annotate data/fileio/CSVFileReader.cpp @ 897:69cc0454ed72

Make it possible to import CSV files directly into Note layers
author Chris Cannam
date Mon, 07 Apr 2014 10:47:15 +0100
parents f5cd33909744
children 59e7fe1b1003
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@631 20 #include "base/StringBits.h"
Chris@148 21 #include "model/SparseOneDimensionalModel.h"
Chris@148 22 #include "model/SparseTimeValueModel.h"
Chris@152 23 #include "model/EditableDenseThreeDimensionalModel.h"
Chris@628 24 #include "model/RegionModel.h"
Chris@897 25 #include "model/NoteModel.h"
Chris@308 26 #include "DataFileReaderFactory.h"
Chris@148 27
Chris@148 28 #include <QFile>
Chris@148 29 #include <QString>
Chris@148 30 #include <QRegExp>
Chris@148 31 #include <QStringList>
Chris@148 32 #include <QTextStream>
Chris@148 33
Chris@148 34 #include <iostream>
Chris@628 35 #include <map>
Chris@148 36
Chris@392 37 CSVFileReader::CSVFileReader(QString path, CSVFormat format,
Chris@392 38 size_t mainModelSampleRate) :
Chris@392 39 m_format(format),
Chris@148 40 m_file(0),
Chris@631 41 m_warnings(0),
Chris@148 42 m_mainModelSampleRate(mainModelSampleRate)
Chris@148 43 {
Chris@148 44 m_file = new QFile(path);
Chris@148 45 bool good = false;
Chris@148 46
Chris@148 47 if (!m_file->exists()) {
Chris@148 48 m_error = QFile::tr("File \"%1\" does not exist").arg(path);
Chris@148 49 } else if (!m_file->open(QIODevice::ReadOnly | QIODevice::Text)) {
Chris@148 50 m_error = QFile::tr("Failed to open file \"%1\"").arg(path);
Chris@148 51 } else {
Chris@148 52 good = true;
Chris@148 53 }
Chris@148 54
Chris@148 55 if (!good) {
Chris@148 56 delete m_file;
Chris@148 57 m_file = 0;
Chris@148 58 }
Chris@148 59 }
Chris@148 60
Chris@148 61 CSVFileReader::~CSVFileReader()
Chris@148 62 {
Chris@690 63 SVDEBUG << "CSVFileReader::~CSVFileReader: file is " << m_file << endl;
Chris@148 64
Chris@148 65 if (m_file) {
Chris@690 66 SVDEBUG << "CSVFileReader::CSVFileReader: Closing file" << endl;
Chris@148 67 m_file->close();
Chris@148 68 }
Chris@148 69 delete m_file;
Chris@148 70 }
Chris@148 71
Chris@148 72 bool
Chris@148 73 CSVFileReader::isOK() const
Chris@148 74 {
Chris@148 75 return (m_file != 0);
Chris@148 76 }
Chris@148 77
Chris@148 78 QString
Chris@148 79 CSVFileReader::getError() const
Chris@148 80 {
Chris@148 81 return m_error;
Chris@148 82 }
Chris@148 83
Chris@631 84 size_t
Chris@631 85 CSVFileReader::convertTimeValue(QString s, int lineno, size_t sampleRate,
Chris@631 86 size_t windowSize) const
Chris@631 87 {
Chris@631 88 QRegExp nonNumericRx("[^0-9eE.,+-]");
Chris@897 89 int warnLimit = 10;
Chris@631 90
Chris@631 91 CSVFormat::TimeUnits timeUnits = m_format.getTimeUnits();
Chris@631 92
Chris@631 93 size_t calculatedFrame = 0;
Chris@631 94
Chris@631 95 bool ok = false;
Chris@631 96 QString numeric = s;
Chris@631 97 numeric.remove(nonNumericRx);
Chris@631 98
Chris@631 99 if (timeUnits == CSVFormat::TimeSeconds) {
Chris@631 100
Chris@631 101 double time = numeric.toDouble(&ok);
Chris@631 102 if (!ok) time = StringBits::stringToDoubleLocaleFree(numeric, &ok);
Chris@631 103 calculatedFrame = int(time * sampleRate + 0.5);
Chris@631 104
Chris@631 105 } else {
Chris@631 106
Chris@631 107 long n = numeric.toLong(&ok);
Chris@631 108 if (n >= 0) calculatedFrame = n;
Chris@631 109
Chris@631 110 if (timeUnits == CSVFormat::TimeWindows) {
Chris@631 111 calculatedFrame *= windowSize;
Chris@631 112 }
Chris@631 113 }
Chris@631 114
Chris@631 115 if (!ok) {
Chris@631 116 if (m_warnings < warnLimit) {
Chris@843 117 cerr << "WARNING: CSVFileReader::load: "
Chris@844 118 << "Bad time format (\"" << s
Chris@631 119 << "\") in data line "
Chris@843 120 << lineno+1 << endl;
Chris@631 121 } else if (m_warnings == warnLimit) {
Chris@843 122 cerr << "WARNING: Too many warnings" << endl;
Chris@631 123 }
Chris@631 124 ++m_warnings;
Chris@631 125 }
Chris@631 126
Chris@631 127 return calculatedFrame;
Chris@631 128 }
Chris@631 129
Chris@148 130 Model *
Chris@148 131 CSVFileReader::load() const
Chris@148 132 {
Chris@148 133 if (!m_file) return 0;
Chris@148 134
Chris@628 135 CSVFormat::ModelType modelType = m_format.getModelType();
Chris@392 136 CSVFormat::TimingType timingType = m_format.getTimingType();
Chris@628 137 CSVFormat::TimeUnits timeUnits = m_format.getTimeUnits();
Chris@392 138 size_t sampleRate = m_format.getSampleRate();
Chris@392 139 size_t windowSize = m_format.getWindowSize();
Chris@631 140 QChar separator = m_format.getSeparator();
Chris@631 141 bool allowQuoting = m_format.getAllowQuoting();
Chris@148 142
Chris@392 143 if (timingType == CSVFormat::ExplicitTiming) {
Chris@611 144 if (modelType == CSVFormat::ThreeDimensionalModel) {
Chris@611 145 // This will be overridden later if more than one line
Chris@611 146 // appears in our file, but we want to choose a default
Chris@611 147 // that's likely to be visible
Chris@611 148 windowSize = 1024;
Chris@611 149 } else {
Chris@611 150 windowSize = 1;
Chris@611 151 }
Chris@392 152 if (timeUnits == CSVFormat::TimeSeconds) {
Chris@148 153 sampleRate = m_mainModelSampleRate;
Chris@148 154 }
Chris@148 155 }
Chris@148 156
Chris@148 157 SparseOneDimensionalModel *model1 = 0;
Chris@148 158 SparseTimeValueModel *model2 = 0;
Chris@628 159 RegionModel *model2a = 0;
Chris@897 160 NoteModel *model2b = 0;
Chris@152 161 EditableDenseThreeDimensionalModel *model3 = 0;
Chris@148 162 Model *model = 0;
Chris@148 163
Chris@148 164 QTextStream in(m_file);
Chris@148 165 in.seek(0);
Chris@148 166
Chris@148 167 unsigned int warnings = 0, warnLimit = 10;
Chris@148 168 unsigned int lineno = 0;
Chris@148 169
Chris@148 170 float min = 0.0, max = 0.0;
Chris@148 171
Chris@148 172 size_t frameNo = 0;
Chris@628 173 size_t duration = 0;
Chris@631 174 size_t endFrame = 0;
Chris@631 175
Chris@631 176 bool haveAnyValue = false;
Chris@631 177 bool haveEndTime = false;
Chris@897 178 bool pitchLooksLikeMIDI = true;
Chris@631 179
Chris@611 180 size_t startFrame = 0; // for calculation of dense model resolution
Chris@631 181 bool firstEverValue = true;
Chris@148 182
Chris@631 183 std::map<QString, int> labelCountMap;
Chris@631 184
Chris@676 185 int valueColumns = 0;
Chris@676 186 for (int i = 0; i < m_format.getColumnCount(); ++i) {
Chris@676 187 if (m_format.getColumnPurpose(i) == CSVFormat::ColumnValue) {
Chris@676 188 ++valueColumns;
Chris@676 189 }
Chris@676 190 }
Chris@676 191
Chris@148 192 while (!in.atEnd()) {
Chris@148 193
Chris@283 194 // QTextStream's readLine doesn't cope with old-style Mac
Chris@283 195 // CR-only line endings. Why did they bother making the class
Chris@283 196 // cope with more than one sort of line ending, if it still
Chris@283 197 // can't be configured to cope with all the common sorts?
Chris@148 198
Chris@283 199 // For the time being we'll deal with this case (which is
Chris@283 200 // relatively uncommon for us, but still necessary to handle)
Chris@283 201 // by reading the entire file using a single readLine, and
Chris@283 202 // splitting it. For CR and CR/LF line endings this will just
Chris@283 203 // read a line at a time, and that's obviously OK.
Chris@148 204
Chris@283 205 QString chunk = in.readLine();
Chris@283 206 QStringList lines = chunk.split('\r', QString::SkipEmptyParts);
Chris@283 207
Chris@897 208 for (int li = 0; li < lines.size(); ++li) {
Chris@148 209
Chris@283 210 QString line = lines[li];
Chris@148 211
Chris@283 212 if (line.startsWith("#")) continue;
Chris@283 213
Chris@631 214 QStringList list = StringBits::split(line, separator, allowQuoting);
Chris@283 215 if (!model) {
Chris@283 216
Chris@283 217 switch (modelType) {
Chris@283 218
Chris@392 219 case CSVFormat::OneDimensionalModel:
Chris@283 220 model1 = new SparseOneDimensionalModel(sampleRate, windowSize);
Chris@283 221 model = model1;
Chris@283 222 break;
Chris@148 223
Chris@392 224 case CSVFormat::TwoDimensionalModel:
Chris@283 225 model2 = new SparseTimeValueModel(sampleRate, windowSize, false);
Chris@283 226 model = model2;
Chris@283 227 break;
Chris@148 228
Chris@628 229 case CSVFormat::TwoDimensionalModelWithDuration:
Chris@628 230 model2a = new RegionModel(sampleRate, windowSize, false);
Chris@628 231 model = model2a;
Chris@628 232 break;
Chris@628 233
Chris@897 234 case CSVFormat::TwoDimensionalModelWithDurationAndPitch:
Chris@897 235 model2b = new NoteModel(sampleRate, windowSize, false);
Chris@897 236 model = model2b;
Chris@897 237 break;
Chris@897 238
Chris@392 239 case CSVFormat::ThreeDimensionalModel:
Chris@535 240 model3 = new EditableDenseThreeDimensionalModel
Chris@535 241 (sampleRate,
Chris@535 242 windowSize,
Chris@676 243 valueColumns,
Chris@535 244 EditableDenseThreeDimensionalModel::NoCompression);
Chris@283 245 model = model3;
Chris@283 246 break;
Chris@283 247 }
Chris@283 248 }
Chris@148 249
Chris@631 250 float value = 0.f;
Chris@897 251 float pitch = 0.f;
Chris@631 252 QString label = "";
Chris@148 253
Chris@631 254 duration = 0.f;
Chris@631 255 haveEndTime = false;
Chris@628 256
Chris@283 257 for (int i = 0; i < list.size(); ++i) {
Chris@148 258
Chris@631 259 QString s = list[i];
Chris@631 260
Chris@631 261 CSVFormat::ColumnPurpose purpose = m_format.getColumnPurpose(i);
Chris@631 262
Chris@631 263 switch (purpose) {
Chris@631 264
Chris@631 265 case CSVFormat::ColumnUnknown:
Chris@631 266 break;
Chris@631 267
Chris@631 268 case CSVFormat::ColumnStartTime:
Chris@631 269 frameNo = convertTimeValue(s, lineno, sampleRate, windowSize);
Chris@631 270 break;
Chris@631 271
Chris@631 272 case CSVFormat::ColumnEndTime:
Chris@631 273 endFrame = convertTimeValue(s, lineno, sampleRate, windowSize);
Chris@631 274 haveEndTime = true;
Chris@631 275 break;
Chris@631 276
Chris@631 277 case CSVFormat::ColumnDuration:
Chris@631 278 duration = convertTimeValue(s, lineno, sampleRate, windowSize);
Chris@631 279 break;
Chris@631 280
Chris@631 281 case CSVFormat::ColumnValue:
Chris@631 282 value = s.toFloat();
Chris@631 283 haveAnyValue = true;
Chris@631 284 break;
Chris@631 285
Chris@897 286 case CSVFormat::ColumnPitch:
Chris@897 287 pitch = s.toFloat();
Chris@897 288 if (pitch < 0.f || pitch > 127.f) {
Chris@897 289 pitchLooksLikeMIDI = false;
Chris@897 290 }
Chris@897 291 break;
Chris@897 292
Chris@631 293 case CSVFormat::ColumnLabel:
Chris@631 294 label = s;
Chris@631 295 ++labelCountMap[label];
Chris@631 296 break;
Chris@283 297 }
Chris@631 298 }
Chris@148 299
Chris@631 300 if (haveEndTime) { // ... calculate duration now all cols read
Chris@631 301 if (endFrame > frameNo) {
Chris@631 302 duration = endFrame - frameNo;
Chris@628 303 }
Chris@283 304 }
Chris@148 305
Chris@392 306 if (modelType == CSVFormat::OneDimensionalModel) {
Chris@148 307
Chris@631 308 SparseOneDimensionalModel::Point point(frameNo, label);
Chris@283 309 model1->addPoint(point);
Chris@148 310
Chris@392 311 } else if (modelType == CSVFormat::TwoDimensionalModel) {
Chris@148 312
Chris@631 313 SparseTimeValueModel::Point point(frameNo, value, label);
Chris@283 314 model2->addPoint(point);
Chris@148 315
Chris@628 316 } else if (modelType == CSVFormat::TwoDimensionalModelWithDuration) {
Chris@628 317
Chris@631 318 RegionModel::Point point(frameNo, value, duration, label);
Chris@628 319 model2a->addPoint(point);
Chris@628 320
Chris@897 321 } else if (modelType == CSVFormat::TwoDimensionalModelWithDurationAndPitch) {
Chris@897 322
Chris@897 323 float level = ((value >= 0.f && value <= 1.f) ? value : 1.f);
Chris@897 324 NoteModel::Point point(frameNo, pitch, duration, level, label);
Chris@897 325 model2b->addPoint(point);
Chris@897 326
Chris@392 327 } else if (modelType == CSVFormat::ThreeDimensionalModel) {
Chris@148 328
Chris@283 329 DenseThreeDimensionalModel::Column values;
Chris@148 330
Chris@631 331 for (int i = 0; i < list.size(); ++i) {
Chris@148 332
Chris@676 333 if (m_format.getColumnPurpose(i) != CSVFormat::ColumnValue) {
Chris@676 334 continue;
Chris@676 335 }
Chris@676 336
Chris@283 337 bool ok = false;
Chris@283 338 float value = list[i].toFloat(&ok);
Chris@611 339
Chris@676 340 values.push_back(value);
Chris@148 341
Chris@631 342 if (firstEverValue || value < min) min = value;
Chris@631 343 if (firstEverValue || value > max) max = value;
Chris@676 344
Chris@631 345 if (firstEverValue) {
Chris@611 346 startFrame = frameNo;
Chris@611 347 model3->setStartFrame(startFrame);
Chris@611 348 } else if (lineno == 1 &&
Chris@611 349 timingType == CSVFormat::ExplicitTiming) {
Chris@611 350 model3->setResolution(frameNo - startFrame);
Chris@611 351 }
Chris@631 352
Chris@631 353 firstEverValue = false;
Chris@148 354
Chris@283 355 if (!ok) {
Chris@283 356 if (warnings < warnLimit) {
Chris@843 357 cerr << "WARNING: CSVFileReader::load: "
Chris@390 358 << "Non-numeric value \""
Chris@844 359 << list[i]
Chris@491 360 << "\" in data line " << lineno+1
Chris@843 361 << ":" << endl;
Chris@843 362 cerr << line << endl;
Chris@283 363 ++warnings;
Chris@283 364 } else if (warnings == warnLimit) {
Chris@843 365 // cerr << "WARNING: Too many warnings" << endl;
Chris@283 366 }
Chris@283 367 }
Chris@283 368 }
Chris@148 369
Chris@690 370 // SVDEBUG << "Setting bin values for count " << lineno << ", frame "
Chris@687 371 // << frameNo << ", time " << RealTime::frame2RealTime(frameNo, sampleRate) << endl;
Chris@148 372
Chris@611 373 model3->setColumn(lineno, values);
Chris@283 374 }
Chris@148 375
Chris@283 376 ++lineno;
Chris@392 377 if (timingType == CSVFormat::ImplicitTiming ||
Chris@283 378 list.size() == 0) {
Chris@283 379 frameNo += windowSize;
Chris@283 380 }
Chris@283 381 }
Chris@148 382 }
Chris@148 383
Chris@631 384 if (!haveAnyValue) {
Chris@631 385 if (model2a) {
Chris@631 386 // assign values for regions based on label frequency; we
Chris@631 387 // have this in our labelCountMap, sort of
Chris@631 388
Chris@631 389 std::map<int, std::map<QString, float> > countLabelValueMap;
Chris@631 390 for (std::map<QString, int>::iterator i = labelCountMap.begin();
Chris@631 391 i != labelCountMap.end(); ++i) {
Chris@631 392 countLabelValueMap[i->second][i->first] = 0.f;
Chris@631 393 }
Chris@631 394
Chris@631 395 float v = 0.f;
Chris@631 396 for (std::map<int, std::map<QString, float> >::iterator i =
Chris@631 397 countLabelValueMap.end(); i != countLabelValueMap.begin(); ) {
Chris@631 398 --i;
Chris@631 399 for (std::map<QString, float>::iterator j = i->second.begin();
Chris@631 400 j != i->second.end(); ++j) {
Chris@631 401 j->second = v;
Chris@631 402 v = v + 1.f;
Chris@631 403 }
Chris@631 404 }
Chris@631 405
Chris@631 406 std::map<RegionModel::Point, RegionModel::Point,
Chris@631 407 RegionModel::Point::Comparator> pointMap;
Chris@631 408 for (RegionModel::PointList::const_iterator i =
Chris@631 409 model2a->getPoints().begin();
Chris@631 410 i != model2a->getPoints().end(); ++i) {
Chris@631 411 RegionModel::Point p(*i);
Chris@631 412 v = countLabelValueMap[labelCountMap[p.label]][p.label];
Chris@631 413 RegionModel::Point pp(p.frame, v, p.duration, p.label);
Chris@631 414 pointMap[p] = pp;
Chris@631 415 }
Chris@631 416
Chris@631 417 for (std::map<RegionModel::Point, RegionModel::Point>::iterator i =
Chris@631 418 pointMap.begin(); i != pointMap.end(); ++i) {
Chris@631 419 model2a->deletePoint(i->first);
Chris@631 420 model2a->addPoint(i->second);
Chris@631 421 }
Chris@631 422 }
Chris@631 423 }
Chris@631 424
Chris@897 425 if (model2b) {
Chris@897 426 if (pitchLooksLikeMIDI) {
Chris@897 427 model2b->setScaleUnits("MIDI Pitch");
Chris@897 428 } else {
Chris@897 429 model2b->setScaleUnits("Hz");
Chris@897 430 }
Chris@897 431 }
Chris@897 432
Chris@392 433 if (modelType == CSVFormat::ThreeDimensionalModel) {
Chris@148 434 model3->setMinimumLevel(min);
Chris@148 435 model3->setMaximumLevel(max);
Chris@148 436 }
Chris@148 437
Chris@148 438 return model;
Chris@148 439 }
Chris@148 440