Merge jieba into master

.gitignore

@@ -28,3 +28,5 @@ Thumb.db
 dist/
 npm-debug.log
 spm_modules/
+tools/dict/error.log
+coverage/

package.json

@@ -1,8 +1,11 @@
 {
   "name": "pinyin",
-  "version": "2.3.3",
+  "version": "2.4.0",
   "description": "汉语拼音转换工具。",
-  "keywords": ["拼音", "Pinyin"],
+  "keywords": [
+    "拼音",
+    "Pinyin"
+  ],
   "homepage": "http://pinyin.hotoo.me/",
   "author": "闲耘 <hotoo.cn@gmail.com>",
   "main": "index.js",
@@ -24,8 +27,8 @@
     "test": "make test"
   },
   "dependencies": {
-    "segment": "0.0.5",
-    "commander": "~1.1.1"
+    "commander": "~1.1.1",
+    "nodejieba": "^0.2.11"
   },
   "devDependencies": {
     "mocha": "1.17.1",

src/pinyin.js

@@ -3,7 +3,7 @@ var isNode = typeof process === "object" &&
   process.toString() === "[object process]";
 
 // 分词模块
-var Segment;
+var jieba;
 var PHRASES_DICT;
 var PINYIN_DICT;
 
@@ -31,11 +31,11 @@ function buildPinyinCache(dict_combo){
 }
 
 if(isNode){
-  Segment = module["require"]("segment").Segment;
-  var segment = new Segment();
-  // 使用默认的识别模块及字典
-  segment.useDefault();
-
+  jieba = module['require']('nodejieba');
+  jieba.loadDict(
+      "./node_modules/nodejieba/dict/jieba.dict.utf8",
+      "./node_modules/nodejieba/dict/hmm_model.utf8",
+      "./node_modules/nodejieba/dict/user.dict.utf8");
 
   // 词语拼音库。
   PHRASES_DICT = module["require"]("./phrases-dict");
@@ -192,13 +192,13 @@ function pinyin(hans, options){
 
   options = extend(DEFAULT_OPTIONS, options || {});
 
-  var phrases = isNode ? segment.doSegment(hans) : hans;
+  var phrases = isNode ? jieba.cutSync(hans) : hans;
   var len = hans.length;
   var pys = [];
 
   for(var i=0,nohans="",firstCharCode,words,l=phrases.length; i<l; i++){
 
-    words = isNode ? phrases[i].w : phrases[i];
+    words = phrases[i];
     firstCharCode = words.charCodeAt(0);
 
     if(PINYIN_DICT[firstCharCode]){

tests/test.js

@@ -87,11 +87,11 @@ var cases = [
     //STYLE_FIRST_LETTER: [["a a"]]
   //} ],
   [ "一 一", {
-    STYLE_NORMAL:       [["yi"],["yi"]],
-    STYLE_TONE:         [["yī"],["yī"]],
-    STYLE_TONE2:        [["yi1"],["yi1"]],
-    STYLE_INITIALS:     [["y"],["y"]],
-    STYLE_FIRST_LETTER: [["y"],["y"]]
+    STYLE_NORMAL:       [["yi"],[' '],["yi"]],
+    STYLE_TONE:         [["yī"],[' '],["yī"]],
+    STYLE_TONE2:        [["yi1"],[' '],["yi1"]],
+    STYLE_INITIALS:     [["y"],[' '],["y"]],
+    STYLE_FIRST_LETTER: [["y"],[' '],["y"]]
   } ],
 
   // 中英混合
@@ -110,7 +110,16 @@ var cases = [
     STYLE_TONE2:        [["zhong1"],["guo2"],["(china)"]],
     STYLE_INITIALS:     [["zh"],["g"],["(china)"]],
     STYLE_FIRST_LETTER: [["z"],["g"],["(china)"]]
-  } ]
+  } ],
+
+  // 中英混合，多音字，单音词。
+  [ "0套价", {
+    STYLE_NORMAL:       [["0"],["tao"],["jia","jie"]],
+    STYLE_TONE:         [["0"],["tào"],["jià","jiè","jie"]],
+    STYLE_TONE2:        [["0"],["tao4"],["jia4","jie4","jie"]],
+    STYLE_INITIALS:     [["0"],["t"],["j"]],
+    STYLE_FIRST_LETTER: [["0"],["t"],["j"]]
+  } ],
 ];
 
 describe('pinyin', function() {
@@ -189,5 +198,3 @@ function deepEquals(a, b){
     break;
   }
 }
-
-console.log("DEEP EQUALS:", deepEquals([["zhong"],["guo"]] , [["zhong"],["guo"]]))