soundex_test.gno

 1package soundex
 2
 3import (
 4	"testing"
 5
 6	"gno.land/p/nt/uassert/v0"
 7)
 8
 9// The canonical examples, including the three that catch a naive
10// implementation: Pfister, Tymczak and Ashcraft.
11func TestEncodeCanonical(t *testing.T) {
12	cases := []struct{ in, want string }{
13		{"Robert", "R163"},
14		{"Rupert", "R163"},
15		{"Rubin", "R150"},
16		{"Ashcraft", "A261"}, // 'h' is transparent: s and c collapse
17		{"Ashcroft", "A261"},
18		{"Tymczak", "T522"}, // vowel separates, so 2 repeats
19		{"Pfister", "P236"}, // first letter's code suppresses the f
20		{"Honeyman", "H555"},
21		{"Jackson", "J250"},
22		{"Washington", "W252"},
23		{"Lee", "L000"},
24		{"Gutierrez", "G362"},
25	}
26	for _, c := range cases {
27		uassert.Equal(t, c.want, Encode(c.in))
28	}
29}
30
31// The first letter keeps its own code for suppression purposes — the rule most
32// often missed. Without it "Pfister" would come out P123.
33func TestFirstLetterSuppresses(t *testing.T) {
34	uassert.Equal(t, "P236", Encode("Pfister"))
35	uassert.Equal(t, "B000", Encode("Bb"))
36}
37
38// 'h'/'w' are transparent; vowels are not.
39func TestTransparentHW(t *testing.T) {
40	uassert.Equal(t, Encode("Ashcraft"), Encode("Ascraft"))
41	// a vowel between the same two codes does NOT collapse them
42	uassert.Equal(t, "T522", Encode("Tymczak"))
43}
44
45func TestCaseAndPunctuationIgnored(t *testing.T) {
46	uassert.Equal(t, Encode("OBrien"), Encode("O'Brien"))
47	uassert.Equal(t, Encode("robert"), Encode("ROBERT"))
48	uassert.Equal(t, "R163", Encode("  Robert!  "))
49}
50
51func TestNoLetters(t *testing.T) {
52	uassert.Equal(t, "", Encode(""))
53	uassert.Equal(t, "", Encode("12345"))
54	uassert.Equal(t, "", Encode("!!!"))
55}
56
57// Two inputs with no letters are not "the same name".
58func TestMatch(t *testing.T) {
59	uassert.True(t, Match("Robert", "Rupert"))
60	uassert.False(t, Match("Robert", "Rubin"))
61	uassert.False(t, Match("", ""))
62	uassert.False(t, Match("123", "456"))
63}
64
65// Always exactly four characters, padded with zeros.
66func TestAlwaysFourChars(t *testing.T) {
67	for _, s := range []string{"A", "Lee", "Robert", "Washington", "Supercalifragilistic"} {
68		uassert.Equal(t, 4, len(Encode(s)))
69	}
70}
71
72func TestEncodeAllAndNormalize(t *testing.T) {
73	got := EncodeAll([]string{"Robert", "Lee"})
74	uassert.Equal(t, 2, len(got))
75	uassert.Equal(t, "R163", got[0])
76	uassert.Equal(t, "L000", got[1])
77	uassert.Equal(t, "OBRIEN", Normalize("O'Brien"))
78}