ((lineno == 0 && i == 0) || value > max) max = value;
Chris@148 250
Chris@283 251 if (!ok) {
Chris@283 252 if (warnings < warnLimit) {
Chris@283 253 std::cerr << "WARNING: CSVFileReader::load: "
Chris@390 254 << "Non-numeric value \""
Chris@390 255 << list[i].toStdString()
Chris@390 256 << "\" in data line " << lineno
Chris@283 257 << ":" << std::endl;
Chris@283 258 std::cerr << line.toStdString() << std::endl;
Chris@283 259 ++warnings;
Chris@283 260 } else if (warnings == warnLimit) {
Chris@390 261 // std::cerr << "WARNING: Too many warnings" << std::endl;
Chris@283 262 }
Chris@283 263 }
Chris@283 264 }
Chris@148 265
Chris@390 266 // std::cerr << "Setting bin values for count " << lineno << ", frame "
Chris@390 267 // << frameNo << ", time " << RealTime::frame2RealTime(frameNo, sampleRate) << std::endl;
Chris@148 268
Chris@283 269 model3->setColumn(frameNo / model3->getResolution(), values);
Chris@283 270 }
Chris@148 271
Chris@283 272 ++lineno;
Chris@392 273 if (timingType == CSVFormat::ImplicitTiming ||
Chris@283 274 list.size() == 0) {
Chris@283 275 frameNo += windowSize;
Chris@283 276 }
Chris@283 277 }
Chris@148 278 }
Chris@148 279
Chris@392 280 if (modelType == CSVFormat::ThreeDimensionalModel) {
Chris@148 281 model3->setMinimumLevel(min);
Chris@148 282 model3->setMaximumLevel(max);
Chris@148 283 }
Chris@148 284
Chris@148 285 return model;
Chris@148 286 }
Chris@148 287