Merge pull request #11 from go-ego/m-pr

add not cut option and export match function code

phrase/paragraph.go

@@ -41,8 +41,15 @@ func Pinyin(p string, segs ...gse.Segmenter) []string {
 // Initial return pinyin initial
 func Initial(p string, segs ...gse.Segmenter) (s string) {
 	a := Pinyin(p, segs...)
+	return Join(a)
+}
+
+// Join []string to string
+func Join(a []string) (s string) {
 	for _, v := range a {
-		s += string([]rune(v)[0])
+		if len(v) > 0 {
+			s += string([]rune(v)[0])
+		}
 	}
 
 	return

phrase/paragraph_test.go

@@ -42,10 +42,22 @@ func TestPinyin(t *testing.T) {
 	seg := gse.New("zh, ../examples/dict.txt")
 	WithGse(seg)
 
+	text := "西雅图都会区, 西雅图太空针"
+
 	AddDict("都会区", "dū huì qū")
-	p := Pinyin("西雅图都会区, 西雅图太空针")
+	p := Pinyin(text)
 	tt.Equal(t, "[xi ya tu du hui qu, xi ya tu tai kong zhen]", p)
 
 	i := Initial("都会区")
 	tt.Equal(t, "dhq", i)
+
+	Cut = false
+	s := seg.Trim(seg.CutAll(text))
+	i += ", "
+	for _, v := range s {
+		i1 := Initial(v)
+		i += i1 + " "
+	}
+	tt.Equal(t, "dhq, xyt dhq xyt tk z ", i)
+	Cut = true
 }

phrase/phrase.go

@@ -10,6 +10,9 @@ import (
 var (
 	seg    gse.Segmenter
 	loaded bool
+
+	// Cut set pinyinPhrase cut
+	Cut = true
 )
 
 // LoadGseDict load the user's gse dict
@@ -42,19 +45,36 @@ func cutWords(s string, segs ...gse.Segmenter) []string {
 	return seg.CutAll(s)
 }
 
+// Match match word pinyin
+func Match(word string) string {
+	match := phraseDict[word]
+	if match == "" {
+		match = DictAdd[word]
+	}
+
+	match = gpy.ToFixed(match, Option)
+	return match
+}
+
+func matchs(s, word string) string {
+	match := Match(word)
+	if match != "" {
+		s = strings.Replace(s, word, " "+match+" ", 1)
+	}
+
+	return s
+}
+
 func pinyinPhrase(s string, segs ...gse.Segmenter) string {
-	words := cutWords(s, segs...)
-	for _, word := range words {
-		match := phraseDict[word]
-		if match == "" {
-			match = DictAdd[word]
+	if Cut {
+		words := cutWords(s, segs...)
+		for _, word := range words {
+			s = matchs(s, word)
 		}
 
-		match = gpy.ToFixed(match, Option)
-		if match != "" {
-			s = strings.Replace(s, word, " "+match+" ", 1)
-		}
+		return s
 	}
 
+	s = matchs(s, s)
 	return s
 }