Merge pull request #106 from evllabs/fix-smash-i

Remove Regressive Smash I Cases.

Author: jnoecker

src/com/jgaap/canonicizers/SmashI.java

@@ -17,18 +17,22 @@
  */
 package com.jgaap.canonicizers;
 
-import com.jgaap.generics.CanonicizationException;
 import com.jgaap.generics.Canonicizer;
 
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
+
 /**
  * Canonicizer for smashing all instances in which "I" is used as a
  * word.
- * 
+ *
  * @author David
  * @since 8.0.0
  */
 public class SmashI extends Canonicizer {
 
+	private static Pattern iWord = Pattern.compile("(^|\\W+)I(\\W+|$)");
+
 	@Override
 	public String displayName() {
 		return "Smash I";
@@ -38,7 +42,7 @@ public String displayName() {
 	public String tooltipText() {
 		return "Converts all uses of \"I\" to lowercase.";
 	}
-	
+
 	@Override
 	public String longDescription() {
 		return "Converts all uses of \"I\" as a word to lowercase.";
@@ -50,40 +54,9 @@ public boolean showInGUI() {
 	}
 
 	@Override
-	public char[] process(char[] procText) throws CanonicizationException {
-		for (int x = 0; x < procText.length; x++) {
-			// Character is not even a potential candidate for smashing if it is
-			// not a capital I.
-			if (procText[x] == 'I' ) {
-				// Left and right flag are for indicating if whitespace was found
-				// on either side of "I."
-				boolean leftFlag = false;
-				boolean rightFlag = false;
-				
-				try {
-					// Check for whitespace on left side.
-					leftFlag = Character.isWhitespace(procText[x - 1]);
-				}
-				catch (ArrayIndexOutOfBoundsException e) {
-					// If the I is at the beginning of the string, set left flag to true.
-					leftFlag = true;
-				}
-				
-				try {
-					// Check for whitespace on the right side.
-					rightFlag = Character.isWhitespace(procText[x + 1]);
-				}
-				catch (ArrayIndexOutOfBoundsException e) {
-					// If the I is at the end of the string, set right flag to true.
-					rightFlag = true;
-				}
-				
-				// Smash character if both flags are set to true.
-				if (leftFlag && rightFlag)
-					procText[x] = Character.toLowerCase(procText[x]);
-			}
-		}
-		return procText;
+	public char[] process(char[] procText) {
+		Matcher m = iWord.matcher(new String(procText));
+		return m.replaceAll("$1i$2").toCharArray();
 	}
-
 }
+

unittests/com/jgaap/canonicizers/SmashITest.java

@@ -17,13 +17,9 @@
  */
 package com.jgaap.canonicizers;
 
-import static org.junit.Assert.assertTrue;
-
-import java.util.Arrays;
-
 import org.junit.Test;
 
-import com.jgaap.generics.CanonicizationException;
+import static org.junit.Assert.assertArrayEquals;
 
 /**
  * Unit test for the Smash I canonicizer.
@@ -33,32 +29,50 @@
  */
 public class SmashITest {
 	@Test
-	public void testProcess() throws CanonicizationException {
+	public void testProcess() {
 		SmashI smashI = new SmashI();
 
 		// Test 1 - I on ends
 		String in = "I don't care if IPad is supposed to be spelled with a lowercase I";
 		char[] correct = "i don't care if IPad is supposed to be spelled with a lowercase i".toCharArray();
 		char[] actual = smashI.process(in.toCharArray());
-		assertTrue(Arrays.equals(correct, actual));
+		assertArrayEquals(correct, actual);
 
 		// Test 2 - I in middle surrounded by spaces
 		in = "Sometimes I cannot think of creative things to write for unit tests.";
 		correct = "Sometimes i cannot think of creative things to write for unit tests.".toCharArray();
 		actual = smashI.process(in.toCharArray());
-		assertTrue(Arrays.equals(correct, actual));
+		assertArrayEquals(correct, actual);
 
 		// Test 3 - I in middle surrounded by tabs
 		in = "Sometimes		I	cannot think of creative things to write for unit tests.";
 		correct = in.toCharArray();
 		correct[11] = 'i';
 		actual = smashI.process(in.toCharArray());
-		assertTrue(Arrays.equals(correct, actual));
+		assertArrayEquals(correct, actual);
 
 		// Test 4 - Bunch of I's next to each other with varying case
 		in = "iIiIiiIIiiiiiiIIiiIiIiIIiiIIiiIiiIiiIIiiiIiiiIiI";
 		correct = in.toCharArray();
 		actual = smashI.process(in.toCharArray());
-		assertTrue(Arrays.equals(correct, actual));
+		assertArrayEquals(correct, actual);
+
+		// Test 5 - Non-whitespaced I
+		in = "\"I am here\", she said.";
+		correct = "\"i am here\", she said.".toCharArray();
+		actual = smashI.process(in.toCharArray());
+		assertArrayEquals(correct, actual);
+
+		// Test 6 - Non-whitespaced I
+		in = "I'm here.";
+		correct = "i'm here.".toCharArray();
+		actual = smashI.process(in.toCharArray());
+		assertArrayEquals(correct, actual);
+
+		// Test 7 - Non-whitespaced I
+		in = "It is I.";
+		correct = "It is i.".toCharArray();
+		actual = smashI.process(in.toCharArray());
+		assertArrayEquals(correct, actual);
 	}
 }