Merge pull request #291 from RavirajBaraiya/MatchTypeEmail

Match type email

client/src/main/java/zingg/client/MatchType.java

@@ -30,6 +30,11 @@ public enum MatchType implements Serializable {
 	 */
 	PINCODE("PINCODE"),
 
+	/**
+	 * an email type which is supposed to look at only the first part of the email and ignore the domain.
+	 */
+	EMAIL("EMAIL"),
+
 	/**
 	 * Long descriptive text, usually more than a couple of words for example
 	 * product descriptions

core/src/main/java/zingg/feature/StringFeature.java

@@ -5,6 +5,7 @@
 import zingg.similarity.function.AJaroWinklerFunction;
 import zingg.similarity.function.AffineGapSimilarityFunction;
 import zingg.similarity.function.CheckBlankOrNullFunction;
+import zingg.similarity.function.EmailMatchTypeFunction;
 import zingg.similarity.function.JaccSimFunction;
 import zingg.similarity.function.JaroWinklerFunction;
 import zingg.similarity.function.NumbersJaccardFunction;
@@ -46,6 +47,9 @@ public void init(FieldDefinition f) {
 		if(f.getMatchType().contains(MatchType.PINCODE)){
 			addSimFunction(new PinCodeMatchTypeFunction());
 		}
+		if(f.getMatchType().contains(MatchType.EMAIL)){
+			addSimFunction(new EmailMatchTypeFunction());
+		}
 		if (f.getMatchType().contains(MatchType.NUMERIC_WITH_UNITS)) {
 			addSimFunction(new ProductCodeFunction());
 		}

core/src/main/java/zingg/similarity/function/AffineGapSimilarityFunction.java

@@ -20,6 +20,11 @@ public AffineGapSimilarityFunction(String s) {
 		super(s);
 		gap = new SAffineGap();
 	}
+
+	@Override
+	public Double call(String first, String second) {
+		return super.call(first,second);
+	}
 
 	@Override
 	 public String getUid() {

core/src/main/java/zingg/similarity/function/EmailMatchTypeFunction.java

@@ -0,0 +1,61 @@
+package zingg.similarity.function;
+
+import org.apache.commons.logging.Log;
+import org.apache.commons.logging.LogFactory;
+import org.apache.spark.ml.util.Identifiable$;
+
+
+public class EmailMatchTypeFunction extends StringSimilarityDistanceFunction {
+
+	public static final Log LOG = LogFactory
+			.getLog(EmailMatchTypeFunction.class);
+	//private static final String String = null;
+
+
+	public EmailMatchTypeFunction() {
+		this("EmailMatchTypeFunction");		
+	}
+
+	public EmailMatchTypeFunction(String s) {
+		super(s);
+		//gap = new SAffineGap();
+	}
+
+	@Override
+	public Double call(String first, String second) {
+		first = first.split("@",0)[0];
+		second = second.split("@",0)[0];
+		return super.call(first,second);
+		/*
+		double score1 = 0.0;
+		double score2 = 0.0;
+		double score = 0.0;
+
+		try {
+			if (first == null || first.equals("")) {
+				score1 = 1.0d;
+			}
+			if (second == null || second.equals("")) {
+				score2 = 1.0d;
+			}
+			if (score1 != 1.0d && score2 != 1.0d) {
+				first = first.split("@",0)[0];
+				second = second.split("@",0)[0];
+				score = first.equalsIgnoreCase(second)? 1.0d : 0.0d;
+			}
+			else {
+				score = 1.0d;
+			}
+			
+			
+		}  catch (Exception e) {
+			e.printStackTrace();
+			LOG.warn("Error processing differences for " + first + "," + second);
+		} finally {
+			if (Double.isNaN(score)) {
+				score = 0.0;
+			}
+			return score;
+			*/
+		}		
+}

core/src/main/java/zingg/similarity/function/StringSimilarityDistanceFunction.java

@@ -1,10 +1,16 @@
 package zingg.similarity.function;
 
+import org.apache.commons.logging.Log;
+import org.apache.commons.logging.LogFactory;
+
 import com.wcohen.ss.*;
 
 public abstract class StringSimilarityDistanceFunction extends StringSimilarityFunction{
 
 	protected AbstractStringDistance gap;
+
+	public static final Log LOG = LogFactory
+			.getLog(StringSimilarityDistanceFunction.class);
 
 	public StringSimilarityDistanceFunction(String name) {
 		super(name);

core/src/main/java/zingg/similarity/function/StringSimilarityFunction.java

@@ -20,8 +20,8 @@ public StringSimilarityFunction(String name) {
 
 	@Override
 	public Double call(String first, String second) {
-		if (first == null || first.trim().length() ==0) return 0d;
-		if (second == null || second.trim().length() ==0) return 0d;
+		if (first == null || first.trim().length() ==0) return 1d;
+		if (second == null || second.trim().length() ==0) return 1d;
 		double score = first.equalsIgnoreCase(second) ? 1d : 0d;
 		return score;		
 	}

core/src/test/java/zingg/similarity/function/TestEmailMatchTypeFunction.java

@@ -0,0 +1,102 @@
+package zingg.similarity.function;
+
+
+import java.util.Arrays;
+
+import org.junit.jupiter.api.Test;
+
+import static org.junit.jupiter.api.Assertions.assertEquals;
+
+public class TestEmailMatchTypeFunction {
+
+
+	@Test
+	public void testFirstEntryNull() {
+		EmailMatchTypeFunction emailMatchFn = new EmailMatchTypeFunction();
+		AffineGapSimilarityFunction expected = new AffineGapSimilarityFunction();
+		String first = null;
+		String second = "xyz321@pqr.co";
+		assertEquals(expected.call(first,second.split("@",0)[0]), emailMatchFn.call(first, second));
+	}
+
+	@Test
+	public void testFirstEntryEmpty() {
+		EmailMatchTypeFunction emailMatchFn = new EmailMatchTypeFunction();
+		AffineGapSimilarityFunction expected = new AffineGapSimilarityFunction();
+		String first = "";
+		String second = "xyz321@pqr.co";
+		assertEquals(expected.call(first.split("@",0)[0],second.split("@",0)[0]), emailMatchFn.call(first, second));
+	}
+
+	@Test
+	public void testSecondEntryNull() {
+		EmailMatchTypeFunction emailMatchFn = new EmailMatchTypeFunction();
+		AffineGapSimilarityFunction expected = new AffineGapSimilarityFunction();
+		String first = "xyz321@pqr.co";
+		String second = null;
+		assertEquals(expected.call(first.split("@",0)[0],second), emailMatchFn.call(first,second));
+	}
+
+	@Test
+	public void testSecondEntryEmpty() {
+		EmailMatchTypeFunction emailMatchFn = new EmailMatchTypeFunction();
+		AffineGapSimilarityFunction expected = new AffineGapSimilarityFunction();
+		String first = "xyz321@pqr.co";
+		String second = "";
+		assertEquals(expected.call(first.split("@",0)[0],second.split("@",0)[0]), emailMatchFn.call(first,second));
+	}
+	@Test
+	public void testBothEmpty() {
+		EmailMatchTypeFunction emailMatchFn = new EmailMatchTypeFunction();
+		AffineGapSimilarityFunction expected = new AffineGapSimilarityFunction();
+		String first = "";
+		String second = "";
+		assertEquals(expected.call(first.split("@",0)[0],second.split("@",0)[0]), emailMatchFn.call(first,second));
+	}
+
+	@Test
+	public void testBothNull() {
+		EmailMatchTypeFunction emailMatchFn = new EmailMatchTypeFunction();
+		//AffineGapSimilarityFunction expected = new AffineGapSimilarityFunction();
+		String first = null;
+		String second = null;
+		assertEquals(1d, emailMatchFn.call(first,second));
+	}
+
+	@Test
+	public void testBothExact() {
+		EmailMatchTypeFunction emailMatchFn = new EmailMatchTypeFunction();
+		//AffineGapSimilarityFunction expected = new AffineGapSimilarityFunction();
+		String first = "xyz321@pqr.co";
+		String second = "xyz321@pqr.co";
+		assertEquals(1d, emailMatchFn.call(first,second));
+	}
+
+	@Test
+	public void testbothDifferent() {
+		EmailMatchTypeFunction emailMatchFn = new EmailMatchTypeFunction();
+		AffineGapSimilarityFunction expected = new AffineGapSimilarityFunction();
+		String first = "xyz321@pqr.co";
+		String second = "pqr981@abc.in";
+		assertEquals(expected.call(first.split("@",0)[0],second.split("@",0)[0]), emailMatchFn.call(first,second));
+	}
+
+	@Test
+	public void testFirstPartMatch() {
+		EmailMatchTypeFunction emailMatchFn = new EmailMatchTypeFunction();
+		//AffineGapSimilarityFunction expected = new AffineGapSimilarityFunction();
+		String first = "pqr981@abc.in";
+		String second = "pqr981@xyz.com";
+		assertEquals(1d, emailMatchFn.call(first,second));
+	}
+
+	@Test
+	public void testFirstPartDifferentSecondPartMatch() {
+		EmailMatchTypeFunction emailMatchFn = new EmailMatchTypeFunction();
+		AffineGapSimilarityFunction expected = new AffineGapSimilarityFunction();
+		String first = "pqr981@xyz.com";
+		String second = "aqr981@xyz.com";
+		assertEquals(expected.call(first.split("@",0)[0],second.split("@",0)[0]), emailMatchFn.call(first,second));
+	}
+
+}