Created refined soundex algorithm, metric, command, and specs.

5 files changed, 205 insertions, 1 deletions

diff --git a/core/source/core/scala/org/hashtree/stringmetric/phonetic/RefinedSoundex.scala b/core/source/core/scala/org/hashtree/stringmetric/phonetic/RefinedSoundex.scala
new file mode 100755
index 0000000..8443a5f
--- /dev/null
+++ b/core/source/core/scala/org/hashtree/stringmetric/phonetic/RefinedSoundex.scala
@@ -0,0 +1,84 @@
+package org.hashtree.stringmetric.phonetic
+
+import org.hashtree.stringmetric.{ StringAlgorithm, StringFilter, StringFilterDelegate }
+import scala.annotation.tailrec
+
+/** An implementation of the refined Soundex [[org.hashtree.stringmetric.StringAlgorithm]]. */
+object RefinedSoundex extends StringAlgorithm {
+	override def compute(charArray: Array[Char])(implicit stringFilter: StringFilter): Option[Array[Char]] = {
+		val ca = stringFilter.filter(charArray)
+
+		if (ca.length == 0) None
+		else {
+			val fc = ca.head.toLower
+
+			if (fc < 97 || fc > 122) None
+			else {
+				Some(
+					transcode(
+						ca,
+						fc, // Pass first letter.
+						Array(fc) // Pass array with first letter.
+					)
+				)
+			}
+		}
+	}
+
+	override def compute(string: String)(implicit stringFilter: StringFilter): Option[String] =
+		compute(stringFilter.filter(string.toCharArray))(new StringFilterDelegate) match {
+			case Some(se) => Some(se.mkString)
+			case None => None
+		}
+
+	@tailrec
+	private[this] def transcode(i: Array[Char], p: Char, o: Array[Char]): Array[Char] = {
+		require(o.length > 0)
+
+		if (i.length == 0) o
+		else {
+			val c = i.head.toLower
+			val m2 = (mc: Char) => mc match {
+				case 'a' | 'e' | 'h' | 'i' | 'o' | 'u' | 'w' | 'y' => '0'
+				case 'b' | 'p' => '1'
+				case 'f' | 'v' => '2'
+				case 'c' | 'k' | 's' => '3'
+				case 'g' | 'j' => '4'
+				case 'q' | 'x' | 'z' => '5'
+				case 'd' | 't' => '6'
+				case 'l' => '7'
+				case 'm' | 'n' => '8'
+				case 'r' => '9'
+				case _ => '\0'
+			}
+			val m1 = (mc: Char, pc: Char) => mc match {
+				case 'a' | 'e' | 'h' | 'i' | 'o' | 'u' | 'w' | 'y' if pc != '0' => '0'
+				case 'b' | 'p' if pc != '1' => '1'
+				case 'f' | 'v' if pc != '2' => '2'
+				case 'c' | 'k' | 's' if pc != '3' => '3'
+				case 'g' | 'j' if pc != '4' => '4'
+				case 'q' | 'x' | 'z' if pc != '5' => '5'
+				case 'd' | 't' if pc != '6' => '6'
+				case 'l' if pc != '7' => '7'
+				case 'm' | 'n' if pc != '8' => '8'
+				case 'r' if pc != '9' => '9'
+				case _ => '\0'
+			}
+			val a =
+				// Code twice.
+				if (o.length == 1)
+					m2(c)
+				// Code once.
+				else
+					m1(
+						c,
+						o.last match {
+							case '0' | '1' | '2' | '3' | '4' | '5' | '6' | '7' | '8' | '9' => o.last
+							case _ => m2(o.last)
+						}
+					)
+
+			transcode(i.tail, c, if (a != '\0') o :+ a else o)
+		}
+	}
+}
\ No newline at end of file
diff --git a/core/source/core/scala/org/hashtree/stringmetric/phonetic/RefinedSoundexMetric.scala b/core/source/core/scala/org/hashtree/stringmetric/phonetic/RefinedSoundexMetric.scala
new file mode 100755
index 0000000..0c233b0
--- /dev/null
+++ b/core/source/core/scala/org/hashtree/stringmetric/phonetic/RefinedSoundexMetric.scala
@@ -0,0 +1,26 @@
+package org.hashtree.stringmetric.phonetic
+
+import org.hashtree.stringmetric.{ StringFilter, StringFilterDelegate, StringMetric }
+
+/** An implementation of the refined Soundex [[org.hashtree.stringmetric.StringMetric]]. */
+object RefinedSoundexMetric extends StringMetric {
+	override def compare(charArray1: Array[Char], charArray2: Array[Char])(implicit stringFilter: StringFilter): Option[Boolean] = {
+		val ca1 = stringFilter.filter(charArray1)
+		val ca2 = stringFilter.filter(charArray2)
+
+		if (ca1.length == 0 || ca2.length == 0) None
+		else {
+			val rse1 = RefinedSoundex.compute(ca1)
+			val rse2 = RefinedSoundex.compute(ca2)
+
+			if (!rse1.isDefined || !rse2.isDefined || (rse1.get.length == 0 && rse2.get.length == 0)) None
+			else Some(rse1.get.sameElements(rse2.get))
+		}
+	}
+
+	override def compare(string1: String, string2: String)(implicit stringFilter: StringFilter): Option[Boolean] =
+		compare(
+			stringFilter.filter(string1.toCharArray),
+			stringFilter.filter(string2.toCharArray)
+		)(new StringFilterDelegate)
+}
\ No newline at end of file
diff --git a/core/source/test/scala/org/hashtree/stringmetric/phonetic/RefinedSoundexMetricSpec.scala b/core/source/test/scala/org/hashtree/stringmetric/phonetic/RefinedSoundexMetricSpec.scala
new file mode 100755
index 0000000..9524835
--- /dev/null
+++ b/core/source/test/scala/org/hashtree/stringmetric/phonetic/RefinedSoundexMetricSpec.scala
@@ -0,0 +1,37 @@
+package org.hashtree.stringmetric.phonetic
+
+import org.hashtree.stringmetric.ScalaTest
+import org.junit.runner.RunWith
+import org.scalatest.junit.JUnitRunner
+
+@RunWith(classOf[JUnitRunner])
+final class RefinedSoundexMetricSpec extends ScalaTest {
+	"RefinedSoundexMetric" should provide {
+		"compare method" when passed {
+			"empty arguments" should returns {
+				"None" in {
+					RefinedSoundexMetric.compare("", "").isDefined should be (false)
+					RefinedSoundexMetric.compare("abc", "").isDefined should be (false)
+					RefinedSoundexMetric.compare("", "xyz").isDefined should be (false)
+				}
+			}
+			"non-phonetic arguments" should returns {
+				"None" in {
+					RefinedSoundexMetric.compare("123", "123").isDefined should be (false)
+					RefinedSoundexMetric.compare("123", "").isDefined should be (false)
+					RefinedSoundexMetric.compare("", "123").isDefined should be (false)
+				}
+			}
+			"phonetically similar arguments" should returns {
+				"Boolean indicating true" in {
+					RefinedSoundexMetric.compare("robert", "rupert").get should be (true)
+				}
+			}
+			"phonetically dissimilar arguments" should returns {
+				"Boolean indicating false" in {
+					RefinedSoundexMetric.compare("robert", "rubin").get should be (false)
+				}
+			}
+		}
+	}
+}
\ No newline at end of file
diff --git a/core/source/test/scala/org/hashtree/stringmetric/phonetic/RefinedSoundexSpec.scala b/core/source/test/scala/org/hashtree/stringmetric/phonetic/RefinedSoundexSpec.scala
new file mode 100755
index 0000000..1531eb0
--- /dev/null
+++ b/core/source/test/scala/org/hashtree/stringmetric/phonetic/RefinedSoundexSpec.scala
@@ -0,0 +1,54 @@
+package org.hashtree.stringmetric.phonetic
+
+import org.hashtree.stringmetric.ScalaTest
+import org.junit.runner.RunWith
+import org.scalatest.junit.JUnitRunner
+
+@RunWith(classOf[JUnitRunner])
+final class RefinedSoundexSpec extends ScalaTest {
+	"RefinedSoundex" should provide {
+		"compute method" when passed {
+			"empty argument" should returns {
+				"None" in {
+					RefinedSoundex.compute("").isDefined should be (false)
+				}
+			}
+			"non-phonetic argument" should returns {
+				"None" in {
+					RefinedSoundex.compute("123").isDefined should be (false)
+				}
+			}
+			"phonetic argument" should returns {
+				"Some" in {
+					RefinedSoundex.compute("x123").get should equal ("x5")
+
+					RefinedSoundex.compute("braz").get should equal ("b1905")
+					RefinedSoundex.compute("broz").get should equal ("b1905")
+					RefinedSoundex.compute("caren").get should equal ("c30908")
+					RefinedSoundex.compute("carren").get should equal ("c30908")
+					RefinedSoundex.compute("coram").get should equal ("c30908")
+					RefinedSoundex.compute("corran").get should equal ("c30908")
+					RefinedSoundex.compute("curreen").get should equal ("c30908")
+					RefinedSoundex.compute("curwen").get should equal ("c30908")
+					RefinedSoundex.compute("hairs").get should equal ("h093")
+					RefinedSoundex.compute("hark").get should equal ("h093")
+					RefinedSoundex.compute("hars").get should equal ("h093")
+					RefinedSoundex.compute("hayers").get should equal ("h093")
+					RefinedSoundex.compute("heers").get should equal ("h093")
+					RefinedSoundex.compute("hiers").get should equal ("h093")
+					RefinedSoundex.compute("lambard").get should equal ("l7081096")
+					RefinedSoundex.compute("lambart").get should equal ("l7081096")
+					RefinedSoundex.compute("lambert").get should equal ("l7081096")
+					RefinedSoundex.compute("lambird").get should equal ("l7081096")
+					RefinedSoundex.compute("lampaert").get should equal ("l7081096")
+					RefinedSoundex.compute("lampart").get should equal ("l7081096")
+					RefinedSoundex.compute("lamport").get should equal ("l7081096")
+					RefinedSoundex.compute("limbert").get should equal ("l7081096")
+					RefinedSoundex.compute("lombard").get should equal ("l7081096")
+					RefinedSoundex.compute("nolton").get should equal ("n807608")
+					RefinedSoundex.compute("noulton").get should equal ("n807608")
+				}
+			}
+		}
+	}
+}
\ No newline at end of file
diff --git a/core/source/test/scala/org/hashtree/stringmetric/phonetic/SoundexSpec.scala b/core/source/test/scala/org/hashtree/stringmetric/phonetic/SoundexSpec.scala
index dbefc51..9f18bc9 100755
--- a/core/source/test/scala/org/hashtree/stringmetric/phonetic/SoundexSpec.scala
+++ b/core/source/test/scala/org/hashtree/stringmetric/phonetic/SoundexSpec.scala
@@ -20,8 +20,11 @@ final class SoundexSpec extends ScalaTest {
 			}
 			"phonetic argument" should returns {
 				"Some" in {
+					Soundex.compute("x123").get should equal ("x000")
+
 					Soundex.compute("abc").get should equal ("a120")
 					Soundex.compute("xyz").get should equal ("x200")
+
 					Soundex.compute("robert").get should equal ("r163")
 					Soundex.compute("rupert").get should equal ("r163")
 					Soundex.compute("rubin").get should equal ("r150")
@@ -43,7 +46,7 @@ final class SoundexSpec extends ScalaTest {
 					Soundex.compute("kant").get should equal ("k530")
 					Soundex.compute("ladd").get should equal ("l300")
 					Soundex.compute("lissajous").get should equal ("l222")
-					Soundex.compute("x123").get should equal ("x000")
+					Soundex.compute("fusedale").get should equal ("f234")
 				}
 			}
 		}