关于文件读取时编码问题 

 1 /**
 2      * 判断文件的编码格式
 3      * @param fileName :file
 4      * @return 
 5      * @return 文件编码格式
 6      * @throws Exception
 7      */
 8     public String codeString(String fileName) throws Exception{
 9         String charset = "GBK";
10         byte[] first3Bytes = new byte[3];
11         try {
12             boolean checked = false;
13             BufferedInputStream bis = new BufferedInputStream(new FileInputStream(fileName));
14             bis.mark(0);
15             int read = bis.read(first3Bytes, 0, 3);
16             if (read == -1) {
17                 return charset; // 文件编码为 ANSI
18             } else if (first3Bytes[0] == (byte) 0xFF && first3Bytes[1] == (byte) 0xFE) {
19                 charset = "UTF-16LE"; // 文件编码为 Unicode
20                 checked = true;
21             } else if (first3Bytes[0] == (byte) 0xFE && first3Bytes[1] == (byte) 0xFF) {
22                 charset = "UTF-16BE"; // 文件编码为 Unicode big endian
23                 checked = true;
24             } else if (first3Bytes[0] == (byte) 0xEF && first3Bytes[1] == (byte) 0xBB
25                     && first3Bytes[2] == (byte) 0xBF) {
26                 charset = "UTF-8"; // 文件编码为 UTF-8
27                 checked = true;
28             }
29             bis.reset();
30             if (!checked) {
31                 int loc = 0;
32                 while ((read = bis.read()) != -1) {
33                     loc++;
34                     if (read >= 0xF0) {
35                         break;
36                     }
37                     if (0x80 <= read && read <= 0xBF) // 单独出现BF以下的，也算是GBK
38                     {
39                         break;
40                     }
41                     if (0xC0 <= read && read <= 0xDF) {
42                         read = bis.read();
43                         if (0x80 <= read && read <= 0xBF) // 双字节 (0xC0 - 0xDF)
44                         // (0x80
45                         // - 0xBF),也可能在GB编码内
46                         {
47                             continue;
48                         } else {
49                             break;
50                         }
51                     } else if (0xE0 <= read && read <= 0xEF) {// 也有可能出错，但是几率较小
52                         read = bis.read();
53                         if (0x80 <= read && read <= 0xBF) {
54                             read = bis.read();
55                             if (0x80 <= read && read <= 0xBF) {
56                                 charset = "UTF-8";
57                                 break;
58                             } else {
59                                 break;
60                             }
61                         } else {
62                             break;
63                         }
64                     }
65                 }
66             }
67             bis.close();
68         } catch (Exception e) {
69             e.printStackTrace();
70         }
71         return charset;
72     }

 

posted @ 2016-07-21 16:39  西瓜加冰  阅读(310)  评论(0)    收藏  举报