fix: 词组分组算法

lib/index.js

@@ -2,6 +2,7 @@
 
 const assign = require("object-assign");
 const PINYIN_DICT = require("../data/dict-zi");
+const util = require("./util");
 const Pinyin = require("./pinyin");
 let jieba;
 let PHRASES_DICT;
@@ -105,11 +106,9 @@ function groupPhrases(phrases) {
     return phrases[0];
   }
 
-  const grouped = phrases.reduce(function(phrase, pys) {
-    return phrase + pys[0];
-  }, "");
+  const grouped = util.combo(phrases);
 
-  return [grouped];
+  return grouped;
 }
 
 const pinyin = new NodePinyin(PINYIN_DICT);

lib/util.js

@@ -0,0 +1,56 @@
+
+
+/**
+ * 组合 2 个拼音数组。
+ * @param {Array<String>} a1 第一个数组，形如 ["zhāo", "cháo"]
+ * @param {Array<String>} a2 字符串型数组。形如 ["yáng"]
+ * @return {Array<String>} 组合后的一维数组，如上可得 ["zhāoyáng", "cháoyáng"]
+ */
+function combo2array(a1, a2) {
+  const result = [];
+  if (!a1.length) {
+    return a2;
+  }
+  if (!a2.length) {
+    return a1;
+  }
+  for (let i = 0, l = a1.length; i < l; i++) {
+    for (let j = 0, m = a2.length; j < m; j++) {
+      result.push(a1[i] + a2[j]);
+    }
+  }
+  return result;
+  // arr.forEach(arrItem => {
+  //   resultItem = '';
+  //   arrItem.forEach(item => {
+  //     resultItem += item;
+  //   });
+  //   result.push(resultItem);
+  // });
+}
+
+/**
+ * 合并二维元祖。
+ * @param {Array<Array<String>>} arr 二维元祖 [["zhāo", "cháo"], ["yáng"], ["dōng"], ["shēng"]]
+ * @return {Array<String>} 返回二维字符串组合数组。形如
+ *  [
+ *    ["zhāoyáng"], ["dōng"], ["shēng"],
+ *    ["cháoyáng"], ["dōng"], ["shēng"]
+ *  ]
+ */
+function combo(arr) {
+  if (arr.length === 0) {
+    return [];
+  }
+  if (arr.length === 1) {
+    return arr[0];
+  }
+  let result = combo2array(arr[0], arr[1]);
+  for (let i = 2, l = arr.length; i < l; i++) {
+    result = combo2array(result, arr[i]);
+  }
+  return result;
+}
+
+exports.combo2array = combo2array;
+exports.combo = combo;

tests/test.js

@@ -277,8 +277,8 @@ describe("pinyin group", function() {
   });
 
   it("groups segments with heteronyms", function() {
-    const han = "我都喜欢";
+    const han = "我都喜欢朝阳";
     const py = pinyin(han, {segment: true, group: true, heteronym: true});
-    expect(py).to.eql([["wǒ"], ["dū", "dōu"], ["xǐhuān"]]);
+    expect(py).to.eql([["wǒ"], ["dū", "dōu"], ["xǐhuān"], ["zhāoyáng", "cháoyáng"]]);
   });
 });

tests/util.test.js

@@ -0,0 +1,70 @@
+"use strict";
+
+const expect = require("expect.js");
+const util = require("../lib/util");
+
+describe("test/util.test.js", function() {
+  describe("combo2array", function() {
+    it("combo2array([], [])", function() {
+      expect(util.combo2array([], [])).to.eql([]);
+    });
+
+    it("combo2array([a], [])", function() {
+      expect(util.combo2array(["a"], [])).to.eql(["a"]);
+    });
+
+    it("combo2array([], [1])", function() {
+      expect(util.combo2array([], ["1"])).to.eql(["1"]);
+    });
+
+    it("combo2array([a], [1])", function() {
+      expect(util.combo2array(["a"], ["1"])).to.eql(["a1"]);
+    });
+
+    it("combo2array([a,b], [1])", function() {
+      expect(util.combo2array(["a", "b"], ["1"])).to.eql(["a1", "b1"]);
+    });
+
+    it("combo2array([a], [1,2])", function() {
+      expect(util.combo2array(["a"], ["1", "2"])).to.eql(["a1", "a2"]);
+    });
+
+    it("combo2array([a,b], [1,2])", function() {
+      expect(util.combo2array(["a", "b"], ["1", "2"])).to.eql(["a1", "a2", "b1", "b2"]);
+    });
+
+    it("combo2array([a,b,c], [1,2,3])", function() {
+      expect(util.combo2array(["a", "b", "c"], ["1", "2", "3"])).to.eql(["a1", "a2", "a3", "b1", "b2", "b3", "c1", "c2", "c3"]);
+    });
+  });
+
+  describe("combo", function() {
+    it("combo([])", function() {
+      expect(util.combo([])).to.eql([]);
+    });
+
+    it("combo([[a]])", function() {
+      expect(util.combo([["a"]])).to.eql([["a"]]);
+    });
+
+    it("combo([[a,b]])", function() {
+      expect(util.combo([["a", "b"]])).to.eql([["a", "b"]]);
+    });
+
+    it("combo([[a,b],[1]])", function() {
+      expect(util.combo([["a", "b"], ["1"]])).to.eql([["a1", "b1"]]);
+    });
+
+    it("combo([[a,b],[1,2]])", function() {
+      expect(util.combo([["a", "b"], ["1", "2"]])).to.eql([["a1", "a2", "b1", "b2"]]);
+    });
+
+    it("combo([[a,b],[1,2],[A]])", function() {
+      expect(util.combo([["a", "b"], ["1", "2"], ["A"]])).to.eql([["a1A", "a2A", "b1A", "b2A"]]);
+    });
+
+    it("combo([[a,b],[1,2],[A,B]])", function() {
+      expect(util.combo([["a", "b"], ["1", "2"], ["A", "B"]])).to.eql([["a1A", "a1B", "a2A", "a2B", "b1A", "b1B", "b2A", "b2B"]]);
+    });
+  });
+});