Android 汉子转换成拼音 

  1 package com.example.test;
  2 
  3 import android.text.TextUtils;
  4 import android.util.Log;
  5 
  6 import java.text.Collator;
  7 import java.util.ArrayList;
  8 import java.util.Locale;
  9 
 10 /**
 11  * An object to convert Chinese character to its corresponding pinyin string.
 12  * For characters with multiple possible pinyin string, only one is selected
 13  * according to collator. Polyphone is not supported in this implementation.
 14  * This class is implemented to achieve the best runtime performance and minimum
 15  * runtime resources with tolerable sacrifice of accuracy. This implementation
 16  * highly depends on zh_CN ICU collation data and must be always synchronized
 17  * with ICU.
 18  * 
 19  * Currently this file is aligned to zh.txt in ICU 4.6 来自android4.2源码
 20  */
 21 public class HanziToPinyin {
 22     private static final String TAG = "HanziToPinyin";
 23 
 24     // Turn on this flag when we want to check internal data structure.
 25     private static final boolean DEBUG = false;
 26 
 27     /**
 28      * Unihans array.
 29      * 
 30      * Each unihans is the first one within same pinyin when collator is zh_CN.
 31      */
 32     public static final char[] UNIHANS = { '\u963f', '\u54ce', '\u5b89',
 33             '\u80ae', '\u51f9', '\u516b', '\u6300', '\u6273', '\u90a6',
 34             '\u52f9', '\u9642', '\u5954', '\u4f3b', '\u5c44', '\u8fb9',
 35             '\u706c', '\u618b', '\u6c43', '\u51ab', '\u7676', '\u5cec',
 36             '\u5693', '\u5072', '\u53c2', '\u4ed3', '\u64a1', '\u518a',
 37             '\u5d7e', '\u66fd', '\u66fe', '\u5c64', '\u53c9', '\u8286',
 38             '\u8fbf', '\u4f25', '\u6284', '\u8f66', '\u62bb', '\u6c88',
 39             '\u6c89', '\u9637', '\u5403', '\u5145', '\u62bd', '\u51fa',
 40             '\u6b3b', '\u63e3', '\u5ddb', '\u5205', '\u5439', '\u65fe',
 41             '\u9034', '\u5472', '\u5306', '\u51d1', '\u7c97', '\u6c46',
 42             '\u5d14', '\u90a8', '\u6413', '\u5491', '\u5446', '\u4e39',
 43             '\u5f53', '\u5200', '\u561a', '\u6265', '\u706f', '\u6c10',
 44             '\u55f2', '\u7538', '\u5201', '\u7239', '\u4e01', '\u4e1f',
 45             '\u4e1c', '\u543a', '\u53be', '\u8011', '\u8968', '\u5428',
 46             '\u591a', '\u59b8', '\u8bf6', '\u5940', '\u97a5', '\u513f',
 47             '\u53d1', '\u5e06', '\u531a', '\u98de', '\u5206', '\u4e30',
 48             '\u8985', '\u4ecf', '\u7d11', '\u4f15', '\u65ee', '\u4f85',
 49             '\u7518', '\u5188', '\u768b', '\u6208', '\u7ed9', '\u6839',
 50             '\u522f', '\u5de5', '\u52fe', '\u4f30', '\u74dc', '\u4e56',
 51             '\u5173', '\u5149', '\u5f52', '\u4e28', '\u5459', '\u54c8',
 52             '\u548d', '\u4f44', '\u592f', '\u8320', '\u8bc3', '\u9ed2',
 53             '\u62eb', '\u4ea8', '\u5677', '\u53ff', '\u9f41', '\u4e6f',
 54             '\u82b1', '\u6000', '\u72bf', '\u5ddf', '\u7070', '\u660f',
 55             '\u5419', '\u4e0c', '\u52a0', '\u620b', '\u6c5f', '\u827d',
 56             '\u9636', '\u5dfe', '\u5755', '\u5182', '\u4e29', '\u51e5',
 57             '\u59e2', '\u5658', '\u519b', '\u5494', '\u5f00', '\u520a',
 58             '\u5ffc', '\u5c3b', '\u533c', '\u808e', '\u52a5', '\u7a7a',
 59             '\u62a0', '\u625d', '\u5938', '\u84af', '\u5bbd', '\u5321',
 60             '\u4e8f', '\u5764', '\u6269', '\u5783', '\u6765', '\u5170',
 61             '\u5577', '\u635e', '\u808b', '\u52d2', '\u5d1a', '\u5215',
 62             '\u4fe9', '\u5941', '\u826f', '\u64a9', '\u5217', '\u62ce',
 63             '\u5222', '\u6e9c', '\u56d6', '\u9f99', '\u779c', '\u565c',
 64             '\u5a08', '\u7567', '\u62a1', '\u7f57', '\u5463', '\u5988',
 65             '\u57cb', '\u5ada', '\u7264', '\u732b', '\u4e48', '\u5445',
 66             '\u95e8', '\u753f', '\u54aa', '\u5b80', '\u55b5', '\u4e5c',
 67             '\u6c11', '\u540d', '\u8c2c', '\u6478', '\u54de', '\u6bea',
 68             '\u55ef', '\u62cf', '\u8149', '\u56e1', '\u56d4', '\u5b6c',
 69             '\u7592', '\u5a1e', '\u6041', '\u80fd', '\u59ae', '\u62c8',
 70             '\u5b22', '\u9e1f', '\u634f', '\u56dc', '\u5b81', '\u599e',
 71             '\u519c', '\u7fba', '\u5974', '\u597b', '\u759f', '\u9ec1',
 72             '\u90cd', '\u5594', '\u8bb4', '\u5991', '\u62cd', '\u7705',
 73             '\u4e53', '\u629b', '\u5478', '\u55b7', '\u5309', '\u4e15',
 74             '\u56e8', '\u527d', '\u6c15', '\u59d8', '\u4e52', '\u948b',
 75             '\u5256', '\u4ec6', '\u4e03', '\u6390', '\u5343', '\u545b',
 76             '\u6084', '\u767f', '\u4eb2', '\u72c5', '\u828e', '\u4e18',
 77             '\u533a', '\u5cd1', '\u7f3a', '\u590b', '\u5465', '\u7a63',
 78             '\u5a06', '\u60f9', '\u4eba', '\u6254', '\u65e5', '\u8338',
 79             '\u53b9', '\u909a', '\u633c', '\u5827', '\u5a51', '\u77a4',
 80             '\u637c', '\u4ee8', '\u6be2', '\u4e09', '\u6852', '\u63bb',
 81             '\u95aa', '\u68ee', '\u50e7', '\u6740', '\u7b5b', '\u5c71',
 82             '\u4f24', '\u5f30', '\u5962', '\u7533', '\u8398', '\u6552',
 83             '\u5347', '\u5c38', '\u53ce', '\u4e66', '\u5237', '\u8870',
 84             '\u95e9', '\u53cc', '\u8c01', '\u542e', '\u8bf4', '\u53b6',
 85             '\u5fea', '\u635c', '\u82cf', '\u72fb', '\u590a', '\u5b59',
 86             '\u5506', '\u4ed6', '\u56fc', '\u574d', '\u6c64', '\u5932',
 87             '\u5fd1', '\u71a5', '\u5254', '\u5929', '\u65eb', '\u5e16',
 88             '\u5385', '\u56f2', '\u5077', '\u51f8', '\u6e4d', '\u63a8',
 89             '\u541e', '\u4e47', '\u7a75', '\u6b6a', '\u5f2f', '\u5c23',
 90             '\u5371', '\u6637', '\u7fc1', '\u631d', '\u4e4c', '\u5915',
 91             '\u8672', '\u4eda', '\u4e61', '\u7071', '\u4e9b', '\u5fc3',
 92             '\u661f', '\u51f6', '\u4f11', '\u5401', '\u5405', '\u524a',
 93             '\u5743', '\u4e2b', '\u6079', '\u592e', '\u5e7a', '\u503b',
 94             '\u4e00', '\u56d9', '\u5e94', '\u54df', '\u4f63', '\u4f18',
 95             '\u625c', '\u56e6', '\u66f0', '\u6655', '\u7b60', '\u7b7c',
 96             '\u5e00', '\u707d', '\u5142', '\u5328', '\u50ae', '\u5219',
 97             '\u8d3c', '\u600e', '\u5897', '\u624e', '\u635a', '\u6cbe',
 98             '\u5f20', '\u957f', '\u9577', '\u4f4b', '\u8707', '\u8d1e',
 99             '\u4e89', '\u4e4b', '\u5cd9', '\u5ea2', '\u4e2d', '\u5dde',
100             '\u6731', '\u6293', '\u62fd', '\u4e13', '\u5986', '\u96b9',
101             '\u5b92', '\u5353', '\u4e72', '\u5b97', '\u90b9', '\u79df',
102             '\u94bb', '\u539c', '\u5c0a', '\u6628', '\u5159', '\u9fc3',
103             '\u9fc4', };
104 
105     /**
106      * Pinyin array.
107      * 
108      * Each pinyin is corresponding to unihans of same offset in the unihans
109      * array.
110      */
111     public static final byte[][] PINYINS = { { 65, 0, 0, 0, 0, 0 },
112             { 65, 73, 0, 0, 0, 0 }, { 65, 78, 0, 0, 0, 0 },
113             { 65, 78, 71, 0, 0, 0 }, { 65, 79, 0, 0, 0, 0 },
114             { 66, 65, 0, 0, 0, 0 }, { 66, 65, 73, 0, 0, 0 },
115             { 66, 65, 78, 0, 0, 0 }, { 66, 65, 78, 71, 0, 0 },
116             { 66, 65, 79, 0, 0, 0 }, { 66, 69, 73, 0, 0, 0 },
117             { 66, 69, 78, 0, 0, 0 }, { 66, 69, 78, 71, 0, 0 },
118             { 66, 73, 0, 0, 0, 0 }, { 66, 73, 65, 78, 0, 0 },
119             { 66, 73, 65, 79, 0, 0 }, { 66, 73, 69, 0, 0, 0 },
120             { 66, 73, 78, 0, 0, 0 }, { 66, 73, 78, 71, 0, 0 },
121             { 66, 79, 0, 0, 0, 0 }, { 66, 85, 0, 0, 0, 0 },
122             { 67, 65, 0, 0, 0, 0 }, { 67, 65, 73, 0, 0, 0 },
123             { 67, 65, 78, 0, 0, 0 }, { 67, 65, 78, 71, 0, 0 },
124             { 67, 65, 79, 0, 0, 0 }, { 67, 69, 0, 0, 0, 0 },
125             { 67, 69, 78, 0, 0, 0 }, { 67, 69, 78, 71, 0, 0 },
126             { 90, 69, 78, 71, 0, 0 }, { 67, 69, 78, 71, 0, 0 },
127             { 67, 72, 65, 0, 0, 0 }, { 67, 72, 65, 73, 0, 0 },
128             { 67, 72, 65, 78, 0, 0 }, { 67, 72, 65, 78, 71, 0 },
129             { 67, 72, 65, 79, 0, 0 }, { 67, 72, 69, 0, 0, 0 },
130             { 67, 72, 69, 78, 0, 0 }, { 83, 72, 69, 78, 0, 0 },
131             { 67, 72, 69, 78, 0, 0 }, { 67, 72, 69, 78, 71, 0 },
132             { 67, 72, 73, 0, 0, 0 }, { 67, 72, 79, 78, 71, 0 },
133             { 67, 72, 79, 85, 0, 0 }, { 67, 72, 85, 0, 0, 0 },
134             { 67, 72, 85, 65, 0, 0 }, { 67, 72, 85, 65, 73, 0 },
135             { 67, 72, 85, 65, 78, 0 }, { 67, 72, 85, 65, 78, 71 },
136             { 67, 72, 85, 73, 0, 0 }, { 67, 72, 85, 78, 0, 0 },
137             { 67, 72, 85, 79, 0, 0 }, { 67, 73, 0, 0, 0, 0 },
138             { 67, 79, 78, 71, 0, 0 }, { 67, 79, 85, 0, 0, 0 },
139             { 67, 85, 0, 0, 0, 0 }, { 67, 85, 65, 78, 0, 0 },
140             { 67, 85, 73, 0, 0, 0 }, { 67, 85, 78, 0, 0, 0 },
141             { 67, 85, 79, 0, 0, 0 }, { 68, 65, 0, 0, 0, 0 },
142             { 68, 65, 73, 0, 0, 0 }, { 68, 65, 78, 0, 0, 0 },
143             { 68, 65, 78, 71, 0, 0 }, { 68, 65, 79, 0, 0, 0 },
144             { 68, 69, 0, 0, 0, 0 }, { 68, 69, 78, 0, 0, 0 },
145             { 68, 69, 78, 71, 0, 0 }, { 68, 73, 0, 0, 0, 0 },
146             { 68, 73, 65, 0, 0, 0 }, { 68, 73, 65, 78, 0, 0 },
147             { 68, 73, 65, 79, 0, 0 }, { 68, 73, 69, 0, 0, 0 },
148             { 68, 73, 78, 71, 0, 0 }, { 68, 73, 85, 0, 0, 0 },
149             { 68, 79, 78, 71, 0, 0 }, { 68, 79, 85, 0, 0, 0 },
150             { 68, 85, 0, 0, 0, 0 }, { 68, 85, 65, 78, 0, 0 },
151             { 68, 85, 73, 0, 0, 0 }, { 68, 85, 78, 0, 0, 0 },
152             { 68, 85, 79, 0, 0, 0 }, { 69, 0, 0, 0, 0, 0 },
153             { 69, 73, 0, 0, 0, 0 }, { 69, 78, 0, 0, 0, 0 },
154             { 69, 78, 71, 0, 0, 0 }, { 69, 82, 0, 0, 0, 0 },
155             { 70, 65, 0, 0, 0, 0 }, { 70, 65, 78, 0, 0, 0 },
156             { 70, 65, 78, 71, 0, 0 }, { 70, 69, 73, 0, 0, 0 },
157             { 70, 69, 78, 0, 0, 0 }, { 70, 69, 78, 71, 0, 0 },
158             { 70, 73, 65, 79, 0, 0 }, { 70, 79, 0, 0, 0, 0 },
159             { 70, 79, 85, 0, 0, 0 }, { 70, 85, 0, 0, 0, 0 },
160             { 71, 65, 0, 0, 0, 0 }, { 71, 65, 73, 0, 0, 0 },
161             { 71, 65, 78, 0, 0, 0 }, { 71, 65, 78, 71, 0, 0 },
162             { 71, 65, 79, 0, 0, 0 }, { 71, 69, 0, 0, 0, 0 },
163             { 71, 69, 73, 0, 0, 0 }, { 71, 69, 78, 0, 0, 0 },
164             { 71, 69, 78, 71, 0, 0 }, { 71, 79, 78, 71, 0, 0 },
165             { 71, 79, 85, 0, 0, 0 }, { 71, 85, 0, 0, 0, 0 },
166             { 71, 85, 65, 0, 0, 0 }, { 71, 85, 65, 73, 0, 0 },
167             { 71, 85, 65, 78, 0, 0 }, { 71, 85, 65, 78, 71, 0 },
168             { 71, 85, 73, 0, 0, 0 }, { 71, 85, 78, 0, 0, 0 },
169             { 71, 85, 79, 0, 0, 0 }, { 72, 65, 0, 0, 0, 0 },
170             { 72, 65, 73, 0, 0, 0 }, { 72, 65, 78, 0, 0, 0 },
171             { 72, 65, 78, 71, 0, 0 }, { 72, 65, 79, 0, 0, 0 },
172             { 72, 69, 0, 0, 0, 0 }, { 72, 69, 73, 0, 0, 0 },
173             { 72, 69, 78, 0, 0, 0 }, { 72, 69, 78, 71, 0, 0 },
174             { 72, 77, 0, 0, 0, 0 }, { 72, 79, 78, 71, 0, 0 },
175             { 72, 79, 85, 0, 0, 0 }, { 72, 85, 0, 0, 0, 0 },
176             { 72, 85, 65, 0, 0, 0 }, { 72, 85, 65, 73, 0, 0 },
177             { 72, 85, 65, 78, 0, 0 }, { 72, 85, 65, 78, 71, 0 },
178             { 72, 85, 73, 0, 0, 0 }, { 72, 85, 78, 0, 0, 0 },
179             { 72, 85, 79, 0, 0, 0 }, { 74, 73, 0, 0, 0, 0 },
180             { 74, 73, 65, 0, 0, 0 }, { 74, 73, 65, 78, 0, 0 },
181             { 74, 73, 65, 78, 71, 0 }, { 74, 73, 65, 79, 0, 0 },
182             { 74, 73, 69, 0, 0, 0 }, { 74, 73, 78, 0, 0, 0 },
183             { 74, 73, 78, 71, 0, 0 }, { 74, 73, 79, 78, 71, 0 },
184             { 74, 73, 85, 0, 0, 0 }, { 74, 85, 0, 0, 0, 0 },
185             { 74, 85, 65, 78, 0, 0 }, { 74, 85, 69, 0, 0, 0 },
186             { 74, 85, 78, 0, 0, 0 }, { 75, 65, 0, 0, 0, 0 },
187             { 75, 65, 73, 0, 0, 0 }, { 75, 65, 78, 0, 0, 0 },
188             { 75, 65, 78, 71, 0, 0 }, { 75, 65, 79, 0, 0, 0 },
189             { 75, 69, 0, 0, 0, 0 }, { 75, 69, 78, 0, 0, 0 },
190             { 75, 69, 78, 71, 0, 0 }, { 75, 79, 78, 71, 0, 0 },
191             { 75, 79, 85, 0, 0, 0 }, { 75, 85, 0, 0, 0, 0 },
192             { 75, 85, 65, 0, 0, 0 }, { 75, 85, 65, 73, 0, 0 },
193             { 75, 85, 65, 78, 0, 0 }, { 75, 85, 65, 78, 71, 0 },
194             { 75, 85, 73, 0, 0, 0 }, { 75, 85, 78, 0, 0, 0 },
195             { 75, 85, 79, 0, 0, 0 }, { 76, 65, 0, 0, 0, 0 },
196             { 76, 65, 73, 0, 0, 0 }, { 76, 65, 78, 0, 0, 0 },
197             { 76, 65, 78, 71, 0, 0 }, { 76, 65, 79, 0, 0, 0 },
198             { 76, 69, 0, 0, 0, 0 }, { 76, 69, 73, 0, 0, 0 },
199             { 76, 69, 78, 71, 0, 0 }, { 76, 73, 0, 0, 0, 0 },
200             { 76, 73, 65, 0, 0, 0 }, { 76, 73, 65, 78, 0, 0 },
201             { 76, 73, 65, 78, 71, 0 }, { 76, 73, 65, 79, 0, 0 },
202             { 76, 73, 69, 0, 0, 0 }, { 76, 73, 78, 0, 0, 0 },
203             { 76, 73, 78, 71, 0, 0 }, { 76, 73, 85, 0, 0, 0 },
204             { 76, 79, 0, 0, 0, 0 }, { 76, 79, 78, 71, 0, 0 },
205             { 76, 79, 85, 0, 0, 0 }, { 76, 85, 0, 0, 0, 0 },
206             { 76, 85, 65, 78, 0, 0 }, { 76, 85, 69, 0, 0, 0 },
207             { 76, 85, 78, 0, 0, 0 }, { 76, 85, 79, 0, 0, 0 },
208             { 77, 0, 0, 0, 0, 0 }, { 77, 65, 0, 0, 0, 0 },
209             { 77, 65, 73, 0, 0, 0 }, { 77, 65, 78, 0, 0, 0 },
210             { 77, 65, 78, 71, 0, 0 }, { 77, 65, 79, 0, 0, 0 },
211             { 77, 69, 0, 0, 0, 0 }, { 77, 69, 73, 0, 0, 0 },
212             { 77, 69, 78, 0, 0, 0 }, { 77, 69, 78, 71, 0, 0 },
213             { 77, 73, 0, 0, 0, 0 }, { 77, 73, 65, 78, 0, 0 },
214             { 77, 73, 65, 79, 0, 0 }, { 77, 73, 69, 0, 0, 0 },
215             { 77, 73, 78, 0, 0, 0 }, { 77, 73, 78, 71, 0, 0 },
216             { 77, 73, 85, 0, 0, 0 }, { 77, 79, 0, 0, 0, 0 },
217             { 77, 79, 85, 0, 0, 0 }, { 77, 85, 0, 0, 0, 0 },
218             { 78, 0, 0, 0, 0, 0 }, { 78, 65, 0, 0, 0, 0 },
219             { 78, 65, 73, 0, 0, 0 }, { 78, 65, 78, 0, 0, 0 },
220             { 78, 65, 78, 71, 0, 0 }, { 78, 65, 79, 0, 0, 0 },
221             { 78, 69, 0, 0, 0, 0 }, { 78, 69, 73, 0, 0, 0 },
222             { 78, 69, 78, 0, 0, 0 }, { 78, 69, 78, 71, 0, 0 },
223             { 78, 73, 0, 0, 0, 0 }, { 78, 73, 65, 78, 0, 0 },
224             { 78, 73, 65, 78, 71, 0 }, { 78, 73, 65, 79, 0, 0 },
225             { 78, 73, 69, 0, 0, 0 }, { 78, 73, 78, 0, 0, 0 },
226             { 78, 73, 78, 71, 0, 0 }, { 78, 73, 85, 0, 0, 0 },
227             { 78, 79, 78, 71, 0, 0 }, { 78, 79, 85, 0, 0, 0 },
228             { 78, 85, 0, 0, 0, 0 }, { 78, 85, 65, 78, 0, 0 },
229             { 78, 85, 69, 0, 0, 0 }, { 78, 85, 78, 0, 0, 0 },
230             { 78, 85, 79, 0, 0, 0 }, { 79, 0, 0, 0, 0, 0 },
231             { 79, 85, 0, 0, 0, 0 }, { 80, 65, 0, 0, 0, 0 },
232             { 80, 65, 73, 0, 0, 0 }, { 80, 65, 78, 0, 0, 0 },
233             { 80, 65, 78, 71, 0, 0 }, { 80, 65, 79, 0, 0, 0 },
234             { 80, 69, 73, 0, 0, 0 }, { 80, 69, 78, 0, 0, 0 },
235             { 80, 69, 78, 71, 0, 0 }, { 80, 73, 0, 0, 0, 0 },
236             { 80, 73, 65, 78, 0, 0 }, { 80, 73, 65, 79, 0, 0 },
237             { 80, 73, 69, 0, 0, 0 }, { 80, 73, 78, 0, 0, 0 },
238             { 80, 73, 78, 71, 0, 0 }, { 80, 79, 0, 0, 0, 0 },
239             { 80, 79, 85, 0, 0, 0 }, { 80, 85, 0, 0, 0, 0 },
240             { 81, 73, 0, 0, 0, 0 }, { 81, 73, 65, 0, 0, 0 },
241             { 81, 73, 65, 78, 0, 0 }, { 81, 73, 65, 78, 71, 0 },
242             { 81, 73, 65, 79, 0, 0 }, { 81, 73, 69, 0, 0, 0 },
243             { 81, 73, 78, 0, 0, 0 }, { 81, 73, 78, 71, 0, 0 },
244             { 81, 73, 79, 78, 71, 0 }, { 81, 73, 85, 0, 0, 0 },
245             { 81, 85, 0, 0, 0, 0 }, { 81, 85, 65, 78, 0, 0 },
246             { 81, 85, 69, 0, 0, 0 }, { 81, 85, 78, 0, 0, 0 },
247             { 82, 65, 78, 0, 0, 0 }, { 82, 65, 78, 71, 0, 0 },
248             { 82, 65, 79, 0, 0, 0 }, { 82, 69, 0, 0, 0, 0 },
249             { 82, 69, 78, 0, 0, 0 }, { 82, 69, 78, 71, 0, 0 },
250             { 82, 73, 0, 0, 0, 0 }, { 82, 79, 78, 71, 0, 0 },
251             { 82, 79, 85, 0, 0, 0 }, { 82, 85, 0, 0, 0, 0 },
252             { 82, 85, 65, 0, 0, 0 }, { 82, 85, 65, 78, 0, 0 },
253             { 82, 85, 73, 0, 0, 0 }, { 82, 85, 78, 0, 0, 0 },
254             { 82, 85, 79, 0, 0, 0 }, { 83, 65, 0, 0, 0, 0 },
255             { 83, 65, 73, 0, 0, 0 }, { 83, 65, 78, 0, 0, 0 },
256             { 83, 65, 78, 71, 0, 0 }, { 83, 65, 79, 0, 0, 0 },
257             { 83, 69, 0, 0, 0, 0 }, { 83, 69, 78, 0, 0, 0 },
258             { 83, 69, 78, 71, 0, 0 }, { 83, 72, 65, 0, 0, 0 },
259             { 83, 72, 65, 73, 0, 0 }, { 83, 72, 65, 78, 0, 0 },
260             { 83, 72, 65, 78, 71, 0 }, { 83, 72, 65, 79, 0, 0 },
261             { 83, 72, 69, 0, 0, 0 }, { 83, 72, 69, 78, 0, 0 },
262             { 88, 73, 78, 0, 0, 0 }, { 83, 72, 69, 78, 0, 0 },
263             { 83, 72, 69, 78, 71, 0 }, { 83, 72, 73, 0, 0, 0 },
264             { 83, 72, 79, 85, 0, 0 }, { 83, 72, 85, 0, 0, 0 },
265             { 83, 72, 85, 65, 0, 0 }, { 83, 72, 85, 65, 73, 0 },
266             { 83, 72, 85, 65, 78, 0 }, { 83, 72, 85, 65, 78, 71 },
267             { 83, 72, 85, 73, 0, 0 }, { 83, 72, 85, 78, 0, 0 },
268             { 83, 72, 85, 79, 0, 0 }, { 83, 73, 0, 0, 0, 0 },
269             { 83, 79, 78, 71, 0, 0 }, { 83, 79, 85, 0, 0, 0 },
270             { 83, 85, 0, 0, 0, 0 }, { 83, 85, 65, 78, 0, 0 },
271             { 83, 85, 73, 0, 0, 0 }, { 83, 85, 78, 0, 0, 0 },
272             { 83, 85, 79, 0, 0, 0 }, { 84, 65, 0, 0, 0, 0 },
273             { 84, 65, 73, 0, 0, 0 }, { 84, 65, 78, 0, 0, 0 },
274             { 84, 65, 78, 71, 0, 0 }, { 84, 65, 79, 0, 0, 0 },
275             { 84, 69, 0, 0, 0, 0 }, { 84, 69, 78, 71, 0, 0 },
276             { 84, 73, 0, 0, 0, 0 }, { 84, 73, 65, 78, 0, 0 },
277             { 84, 73, 65, 79, 0, 0 }, { 84, 73, 69, 0, 0, 0 },
278             { 84, 73, 78, 71, 0, 0 }, { 84, 79, 78, 71, 0, 0 },
279             { 84, 79, 85, 0, 0, 0 }, { 84, 85, 0, 0, 0, 0 },
280             { 84, 85, 65, 78, 0, 0 }, { 84, 85, 73, 0, 0, 0 },
281             { 84, 85, 78, 0, 0, 0 }, { 84, 85, 79, 0, 0, 0 },
282             { 87, 65, 0, 0, 0, 0 }, { 87, 65, 73, 0, 0, 0 },
283             { 87, 65, 78, 0, 0, 0 }, { 87, 65, 78, 71, 0, 0 },
284             { 87, 69, 73, 0, 0, 0 }, { 87, 69, 78, 0, 0, 0 },
285             { 87, 69, 78, 71, 0, 0 }, { 87, 79, 0, 0, 0, 0 },
286             { 87, 85, 0, 0, 0, 0 }, { 88, 73, 0, 0, 0, 0 },
287             { 88, 73, 65, 0, 0, 0 }, { 88, 73, 65, 78, 0, 0 },
288             { 88, 73, 65, 78, 71, 0 }, { 88, 73, 65, 79, 0, 0 },
289             { 88, 73, 69, 0, 0, 0 }, { 88, 73, 78, 0, 0, 0 },
290             { 88, 73, 78, 71, 0, 0 }, { 88, 73, 79, 78, 71, 0 },
291             { 88, 73, 85, 0, 0, 0 }, { 88, 85, 0, 0, 0, 0 },
292             { 88, 85, 65, 78, 0, 0 }, { 88, 85, 69, 0, 0, 0 },
293             { 88, 85, 78, 0, 0, 0 }, { 89, 65, 0, 0, 0, 0 },
294             { 89, 65, 78, 0, 0, 0 }, { 89, 65, 78, 71, 0, 0 },
295             { 89, 65, 79, 0, 0, 0 }, { 89, 69, 0, 0, 0, 0 },
296             { 89, 73, 0, 0, 0, 0 }, { 89, 73, 78, 0, 0, 0 },
297             { 89, 73, 78, 71, 0, 0 }, { 89, 79, 0, 0, 0, 0 },
298             { 89, 79, 78, 71, 0, 0 }, { 89, 79, 85, 0, 0, 0 },
299             { 89, 85, 0, 0, 0, 0 }, { 89, 85, 65, 78, 0, 0 },
300             { 89, 85, 69, 0, 0, 0 }, { 89, 85, 78, 0, 0, 0 },
301             { 74, 85, 78, 0, 0, 0 }, { 89, 85, 78, 0, 0, 0 },
302             { 90, 65, 0, 0, 0, 0 }, { 90, 65, 73, 0, 0, 0 },
303             { 90, 65, 78, 0, 0, 0 }, { 90, 65, 78, 71, 0, 0 },
304             { 90, 65, 79, 0, 0, 0 }, { 90, 69, 0, 0, 0, 0 },
305             { 90, 69, 73, 0, 0, 0 }, { 90, 69, 78, 0, 0, 0 },
306             { 90, 69, 78, 71, 0, 0 }, { 90, 72, 65, 0, 0, 0 },
307             { 90, 72, 65, 73, 0, 0 }, { 90, 72, 65, 78, 0, 0 },
308             { 90, 72, 65, 78, 71, 0 }, { 67, 72, 65, 78, 71, 0 },
309             { 90, 72, 65, 78, 71, 0 }, { 90, 72, 65, 79, 0, 0 },
310             { 90, 72, 69, 0, 0, 0 }, { 90, 72, 69, 78, 0, 0 },
311             { 90, 72, 69, 78, 71, 0 }, { 90, 72, 73, 0, 0, 0 },
312             { 83, 72, 73, 0, 0, 0 }, { 90, 72, 73, 0, 0, 0 },
313             { 90, 72, 79, 78, 71, 0 }, { 90, 72, 79, 85, 0, 0 },
314             { 90, 72, 85, 0, 0, 0 }, { 90, 72, 85, 65, 0, 0 },
315             { 90, 72, 85, 65, 73, 0 }, { 90, 72, 85, 65, 78, 0 },
316             { 90, 72, 85, 65, 78, 71 }, { 90, 72, 85, 73, 0, 0 },
317             { 90, 72, 85, 78, 0, 0 }, { 90, 72, 85, 79, 0, 0 },
318             { 90, 73, 0, 0, 0, 0 }, { 90, 79, 78, 71, 0, 0 },
319             { 90, 79, 85, 0, 0, 0 }, { 90, 85, 0, 0, 0, 0 },
320             { 90, 85, 65, 78, 0, 0 }, { 90, 85, 73, 0, 0, 0 },
321             { 90, 85, 78, 0, 0, 0 }, { 90, 85, 79, 0, 0, 0 },
322             { 0, 0, 0, 0, 0, 0 }, { 83, 72, 65, 78, 0, 0 },
323             { 0, 0, 0, 0, 0, 0 }, };
324 
325     /**
326      * First and last Chinese character with known Pinyin according to zh
327      * collation
328      */
329     private static final String FIRST_PINYIN_UNIHAN = "\u963F";
330     private static final String LAST_PINYIN_UNIHAN = "\u9FFF";
331 
332     private static final Collator COLLATOR = Collator.getInstance(Locale.CHINA);
333 
334     private static HanziToPinyin sInstance;
335     private final boolean mHasChinaCollator;
336 
337     public static class Token {
338         /**
339          * Separator between target string for each source char
340          */
341         public static final String SEPARATOR = " ";
342 
343         public static final int LATIN = 1;
344         public static final int PINYIN = 2;
345         public static final int UNKNOWN = 3;
346 
347         public Token() {
348         }
349 
350         public Token(int type, String source, String target) {
351             this.type = type;
352             this.source = source;
353             this.target = target;
354         }
355 
356         /**
357          * Type of this token, ASCII, PINYIN or UNKNOWN.
358          */
359         public int type;
360         /**
361          * Original string before translation.
362          */
363         public String source;
364         /**
365          * Translated string of source. For Han, target is corresponding Pinyin.
366          * Otherwise target is original string in source.
367          */
368         public String target;
369     }
370 
371     protected HanziToPinyin(boolean hasChinaCollator) {
372         mHasChinaCollator = hasChinaCollator;
373     }
374 
375     public static HanziToPinyin getInstance() {
376         synchronized (HanziToPinyin.class) {
377             if (sInstance != null) {
378                 return sInstance;
379             }
380             // Check if zh_CN collation data is available
381             final Locale locale[] = Collator.getAvailableLocales();
382             for (int i = 0; i < locale.length; i++) {
383                 if (locale[i].equals(Locale.CHINA)) {
384                     // Do self validation just once.
385                     if (DEBUG) {
386                         Log.d(TAG, "Self validation. Result: "
387                                 + doSelfValidation());
388                     }
389                     sInstance = new HanziToPinyin(true);
390                     return sInstance;
391                 }
392             }
393             Log.w(TAG,
394                     "There is no Chinese collator, HanziToPinyin is disabled");
395             sInstance = new HanziToPinyin(false);
396             return sInstance;
397         }
398     }
399 
400     /**
401      * Validate if our internal table has some wrong value.
402      * 
403      * @return true when the table looks correct.
404      */
405     private static boolean doSelfValidation() {
406         char lastChar = UNIHANS[0];
407         String lastString = Character.toString(lastChar);
408         for (char c : UNIHANS) {
409             if (lastChar == c) {
410                 continue;
411             }
412             final String curString = Character.toString(c);
413             int cmp = COLLATOR.compare(lastString, curString);
414             if (cmp >= 0) {
415                 Log.e(TAG, "Internal error in Unihan table. "
416                         + "The last string \"" + lastString
417                         + "\" is greater than current string \"" + curString
418                         + "\".");
419                 return false;
420             }
421             lastString = curString;
422         }
423         return true;
424     }
425 
426     private Token getToken(char character) {
427         Token token = new Token();
428         final String letter = Character.toString(character);
429         token.source = letter;
430         int offset = -1;
431         int cmp;
432         if (character < 256) {
433             token.type = Token.LATIN;
434             token.target = letter;
435             return token;
436         } else {
437             cmp = COLLATOR.compare(letter, FIRST_PINYIN_UNIHAN);
438             if (cmp < 0) {
439                 token.type = Token.UNKNOWN;
440                 token.target = letter;
441                 return token;
442             } else if (cmp == 0) {
443                 token.type = Token.PINYIN;
444                 offset = 0;
445             } else {
446                 cmp = COLLATOR.compare(letter, LAST_PINYIN_UNIHAN);
447                 if (cmp > 0) {
448                     token.type = Token.UNKNOWN;
449                     token.target = letter;
450                     return token;
451                 } else if (cmp == 0) {
452                     token.type = Token.PINYIN;
453                     offset = UNIHANS.length - 1;
454                 }
455             }
456         }
457 
458         token.type = Token.PINYIN;
459         if (offset < 0) {
460             int begin = 0;
461             int end = UNIHANS.length - 1;
462             while (begin <= end) {
463                 offset = (begin + end) / 2;
464                 final String unihan = Character.toString(UNIHANS[offset]);
465                 cmp = COLLATOR.compare(letter, unihan);
466                 if (cmp == 0) {
467                     break;
468                 } else if (cmp > 0) {
469                     begin = offset + 1;
470                 } else {
471                     end = offset - 1;
472                 }
473             }
474         }
475         if (cmp < 0) {
476             offset--;
477         }
478         StringBuilder pinyin = new StringBuilder();
479         for (int j = 0; j < PINYINS[offset].length && PINYINS[offset][j] != 0; j++) {
480             pinyin.append((char) PINYINS[offset][j]);
481         }
482         token.target = pinyin.toString();
483         if (TextUtils.isEmpty(token.target)) {
484             token.type = Token.UNKNOWN;
485             token.target = token.source;
486         }
487         return token;
488     }
489 
490     /**
491      * Convert the input to a array of tokens. The sequence of ASCII or Unknown
492      * characters without space will be put into a Token, One Hanzi character
493      * which has pinyin will be treated as a Token. If these is no China
494      * collator, the empty token array is returned.
495      */
496     public ArrayList<Token> get(final String input) {
497         ArrayList<Token> tokens = new ArrayList<Token>();
498         if (!mHasChinaCollator || TextUtils.isEmpty(input)) {
499             // return empty tokens.
500             return tokens;
501         }
502         final int inputLength = input.length();
503         final StringBuilder sb = new StringBuilder();
504         int tokenType = Token.LATIN;
505         // Go through the input, create a new token when
506         // a. Token type changed
507         // b. Get the Pinyin of current charater.
508         // c. current character is space.
509         for (int i = 0; i < inputLength; i++) {
510             final char character = input.charAt(i);
511             if (character == ' ') {
512                 if (sb.length() > 0) {
513                     addToken(sb, tokens, tokenType);
514                 }
515             } else if (character < 256) {
516                 if (tokenType != Token.LATIN && sb.length() > 0) {
517                     addToken(sb, tokens, tokenType);
518                 }
519                 tokenType = Token.LATIN;
520                 sb.append(character);
521             } else {
522                 Token t = getToken(character);
523                 if (t.type == Token.PINYIN) {
524                     if (sb.length() > 0) {
525                         addToken(sb, tokens, tokenType);
526                     }
527                     tokens.add(t);
528                     tokenType = Token.PINYIN;
529                 } else {
530                     if (tokenType != t.type && sb.length() > 0) {
531                         addToken(sb, tokens, tokenType);
532                     }
533                     tokenType = t.type;
534                     sb.append(character);
535                 }
536             }
537         }
538         if (sb.length() > 0) {
539             addToken(sb, tokens, tokenType);
540         }
541         return tokens;
542     }
543 
544     private void addToken(final StringBuilder sb,
545             final ArrayList<Token> tokens, final int tokenType) {
546         String str = sb.toString();
547         tokens.add(new Token(tokenType, str, str));
548         sb.setLength(0);
549     }
550 }

方法调用：

package com.example.test;

import java.util.ArrayList;

import com.example.test.HanziToPinyin.Token;

import android.os.Bundle;
import android.app.Activity;
import android.widget.TextView;

public class MainActivity extends Activity {

	@Override
	protected void onCreate(Bundle savedInstanceState) {
		super.onCreate(savedInstanceState);
		setContentView(R.layout.activity_main);
		TextView txt = (TextView) findViewById(R.id.txtContent);
		txt.append(getPinYin("令狐冲") + " -- " + getPinYin("张无忌") + " -- "
				+ getPinYin("任盈盈"));
	}

	/**
	 * 汉字返回拼音，字母原样返回，都转换为小写
	 * 
	 * @param input
	 * @return
	 */
	public static String getPinYin(String str) {
		ArrayList<Token> tokens = HanziToPinyin.getInstance().get(str);
		StringBuilder sb = new StringBuilder();
		if (tokens != null && tokens.size() > 0) {
			for (Token token : tokens) {
				if (Token.PINYIN == token.type) {
					sb.append(token.target);
				} else {
					sb.append(token.source);
				}
			}
		}
		return sb.toString().toLowerCase();
	}
}

　　

posted @ 2015-12-17 08:36  宇少vlog  阅读(1112)  评论(0)    收藏  举报