unify Cut method, return channel instead of array

finalseg/finalseg.go

@@ -9,80 +9,86 @@ var (
 	reSkip = regexp.MustCompile(`(\d+\.\d+|[a-zA-Z0-9]+)`)
 )
 
-func cutHan(sentence string) []string {
+func cutHan(sentence string) chan string {
-	runes := []rune(sentence)
+	result := make(chan string)
-	result := make([]string, 0)
+	go func() {
-	_, pos_list := viterbi(runes, []byte{'B', 'M', 'E', 'S'})
+		runes := []rune(sentence)
-	begin, next := 0, 0
+		_, pos_list := viterbi(runes, []byte{'B', 'M', 'E', 'S'})
-	for i, char := range runes {
+		begin, next := 0, 0
-		pos := pos_list[i]
+		for i, char := range runes {
-		switch pos {
+			pos := pos_list[i]
-		case 'B':
+			switch pos {
-			begin = i
+			case 'B':
-		case 'E':
+				begin = i
-			result = append(result, string(runes[begin:i+1]))
+			case 'E':
-			next = i + 1
+				result <- string(runes[begin : i+1])
-		case 'S':
+				next = i + 1
-			result = append(result, string(char))
+			case 'S':
-			next = i + 1
+				result <- string(char)
+				next = i + 1
+			}
 		}
-	}
+		if next < len(runes) {
-	if next < len(runes) {
+			result <- string(runes[next:])
-		result = append(result, string(runes[next:]))
+		}
-	}
+		close(result)
+	}()
 	return result
 }
 
-func Cut(sentence string) []string {
+func Cut(sentence string) chan string {
-	result := make([]string, 0)
+	result := make(chan string)
 	s := sentence
 	var hans string
 	var hanLoc []int
 	var nonhanLoc []int
-	for {
+	go func() {
-		hanLoc = reHan.FindStringIndex(s)
+		for {
-		if hanLoc == nil {
+			hanLoc = reHan.FindStringIndex(s)
-			if len(s) == 0 {
+			if hanLoc == nil {
-				break
+				if len(s) == 0 {
-			}
+					break
-		} else if hanLoc[0] == 0 {
+				}
-			hans = s[hanLoc[0]:hanLoc[1]]
+			} else if hanLoc[0] == 0 {
-			s = s[hanLoc[1]:]
+				hans = s[hanLoc[0]:hanLoc[1]]
-			for _, han := range cutHan(hans) {
+				s = s[hanLoc[1]:]
-				result = append(result, han)
+				for han := range cutHan(hans) {
-			}
+					result <- han
-			continue
+				}
-		}
-		nonhanLoc = reSkip.FindStringIndex(s)
-		if nonhanLoc == nil {
-			if len(s) == 0 {
-				break
-			}
-		} else if nonhanLoc[0] == 0 {
-			nonhans := s[nonhanLoc[0]:nonhanLoc[1]]
-			s = s[nonhanLoc[1]:]
-			if nonhans != "" {
-				result = append(result, nonhans)
 				continue
 			}
-		}
+			nonhanLoc = reSkip.FindStringIndex(s)
-		var loc []int
+			if nonhanLoc == nil {
-		if hanLoc == nil && nonhanLoc == nil {
+				if len(s) == 0 {
-			if len(s) > 0 {
+					break
-				result = append(result, s)
+				}
-				break
+			} else if nonhanLoc[0] == 0 {
+				nonhans := s[nonhanLoc[0]:nonhanLoc[1]]
+				s = s[nonhanLoc[1]:]
+				if nonhans != "" {
+					result <- nonhans
+					continue
+				}
 			}
-		} else if hanLoc == nil {
+			var loc []int
-			loc = nonhanLoc
+			if hanLoc == nil && nonhanLoc == nil {
-		} else if nonhanLoc == nil {
+				if len(s) > 0 {
-			loc = hanLoc
+					result <- s
-		} else if hanLoc[0] < nonhanLoc[0] {
+					break
-			loc = hanLoc
+				}
-		} else {
+			} else if hanLoc == nil {
-			loc = nonhanLoc
+				loc = nonhanLoc
+			} else if nonhanLoc == nil {
+				loc = hanLoc
+			} else if hanLoc[0] < nonhanLoc[0] {
+				loc = hanLoc
+			} else {
+				loc = nonhanLoc
+			}
+			result <- s[:loc[0]]
+			s = s[loc[0]:]
 		}
-		result = append(result, s[:loc[0]])
+		close(result)
-		s = s[loc[0]:]
+	}()
-	}
 	return result
 }

finalseg/finalseg_test.go

@@ -5,6 +5,14 @@ import (
 	"testing"
 )
 
+func chanToArray(ch chan string) []string {
+	result := make([]string, 0)
+	for word := range ch {
+		result = append(result, word)
+	}
+	return result
+}
+
 func TestViterbi(t *testing.T) {
 	obs := "我们是程序员"
 	states := []byte{'B', 'M', 'E', 'S'}
@@ -21,7 +29,7 @@ func TestViterbi(t *testing.T) {
 
 func TestCutHan(t *testing.T) {
 	obs := "我们是程序员"
-	result := cutHan(obs)
+	result := chanToArray(cutHan(obs))
 	if len(result) != 3 {
 		t.Error(result)
 	}
@@ -38,7 +46,7 @@ func TestCutHan(t *testing.T) {
 
 func TestCut(t *testing.T) {
 	sentence := "我们是程序员"
-	result := Cut(sentence)
+	result := chanToArray(Cut(sentence))
 	if len(result) != 3 {
 		t.Error(len(result))
 	}
@@ -51,11 +59,11 @@ func TestCut(t *testing.T) {
 	if result[2] != "程序员" {
 		t.Error(result[2])
 	}
-	result2 := Cut("I'm a programmer!")
+	result2 := chanToArray(Cut("I'm a programmer!"))
 	if len(result2) != 8 {
 		t.Error(result2)
 	}
-	result3 := Cut("程序员average年龄28.6岁。")
+	result3 := chanToArray(Cut("程序员average年龄28.6岁。"))
 	if len(result3) != 6 {
 		t.Error(result3)
 	}

jieba.go

@@ -154,8 +154,7 @@ func cutDAG(sentence string) []string {
 				} else {
 					bufString := string(buf)
 					if v, ok := Trie.Freq[bufString]; !ok || v == 0.0 {
-						recognized := finalseg.Cut(bufString)
+						for t := range finalseg.Cut(bufString) {
-						for _, t := range recognized {
 							result = append(result, t)
 						}
 					} else {
@@ -177,8 +176,7 @@ func cutDAG(sentence string) []string {
 		} else {
 			bufString := string(buf)
 			if v, ok := Trie.Freq[bufString]; !ok || v == 0.0 {
-				recognized := finalseg.Cut(bufString)
+				for t := range finalseg.Cut(bufString) {
-				for _, t := range recognized {
 					result = append(result, t)
 				}
 			} else {