/***********************************************************************
This file is part of KEEL-software, the Data Mining tool for regression,
classification, clustering, pattern mining and so on.
Copyright (C) 2004-2010
F. Herrera (herrera@decsai.ugr.es)
L. S�nchez (luciano@uniovi.es)
J. Alcal�-Fdez (jalcala@decsai.ugr.es)
S. Garc�a (sglopez@ujaen.es)
A. Fern�ndez (alberto.fernandez@ujaen.es)
J. Luengo (julianlm@decsai.ugr.es)
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see http://www.gnu.org/licenses/
**********************************************************************/
/*
* WekaToKeel.java
*/
package keel.Algorithms.Preprocess.Converter;
import keel.Dataset.*;
import java.io.BufferedReader;
import java.io.BufferedWriter;
import java.io.FileReader;
import java.io.FileWriter;
import java.io.File;
import java.util.*;
import java.util.regex.Matcher;
import java.util.regex.Pattern;
import com.Ostermiller.util.CSVParser;
/**
* <p>
* <b> WekaToKeel </b>
* </p>
*
* Clase extendida de la clase Importer. Esta clase se utiliza
* para leer datos localizados en ficheros con formato Weka
* (fichero relacion-atributo) y convertirlos a formato keel.
*
* @author Teresa Prieto López (UCO)
* @version 1.0
*/
public class WekaToKeel extends Importer {
/* Metodo utilizado para convertir los datos del fichero indicado
* mediante la variable pathnameInput a formato keel en el fichero
* indicado por la ruta pathnameOutput
*
* @param pathnameInput ruta con los datos en formato Weka
* @param pathnameOutput ruta para el fichero de datos Keel.
*
* @throws Exception */
public void Start(String pathnameInput, String pathnameOutput) throws Exception {
BufferedReader reader;
Pattern p;
Matcher m;
File f;
StringTokenizer token;
String line = new String();
String tokenInitial = new String();
String nameAttribute = new String();
String element = new String();
String lineReduced = new String();
String filename = "tempOf";
int actualValueInt;
double actualValue;
double min;
double max;
int i;
int j;
int indexInitial = 0;
int indexSecond = 0;
int type = -1;
File fileInput = new File(pathnameInput);
filename = filename.concat(fileInput.getName());
reader = new BufferedReader(new FileReader(pathnameInput));
BufferedWriter auxFile = new BufferedWriter(new FileWriter(filename));
while ((line = reader.readLine()) != null) {
p = Pattern.compile("\\s*,\\s*");
m = p.matcher(line);
line = m.replaceAll(",");
p = Pattern.compile("^\\s+");
m = p.matcher(line);
line = m.replaceAll("");
p = Pattern.compile("\\s+$");
m = p.matcher(line);
line = m.replaceAll("");
p = Pattern.compile("\\s+");
m = p.matcher(line);
line = m.replaceAll(" ");
auxFile.write(line + "\n");
}
auxFile.close();
reader.close();
reader = new BufferedReader(new FileReader(filename));
/* Contamos el número de atributos que existen*/
line = reader.readLine();
token = new StringTokenizer(line, " ");
while (!line.equalsIgnoreCase("@data")) {
if (line.startsWith("@")) {
tokenInitial = token.nextToken().toLowerCase();
if (tokenInitial.equals("@attribute")) {
numAttributes++;
}
if (tokenInitial.equals("@relation")) {
nameRelation = token.nextToken();
if (nameRelation.startsWith("'")) {
indexInitial = line.indexOf("\'");
indexSecond = line.indexOf("\'", indexInitial + 1);
nameRelation = line.substring(indexInitial, indexSecond + 1);
}
p = Pattern.compile("\\s+");
m = p.matcher(nameRelation);
nameRelation = m.replaceAll("");
}
}
line = reader.readLine();
token = new StringTokenizer(line, " ");
}// end while()
reader.close();
/* Reservamos memoria para guardar la informacion de los atributos*/
attribute = new Attribute[numAttributes];
data = new Vector[numAttributes];
for (i = 0; i < numAttributes; i++) {
attribute[i] = new Attribute();
data[i] = new Vector();
}
// Insertamos la definición de los atributos en Attribute
reader = new BufferedReader(new FileReader(filename));
line = reader.readLine();
i = -1;
while (!(line.equalsIgnoreCase("@data"))) {
if (line.startsWith("@")) {
token = new StringTokenizer(line, " ");
tokenInitial = token.nextToken();
if (tokenInitial.equalsIgnoreCase("@attribute")) {
i++;
nameAttribute = token.nextToken();
if (nameAttribute.startsWith("'")) {
indexInitial = line.indexOf("\'");
indexSecond = line.indexOf("\'", indexInitial + 1) + 1;
nameAttribute = line.substring(indexInitial, indexSecond);
} else {
if (nameAttribute.contains("{")) {
nameAttribute = nameAttribute.substring(0, nameAttribute.indexOf("{"));
}
}
indexSecond = line.indexOf(nameAttribute) + nameAttribute.length();
nameAttribute = nameAttribute.replace("'", "");
p = Pattern.compile("\\s+");
m = p.matcher(nameAttribute);
nameAttribute = m.replaceAll(" ");
if (nameAttribute.contains(" ")) {
StringTokenizer tokenUcfirts = new StringTokenizer(nameAttribute, " ");
String lineUcfirts = "";
if (tokenUcfirts.hasMoreTokens()) {
lineUcfirts = tokenUcfirts.nextToken();
}
while (tokenUcfirts.hasMoreTokens()) {
lineUcfirts = lineUcfirts.concat(UcFirst(tokenUcfirts.nextToken()));
}
nameAttribute = lineUcfirts;
}
attribute[i].setName(nameAttribute);
lineReduced = line.substring(indexSecond + 1, line.length());
p = Pattern.compile("^\\s+");
m = p.matcher(lineReduced);
lineReduced = m.replaceAll("");
p = Pattern.compile("\\s+$");
m = p.matcher(lineReduced);
lineReduced = m.replaceAll("");
String lineReducedLower = lineReduced.toLowerCase();
if (lineReducedLower.startsWith("numeric") || lineReducedLower.startsWith("real")) {
attribute[i].setType(REAL);
} else {
if (lineReducedLower.startsWith("integer")) {
attribute[i].setType(INTEGER);
} else {
if (lineReducedLower.startsWith("string") || lineReducedLower.startsWith("date")) {
attribute[i].setType(NOMINAL);
} else {
attribute[i].setType(NOMINAL);
if (line.contains("{") && line.contains("}")) {
lineReduced = line.substring(line.indexOf("{") + 1, line.indexOf("}"));
p = Pattern.compile("^\\s+");
m = p.matcher(lineReduced);
lineReduced = m.replaceAll("");
p = Pattern.compile("\\s+$");
m = p.matcher(lineReduced);
lineReduced = m.replaceAll("");
if (lineReduced != "") {
StringTokenizer listValues = new StringTokenizer(lineReduced, ",");
while (listValues.hasMoreTokens()) {
element = listValues.nextToken();
element = element.replace("\"", "");
p = Pattern.compile("[^A-ZÑa-zñ0-9_-]+");
m = p.matcher(element);
/**
* Cambio hecho para que los nominales con espacios en blanco se dejen
* con subrayado bajo "_" y sin comillas simples. Se añade la siguiente linea
*/
element = element.replace(" ", "_");
if (m.find() && !element.startsWith("'") && !element.endsWith("'") && !element.equals("?")) /**
* Cambio hecho para que los nominales con espacios en blanco se dejen
* con subrayado bajo "_" y sin comillas simples. Se comenta la siguiente linea
*/
/*
//element="'"+element+"'";
*/ {
if (element.equalsIgnoreCase("<null>")) {
element = "?";
}
}
attribute[i].addNominalValue(element);
}
}
}//end if
}//end else()
}//end else()
}//end else()
type = attribute[i].getType();
if (type == REAL || type == INTEGER) {
if (line.contains("[") && line.contains("]")) {
lineReduced = line.substring(line.indexOf("[") + 1, line.indexOf("]"));
p = Pattern.compile("^\\s+");
m = p.matcher(lineReduced);
lineReduced = m.replaceAll("");
p = Pattern.compile("\\s+$");
m = p.matcher(lineReduced);
lineReduced = m.replaceAll("");
if (lineReduced != "") {
StringTokenizer range = new StringTokenizer(lineReduced, ",");
if (type == REAL) {
attribute[i].setBounds(Double.valueOf(range.nextToken()), Double.valueOf(range.nextToken()));
}
if (type == INTEGER) {
attribute[i].setBounds(Integer.valueOf(range.nextToken()), Integer.valueOf(range.nextToken()));
}
}
}
}
}//end if
}//end if()*/
line = reader.readLine();
}//end while()
/* Almacenamos los datos en un fichero temporal para luego poder
ser parseado con CSVParser por ','
*/
BufferedWriter writer = new BufferedWriter(new FileWriter("temp"));
while ((line = reader.readLine()) != null) {
// Saltamos las líneas comentadas
if (!line.startsWith("%") && !line.equals("\n") && !line.equals("\r") && !line.equals("")) {
writer.write(line + "\n");
}
}
writer.close();
reader.close();
FileReader filereader = new FileReader("temp");
String[][] values = CSVParser.parse(filereader, ',');
filereader.close();
for (i = 0; i < values.length; i++) {
for (j = 0; j < numAttributes; j++) {
element = values[i][j];
type = attribute[j].getType();
p = Pattern.compile("^\\s+");
m = p.matcher(element);
element = m.replaceAll("");
p = Pattern.compile("\\s+$");
m = p.matcher(element);
element = m.replaceAll("");
element = element.replace("\"", "");
if (type == NOMINAL) {
p = Pattern.compile("[^A-ZÑa-zñ0-9_-]+");
m = p.matcher(element);
/**
* Cambio hecho para que los nominales con espacios en blanco se dejen
* con subrayado bajo "_" y sin comillas simples. Se añade la siguiente linea
*/
element = element.replace(" ", "_");
if (m.find() && !element.startsWith("'") && !element.endsWith("'") && !element.equals("?")) {
/**
* Cambio hecho para que los nominales con espacios en blanco se dejen
* con subrayado bajo "_" y sin comillas simples. Se comenta la siguiente linea
*/
/*
//element="'"+element+"'";
*/
}
}
if (element.equalsIgnoreCase("<null>")) {
element = "?";
}
data[j].addElement(element);
}
}
/* Recogemos la lista de valores nominales de los datos, para aquellos atributos que
no hayan definido la lista en la declaración */
for (i = 0; i < data.length; i++) {
type = attribute[i].getType();
if (type == NOMINAL && attribute[i].getNumNominalValues() == 0) {
for (j = 0; j < data[0].size(); j++) {
element = (String) data[i].elementAt(j);
if (!(attribute[i].isNominalValue(element)) && !element.equals("?")) {
attribute[i].addNominalValue(element);
}
}
}
}
/* Leemos el rango de los datos, para aquellos atributos que no lo hayan definido
en la lista en la declaración */
for (i = 0; i < data[0].size(); i++) {
for (j = 0; j < numAttributes; j++) {
type = attribute[j].getType();
if (type == INTEGER) {
element = (String) data[j].elementAt(i);
if (!element.equals("?")) {
actualValueInt = Integer.valueOf(element);
if ((attribute[j].getFixedBounds()) == false) {
attribute[j].setBounds(actualValueInt, actualValueInt);
} else {
min = attribute[j].getMinAttribute();
max = attribute[j].getMaxAttribute();
if (actualValueInt < min) {
attribute[j].setBounds(actualValueInt, max);
}
if (actualValueInt > max) {
attribute[j].setBounds(min, actualValueInt);
}
}
}
}
if (type == REAL) {
element = (String) data[j].elementAt(i);
if (!element.equals("?")) {
actualValue = Double.valueOf(element);
if ((attribute[j].getFixedBounds()) == false) {
attribute[j].setBounds(actualValue, actualValue);
} else {
min = attribute[j].getMinAttribute();
max = attribute[j].getMaxAttribute();
if (actualValue < min) {
attribute[j].setBounds(actualValue, max);
}
if (actualValue > max) {
attribute[j].setBounds(min, actualValue);
}
}
}
}
}// end for
}//end for
f = new File(filename);
f.delete();
f = new File("temp");
f.delete();
super.Save(pathnameOutput);
}//end start()
}//end WekaToKeel()