Laboratoire ICAR » Plateforme TXM

import java.util.HashSet;

import org.txm.utils.BiHashMap;

 * eg. : "Error while computing." => errorWhileComputing

 * @author sjacquot

	public static boolean debug = true;

	public static BiHashMap<String, String> normalize(BiHashMap<String, String> messages)	{

		for (String key : messages.getKeys()) {

			messages.put(normalize(messages.get(key)), messages.get(key));

		}

		return messages;

	/**

	 * 

	 * @param strings

	 * @return

	 */

	public static HashSet<String> normalize(HashSet<String> strings)	{

		for (String str : strings) {

			if(debug)	{

				System.out.println(str);

				System.out.println("   => " + normalize(str));

			}

		}

		return strings;

	}

	/**

	 * 

	 * @param str

	 */

	public static String normalize(String str)	{

		// log

		if(debug)	{

			System.out.println(str);

		}

		str = str.replaceAll("[^a-zA-Z0-9 ]", "").trim();

		// empty string

		if(str.length() == 0)	{

			System.err.println("NormalizeKeys.normalize(): warning: empty string.");

		}

		else	{

			String[] words = str.split(" ");

			str = "";

			for (int i = 0; i < words.length; i++) {

				str += words[i].substring(0, 1).toUpperCase();

				try {

					str += words[i].substring(1);

				}

				catch (Exception e) {

					// nothing to do

				}

			}

			// uncapitalize first letter

			String tmpString = str.substring(0, 1).toLowerCase();

			try {

				tmpString += str.substring(1);

			}

			catch (Exception e) {

				// nothing to do

			}

			str = tmpString;

		}

		// log

		if(debug)	{

			System.out.println("   => " + str);

		}

		return str;

	}

	/**

	 * 

	 * @param args

	 */

		// tests

		BiHashMap<String, String> messages = new BiHashMap<String, String>();

		messages.put("Convert5To6_46", "Error while computing."); 

		messages.put("Convert5To6_46", "** Error: the connexion to the server failed: wrong port format");

		messages.put("Convert5To6_6", "** Error: ''{0}'' corpus directory is not conformant to TXM corpus binary format: corpus skipped.");

		messages.put("Convert5To6_27", "T");

		messages.put("Convert5To6_28", "");

		messages = normalize(messages);

		System.out.println("NormalizeKeys.main()");

//		HashSet<String> strings = new HashSet<String>();

//		strings.add("Error while computing.");

//		strings.add("** Error: the connexion to the server failed: wrong port format");

//		strings.add("** Error: ''{0}'' corpus directory is not conformant to TXM corpus binary format: corpus skipped.");

//		strings.add("T");

//		strings.add("");

//		normalize(strings);