/**
* Reads schema from the KEEL file
*
* @param jobFilename Name of the KEEL dataset file
*/
private static byte[] readSchema(String fileName) throws IOException, DatasetException {
KeelDataSet dataset = new KeelDataSet(fileName);
dataset.open();
File file = new File(fileName);
List<String> inputIds = new ArrayList<String>();
List<String> outputIds = new ArrayList<String>();
Reader reader = new BufferedReader(new FileReader(file));
String line = ((BufferedReader) reader).readLine();
line = line.replace("real[", "real [");
line = line.replace("integer[", "integer [");
line = line.replace("{", " {");
StringTokenizer elementLine = new StringTokenizer(line);
String element = elementLine.nextToken();
while (!element.equalsIgnoreCase("@data")) {
if (element.equalsIgnoreCase("@inputs")) {
while (elementLine.hasMoreTokens()) {
StringTokenizer commaTokenizer = new StringTokenizer(elementLine.nextToken(), ",");
while (commaTokenizer.hasMoreTokens()) inputIds.add(commaTokenizer.nextToken());
}
} else if (element.equalsIgnoreCase("@outputs")) {
while (elementLine.hasMoreTokens()) {
StringTokenizer commaTokenizer = new StringTokenizer(elementLine.nextToken(), ",");
while (commaTokenizer.hasMoreTokens()) outputIds.add(commaTokenizer.nextToken());
}
}
// Next line of the file
line = ((BufferedReader) reader).readLine();
while (line.startsWith("%") || line.equalsIgnoreCase(""))
line = ((BufferedReader) reader).readLine();
line = line.replace("real[", "real [");
line = line.replace("integer[", "integer [");
line = line.replace("{", " {");
elementLine = new StringTokenizer(line);
element = elementLine.nextToken();
}
IMetadata metadata = dataset.getMetadata();
byte[] schema = new byte[metadata.numberOfAttributes()];
if (inputIds.isEmpty() || outputIds.isEmpty()) {
for (int i = 0; i < schema.length; i++) {
if (i != (schema.length - 1)) schema[i] = 1;
else {
IAttribute outputAttribute = metadata.getAttribute(i);
schema[i] = 2;
consoleReporter.setOutputAttribute(outputAttribute);
}
}
} else {
for (int i = 0; i < schema.length; i++) {
if (inputIds.contains(metadata.getAttribute(i).getName())) schema[i] = 1;
else if (outputIds.contains(metadata.getAttribute(i).getName())) {
IAttribute outputAttribute = metadata.getAttribute(i);
schema[i] = 2;
consoleReporter.setOutputAttribute(outputAttribute);
} else schema[i] = -1;
}
}
StringBuffer header = new StringBuffer();
header.append("@relation " + dataset.getName() + "\n");
for (int i = 0; i < metadata.numberOfAttributes(); i++) {
IAttribute attribute = metadata.getAttribute(i);
header.append("@attribute " + attribute.getName() + " ");
if (attribute.getType() == AttributeType.Categorical) {
CategoricalAttribute catAtt = (CategoricalAttribute) attribute;
Interval interval = catAtt.intervalValues();
header.append("{");
for (int j = (int) interval.getLeft(); j <= interval.size() + 1; j++) {
header.append(catAtt.show(j) + (j != interval.size() + 1 ? "," : "}\n"));
}
} else if (attribute.getType() == AttributeType.IntegerNumerical) {
IntegerNumericalAttribute intAtt = (IntegerNumericalAttribute) attribute;
header.append(
"integer["
+ (int) intAtt.intervalValues().getLeft()
+ ","
+ (int) intAtt.intervalValues().getRight()
+ "]\n");
} else if (attribute.getType() == AttributeType.DoubleNumerical) {
RealNumericalAttribute doubleAtt = (RealNumericalAttribute) attribute;
header.append(
"real["
+ doubleAtt.intervalValues().getLeft()
+ ","
+ doubleAtt.intervalValues().getRight()
+ "]\n");
}
}
header.append("@data\n");
consoleReporter.setHeader(header.toString());
dataset.close();
return schema;
}
