Skip to content

Commit 229ed00

Browse files
netominetomi
committed
[CODEC-63] Added explanation for different results to dropby.com, Raised CC to 100/100
git-svn-id: https://svn.apache.org/repos/asf/commons/proper/codec/trunk@1298576 13f79535-47bb-0310-9956-ffa450edef68
1 parent 805103c commit 229ed00

1 file changed

Lines changed: 82 additions & 9 deletions

File tree

‎src/test/java/org/apache/commons/codec/language/NysiisTest.java‎

Lines changed: 82 additions & 9 deletions
Original file line numberDiff line numberDiff line change
@@ -48,6 +48,15 @@ private void encodeAll(String[] strings, String expectedEncoding) throws Encoder
4848
}
4949
}
5050

51+
@Test
52+
public void testTrueVariant() {
53+
Nysiis encoder = new Nysiis(true);
54+
55+
String encoded = encoder.encode("WESTERLUND");
56+
Assert.assertTrue(encoded.length() <= 6);
57+
Assert.assertEquals("WASTAR", encoded);
58+
}
59+
5160
@Test
5261
public void testBran() throws EncoderException {
5362
encodeAll(new String[] { "Brian", "Brown", "Brun" }, "BRAN");
@@ -70,6 +79,17 @@ public void testDan() throws EncoderException {
7079
this.encodeAll(new String[] { "Dane", "Dean", "Dionne" }, "DAN");
7180
}
7281

82+
@Test
83+
public void testSpecialBranches() throws EncoderException {
84+
this.encodeAll(new String[] { "Kobwick" }, "CABWAC");
85+
this.encodeAll(new String[] { "Kocher" }, "CACAR");
86+
this.encodeAll(new String[] { "Fesca" }, "FASC");
87+
this.encodeAll(new String[] { "Shom" }, "SAN");
88+
this.encodeAll(new String[] { "Ohlo" }, "OL");
89+
this.encodeAll(new String[] { "Uhu" }, "UH");
90+
this.encodeAll(new String[] { "Um" }, "UN");
91+
}
92+
7393
@Test
7494
public void testDropBy() throws EncoderException {
7595
List<String[]> testValues =
@@ -112,16 +132,62 @@ public void testDropBy() throws EncoderException {
112132
*/
113133
@Test
114134
public void testDropBy2() throws EncoderException {
135+
// Explanation of differences between this implementation and the one at dropby.com.
136+
//
137+
// Algorithm (taken from www.dropby.com/NYSIIS.html):
138+
//
139+
// 1. Transcode first characters of name:
140+
// MAC » MCC
141+
// KN » NN
142+
// K » C
143+
// PH » FF
144+
// PF » FF
145+
// SCH » SSS
146+
//
147+
// 2. Transcode last characters of name:
148+
// EE, IE » Y
149+
// DT,RT,RD,NT,ND » D
150+
//
151+
// 3. First character of key = first character of name.
152+
//
153+
// 4. Transcode remaining characters by following these rules, incrementing by one character each time:
154+
// 4a. EV » AF else A,E,I,O,U » A
155+
// 4b. Q » G
156+
// 4c. Z » S
157+
// 4d. M » N
158+
// 4e. KN » N else K » C
159+
// 4f. SCH » SSS
160+
// 4g. PH » FF
161+
// 4h. H » If previous or next is nonvowel, previous
162+
// 4i. W » If previous is vowel, previous
163+
// 4j. Add current to key if current != last key character
164+
//
165+
// 5. If last character is S, remove it
166+
// 6. If last characters are AY, replace with Y
167+
// 7. If last character is A, remove it
168+
// 8. Collapse all strings of repeated characters
169+
// 9. Add original first character of name as first character of key
170+
115171
List<String[]> testValues =
116172
Arrays.asList(
117173
// http://www.dropby.com/indexLF.html?content=/NYSIIS.html
118174
// 1. Transcode first characters of name
119175
new String[] { "MACINTOSH", "MCANT" },
120-
//new String[] { "KNUTH", "NNATH" }, // Original: NNAT; modified: NATH
121-
//new String[] { "KOEHN", "C" },
122-
//new String[] { "PHILLIPSON", "FFALAP" },
123-
//new String[] { "PFEISTER", "FFASTA" },
124-
//new String[] { "SCHOENHOEFT", "SSANAF" },
176+
// violates 4j: the second N should not be added, as the first
177+
// key char is already a N
178+
new String[] { "KNUTH", "NAT" }, // Original: NNAT; modified: NATH
179+
// O and E are transcoded to A because of rule 4a
180+
// H also to A because of rule 4h
181+
// the N gets mysteriously lost, maybe because of a wrongly implemented rule 4h
182+
// that skips the next char in such a case?
183+
// the remaining A is removed because of rule 7
184+
new String[] { "KOEHN", "CAN" }, // Original: C
185+
// violates 4j: see also KNUTH
186+
new String[] { "PHILLIPSON", "FALAPSAN" }, // Original: FFALAP[SAN]
187+
// violates 4j: see also KNUTH
188+
new String[] { "PFEISTER", "FASTAR" }, // Original: FFASTA[R]
189+
// violoates 4j: see also KNUTH
190+
new String[] { "SCHOENHOEFT", "SANAFT" }, // Original: SSANAF[T]
125191
// http://www.dropby.com/indexLF.html?content=/NYSIIS.html
126192
// 2.Transcode last characters of name:
127193
new String[] { "MCKEE", "MCY" },
@@ -139,14 +205,21 @@ public void testDropBy2() throws EncoderException {
139205
new String[] { "BOWMAN", "BANAN" },
140206
new String[] { "MCKNIGHT", "MCNAGT" },
141207
new String[] { "RICKERT", "RACAD" },
142-
//new String[] { "DEUTSCH", "DATS" },
208+
// violates 5: the last S is not removed
209+
// when comparing to DEUTS, which is phonetically similar
210+
// the result it also DAT, which is correct for DEUTSCH too imo
211+
new String[] { "DEUTSCH", "DAT" }, // Original: DATS
143212
new String[] { "WESTPHAL", "WASTFAL" },
144-
//new String[] { "SHRIVER", "SHRAVA" },
145-
//new String[] { "KUHL", "C" },
213+
// violates 4h: the H should be transcoded to S and thus ignored as
214+
// the first key character is also S
215+
new String[] { "SHRIVER", "SRAVAR" }, // Original: SHRAVA[R]
216+
// same as KOEHN, the L gets mysteriously lost, the correct one
217+
new String[] { "KUHL", "CAL" }, // Original: C
146218
new String[] { "RAWSON", "RASAN" },
147219
// If last character is S, remove it
148220
new String[] { "JILES", "JAL" },
149-
//new String[] { "CARRAWAY", "CARAY" },
221+
// violates 6: if the last two characters are AY, remove A
222+
new String[] { "CARRAWAY", "CARY" }, // Original: CARAY
150223
new String[] { "YAMADA", "YANAD" });
151224

152225
for (String[] arr : testValues) {

0 commit comments

Comments
 (0)