annotate data/fileio/CSVFileReader.cpp @ 228:f02989f7e160

...
author Chris Cannam
date Mon, 12 Feb 2007 13:09:31 +0000
parents f75f8a1cd7b1
children 9c85517ff0f5
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@148 23
Chris@148 24 #include <QFile>
Chris@148 25 #include <QString>
Chris@148 26 #include <QRegExp>
Chris@148 27 #include <QStringList>
Chris@148 28 #include <QTextStream>
Chris@148 29 #include <QFrame>
Chris@148 30 #include <QGridLayout>
Chris@148 31 #include <QPushButton>
Chris@148 32 #include <QHBoxLayout>
Chris@148 33 #include <QVBoxLayout>
Chris@148 34 #include <QTableWidget>
Chris@148 35 #include <QComboBox>
Chris@148 36 #include <QLabel>
Chris@148 37
Chris@148 38 #include <iostream>
Chris@148 39
Chris@148 40 CSVFileReader::CSVFileReader(QString path, size_t mainModelSampleRate) :
Chris@148 41 m_file(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@148 63 std::cerr << "CSVFileReader::~CSVFileReader: file is " << m_file << std::endl;
Chris@148 64
Chris@148 65 if (m_file) {
Chris@148 66 std::cerr << "CSVFileReader::CSVFileReader: Closing file" << std::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@148 84 Model *
Chris@148 85 CSVFileReader::load() const
Chris@148 86 {
Chris@148 87 if (!m_file) return 0;
Chris@148 88
Chris@148 89 CSVFormatDialog *dialog = new CSVFormatDialog
Chris@148 90 (0, m_file, m_mainModelSampleRate);
Chris@148 91
Chris@148 92 if (dialog->exec() == QDialog::Rejected) {
Chris@148 93 delete dialog;
Chris@148 94 return 0;
Chris@148 95 }
Chris@148 96
Chris@148 97 CSVFormatDialog::ModelType modelType = dialog->getModelType();
Chris@148 98 CSVFormatDialog::TimingType timingType = dialog->getTimingType();
Chris@148 99 CSVFormatDialog::TimeUnits timeUnits = dialog->getTimeUnits();
Chris@148 100 QString separator = dialog->getSeparator();
Chris@148 101 size_t sampleRate = dialog->getSampleRate();
Chris@148 102 size_t windowSize = dialog->getWindowSize();
Chris@148 103
Chris@148 104 delete dialog;
Chris@148 105
Chris@148 106 if (timingType == CSVFormatDialog::ExplicitTiming) {
Chris@148 107 windowSize = 1;
Chris@148 108 if (timeUnits == CSVFormatDialog::TimeSeconds) {
Chris@148 109 sampleRate = m_mainModelSampleRate;
Chris@148 110 }
Chris@148 111 }
Chris@148 112
Chris@148 113 SparseOneDimensionalModel *model1 = 0;
Chris@148 114 SparseTimeValueModel *model2 = 0;
Chris@152 115 EditableDenseThreeDimensionalModel *model3 = 0;
Chris@148 116 Model *model = 0;
Chris@148 117
Chris@148 118 QTextStream in(m_file);
Chris@148 119 in.seek(0);
Chris@148 120
Chris@148 121 unsigned int warnings = 0, warnLimit = 10;
Chris@148 122 unsigned int lineno = 0;
Chris@148 123
Chris@148 124 float min = 0.0, max = 0.0;
Chris@148 125
Chris@148 126 size_t frameNo = 0;
Chris@148 127
Chris@148 128 while (!in.atEnd()) {
Chris@148 129
Chris@148 130 QString line = in.readLine().trimmed();
Chris@148 131 if (line.startsWith("#")) continue;
Chris@148 132
Chris@148 133 QStringList list = line.split(separator);
Chris@148 134
Chris@148 135 if (!model) {
Chris@148 136
Chris@148 137 switch (modelType) {
Chris@148 138
Chris@148 139 case CSVFormatDialog::OneDimensionalModel:
Chris@148 140 model1 = new SparseOneDimensionalModel(sampleRate, windowSize);
Chris@148 141 model = model1;
Chris@148 142 break;
Chris@148 143
Chris@148 144 case CSVFormatDialog::TwoDimensionalModel:
Chris@148 145 model2 = new SparseTimeValueModel(sampleRate, windowSize,
Chris@148 146 0.0, 0.0,
Chris@148 147 false);
Chris@148 148 model = model2;
Chris@148 149 break;
Chris@148 150
Chris@148 151 case CSVFormatDialog::ThreeDimensionalModel:
Chris@152 152 model3 = new EditableDenseThreeDimensionalModel(sampleRate,
Chris@152 153 windowSize,
Chris@152 154 list.size());
Chris@148 155 model = model3;
Chris@148 156 break;
Chris@148 157 }
Chris@148 158 }
Chris@148 159
Chris@148 160 QStringList tidyList;
Chris@148 161 QRegExp nonNumericRx("[^0-9.,+-]");
Chris@148 162
Chris@148 163 for (int i = 0; i < list.size(); ++i) {
Chris@148 164
Chris@148 165 QString s(list[i].trimmed());
Chris@148 166
Chris@148 167 if (s.length() >= 2 && s.startsWith("\"") && s.endsWith("\"")) {
Chris@148 168 s = s.mid(1, s.length() - 2);
Chris@148 169 } else if (s.length() >= 2 && s.startsWith("'") && s.endsWith("'")) {
Chris@148 170 s = s.mid(1, s.length() - 2);
Chris@148 171 }
Chris@148 172
Chris@148 173 if (i == 0 && timingType == CSVFormatDialog::ExplicitTiming) {
Chris@148 174
Chris@148 175 bool ok = false;
Chris@148 176 QString numeric = s;
Chris@148 177 numeric.remove(nonNumericRx);
Chris@148 178
Chris@148 179 if (timeUnits == CSVFormatDialog::TimeSeconds) {
Chris@148 180
Chris@148 181 double time = numeric.toDouble(&ok);
Chris@148 182 frameNo = int(time * sampleRate + 0.00001);
Chris@148 183
Chris@148 184 } else {
Chris@148 185
Chris@148 186 frameNo = numeric.toInt(&ok);
Chris@148 187
Chris@148 188 if (timeUnits == CSVFormatDialog::TimeWindows) {
Chris@148 189 frameNo *= windowSize;
Chris@148 190 }
Chris@148 191 }
Chris@148 192
Chris@148 193 if (!ok) {
Chris@148 194 if (warnings < warnLimit) {
Chris@148 195 std::cerr << "WARNING: CSVFileReader::load: "
Chris@148 196 << "Bad time format (\"" << s.toStdString()
Chris@148 197 << "\") in data line "
Chris@148 198 << lineno << ":" << std::endl;
Chris@148 199 std::cerr << line.toStdString() << std::endl;
Chris@148 200 } else if (warnings == warnLimit) {
Chris@148 201 std::cerr << "WARNING: Too many warnings" << std::endl;
Chris@148 202 }
Chris@148 203 ++warnings;
Chris@148 204 }
Chris@148 205 } else {
Chris@148 206 tidyList.push_back(s);
Chris@148 207 }
Chris@148 208 }
Chris@148 209
Chris@148 210 if (modelType == CSVFormatDialog::OneDimensionalModel) {
Chris@148 211
Chris@148 212 SparseOneDimensionalModel::Point point
Chris@148 213 (frameNo,
Chris@148 214 tidyList.size() > 0 ? tidyList[tidyList.size()-1] :
Chris@148 215 QString("%1").arg(lineno));
Chris@148 216
Chris@148 217 model1->addPoint(point);
Chris@148 218
Chris@148 219 } else if (modelType == CSVFormatDialog::TwoDimensionalModel) {
Chris@148 220
Chris@148 221 SparseTimeValueModel::Point point
Chris@148 222 (frameNo,
Chris@148 223 tidyList.size() > 0 ? tidyList[0].toFloat() : 0.0,
Chris@148 224 tidyList.size() > 1 ? tidyList[1] : QString("%1").arg(lineno));
Chris@148 225
Chris@148 226 model2->addPoint(point);
Chris@148 227
Chris@148 228 } else if (modelType == CSVFormatDialog::ThreeDimensionalModel) {
Chris@148 229
Chris@182 230 DenseThreeDimensionalModel::Column values;
Chris@148 231
Chris@148 232 for (int i = 0; i < tidyList.size(); ++i) {
Chris@148 233
Chris@148 234 bool ok = false;
Chris@148 235 float value = list[i].toFloat(&ok);
Chris@148 236 values.push_back(value);
Chris@148 237
Chris@148 238 if ((lineno == 0 && i == 0) || value < min) min = value;
Chris@148 239 if ((lineno == 0 && i == 0) || value > max) max = value;
Chris@148 240
Chris@148 241 if (!ok) {
Chris@148 242 if (warnings < warnLimit) {
Chris@148 243 std::cerr << "WARNING: CSVFileReader::load: "
Chris@148 244 << "Non-numeric value in data line " << lineno
Chris@148 245 << ":" << std::endl;
Chris@148 246 std::cerr << line.toStdString() << std::endl;
Chris@148 247 ++warnings;
Chris@148 248 } else if (warnings == warnLimit) {
Chris@148 249 std::cerr << "WARNING: Too many warnings" << std::endl;
Chris@148 250 }
Chris@148 251 }
Chris@148 252 }
Chris@148 253
Chris@148 254 std::cerr << "Setting bin values for count " << lineno << ", frame "
Chris@148 255 << frameNo << ", time " << RealTime::frame2RealTime(frameNo, sampleRate) << std::endl;
Chris@148 256
Chris@182 257 model3->setColumn(frameNo / model3->getResolution(), values);
Chris@148 258 }
Chris@148 259
Chris@148 260 ++lineno;
Chris@148 261 if (timingType == CSVFormatDialog::ImplicitTiming ||
Chris@148 262 list.size() == 0) {
Chris@148 263 frameNo += windowSize;
Chris@148 264 }
Chris@148 265 }
Chris@148 266
Chris@148 267 if (modelType == CSVFormatDialog::ThreeDimensionalModel) {
Chris@148 268 model3->setMinimumLevel(min);
Chris@148 269 model3->setMaximumLevel(max);
Chris@148 270 }
Chris@148 271
Chris@148 272 return model;
Chris@148 273 }
Chris@148 274
Chris@148 275
Chris@148 276 CSVFormatDialog::CSVFormatDialog(QWidget *parent, QFile *file,
Chris@148 277 size_t defaultSampleRate) :
Chris@148 278 QDialog(parent),
Chris@148 279 m_modelType(OneDimensionalModel),
Chris@148 280 m_timingType(ExplicitTiming),
Chris@148 281 m_timeUnits(TimeAudioFrames),
Chris@148 282 m_separator("")
Chris@148 283 {
Chris@148 284 setModal(true);
Chris@148 285 setWindowTitle(tr("Select Data Format"));
Chris@148 286
Chris@148 287 (void)guessFormat(file);
Chris@148 288
Chris@148 289 QGridLayout *layout = new QGridLayout;
Chris@148 290
Chris@148 291 layout->addWidget(new QLabel(tr("\nPlease select the correct data format for this file.\n")),
Chris@148 292 0, 0, 1, 4);
Chris@148 293
Chris@148 294 layout->addWidget(new QLabel(tr("Each row specifies:")), 1, 0);
Chris@148 295
Chris@148 296 m_modelTypeCombo = new QComboBox;
Chris@148 297 m_modelTypeCombo->addItem(tr("A point in time"));
Chris@148 298 m_modelTypeCombo->addItem(tr("A value at a time"));
Chris@148 299 m_modelTypeCombo->addItem(tr("A set of values"));
Chris@148 300 layout->addWidget(m_modelTypeCombo, 1, 1, 1, 2);
Chris@148 301 connect(m_modelTypeCombo, SIGNAL(activated(int)),
Chris@148 302 this, SLOT(modelTypeChanged(int)));
Chris@148 303 m_modelTypeCombo->setCurrentIndex(int(m_modelType));
Chris@148 304
Chris@148 305 layout->addWidget(new QLabel(tr("The first column contains:")), 2, 0);
Chris@148 306
Chris@148 307 m_timingTypeCombo = new QComboBox;
Chris@148 308 m_timingTypeCombo->addItem(tr("Time, in seconds"));
Chris@148 309 m_timingTypeCombo->addItem(tr("Time, in audio sample frames"));
Chris@148 310 m_timingTypeCombo->addItem(tr("Data (rows are consecutive in time)"));
Chris@148 311 layout->addWidget(m_timingTypeCombo, 2, 1, 1, 2);
Chris@148 312 connect(m_timingTypeCombo, SIGNAL(activated(int)),
Chris@148 313 this, SLOT(timingTypeChanged(int)));
Chris@148 314 m_timingTypeCombo->setCurrentIndex(m_timingType == ExplicitTiming ?
Chris@148 315 m_timeUnits == TimeSeconds ? 0 : 1 : 2);
Chris@148 316
Chris@148 317 m_sampleRateLabel = new QLabel(tr("Audio sample rate (Hz):"));
Chris@148 318 layout->addWidget(m_sampleRateLabel, 3, 0);
Chris@148 319
Chris@148 320 size_t sampleRates[] = {
Chris@148 321 8000, 11025, 12000, 22050, 24000, 32000,
Chris@148 322 44100, 48000, 88200, 96000, 176400, 192000
Chris@148 323 };
Chris@148 324
Chris@148 325 m_sampleRateCombo = new QComboBox;
Chris@148 326 m_sampleRate = defaultSampleRate;
Chris@148 327 for (size_t i = 0; i < sizeof(sampleRates) / sizeof(sampleRates[0]); ++i) {
Chris@148 328 m_sampleRateCombo->addItem(QString("%1").arg(sampleRates[i]));
Chris@148 329 if (sampleRates[i] == m_sampleRate) m_sampleRateCombo->setCurrentIndex(i);
Chris@148 330 }
Chris@148 331 m_sampleRateCombo->setEditable(true);
Chris@148 332
Chris@148 333 layout->addWidget(m_sampleRateCombo, 3, 1);
Chris@148 334 connect(m_sampleRateCombo, SIGNAL(activated(QString)),
Chris@148 335 this, SLOT(sampleRateChanged(QString)));
Chris@148 336 connect(m_sampleRateCombo, SIGNAL(editTextChanged(QString)),
Chris@148 337 this, SLOT(sampleRateChanged(QString)));
Chris@148 338
Chris@148 339 m_windowSizeLabel = new QLabel(tr("Frame increment between rows:"));
Chris@148 340 layout->addWidget(m_windowSizeLabel, 4, 0);
Chris@148 341
Chris@148 342 m_windowSizeCombo = new QComboBox;
Chris@148 343 m_windowSize = 1024;
Chris@148 344 for (int i = 0; i <= 16; ++i) {
Chris@148 345 int value = 1 << i;
Chris@148 346 m_windowSizeCombo->addItem(QString("%1").arg(value));
Chris@148 347 if (value == m_windowSize) m_windowSizeCombo->setCurrentIndex(i);
Chris@148 348 }
Chris@148 349 m_windowSizeCombo->setEditable(true);
Chris@148 350
Chris@148 351 layout->addWidget(m_windowSizeCombo, 4, 1);
Chris@148 352 connect(m_windowSizeCombo, SIGNAL(activated(QString)),
Chris@148 353 this, SLOT(windowSizeChanged(QString)));
Chris@148 354 connect(m_windowSizeCombo, SIGNAL(editTextChanged(QString)),
Chris@148 355 this, SLOT(windowSizeChanged(QString)));
Chris@148 356
Chris@148 357 layout->addWidget(new QLabel(tr("\nExample data from file:")), 5, 0, 1, 4);
Chris@148 358
Chris@148 359 m_exampleWidget = new QTableWidget
Chris@148 360 (std::min(10, m_example.size()), m_maxExampleCols);
Chris@148 361
Chris@148 362 layout->addWidget(m_exampleWidget, 6, 0, 1, 4);
Chris@148 363 layout->setColumnStretch(3, 10);
Chris@148 364 layout->setRowStretch(4, 10);
Chris@148 365
Chris@148 366 QPushButton *ok = new QPushButton(tr("OK"));
Chris@148 367 connect(ok, SIGNAL(clicked()), this, SLOT(accept()));
Chris@148 368 ok->setDefault(true);
Chris@148 369
Chris@148 370 QPushButton *cancel = new QPushButton(tr("Cancel"));
Chris@148 371 connect(cancel, SIGNAL(clicked()), this, SLOT(reject()));
Chris@148 372
Chris@148 373 QHBoxLayout *buttonLayout = new QHBoxLayout;
Chris@148 374 buttonLayout->addStretch(1);
Chris@148 375 buttonLayout->addWidget(ok);
Chris@148 376 buttonLayout->addWidget(cancel);
Chris@148 377
Chris@148 378 QVBoxLayout *mainLayout = new QVBoxLayout;
Chris@148 379 mainLayout->addLayout(layout);
Chris@148 380 mainLayout->addLayout(buttonLayout);
Chris@148 381
Chris@148 382 setLayout(mainLayout);
Chris@148 383
Chris@148 384 timingTypeChanged(m_timingTypeCombo->currentIndex());
Chris@148 385 }
Chris@148 386
Chris@148 387 CSVFormatDialog::~CSVFormatDialog()
Chris@148 388 {
Chris@148 389 }
Chris@148 390
Chris@148 391 void
Chris@148 392 CSVFormatDialog::populateExample()
Chris@148 393 {
Chris@148 394 m_exampleWidget->setColumnCount
Chris@148 395 (m_timingType == ExplicitTiming ?
Chris@148 396 m_maxExampleCols - 1 : m_maxExampleCols);
Chris@148 397
Chris@148 398 m_exampleWidget->setHorizontalHeaderLabels(QStringList());
Chris@148 399
Chris@148 400 for (int i = 0; i < m_example.size(); ++i) {
Chris@148 401 for (int j = 0; j < m_example[i].size(); ++j) {
Chris@148 402
Chris@148 403 QTableWidgetItem *item = new QTableWidgetItem(m_example[i][j]);
Chris@148 404
Chris@148 405 if (j == 0) {
Chris@148 406 if (m_timingType == ExplicitTiming) {
Chris@148 407 m_exampleWidget->setVerticalHeaderItem(i, item);
Chris@148 408 continue;
Chris@148 409 } else {
Chris@148 410 QTableWidgetItem *header =
Chris@148 411 new QTableWidgetItem(QString("%1").arg(i));
Chris@148 412 header->setFlags(Qt::ItemIsEnabled);
Chris@148 413 m_exampleWidget->setVerticalHeaderItem(i, header);
Chris@148 414 }
Chris@148 415 }
Chris@148 416 int index = j;
Chris@148 417 if (m_timingType == ExplicitTiming) --index;
Chris@148 418 item->setFlags(Qt::ItemIsEnabled);
Chris@148 419 m_exampleWidget->setItem(i, index, item);
Chris@148 420 }
Chris@148 421 }
Chris@148 422 }
Chris@148 423
Chris@148 424 void
Chris@148 425 CSVFormatDialog::modelTypeChanged(int type)
Chris@148 426 {
Chris@148 427 m_modelType = (ModelType)type;
Chris@148 428
Chris@148 429 if (m_modelType == ThreeDimensionalModel) {
Chris@148 430 // We can't load 3d models with explicit timing, because the 3d
Chris@148 431 // model is dense so we need a fixed sample increment
Chris@148 432 m_timingTypeCombo->setCurrentIndex(2);
Chris@148 433 timingTypeChanged(2);
Chris@148 434 }
Chris@148 435 }
Chris@148 436
Chris@148 437 void
Chris@148 438 CSVFormatDialog::timingTypeChanged(int type)
Chris@148 439 {
Chris@148 440 switch (type) {
Chris@148 441
Chris@148 442 case 0:
Chris@148 443 m_timingType = ExplicitTiming;
Chris@148 444 m_timeUnits = TimeSeconds;
Chris@148 445 m_sampleRateCombo->setEnabled(false);
Chris@148 446 m_sampleRateLabel->setEnabled(false);
Chris@148 447 m_windowSizeCombo->setEnabled(false);
Chris@148 448 m_windowSizeLabel->setEnabled(false);
Chris@148 449 if (m_modelType == ThreeDimensionalModel) {
Chris@148 450 m_modelTypeCombo->setCurrentIndex(1);
Chris@148 451 modelTypeChanged(1);
Chris@148 452 }
Chris@148 453 break;
Chris@148 454
Chris@148 455 case 1:
Chris@148 456 m_timingType = ExplicitTiming;
Chris@148 457 m_timeUnits = TimeAudioFrames;
Chris@148 458 m_sampleRateCombo->setEnabled(true);
Chris@148 459 m_sampleRateLabel->setEnabled(true);
Chris@148 460 m_windowSizeCombo->setEnabled(false);
Chris@148 461 m_windowSizeLabel->setEnabled(false);
Chris@148 462 if (m_modelType == ThreeDimensionalModel) {
Chris@148 463 m_modelTypeCombo->setCurrentIndex(1);
Chris@148 464 modelTypeChanged(1);
Chris@148 465 }
Chris@148 466 break;
Chris@148 467
Chris@148 468 case 2:
Chris@148 469 m_timingType = ImplicitTiming;
Chris@148 470 m_timeUnits = TimeWindows;
Chris@148 471 m_sampleRateCombo->setEnabled(true);
Chris@148 472 m_sampleRateLabel->setEnabled(true);
Chris@148 473 m_windowSizeCombo->setEnabled(true);
Chris@148 474 m_windowSizeLabel->setEnabled(true);
Chris@148 475 break;
Chris@148 476 }
Chris@148 477
Chris@148 478 populateExample();
Chris@148 479 }
Chris@148 480
Chris@148 481 void
Chris@148 482 CSVFormatDialog::sampleRateChanged(QString rateString)
Chris@148 483 {
Chris@148 484 bool ok = false;
Chris@148 485 int sampleRate = rateString.toInt(&ok);
Chris@148 486 if (ok) m_sampleRate = sampleRate;
Chris@148 487 }
Chris@148 488
Chris@148 489 void
Chris@148 490 CSVFormatDialog::windowSizeChanged(QString sizeString)
Chris@148 491 {
Chris@148 492 bool ok = false;
Chris@148 493 int size = sizeString.toInt(&ok);
Chris@148 494 if (ok) m_windowSize = size;
Chris@148 495 }
Chris@148 496
Chris@148 497 bool
Chris@148 498 CSVFormatDialog::guessFormat(QFile *file)
Chris@148 499 {
Chris@148 500 QTextStream in(file);
Chris@148 501 in.seek(0);
Chris@148 502
Chris@148 503 unsigned int lineno = 0;
Chris@148 504
Chris@148 505 bool nonIncreasingPrimaries = false;
Chris@148 506 bool nonNumericPrimaries = false;
Chris@148 507 bool floatPrimaries = false;
Chris@148 508 bool variableItemCount = false;
Chris@148 509 int itemCount = 1;
Chris@148 510 int earliestNonNumericItem = -1;
Chris@148 511
Chris@148 512 float prevPrimary = 0.0;
Chris@148 513
Chris@148 514 m_maxExampleCols = 0;
Chris@148 515
Chris@148 516 while (!in.atEnd()) {
Chris@148 517
Chris@148 518 QString line = in.readLine().trimmed();
Chris@148 519 if (line.startsWith("#")) continue;
Chris@148 520
Chris@148 521 if (m_separator == "") {
Chris@148 522 //!!! to do: ask the user
Chris@148 523 if (line.split(",").size() >= 2) m_separator = ",";
Chris@148 524 else if (line.split("\t").size() >= 2) m_separator = "\t";
Chris@148 525 else if (line.split("|").size() >= 2) m_separator = "|";
Chris@148 526 else if (line.split("/").size() >= 2) m_separator = "/";
Chris@148 527 else if (line.split(":").size() >= 2) m_separator = ":";
Chris@148 528 else m_separator = " ";
Chris@148 529 }
Chris@148 530
Chris@148 531 QStringList list = line.split(m_separator);
Chris@148 532 QStringList tidyList;
Chris@148 533
Chris@148 534 for (int i = 0; i < list.size(); ++i) {
Chris@148 535
Chris@148 536 QString s(list[i]);
Chris@148 537 bool numeric = false;
Chris@148 538
Chris@148 539 if (s.length() >= 2 && s.startsWith("\"") && s.endsWith("\"")) {
Chris@148 540 s = s.mid(1, s.length() - 2);
Chris@148 541 } else if (s.length() >= 2 && s.startsWith("'") && s.endsWith("'")) {
Chris@148 542 s = s.mid(1, s.length() - 2);
Chris@148 543 } else {
Chris@148 544 (void)s.toFloat(&numeric);
Chris@148 545 }
Chris@148 546
Chris@148 547 tidyList.push_back(s);
Chris@148 548
Chris@148 549 if (lineno == 0 || (list.size() < itemCount)) {
Chris@148 550 itemCount = list.size();
Chris@148 551 } else {
Chris@148 552 if (itemCount != list.size()) {
Chris@148 553 variableItemCount = true;
Chris@148 554 }
Chris@148 555 }
Chris@148 556
Chris@148 557 if (i == 0) { // primary
Chris@148 558
Chris@148 559 if (numeric) {
Chris@148 560
Chris@148 561 float primary = s.toFloat();
Chris@148 562
Chris@148 563 if (lineno > 0 && primary <= prevPrimary) {
Chris@148 564 nonIncreasingPrimaries = true;
Chris@148 565 }
Chris@148 566
Chris@148 567 if (s.contains(".") || s.contains(",")) {
Chris@148 568 floatPrimaries = true;
Chris@148 569 }
Chris@148 570
Chris@148 571 prevPrimary = primary;
Chris@148 572
Chris@148 573 } else {
Chris@148 574 nonNumericPrimaries = true;
Chris@148 575 }
Chris@148 576 } else { // secondary
Chris@148 577
Chris@148 578 if (!numeric) {
Chris@148 579 if (earliestNonNumericItem < 0 ||
Chris@148 580 i < earliestNonNumericItem) {
Chris@148 581 earliestNonNumericItem = i;
Chris@148 582 }
Chris@148 583 }
Chris@148 584 }
Chris@148 585 }
Chris@148 586
Chris@148 587 if (lineno < 10) {
Chris@148 588 m_example.push_back(tidyList);
Chris@148 589 if (lineno == 0 || tidyList.size() > m_maxExampleCols) {
Chris@148 590 m_maxExampleCols = tidyList.size();
Chris@148 591 }
Chris@148 592 }
Chris@148 593
Chris@148 594 ++lineno;
Chris@148 595
Chris@148 596 if (lineno == 50) break;
Chris@148 597 }
Chris@148 598
Chris@148 599 if (nonNumericPrimaries || nonIncreasingPrimaries) {
Chris@148 600
Chris@148 601 // Primaries are probably not a series of times
Chris@148 602
Chris@148 603 m_timingType = ImplicitTiming;
Chris@148 604 m_timeUnits = TimeWindows;
Chris@148 605
Chris@148 606 if (nonNumericPrimaries) {
Chris@148 607 m_modelType = OneDimensionalModel;
Chris@148 608 } else if (itemCount == 1 || variableItemCount ||
Chris@148 609 (earliestNonNumericItem != -1)) {
Chris@148 610 m_modelType = TwoDimensionalModel;
Chris@148 611 } else {
Chris@148 612 m_modelType = ThreeDimensionalModel;
Chris@148 613 }
Chris@148 614
Chris@148 615 } else {
Chris@148 616
Chris@148 617 // Increasing numeric primaries -- likely to be time
Chris@148 618
Chris@148 619 m_timingType = ExplicitTiming;
Chris@148 620
Chris@148 621 if (floatPrimaries) {
Chris@148 622 m_timeUnits = TimeSeconds;
Chris@148 623 } else {
Chris@148 624 m_timeUnits = TimeAudioFrames;
Chris@148 625 }
Chris@148 626
Chris@148 627 if (itemCount == 1) {
Chris@148 628 m_modelType = OneDimensionalModel;
Chris@148 629 } else if (variableItemCount || (earliestNonNumericItem != -1)) {
Chris@148 630 if (earliestNonNumericItem != -1 && earliestNonNumericItem < 2) {
Chris@148 631 m_modelType = OneDimensionalModel;
Chris@148 632 } else {
Chris@148 633 m_modelType = TwoDimensionalModel;
Chris@148 634 }
Chris@148 635 } else {
Chris@148 636 m_modelType = ThreeDimensionalModel;
Chris@148 637 }
Chris@148 638 }
Chris@148 639
Chris@148 640 std::cerr << "Estimated model type: " << m_modelType << std::endl;
Chris@148 641 std::cerr << "Estimated timing type: " << m_timingType << std::endl;
Chris@148 642 std::cerr << "Estimated units: " << m_timeUnits << std::endl;
Chris@148 643
Chris@148 644 in.seek(0);
Chris@148 645 return true;
Chris@148 646 }