func Test_r1r2(t *testing.T) {
var wordTests = []struct {
word string
r1 string
r2 string
}{
{"crepuscular", "uscular", "cular"},
{"beautiful", "iful", "ul"},
{"beauty", "y", ""},
{"eucharist", "harist", "ist"},
{"animadversion", "imadversion", "adversion"},
{"mistresses", "tresses", "ses"},
{"sprinkled", "kled", ""},
// Special cases below
{"communism", "ism", "m"},
{"arsenal", "al", ""},
{"generalities", "alities", "ities"},
{"embed", "bed", ""},
}
for _, testCase := range wordTests {
w := snowballword.New(testCase.word)
r1start, r2start := r1r2(w)
w.R1start = r1start
w.R2start = r2start
if w.R1String() != testCase.r1 || w.R2String() != testCase.r2 {
t.Errorf("Expected \"{%v, %v}\", but got \"{%v, %v}\"", testCase.r1, testCase.r2, w.R1String(), w.R2String())
}
}
}
// Stem an English word. This is the only exported
// function in this package.
//
func Stem(word string, stemStopwWords bool) string {
word = strings.ToLower(strings.TrimSpace(word))
// Return small words and stop words
if len(word) <= 2 || (stemStopwWords == false && isStopWord(word)) {
return word
}
// Return special words immediately
if specialVersion := stemSpecialWord(word); specialVersion != "" {
word = specialVersion
return word
}
w := snowballword.New(word)
// Stem the word. Note, each of these
// steps will alter `w` in place.
//
preprocess(w)
step0(w)
step1a(w)
step1b(w)
step1c(w)
step2(w)
step3(w)
step4(w)
step5(w)
postprocess(w)
return w.String()
}
// Stem an Spanish word. This is the only exported
// function in this package.
//
func Stem(word string, stemStopwWords bool) string {
word = strings.ToLower(strings.TrimSpace(word))
// Return small words and stop words
if len(word) <= 2 || (stemStopwWords == false && isStopWord(word)) {
return word
}
w := snowballword.New(word)
// Stem the word. Note, each of these
// steps will alter `w` in place.
//
preprocess(w)
step0(w)
changeInStep1 := step1(w)
if changeInStep1 == false {
changeInStep2a := step2a(w)
if changeInStep2a == false {
step2b(w)
}
}
step3(w)
postprocess(w)
return w.String()
}
func runStepTest(t *testing.T, f stepFunc, tcs []stepTest) {
for _, testCase := range tcs {
w := snowballword.New(testCase.wordIn)
w.R1start = testCase.r1start
w.R2start = testCase.r2start
_ = f(w)
if w.String() != testCase.wordOut || w.R1String() != testCase.r1out || w.R2String() != testCase.r2out {
t.Errorf("Expected \"{%v, %v, %v}\", but got \"{%v, %v, %v}\"", testCase.wordOut, testCase.r1out, testCase.r2out, w.String(), w.R1String(), w.R2String())
}
}
}
// Test isLowerVowel for things we know should be true
// or false.
//
func RunFindRegionsTest(t *testing.T, f func(*snowballword.SnowballWord) (int, int, int), tcs []FindRegionsTestCase) {
for _, testCase := range tcs {
w := snowballword.New(testCase.Word)
r1start, r2start, rvstart := f(w)
if r1start != testCase.R1start || r2start != testCase.R2start || rvstart != testCase.RVstart {
t.Errorf("Expect \"%v\" -> %v, %v, %v, but got %v, %v, %v",
testCase.Word, testCase.R1start, testCase.R2start, testCase.RVstart,
r1start, r2start, rvstart,
)
}
}
}
func Test_capitalizeYs(t *testing.T) {
var wordTests = []struct {
in string
out string
}{
{"ysdcsdeysdfsysdfsdiyoyyyxyxayxey", "YsdcsdeYsdfsysdfsdiYoYyYxyxaYxeY"},
}
for _, wt := range wordTests {
w := snowballword.New(wt.in)
capitalizeYs(w)
if w.String() != wt.out {
t.Errorf("Expected \"%v\", not \"%v\"", wt.out, w.String())
}
}
}
func RunStepTest(t *testing.T, f stepFunc, tcs []StepTestCase) {
for _, testCase := range tcs {
w := snowballword.New(testCase.WordIn)
w.R1start = testCase.R1start
w.R2start = testCase.R2start
w.RVstart = testCase.RVstart
retval := f(w)
if retval != testCase.Changed || w.String() != testCase.WordOut || w.R1start != testCase.R1startOut || w.R2start != testCase.R2startOut || w.RVstart != testCase.RVstartOut {
t.Errorf("Expected %v -> \"{%v, %v, %v, %v, %v}\", but got \"{%v, %v, %v, %v, %v}\"", testCase.WordIn, testCase.WordOut, testCase.R1startOut, testCase.R2startOut, testCase.RVstartOut, testCase.Changed, w.String(), w.R1start, w.R2start, w.RVstart, retval)
}
if w.String() != testCase.WordOut {
fmt.Printf("{\"%v\", %v, %v, %v, true, \"%v\", %v, %v, %v},\n", testCase.WordIn, testCase.R1start, testCase.R2start, testCase.RVstart, testCase.WordOut, w.R1start, w.R2start, w.RVstart)
}
}
}
// Stem an French word. This is the only exported
// function in this package.
//
func Stem(word string, stemStopwWords bool) string {
word = strings.ToLower(strings.TrimSpace(word))
// Return small words and stop words
if len(word) <= 2 || (stemStopwWords == false && isStopWord(word)) {
return word
}
w := snowballword.New(word)
// Stem the word. Note, each of these
// steps will alter `w` in place.
//
preprocess(w)
var (
changeInStep1 bool
changeInStep2a bool
changeInStep2b bool
)
changeInStep1 = step1(w)
if changeInStep1 == false {
changeInStep2a = step2a(w)
if changeInStep2a == false {
changeInStep2b = step2b(w)
}
}
// If the last step was successful, do step 3. Note that,
// since we only do 2a if 1 is unsuccessful, the following
// "if" condition tests to see if the previous step was
// successful.
//
if changeInStep1 || changeInStep2a || changeInStep2b {
step3(w)
} else {
step4(w)
}
step5(w)
step6(w)
postprocess(w)
return w.String()
}
func Test_normalizeApostrophes(t *testing.T) {
variants := [...]string{
"\u2019xxx\u2019",
"\u2018xxx\u2018",
"\u201Bxxx\u201B",
"’xxx’",
"‘xxx‘",
"‛xxx‛",
}
for _, v := range variants {
w := snowballword.New(v)
normalizeApostrophes(w)
if w.String() != "'xxx'" {
t.Errorf("Expected \"'xxx'\", not \"%v\"", w.String())
}
}
}
func Test_vnvSuffix(t *testing.T) {
var wordTests = []struct {
word string
start int
pos int
}{
{"crepuscular", 0, 4},
{"uscular", 0, 2},
}
for _, tc := range wordTests {
w := snowballword.New(tc.word)
pos := romance.VnvSuffix(w, isLowerVowel, tc.start)
if pos != tc.pos {
t.Errorf("Expected %v, but got %v", tc.pos, pos)
}
}
}
func Test_preprocess(t *testing.T) {
var wordTests = []struct {
in string
out string
}{
{"arguing", "arguing"},
{"'catty", "catty"},
{"kyle’s", "kyle's"},
{"toy", "toY"},
}
for _, wt := range wordTests {
w := snowballword.New(wt.in)
preprocess(w)
if w.String() != wt.out {
t.Errorf("Expected \"%v\", not \"%v\"", wt.out, w.String())
}
}
}
// Stem an Russian word. This is the only exported
// function in this package.
//
func Stem(word string, stemStopwWords bool) string {
word = strings.ToLower(strings.TrimSpace(word))
w := snowballword.New(word)
// Return small words and stop words
if len(w.RS) <= 2 || (stemStopwWords == false && isStopWord(word)) {
return word
}
preprocess(w)
step1(w)
step2(w)
step3(w)
step4(w)
return w.String()
}
// Test capitalization of vowels acting as non-vowels.
//
func Test_capitalizeYUI(t *testing.T) {
testCases := []struct {
wordIn string
wordOut string
}{
{"jouer", "joUer"},
{"ennuie", "ennuIe"},
{"yeux", "Yeux"},
{"quand", "qUand"},
}
for _, testCase := range testCases {
w := snowballword.New(testCase.wordIn)
capitalizeYUI(w)
if w.String() != testCase.wordOut {
t.Errorf("Expect %v -> %v, but got %v", testCase.wordIn, testCase.wordOut, w.String())
}
}
}
func Test_isShortWord(t *testing.T) {
var testCases = []struct {
word string
isShort bool
}{
{"bed", true},
{"shed", true},
{"shred", true},
{"bead", false},
{"embed", false},
{"beds", false},
}
for _, testCase := range testCases {
w := snowballword.New(testCase.word)
r1start, r2start := r1r2(w)
w.R1start = r1start
w.R2start = r2start
isShort := isShortWord(w)
if isShort != testCase.isShort {
t.Errorf("Expected %v, but got %v for \"{%v, %v}\"", testCase.isShort, isShort, testCase.word, w.R1String())
}
}
}
func Test_endsShortSyllable(t *testing.T) {
var testCases = []struct {
word string
pos int
result bool
}{
{"absolute", 7, true},
{"ape", 2, true},
{"rap", 3, true},
{"trap", 4, true},
{"entrap", 6, true},
{"uproot", 6, false},
{"bestow", 6, false},
{"disturb", 7, false},
}
for _, testCase := range testCases {
w := snowballword.New(testCase.word)
result := endsShortSyllable(w, testCase.pos)
if result != testCase.result {
t.Errorf("Expected endsShortSyllable(%v, %v) to return %v, not %v", testCase.word, testCase.pos, testCase.result, result)
}
}
}
