D-Net

package eu.dnetlib.pace.distance;

import static java.lang.String.format;

import org.apache.commons.lang.StringUtils;

import org.junit.Before;

import org.junit.Test;

import com.google.common.collect.Lists;

import com.wcohen.ss.JaroWinkler;

/**

 * The Class DistanceAlgoTest.

 */

public class DistanceAlgoTest {

	/** The Constant alpha. */

	private static final String alpha = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789";

	/** The Constant aliases_from. */

	private static final String aliases_from = "⁰¹²³⁴⁵⁶⁷⁸⁹⁺⁻⁼⁽⁾ⁿ₀₁₂₃₄₅₆₇₈₉₊₋₌₍₎";

	/** The Constant aliases_to. */

	private static final String aliases_to = "0123456789+-=()n0123456789+-=()";

	/** The Constant TEST_STRING. */

	private final static String TEST_STRING = "Toshiba NB550D: netbook su piattaforma AMD Fusion⁽¹²⁾.";

	/**

	 * Setup.

	 */

	@Before

	public void setup() {

		System.out.println("****************************************************************");

		System.out.println("Test String    : " + TEST_STRING);

	}

	/**

	 * Test get numbers.

	 */

	@Test

	public void testGetNumbers() {

		System.out.println("Numbers        : " + getNumbers(TEST_STRING));

	}

	/**

	 * Test remove symbols.

	 */

	@Test

	public void testRemoveSymbols() {

		System.out.println("Without symbols: " + removeSymbols(TEST_STRING));

	}

	/**

	 * Test fix aliases.

	 */

	@Test

	public void testFixAliases() {

		System.out.println("Fixed aliases  : " + fixAliases(TEST_STRING));

	}

	/**

	 * Test jaro winkler.

	 */

	@Test

	public void testJaroWinkler() {

		JaroWinkler algo = new JaroWinkler();

		double d = algo.score("Artin Bardi Manghi", "Artini Bardi Manghi");

		System.out.println(format("d -> %s", d));

	}

	/**

	 * Gets the numbers.

	 * 

	 * @param s

	 *            the s

	 * @return the numbers

	 */

	private String getNumbers(final String s) {

		return s.replaceAll("\\D", "");

	}

	/**

	 * Removes the symbols.

	 * 

	 * @param s

	 *            the s

	 * @return the string

	 */

	private String removeSymbols(final String s) {

		final StringBuilder sb = new StringBuilder();

		for (char ch : Lists.charactersOf(s)) {

			sb.append(StringUtils.contains(alpha, ch) ? ch : " ");

		}

		return sb.toString().replaceAll("\\s+", " ");

	}

	/**

	 * Fix aliases.

	 * 

	 * @param s

	 *            the s

	 * @return the string

	 */

	private String fixAliases(final String s) {

		final StringBuilder sb = new StringBuilder();

		for (char ch : Lists.charactersOf(s)) {

			int i = StringUtils.indexOf(aliases_from, ch);

			sb.append(i >= 0 ? aliases_to.charAt(i) : ch);

		}

		return sb.toString();

	}

}