T handling

NaturalNode · Feb 14, 2012 · 18c4aff · 18c4aff
1 parent 257eb7f
commit 18c4aff
Show file tree

Hide file tree

Showing 2 changed files with 53 additions and 2 deletions.
diff --git a/lib/natural/phonetics/double_metaphone.js b/lib/natural/phonetics/double_metaphone.js
@@ -123,8 +123,26 @@ function process(token, spyCallback) {
     }
 
     function handleT() {
-        if(token.substring(pos + 1, pos + 4) == 'ION' || subMatch(1, 3, ['IA', 'CH'])) {
+        if(token.substring(pos + 1, pos + 4) == 'ION') {
+            add('XN');
+            pos += 3;
+        } else if(subMatch(1, 3, ['IA', 'CH'])) {
             add('X');
+            pos += 2;
+        } else if(token[pos + 1] == 'H' 
+                || token.substring(1, 2) == 'TH') {
+            if(subMatch(2, 4, ['OM', 'AM']) 
+                    || ['VAN ', 'VON '].indexOf(token.substring(0, 4)) > -1
+                    || token.substring(0, 3) == 'SCH') {
+                add('T');            
+            } else
+                addSecondary('0', 'T');
+            pos++;
+        } else {
+            addCompressedDouble('T');
+
+            if(token[pos + 1] == 'D')
+                pos++;
         }
     }
 

diff --git a/spec/double_metaphone_spec.js b/spec/double_metaphone_spec.js
@@ -257,9 +257,42 @@ describe('double metaphone', function() {
     });
 
     describe('T', function() {
-    	it('should encode T', function() {
+    	it('should encode TION to XN', function() {
 	    	var encodings = doubleMetaphone.process('nation');
 	    	expect(encodings[0]).toMatch(/.*XN$/);
+	    	expect(encodings[1]).toMatch(/.*XN$/);	    	
+    	});
+
+    	it('should encode CH sounds to X', function() {
+	    	var encodings = doubleMetaphone.process('thatch');
+	    	expect(encodings[0]).toMatch(/.*X$/);
+	    	expect(encodings[1]).toMatch(/.*X$/);
+    	});
+
+    	it('should encode hard TH to T', function() {
+	    	var encodings = doubleMetaphone.process('thomas');
+	    	expect(encodings[0]).toMatch(/^T.*/);
+	    	expect(encodings[1]).toMatch(/^T.*/);
+    	});
+
+    	it('should encode soft TH to 0,T', function() {
+	    	var encodings = doubleMetaphone.process('this');
+	    	expect(encodings[0]).toMatch(/^0.*/);
+	    	expect(encodings[1]).toMatch(/^T.*/);
+    	});
+
+    	it('should encode TT to T', function() {
+	    	var encodings = doubleMetaphone.process('matta');
+	    	expect(encodings[0]).toMatch(/[^T]T/);
+	    	expect(encodings[1]).toMatch(/[^T]T/);	    	
+    	}); 
+
+    	it('should encode TD to T', function() {
+	    	var encodings = doubleMetaphone.process('countdown');
+	    	expect(encodings[0]).toContain('T');
+	    	expect(encodings[0]).toNotContain('D');
+	    	expect(encodings[1]).toContain('T');
+	    	expect(encodings[1]).toNotContain('D');
     	});
     });