Resolved the issue of undetected English words

fluffy-melli · Jan 16, 2025 · 84a0212 · 84a0212
1 parent 4eff944
commit 84a0212
Show file tree

Hide file tree

Showing 4 changed files with 31 additions and 40 deletions.
diff --git a/go.mod b/go.mod
@@ -2,4 +2,4 @@ module github.com/fluffy-melli/korcen-go
 
 go 1.22.2
 
-require github.com/gyarang/gohangul v0.1.3 // indirect
+require github.com/gyarang/gohangul v0.1.3
diff --git a/korcen.go b/korcen.go
@@ -58,6 +58,9 @@ func CList(list []string, types int, input string, continues bool) (bool, []Inde
 			}
 		}
 		item = KoToEnglish(item)
+		if len(item) <= 1 {
+			continue
+		}
 		in = strings.Index(input, item)
 		if in != -1 {
 			indexs = append(indexs, IndexOF{
@@ -184,13 +187,12 @@ func Politics(input string, continues bool) (bool, []IndexOF) {
 //	bool: 비속어가 포함된 경우 true, 그렇지 않으면 false.
 //	string: 감지된 비속어가 있으면 해당 비속어를, 없으면 빈 문자열("")을 반환.
 func English(input string, continues bool) (bool, []IndexOF) {
-	newtext := ChangeUnicode(input)
 	prof := NewProfanity(cache.English)
-	df, pr := prof.Censor(newtext)
+	df, pr := prof.Censor(input)
 	if !df {
 		return false, make([]IndexOF, 0)
 	}
-	in := strings.Index(newtext, pr)
+	in := strings.Index(input, pr)
 	return true, []IndexOF{
 		{
 			Swear: pr,

diff --git a/replace.go b/replace.go
@@ -9,6 +9,7 @@ import (
 
 func ChangeUnicode(unicode string) string {
 	//unicode = strings.ToLower(unicode)
+	unicode = RemoveDomain(unicode)
 	unicode = strings.ReplaceAll(unicode, "ㅿ", "ㅅ")
 	unicode = strings.ReplaceAll(unicode, "^", "ㅅ")
 	unicode = strings.ReplaceAll(unicode, "^", "ㅅ")
@@ -118,10 +119,10 @@ func ChangeUnicode(unicode string) string {
 	unicode = strings.ReplaceAll(unicode, "b", "ㅂ")
 	unicode = strings.ReplaceAll(unicode, "u", "ㅂ")
 	unicode = strings.ReplaceAll(unicode, "v", "ㅂ")
-	unicode = RemoveDomain(unicode)
 	unicode = RemoveDuplicate(unicode)
 	unicode = RemoveNumber(unicode)
 	unicode = RemoveEnglish(unicode)
+	unicode = After(unicode)
 	dism := gohangul.Disassemble(unicode)
 	unicode = gohangul.Assemble(dism)
 	unicode = PreGeneral(unicode)
@@ -130,7 +131,6 @@ func ChangeUnicode(unicode string) string {
 	unicode = PreBelittle(unicode)
 	unicode = PrePolitics(unicode)
 	unicode = RemoveJ(unicode)
-	unicode = After(unicode)
 	return unicode
 }
 
@@ -187,6 +187,7 @@ func RemoveDomain(input string) string {
 	input = strings.ReplaceAll(input, ".me", "")
 	input = strings.ReplaceAll(input, ".xyz", "")
 	input = strings.ReplaceAll(input, ".online", "")
+	input = strings.ReplaceAll(input, "?v=", "")
 	return input
 }
 
@@ -226,44 +227,32 @@ func RemoveJ(input string) string {
 }
 
 func RemoveEnglish(input string) string {
+	input = strings.ReplaceAll(input, "ha", "")
+	input = strings.ReplaceAll(input, "hi", "")
+	input = strings.ReplaceAll(input, "hag", "")
+	input = strings.ReplaceAll(input, "ham", "")
+	input = strings.ReplaceAll(input, "hat", "")
+	input = strings.ReplaceAll(input, "hem", "")
+	input = strings.ReplaceAll(input, "her", "")
+	input = strings.ReplaceAll(input, "hex", "")
+	input = strings.ReplaceAll(input, "hic", "")
+	input = strings.ReplaceAll(input, "him", "")
+	input = strings.ReplaceAll(input, "his", "")
+	input = strings.ReplaceAll(input, "hiv", "")
+	input = strings.ReplaceAll(input, "hmo", "")
+	input = strings.ReplaceAll(input, "hod", "")
+	input = strings.ReplaceAll(input, "hog", "")
+	input = strings.ReplaceAll(input, "hot", "")
+	input = strings.ReplaceAll(input, "hub", "")
+	input = strings.ReplaceAll(input, "hug", "")
+	input = strings.ReplaceAll(input, "hum", "")
+	input = strings.ReplaceAll(input, "ssh", "")
 	input = strings.ReplaceAll(input, "cat", "")
+	input = strings.ReplaceAll(input, "watch", "")
 	input = strings.ReplaceAll(input, "status", "")
 	return input
 }
 
-//func EnglishToKo(input string) string {
-//	var result strings.Builder
-//	runes := []rune(input)
-//	i := 0
-//	for i < len(runes) {
-//		ch := runes[i]
-//		if isEnglishLetter(ch) {
-//			if cache.Regexinitial.MatchString(string(ch)) {
-//				initial := cache.KRinitial[string(ch)]
-//				i++
-//				if i < len(runes) && cache.Regexmedial.MatchString(string(runes[i])) {
-//					medial := cache.KRmedial[string(runes[i])]
-//					i++
-//					if i < len(runes) && cache.Regexfinal.MatchString(string(runes[i])) {
-//						final := cache.KRfinal[string(runes[i])]
-//						result.WriteString(initial + medial + final)
-//						i++
-//					} else {
-//						result.WriteString(initial + medial)
-//					}
-//				} else {
-//					result.WriteString(string(runes[i-1]))
-//				}
-//			}
-//		} else {
-//			result.WriteString(string(ch))
-//			i++
-//		}
-//	}
-//
-//	return result.String()
-//}
-
 func KoToEnglish(input string) string {
 	dism := gohangul.Disassemble(input)
 	input = string(dism)

diff --git a/test/speed_test.go b/test/speed_test.go
@@ -15,5 +15,5 @@ func Benchmark0(b *testing.B) {
 }
 
 func Test0(t *testing.T) {
-	fmt.Println(korcen.Check("gpdmdmd"))
+	fmt.Println(korcen.Check(""))
 }
Original file line number	Diff line number	Diff line change
Expand Up		@@ -2,4 +2,4 @@ module github.com/fluffy-melli/korcen-go

		go 1.22.2

		require github.com/gyarang/gohangul v0.1.3 // indirect
		require github.com/gyarang/gohangul v0.1.3