Mapped c to greek letter ψ in the greeklish generator

src/main/java/org/elasticsearch/index/analysis/GreeklishConverter.java

@@ -59,7 +59,7 @@ public class GreeklishConverter {
 	private final boolean generateGreekVariants;
 
 	// Constructor
-	public GreeklishConverter(int maxExpansions, boolean generateGreekVariants) {
+	public GreeklishConverter(int maxExpansions, boolean generateGreekVariants, boolean useSpecialMapping) {
 
 		// Initialize greekWords list
 		this.greekWords = new ArrayList<String>();
@@ -68,7 +68,7 @@ public GreeklishConverter(int maxExpansions, boolean generateGreekVariants) {
 		this.reverseStemmer = new GreekReverseStemmer();
 
 		// Initialize greeklish generator
-		this.greeklishGenerator = new GreeklishGenerator(maxExpansions);
+		this.greeklishGenerator = new GreeklishGenerator(maxExpansions, useSpecialMapping);
 
 		// Initialize setting for generating greek variants
 		this.generateGreekVariants = generateGreekVariants;

src/main/java/org/elasticsearch/index/analysis/GreeklishGenerator.java

@@ -72,6 +72,13 @@ public class GreeklishGenerator {
 			{ "φ", "f", "ph" }, { "χ", "x", "h", "ch" }, { "ψ", "ps" },
 			{ "ω", "w", "o", "v" } };
 
+	/**
+	 * The possible string conversions for special cases.
+	 */
+	private static final String[][] specialConvertStrings = new String[][] {
+		{ "ψ", "c", "ps"}
+	};
+
 	/**
 	 * The maximum greeklish expansions per greek token.
 	 */
@@ -102,7 +109,7 @@ public class GreeklishGenerator {
 	private String initialToken;
 
 	// Constructor
-	public GreeklishGenerator(int maxExpansions) {
+	public GreeklishGenerator(int maxExpansions, boolean useSpecialMapping) {
 
 		this.maxExpansions = maxExpansions;
 
@@ -120,6 +127,13 @@ public GreeklishGenerator(int maxExpansions) {
 			conversions.put(convertString[0].charAt(0),
 					Arrays.copyOfRange(convertString, 1, convertString.length));
 		}
+
+		if(useSpecialMapping) {
+			for (String[] convertString : specialConvertStrings) {
+				conversions.put(convertString[0].charAt(0),
+						Arrays.copyOfRange(convertString, 1, convertString.length));
+			}
+		}
 	}
 
 	/**

src/main/java/org/elasticsearch/index/analysis/GreeklishTokenFilter.java

@@ -49,9 +49,9 @@ public class GreeklishTokenFilter extends TokenFilter {
 	private GreeklishConverter greeklishConverter;
 
 	// Constructor
-	public GreeklishTokenFilter(TokenStream tokenStream, int maxExpansions, boolean generateGreekVariants) {
+	public GreeklishTokenFilter(TokenStream tokenStream, int maxExpansions, boolean generateGreekVariants, boolean useSpecialMapping) {
 		super(tokenStream);
-		this.greeklishConverter = new GreeklishConverter(maxExpansions, generateGreekVariants);
+		this.greeklishConverter = new GreeklishConverter(maxExpansions, generateGreekVariants, useSpecialMapping);
 	}
 
 	@Override

src/main/java/org/elasticsearch/index/analysis/GreeklishTokenFilterFactory.java

@@ -11,6 +11,7 @@
 public class GreeklishTokenFilterFactory  extends AbstractTokenFilterFactory {
 	private final int maxExpansions;
 	private final boolean generateGreekVariants;
+	private final boolean useSpecialMapping;
 
 	@Inject
 	public GreeklishTokenFilterFactory(IndexSettings indexSettings,
@@ -21,11 +22,12 @@ public GreeklishTokenFilterFactory(IndexSettings indexSettings,
       		super(indexSettings, name, settings);
             	this.maxExpansions = settings.getAsInt("max_expansions", 20);
             	this.generateGreekVariants = settings.getAsBoolean("greek_variants", true);
+				this.useSpecialMapping = settings.getAsBoolean("use_special_mapping", false);
 	}
 
 	@Override
 	public TokenStream create(TokenStream tokenStream) {
-		return new GreeklishTokenFilter(tokenStream, maxExpansions, generateGreekVariants);
+		return new GreeklishTokenFilter(tokenStream, maxExpansions, generateGreekVariants, useSpecialMapping);
     }
 
 }

src/test/java/org/elasticsearch/index/analysis/GreeklishConverterTest.java

@@ -14,6 +14,9 @@ public class GreeklishConverterTest {
 
 	private static final boolean GENERATE_GREEK_VARIANTS = true;
 
+	private static final boolean USE_SPECIAL_MAPPING_ON = true;
+	private static final boolean USE_SPECIAL_MAPPING_OFF = false;
+
 	private GreeklishConverter converter;
 
 	/**
@@ -23,6 +26,12 @@ public class GreeklishConverterTest {
 	private static final String[] greekWords = { "αυτοκινητο", "ομπρελα",
 			"ξεσκεπαστοσ"};
 
+	/**
+	 * a sample of greek words to generate their greeklish
+	 * counterparts.
+	 */
+	private static final String[] greekWordsSpecial = { "ωιψηυ"};
+
 	/**
 	 * the greeklish counterparts that should be generated from the
 	 * greek words.
@@ -35,6 +44,19 @@ public class GreeklishConverterTest {
 			{"kseskepastos", "xeskepastos", "kseskepastou", "xeskepastwn", "kseskepastoi"}
 	};
 
+	/**
+	 * the greeklish counterparts that should be generated from the
+	 * greek words.
+	 */
+	private static final String[][] generatedGreeklishWordsSpecial = {
+			{
+					"oichu", "wichi", "wichu", "vipsiy",
+					"oipsiy", "wipsiy", "viciy", "oiciy",
+					"wiciy", "vipshy", "oipshy", "wipshy",
+					"vichy", "oichy", "wichy"
+			}
+	};
+
 	/**
 	 * these words should not be processed by the converter.
 	 */
@@ -47,7 +69,7 @@ public class GreeklishConverterTest {
 
 	@BeforeClass
 	public void setUp() {
-		this.converter = new GreeklishConverter(MAX_EXPANSIONS, GENERATE_GREEK_VARIANTS);
+		this.converter = new GreeklishConverter(MAX_EXPANSIONS, GENERATE_GREEK_VARIANTS, USE_SPECIAL_MAPPING_OFF);
 	}
 
 	@BeforeMethod
@@ -85,11 +107,32 @@ public void testGreekTokenConversionForValidWords() {
 		}
 	}
 
+	@Test
+	public void testGreekTokenConversionForValidWordsSpecial() {
+		int newMaxExpansions = 20;
+		converter = new GreeklishConverter(newMaxExpansions, GENERATE_GREEK_VARIANTS, USE_SPECIAL_MAPPING_ON);
+		for (int i = 0; i < greekWordsSpecial.length; i++) {
+			greeklishWords = converter.convert(greekWordsSpecial[i].toCharArray(),
+					greekWordsSpecial[i].length());
+
+			populateConvertedStringsList();
+
+			Assert.assertFalse(greeklishWords.isEmpty(),
+					"Greeklish words should be generated");
+
+			for (String greeklishWord : generatedGreeklishWordsSpecial[i]) {
+				Assert.assertTrue(convertedGreeklishStrings
+								.contains(greeklishWord),
+						"It should contain greeklish word: " + greeklishWord);
+			}
+		}
+	}
+
 	@Test
 	public void testMaxGreeklishExpansions() {
 		int newMaxExpansions = 2;
 		boolean generateGreekVariants = false;
-		converter = new GreeklishConverter(newMaxExpansions, generateGreekVariants);
+		converter = new GreeklishConverter(newMaxExpansions, generateGreekVariants, USE_SPECIAL_MAPPING_OFF);
 
 		greeklishWords = converter.convert(greekWords[0].toCharArray(),
 				greekWords[0].length());
@@ -114,7 +157,7 @@ public void testMaxGreeklishExpansions() {
 	public void testGreekVariantsGeneration() {
 		int newMaxExpansions = 1;
 		boolean generateGreekVariants = false;
-		converter = new GreeklishConverter(newMaxExpansions, generateGreekVariants);
+		converter = new GreeklishConverter(newMaxExpansions, generateGreekVariants, USE_SPECIAL_MAPPING_OFF);
 
 		greeklishWords = converter.convert(greekWords[0].toCharArray(),
 				greekWords[0].length());

src/test/java/org/elasticsearch/index/analysis/GreeklishGeneratorTest.java

@@ -11,6 +11,8 @@
 public class GreeklishGeneratorTest {
 
 	private static final int MAX_EXPANSIONS = 10;
+	private static final boolean USE_SPECIAL_MAPPING_ON = true;
+	private static final boolean USE_SPECIAL_MAPPING_OFF = false;
 
 	/**
 	 * a sample of greek words to generate their greeklish
@@ -19,6 +21,12 @@ public class GreeklishGeneratorTest {
 	private static final String[] greekWords = { "αυτοκινητο", "ομπρελα",
 			"ξεσκεπαστοσ", };
 
+	/**
+	 * a special sample of greek words to generate their greeklish
+	 * counterparts.
+	 */
+	private static final String[] greekWordsSpecial = { "ωιψηυ" };
+
 	/**
 	 * the greeklish counterparts that should be generated from the
 	 * greek words.
@@ -29,6 +37,17 @@ public class GreeklishGeneratorTest {
 			"omprela", "obrela", "kseskepastos", "xeskepastos"
 	};
 
+	/**
+	 * the special greeklish counterparts that should be generated from the
+	 * greek words.
+	 */
+	private static final String[] generatedGreeklishWordsSpecial = {
+			"oichu", "wichi", "wichu", "vipsiy",
+			"oipsiy", "wipsiy", "viciy", "oiciy",
+			"wiciy", "vipshy", "oipshy", "wipshy",
+			"vichy", "oichy", "wichy"
+	};
+
 	private GreeklishGenerator generator;
 
 	private List<String> inputGreekList = new ArrayList<String>();
@@ -39,7 +58,7 @@ public class GreeklishGeneratorTest {
 
 	@BeforeClass
 	public void populateInputGreekList() {
-		this.generator = new GreeklishGenerator(MAX_EXPANSIONS);
+		this.generator = new GreeklishGenerator(MAX_EXPANSIONS, USE_SPECIAL_MAPPING_OFF);
 
 		for (String word : greekWords) {
 			inputGreekList.add(word);
@@ -72,7 +91,7 @@ public void testGreekTokenConversionForValidWords() {
 	@Test
 	public void testMaxGreeklishWordGenerations() {
 		int newMaxExpansions = 2;
-		generator = new GreeklishGenerator(newMaxExpansions);
+		generator = new GreeklishGenerator(newMaxExpansions, USE_SPECIAL_MAPPING_OFF);
 
 		greeklishWords = generator.generateGreeklishWords(inputGreekList);
 
@@ -82,6 +101,31 @@ public void testMaxGreeklishWordGenerations() {
 
 	}
 
+	@Test
+	public void testGreekTokenConversionForValidWordsSpecial() {
+		inputGreekList.clear();
+		for (String word : greekWordsSpecial) {
+			inputGreekList.add(word);
+		}
+
+		for (int i = 0; i < greekWordsSpecial.length; i++) {
+			int newMaxExpansions = 20;
+			generator = new GreeklishGenerator(newMaxExpansions, USE_SPECIAL_MAPPING_ON);
+			greeklishWords = generator.generateGreeklishWords(inputGreekList);
+
+			populateConvertedStringsList();
+
+			Assert.assertFalse(greeklishWords.isEmpty(),
+					"Greeklish words should be generated");
+			for (String greeklishWord : generatedGreeklishWordsSpecial) {
+				Assert.assertTrue(
+						convertedGreeklishStrings.contains(greeklishWord),
+						"It should contain the greeklish word: "
+								+ greeklishWord);
+			}
+		}
+	}
+
 	private final void populateConvertedStringsList() {
 		for (StringBuilder word : greeklishWords) {
 			convertedGreeklishStrings.add(word.toString());