HTMLDecoder，&#开头的编码换转成中文

HTMLDecoder，&#开头的编码换转成中文

HTMLDecoder，中文转换成&#开头的编码实现

1. package test.com.gjob.services;
2. import java.util.HashMap;  
3. /** 
4.  * 替换HTMl里面的字符 e.g.: < > " å И 水 
5.  *  
6.  * @author 老紫竹(laozizhu.com) 
7.  */ 
8. public class HTMLDecoder {  
9.    
10.     public static void main(String args[]){
11.            String tt1 = "你好，我想给你说一个事情";
12.           
13.            System.out.println(HTMLDecoder.decode(tt1));
14.         System.out.println(HTMLDecoder.decode("中国"));
15.         System.out.println(HTMLDecoder.compact("\u4F60\u597D"));
16.        
17.     }
18.  
19.   public static final HashMap<String, Character> charTable;  
20.  
21.   public static String decode(String s) {  
22.     String t;  
23.     Character ch;  
24.     int tmpPos, i;  
25.  
26.     int maxPos = s.length();  
27.     StringBuffer sb = new StringBuffer(maxPos);  
28.     int curPos = 0;  
29.     while (curPos < maxPos) {  
30.       char c = s.charAt(curPos++);  
31.       if (c == '&') {  
32.         tmpPos = curPos;  
33.         if (tmpPos < maxPos) {  
34.           char d = s.charAt(tmpPos++);  
35.           if (d == '#') {  
36.             if (tmpPos < maxPos) {  
37.               d = s.charAt(tmpPos++);  
38.               if ((d == 'x') || (d == 'X')) {  
39.                 if (tmpPos < maxPos) {  
40.                   d = s.charAt(tmpPos++);  
41.                   if (isHexDigit(d)) {  
42.                     while (tmpPos < maxPos) {  
43.                       d = s.charAt(tmpPos++);  
44.                       if (!isHexDigit(d)) {  
45.                         if (d == ';') {  
46.                           t = s.substring(curPos + 2, tmpPos - 1);  
47.                           try {  
48.                             i = Integer.parseInt(t, 16);  
49.                             if ((i >= 0) && (i < 65536)) {  
50.                               c = (char) i;  
51.                               curPos = tmpPos;  
52.                             }  
53.                           } catch (NumberFormatException e) {  
54.                           }  
55.                         }  
56.                         break;  
57.                       }  
58.                     }  
59.                   }  
60.                 }  
61.               } else if (isDigit(d)) {  
62.                 while (tmpPos < maxPos) {  
63.                   d = s.charAt(tmpPos++);  
64.                   if (!isDigit(d)) {  
65.                     if (d == ';') {  
66.                       t = s.substring(curPos + 1, tmpPos - 1);  
67.                       try {  
68.                         i = Integer.parseInt(t);  
69.                         if ((i >= 0) && (i < 65536)) {  
70.                           c = (char) i;  
71.                           curPos = tmpPos;  
72.                         }  
73.                       } catch (NumberFormatException e) {  
74.                       }  
75.                     }  
76.                     break;  
77.                   }  
78.                 }  
79.               }  
80.             }  
81.           } else if (isLetter(d)) {  
82.             while (tmpPos < maxPos) {  
83.               d = s.charAt(tmpPos++);  
84.               if (!isLetterOrDigit(d)) {  
85.                 if (d == ';') {  
86.                   t = s.substring(curPos, tmpPos - 1);  
87.                   ch = (Character) charTable.get(t);  
88.                   if (ch != null) {  
89.                     c = ch.charValue();  
90.                     curPos = tmpPos;  
91.                   }  
92.                 }  
93.                 break;  
94.               }  
95.             }  
96.           }  
97.         }  
98.       }  
99.       sb.append(c);  
100.     }  
101.     return sb.toString();  
102.   }  
103.  
104.   private static boolean isLetterOrDigit(char c) {  
105.     return isLetter(c) || isDigit(c);  
106.   }  
107.  
108.   private static boolean isHexDigit(char c) {  
109.     return isHexLetter(c) || isDigit(c);  
110.   }  
111.  
112.   private static boolean isLetter(char c) {  
113.     return ((c >= 'a') && (c <= 'z')) || ((c >= 'A') && (c <= 'Z'));  
114.   }  
115.  
116.   private static boolean isHexLetter(char c) {  
117.     return ((c >= 'a') && (c <= 'f')) || ((c >= 'A') && (c <= 'F'));  
118.   }  
119.  
120.   private static boolean isDigit(char c) {  
121.     return (c >= '0') && (c <= '9');  
122.   }  
123.  
124.   public static String compact(String s) {  
125.     int maxPos = s.length();  
126.     StringBuffer sb = new StringBuffer(maxPos);  
127.     int curPos = 0;  
128.     while (curPos < maxPos) {  
129.       char c = s.charAt(curPos++);  
130.       if (isWhitespace(c)) {  
131.         while ((curPos < maxPos) && isWhitespace(s.charAt(curPos))) {  
132.           curPos++;  
133.         }  
134.         c = '\u0020';  
135.       }  
136.       sb.append(c);  
137.     }  
138.     return sb.toString();  
139.   }  
140.  
141.   // HTML is very particular about what constitutes white space.  
142.   public static boolean isWhitespace(char ch) {  
143.     return (ch == '\u0020') || (ch == '\r') || (ch == '\n') || (ch == '\u0009') || (ch == '\u000c') || (ch == '\u200b');  
144.   }  
145.  
146.   static {  
147.     charTable = new HashMap<String, Character>();  
148.     charTable.put("quot", new Character((char) 34));  
149.     charTable.put("amp", new Character((char) 38));  
150.     charTable.put("apos", new Character((char) 39));  
151.     charTable.put("lt", new Character((char) 60));  
152.     charTable.put("gt", new Character((char) 62));  
153.     charTable.put("nbsp", new Character((char) 160));  
154.     charTable.put("iexcl", new Character((char) 161));  
155.     charTable.put("cent", new Character((char) 162));  
156.     charTable.put("pound", new Character((char) 163));  
157.     charTable.put("curren", new Character((char) 164));  
158.     charTable.put("yen", new Character((char) 165));  
159.     charTable.put("brvbar", new Character((char) 166));  
160.     charTable.put("sect", new Character((char) 167));  
161.     charTable.put("uml", new Character((char) 168));  
162.     charTable.put("copy", new Character((char) 169));  
163.     charTable.put("ordf", new Character((char) 170));  
164.     charTable.put("laquo", new Character((char) 171));  
165.     charTable.put("not", new Character((char) 172));  
166.     charTable.put("shy", new Character((char) 173));  
167.     charTable.put("reg", new Character((char) 174));  
168.     charTable.put("macr", new Character((char) 175));  
169.     charTable.put("deg", new Character((char) 176));  
170.     charTable.put("plusmn", new Character((char) 177));  
171.     charTable.put("sup2", new Character((char) 178));  
172.     charTable.put("sup3", new Character((char) 179));  
173.     charTable.put("acute", new Character((char) 180));  
174.     charTable.put("micro", new Character((char) 181));  
175.     charTable.put("para", new Character((char) 182));  
176.     charTable.put("middot", new Character((char) 183));  
177.     charTable.put("cedil", new Character((char) 184));  
178.     charTable.put("sup1", new Character((char) 185));  
179.     charTable.put("ordm", new Character((char) 186));  
180.     charTable.put("raquo", new Character((char) 187));  
181.     charTable.put("frac14", new Character((char) 188));  
182.     charTable.put("frac12", new Character((char) 189));  
183.     charTable.put("frac34", new Character((char) 190));  
184.     charTable.put("iquest", new Character((char) 191));  
185.     charTable.put("Agrave", new Character((char) 192));  
186.     charTable.put("Aacute", new Character((char) 193));  
187.     charTable.put("Acirc", new Character((char) 194));  
188.     charTable.put("Atilde", new Character((char) 195));  
189.     charTable.put("Auml", new Character((char) 196));  
190.     charTable.put("Aring", new Character((char) 197));  
191.     charTable.put("AElig", new Character((char) 198));  
192.     charTable.put("Ccedil", new Character((char) 199));  
193.     charTable.put("Egrave", new Character((char) 200));  
194.     charTable.put("Eacute", new Character((char) 201));  
195.     charTable.put("Ecirc", new Character((char) 202));  
196.     charTable.put("Euml", new Character((char) 203));  
197.     charTable.put("Igrave", new Character((char) 204));  
198.     charTable.put("Iacute", new Character((char) 205));  
199.     charTable.put("Icirc", new Character((char) 206));  
200.     charTable.put("Iuml", new Character((char) 207));  
201.     charTable.put("ETH", new Character((char) 208));  
202.     charTable.put("Ntilde", new Character((char) 209));  
203.     charTable.put("Ograve", new Character((char) 210));  
204.     charTable.put("Oacute", new Character((char) 211));  
205.     charTable.put("Ocirc", new Character((char) 212));  
206.     charTable.put("Otilde", new Character((char) 213));  
207.     charTable.put("Ouml", new Character((char) 214));  
208.     charTable.put("times", new Character((char) 215));  
209.     charTable.put("Oslash", new Character((char) 216));  
210.     charTable.put("Ugrave", new Character((char) 217));  
211.     charTable.put("Uacute", new Character((char) 218));  
212.     charTable.put("Ucirc", new Character((char) 219));  
213.     charTable.put("Uuml", new Character((char) 220));  
214.     charTable.put("Yacute", new Character((char) 221));  
215.     charTable.put("THORN", new Character((char) 222));  
216.     charTable.put("szlig", new Character((char) 223));  
217.     charTable.put("agrave", new Character((char) 224));  
218.     charTable.put("aacute", new Character((char) 225));  
219.     charTable.put("acirc", new Character((char) 226));  
220.     charTable.put("atilde", new Character((char) 227));  
221.     charTable.put("auml", new Character((char) 228));  
222.     charTable.put("aring", new Character((char) 229));  
223.     charTable.put("aelig", new Character((char) 230));  
224.     charTable.put("ccedil", new Character((char) 231));  
225.     charTable.put("egrave", new Character((char) 232));  
226.     charTable.put("eacute", new Character((char) 233));  
227.     charTable.put("ecirc", new Character((char) 234));  
228.     charTable.put("euml", new Character((char) 235));  
229.     charTable.put("igrave", new Character((char) 236));  
230.     charTable.put("iacute", new Character((char) 237));  
231.     charTable.put("icirc", new Character((char) 238));  
232.     charTable.put("iuml", new Character((char) 239));  
233.     charTable.put("eth", new Character((char) 240));  
234.     charTable.put("ntilde", new Character((char) 241));  
235.     charTable.put("ograve", new Character((char) 242));  
236.     charTable.put("oacute", new Character((char) 243));  
237.     charTable.put("ocirc", new Character((char) 244));  
238.     charTable.put("otilde", new Character((char) 245));  
239.     charTable.put("ouml", new Character((char) 246));  
240.     charTable.put("divide", new Character((char) 247));  
241.     charTable.put("oslash", new Character((char) 248));  
242.     charTable.put("ugrave", new Character((char) 249));  
243.     charTable.put("uacute", new Character((char) 250));  
244.     charTable.put("ucirc", new Character((char) 251));  
245.     charTable.put("uuml", new Character((char) 252));  
246.     charTable.put("yacute", new Character((char) 253));  
247.     charTable.put("thorn", new Character((char) 254));  
248.     charTable.put("yuml", new Character((char) 255));  
249.     charTable.put("OElig", new Character((char) 338));  
250.     charTable.put("oelig", new Character((char) 339));  
251.     charTable.put("Scaron", new Character((char) 352));  
252.     charTable.put("scaron", new Character((char) 353));  
253.     charTable.put("fnof", new Character((char) 402));  
254.     charTable.put("circ", new Character((char) 710));  
255.     charTable.put("tilde", new Character((char) 732));  
256.     charTable.put("Alpha", new Character((char) 913));  
257.     charTable.put("Beta", new Character((char) 914));  
258.     charTable.put("Gamma", new Character((char) 915));  
259.     charTable.put("Delta", new Character((char) 916));  
260.     charTable.put("Epsilon", new Character((char) 917));  
261.     charTable.put("Zeta", new Character((char) 918));  
262.     charTable.put("Eta", new Character((char) 919));  
263.     charTable.put("Theta", new Character((char) 920));  
264.     charTable.put("Iota", new Character((char) 921));  
265.     charTable.put("Kappa", new Character((char) 922));  
266.     charTable.put("Lambda", new Character((char) 923));  
267.     charTable.put("Mu", new Character((char) 924));  
268.     charTable.put("Nu", new Character((char) 925));  
269.     charTable.put("Xi", new Character((char) 926));  
270.     charTable.put("Omicron", new Character((char) 927));  
271.     charTable.put("Pi", new Character((char) 928));  
272.     charTable.put("Rho", new Character((char) 929));  
273.     charTable.put("Sigma", new Character((char) 931));  
274.     charTable.put("Tau", new Character((char) 932));  
275.     charTable.put("Upsilon", new Character((char) 933));  
276.     charTable.put("Phi", new Character((char) 934));  
277.     charTable.put("Chi", new Character((char) 935));  
278.     charTable.put("Psi", new Character((char) 936));  
279.     charTable.put("Omega", new Character((char) 937));  
280.     charTable.put("alpha", new Character((char) 945));  
281.     charTable.put("beta", new Character((char) 946));  
282.     charTable.put("gamma", new Character((char) 947));  
283.     charTable.put("delta", new Character((char) 948));  
284.     charTable.put("epsilon", new Character((char) 949));  
285.     charTable.put("zeta", new Character((char) 950));  
286.     charTable.put("eta", new Character((char) 951));  
287.     charTable.put("theta", new Character((char) 952));  
288.     charTable.put("iota", new Character((char) 953));  
289.     charTable.put("kappa", new Character((char) 954));  
290.     charTable.put("lambda", new Character((char) 955));  
291.     charTable.put("mu", new Character((char) 956));  
292.     charTable.put("nu", new Character((char) 957));  
293.     charTable.put("xi", new Character((char) 958));  
294.     charTable.put("omicron", new Character((char) 959));  
295.     charTable.put("pi", new Character((char) 960));  
296.     charTable.put("rho", new Character((char) 961));  
297.     charTable.put("sigmaf", new Character((char) 962));  
298.     charTable.put("sigma", new Character((char) 963));  
299.     charTable.put("tau", new Character((char) 964));  
300.     charTable.put("upsilon", new Character((char) 965));  
301.     charTable.put("phi", new Character((char) 966));  
302.     charTable.put("chi", new Character((char) 967));  
303.     charTable.put("psi", new Character((char) 968));  
304.     charTable.put("omega", new Character((char) 969));  
305.     charTable.put("thetasym", new Character((char) 977));  
306.     charTable.put("upsih", new Character((char) 978));  
307.     charTable.put("piv", new Character((char) 982));  
308.     charTable.put("ensp", new Character((char) 8194));  
309.     charTable.put("emsp", new Character((char) 8195));  
310.     charTable.put("thinsp", new Character((char) 8201));  
311.     charTable.put("zwnj", new Character((char) 8204));  
312.     charTable.put("zwj", new Character((char) 8205));  
313.     charTable.put("lrm", new Character((char) 8206));  
314.     charTable.put("rlm", new Character((char) 8207));  
315.     charTable.put("ndash", new Character((char) 8211));  
316.     charTable.put("mdash", new Character((char) 8212));  
317.     charTable.put("lsquo", new Character((char) 8216));  
318.     charTable.put("rsquo", new Character((char) 8217));  
319.     charTable.put("sbquo", new Character((char) 8218));  
320.     charTable.put("ldquo", new Character((char) 8220));  
321.     charTable.put("rdquo", new Character((char) 8221));  
322.     charTable.put("bdquo", new Character((char) 8222));  
323.     charTable.put("dagger", new Character((char) 8224));  
324.     charTable.put("Dagger", new Character((char) 8225));  
325.     charTable.put("bull", new Character((char) 8226));  
326.     charTable.put("hellip", new Character((char) 8230));  
327.     charTable.put("permil", new Character((char) 8240));  
328.     charTable.put("prime", new Character((char) 8242));  
329.     charTable.put("Prime", new Character((char) 8243));  
330.     charTable.put("lsaquo", new Character((char) 8249));  
331.     charTable.put("rsaquo", new Character((char) 8250));  
332.     charTable.put("oline", new Character((char) 8254));  
333.     charTable.put("frasl", new Character((char) 8260));  
334.     charTable.put("euro", new Character((char) 8364));  
335.     charTable.put("image", new Character((char) 8465));  
336.     charTable.put("weierp", new Character((char) 8472));  
337.     charTable.put("real", new Character((char) 8476));  
338.     charTable.put("trade", new Character((char) 8482));  
339.     charTable.put("alefsym", new Character((char) 8501));  
340.     charTable.put("larr", new Character((char) 8592));  
341.     charTable.put("uarr", new Character((char) 8593));  
342.     charTable.put("rarr", new Character((char) 8594));  
343.     charTable.put("darr", new Character((char) 8595));  
344.     charTable.put("harr", new Character((char) 8596));  
345.     charTable.put("crarr", new Character((char) 8629));  
346.     charTable.put("lArr", new Character((char) 8656));  
347.     charTable.put("uArr", new Character((char) 8657));  
348.     charTable.put("rArr", new Character((char) 8658));  
349.     charTable.put("dArr", new Character((char) 8659));  
350.     charTable.put("hArr", new Character((char) 8660));  
351.     charTable.put("forall", new Character((char) 8704));  
352.     charTable.put("part", new Character((char) 8706));  
353.     charTable.put("exist", new Character((char) 8707));  
354.     charTable.put("empty", new Character((char) 8709));  
355.     charTable.put("nabla", new Character((char) 8711));  
356.     charTable.put("isin", new Character((char) 8712));  
357.     charTable.put("notin", new Character((char) 8713));  
358.     charTable.put("ni", new Character((char) 8715));  
359.     charTable.put("prod", new Character((char) 8719));  
360.     charTable.put("sum", new Character((char) 8721));  
361.     charTable.put("minus", new Character((char) 8722));  
362.     charTable.put("lowast", new Character((char) 8727));  
363.     charTable.put("radic", new Character((char) 8730));  
364.     charTable.put("prop", new Character((char) 8733));  
365.     charTable.put("infin", new Character((char) 8734));  
366.     charTable.put("ang", new Character((char) 8736));  
367.     charTable.put("and", new Character((char) 8743));  
368.     charTable.put("or", new Character((char) 8744));  
369.     charTable.put("cap", new Character((char) 8745));  
370.     charTable.put("cup", new Character((char) 8746));  
371.     charTable.put("int", new Character((char) 8747));  
372.     charTable.put("there4", new Character((char) 8756));  
373.     charTable.put("sim", new Character((char) 8764));  
374.     charTable.put("cong", new Character((char) 8773));  
375.     charTable.put("asymp", new Character((char) 8776));  
376.     charTable.put("ne", new Character((char) 8800));  
377.     charTable.put("equiv", new Character((char) 8801));  
378.     charTable.put("le", new Character((char) 8804));  
379.     charTable.put("ge", new Character((char) 8805));  
380.     charTable.put("sub", new Character((char) 8834));  
381.     charTable.put("sup", new Character((char) 8835));  
382.     charTable.put("nsub", new Character((char) 8836));  
383.     charTable.put("sube", new Character((char) 8838));  
384.     charTable.put("supe", new Character((char) 8839));  
385.     charTable.put("oplus", new Character((char) 8853));  
386.     charTable.put("otimes", new Character((char) 8855));  
387.     charTable.put("perp", new Character((char) 8869));  
388.     charTable.put("sdot", new Character((char) 8901));  
389.     charTable.put("lceil", new Character((char) 8968));  
390.     charTable.put("rceil", new Character((char) 8969));  
391.     charTable.put("lfloor", new Character((char) 8970));  
392.     charTable.put("rfloor", new Character((char) 8971));  
393.     charTable.put("lang", new Character((char) 9001));  
394.     charTable.put("rang", new Character((char) 9002));  
395.     charTable.put("loz", new Character((char) 9674));  
396.     charTable.put("spades", new Character((char) 9824));  
397.     charTable.put("clubs", new Character((char) 9827));  
398.     charTable.put("hearts", new Character((char) 9829));  
399.     charTable.put("diams", new Character((char) 9830));  
400.   }  
401. }