UTF-8 <==> unicode(WCHAR)

 1 static int fetchWordFromUTF8(const chConstStringA& strText, WCHAR& result)
 2 {
 3     int nLength = strText.length();
 4     if(nLength <= 0) return 0;
 5 
 6     LPCSTR lpszTextA = strText.c_ptr();
 7     BYTE byte = *lpszTextA;
 8     if ((byte >> 7) == 0x00)
 9     {
10         result = (WCHAR)(BYTE)*lpszTextA;
11         return 1;
12     }
13     else if ((byte >> 5) == 0x06)
14     {
15         if(nLength >= 2)
16         {
17             DWORD dw0 = (DWORD)(BYTE)lpszTextA[0];
18             DWORD dw1 = (DWORD)(BYTE)lpszTextA[1];
19             result = (WCHAR)((dw1&0x3F) | ((dw0&0x1F) << 6));
20             return 2;
21         }
22     }
23     else if ((byte >> 4) == 0x0E)
24     {
25         if(nLength >= 3)
26         {
27             DWORD dw0 = (DWORD)(BYTE)lpszTextA[0];
28             DWORD dw1 = (DWORD)(BYTE)lpszTextA[1];
29             DWORD dw2 = (DWORD)(BYTE)lpszTextA[2];
30             result = (WCHAR)((dw2&0x3F) | ((dw1&0x3F) << 6) | ((dw0&0x0F)<<12));
31             return 3;
32         }
33     }
34     result = '?';
35     return 1;
36 }
37 
38 static int fetchUTF8FromWord(const UINT& word, LPSTR szResult)
39 {
40     if (word < 0x0080)
41     {
42         szResult[0] = (ACHAR)word;
43         return 1;
44     }
45     else if (word < 0x0800)
46     {
47         szResult[0] = (ACHAR)(0xC0 | (BYTE)(word>>6));
48         szResult[1] = (ACHAR)(0x80 | (BYTE)(word&0x003F));
49         return 2;
50     }
51     else// if (word <= 0xFFFF)
52     {
53         szResult[0] = (ACHAR)(0xE0 | (BYTE)(word>>12));
54         szResult[1] = (ACHAR)(0x80 | (BYTE)((word>>6) & 0x3F));
55         szResult[2] = (ACHAR)(0x80 | (BYTE)(word&0x3F));
56         return 3;
57     }
58     szResult[0] = '?';
59     return 1;
60 }
posted on 2015-05-14 09:52 叶城宇阅读(401) 评论(0) 编辑收藏举报