unify Cut method, return channel instead of array

jieba.go

@@ -127,16 +127,17 @@ func Calc(sentence string, dag map[int][]int) map[int]*Route {
 	return routes
 }
 
-type cutFunc func(sentence string) []string
+type cutFunc func(sentence string) chan string
 
-func cutDAG(sentence string) []string {
+func cutDAG(sentence string) chan string {
+	result := make(chan string)
+	go func() {
 		dag := DAG(sentence)
 		routes := Calc(sentence, dag)
 		x := 0
 		var y int
 		runes := []rune(sentence)
 		length := len(runes)
-	result := make([]string, 0)
 		buf := make([]rune, 0)
 		for {
 			if x >= length {
@@ -149,49 +150,52 @@ func cutDAG(sentence string) []string {
 			} else {
 				if len(buf) > 0 {
 					if len(buf) == 1 {
-					result = append(result, string(buf))
+						result <- string(buf)
 						buf = make([]rune, 0)
 					} else {
 						bufString := string(buf)
 						if v, ok := Trie.Freq[bufString]; !ok || v == 0.0 {
-						for t := range finalseg.Cut(bufString) {
+							for x := range finalseg.Cut(bufString) {
-							result = append(result, t)
+								result <- x
 							}
 						} else {
 							for _, elem := range buf {
-							result = append(result, string(elem)) // TODO: I don't get this?
+								result <- string(elem) // TODO: I don't get this?
 							}
 						}
 						buf = make([]rune, 0)
 					}
 				}
-			result = append(result, string(l_word))
+				result <- string(l_word)
 			}
 			x = y
 		}
 
 		if len(buf) > 0 {
 			if len(buf) == 1 {
-			result = append(result, string(buf))
+				result <- string(buf)
 			} else {
 				bufString := string(buf)
 				if v, ok := Trie.Freq[bufString]; !ok || v == 0.0 {
 					for t := range finalseg.Cut(bufString) {
-					result = append(result, t)
+						result <- t
 					}
 				} else {
 					for _, elem := range buf {
-					result = append(result, string(elem)) // TODO: I don't get this?
+						result <- string(elem) // TODO: I don't get this?
 					}
 				}
 			}
 		}
+		close(result)
+	}()
 	return result
 }
 
-func cutDAGNoHMM(sentence string) []string {
+func cutDAGNoHMM(sentence string) chan string {
-	result := make([]string, 0)
+	result := make(chan string)
 
+	go func() {
 		dag := DAG(sentence)
 		routes := Calc(sentence, dag)
 		x := 0
@@ -210,22 +214,26 @@ func cutDAGNoHMM(sentence string) []string {
 				x = y
 			} else {
 				if len(buf) > 0 {
-				result = append(result, string(buf))
+					result <- string(buf)
 					buf = make([]rune, 0)
 				}
-			result = append(result, string(l_word))
+				result <- string(l_word)
 				x = y
 			}
 		}
 		if len(buf) > 0 {
-		result = append(result, string(buf))
+			result <- string(buf)
 			buf = make([]rune, 0)
 		}
+		close(result)
+	}()
 	return result
 }
 
-func cutAll(sentence string) []string {
+func cutAll(sentence string) chan string {
-	result := make([]string, 0)
+	result := make(chan string)
+
+	go func() {
 		runes := []rune(sentence)
 		dag := DAG(sentence)
 		old_j := -1
@@ -237,17 +245,19 @@ func cutAll(sentence string) []string {
 		for k := range ks {
 			l := dag[k]
 			if len(l) == 1 && k > old_j {
-			result = append(result, string(runes[k:l[0]+1]))
+				result <- string(runes[k : l[0]+1])
 				old_j = l[0]
 			} else {
 				for _, j := range l {
 					if j > k {
-					result = append(result, string(runes[k:j+1]))
+						result <- string(runes[k : j+1])
 						old_j = j
 					}
 				}
 			}
 		}
+		close(result)
+	}()
 	return result
 }
 
@@ -277,8 +287,8 @@ func Cut(sentence string, isCutAll bool, HMM bool) chan string {
 				continue
 			}
 			if reHan.MatchString(blk) {
-				for _, word := range cut(blk) {
+				for x := range cut(blk) {
-					result <- word
+					result <- x
 				}
 			} else {
 				type skipSplitFunc func(sentence string) []string

jieba_test.go

@@ -621,15 +621,23 @@ func init() {
 	SetDictionary("dict.txt")
 }
 
+func chanToArray(ch chan string) []string {
+	result := make([]string, 0)
+	for word := range ch {
+		result = append(result, word)
+	}
+	return result
+}
+
 func TestCutDAG(t *testing.T) {
-	result := cutDAG("BP神经网络如何训练才能在分类时增加区分度？")
+	result := chanToArray(cutDAG("BP神经网络如何训练才能在分类时增加区分度？"))
 	if len(result) != 11 {
 		t.Error(result)
 	}
 }
 
 func TestCutDAGNoHmm(t *testing.T) {
-	result := cutDAGNoHMM("BP神经网络如何训练才能在分类时增加区分度？")
+	result := chanToArray(cutDAGNoHMM("BP神经网络如何训练才能在分类时增加区分度？"))
 	if len(result) != 11 {
 		t.Error(result)
 	}
@@ -648,14 +656,6 @@ func TestRegexpSplit(t *testing.T) {
 	}
 }
 
-func chanToArray(ch chan string) []string {
-	result := make([]string, 0)
-	for word := range ch {
-		result = append(result, word)
-	}
-	return result
-}
-
 func TestDefaultCut(t *testing.T) {
 	var result []string
 	for index, content := range test_contents {