public void loadDataset(String fileName, int maxCount) {
// Clearing the database
if (database == null) {
database = new SequenceDatabase();
} else {
database.clear();
}
// Tries to guess the format if it is a predefined dataset
try {
Format datasetFormat = Format.valueOf(fileName);
loadPredefinedDataset(datasetFormat, maxCount);
} catch (IllegalArgumentException e) {
loadCustomDataset(fileName, maxCount);
}
// Shuffling the database
Collections.shuffle(database.getSequences());
}
private void loadCustomDataset(String fileName, int maxCount) {
try {
database.loadFileCustomFormat(
fileToPath(fileName),
maxCount,
Profile.paramInt("sequenceMinSize"),
Profile.paramInt("sequenceMaxSize"));
} catch (IOException e) {
System.out.println("Could not load dataset, IOExeption");
e.printStackTrace();
}
}
/** Loads a predefined dataset -- see full list in DatabaseHelper.Format */
private void loadPredefinedDataset(Format format, int maxCount) {
// Loading the specified dataset (according to the format)
try {
switch (format) {
case BMS:
database.loadFileBMSFormat(
fileToPath("BMS.dat"),
maxCount,
Profile.paramInt("sequenceMinSize"),
Profile.paramInt("sequenceMaxSize"));
break;
case KOSARAK:
database.loadFileCustomFormat(
fileToPath("kosarak.dat"),
maxCount,
Profile.paramInt("sequenceMinSize"),
Profile.paramInt("sequenceMaxSize"));
break;
case FIFA:
database.loadFileFIFAFormat(
fileToPath("FIFA_large.dat"),
maxCount,
Profile.paramInt("sequenceMinSize"),
Profile.paramInt("sequenceMaxSize"));
break;
case MSNBC:
database.loadFileMsnbsFormat(
fileToPath("msnbc.seq"),
maxCount,
Profile.paramInt("sequenceMinSize"),
Profile.paramInt("sequenceMaxSize"));
break;
case SIGN:
database.loadFileSignLanguage(
fileToPath("sign_language.txt"),
maxCount,
Profile.paramInt("sequenceMinSize"),
Profile.paramInt("sequenceMaxSize"));
break;
case CANADARM1:
database.loadFileSPMFFormat(
fileToPath("Canadarm1_actions.txt"),
maxCount,
Profile.paramInt("sequenceMinSize"),
Profile.paramInt("sequenceMaxSize"));
break;
case CANADARM2:
database.loadFileSPMFFormat(
fileToPath("Canadarm2_states.txt"),
maxCount,
Profile.paramInt("sequenceMinSize"),
Profile.paramInt("sequenceMaxSize"));
break;
case SNAKE:
database.loadSnakeDataset(
fileToPath("snake.dat"),
maxCount,
Profile.paramInt("sequenceMinSize"),
Profile.paramInt("sequenceMaxSize"));
break;
case BIBLE_CHAR:
database.loadFileLargeTextFormatAsCharacter(
fileToPath("Bible.txt"),
maxCount,
Profile.paramInt("sequenceMinSize"),
Profile.paramInt("sequenceMaxSize"));
break;
case BIBLE_WORD:
database.loadFileLargeTextFormatAsWords(
fileToPath("Bible.txt"),
maxCount,
Profile.paramInt("sequenceMinSize"),
Profile.paramInt("sequenceMaxSize"),
true);
break;
case KORAN_WORD:
database.loadFileLargeTextFormatAsWords(
fileToPath("koran.txt"),
maxCount,
Profile.paramInt("sequenceMinSize"),
Profile.paramInt("sequenceMaxSize"),
false);
break;
case LEVIATHAN_WORD:
database.loadFileLargeTextFormatAsWords(
fileToPath("leviathan.txt"),
maxCount,
Profile.paramInt("sequenceMinSize"),
Profile.paramInt("sequenceMaxSize"),
false);
break;
default:
System.out.println("Could not load dataset, unknown format.");
}
} catch (IOException e) {
System.out.println("Could not load dataset, IOExeption");
e.printStackTrace();
}
}
