VC/MFC 获取汉字拼音首字母（unicode编码）

原文链接：http://blog.csdn.net/china200_ok/article/details/50485849

CString GetFirstLetter(LPCTSTR strName)

{
static int li_SecPosValue[] = {
1601, 1637, 1833, 2078, 2274, 2302, 2433, 2594, 2787, 3106, 3212,
3472, 3635, 3722, 3730, 3858, 4027, 4086, 4390, 4558, 4684, 4925, 5249
};
static char* lc_FirstLetter[] = {
"A", "B", "C", "D", "E", "F", "G", "H", "J", "K", "L", "M", "N", "O",
"P", "Q", "R", "S", "T", "W", "X", "Y", "Z"
};
static char* ls_SecondSecTable =
"CJWGNSPGCGNE[Y[BTYYZDXYKYGT[JNNJQMBSGZSCYJSYY[PGKBZGY[YWJKGKLJYWKPJQHY[W[DZLSGMRYPYWWCCKZNKYYGTTNJJNYKKZYTCJNMCYLQLYPYQFQRPZSLWBTGKJFYXJWZLTBNCXJJJJTXDTTSQZYCDXXHGCK[PHFFSS[YBGXLPPBYLL[HLXS[ZM[JHSOJNGHDZQYKLGJHSGQZHXQGKEZZWYSCSCJXYEYXADZPMDSSMZJZQJYZC[J[WQJBYZPXGZNZCPWHKXHQKMWFBPBYDTJZZKQHY"
"LYGXFPTYJYYZPSZLFCHMQSHGMXXSXJ[[DCSBBQBEFSJYHXWGZKPYLQBGLDLCCTNMAYDDKSSNGYCSGXLYZAYBNPTSDKDYLHGYMYLCXPY[JNDQJWXQXFYYFJLEJPZRXCCQWQQSBNKYMGPLBMJRQCFLNYMYQMSQYRBCJTHZTQFRXQHXMJJCJLXQGJMSHZKBSWYEMYLTXFSYDSWLYCJQXSJNQBSCTYHBFTDCYZDJWYGHQFRXWCKQKXEBPTLPXJZSRMEBWHJLBJSLYYSMDXLCLQKXLHXJRZJMFQHXHWY"
"WSBHTRXXGLHQHFNM[YKLDYXZPYLGG[MTCFPAJJZYLJTYANJGBJPLQGDZYQYAXBKYSECJSZNSLYZHSXLZCGHPXZHZNYTDSBCJKDLZAYFMYDLEBBGQYZKXGLDNDNYSKJSHDLYXBCGHXYPKDJMMZNGMMCLGWZSZXZJFZNMLZZTHCSYDBDLLSCDDNLKJYKJSYCJLKWHQASDKNHCSGANHDAASHTCPLCPQYBSDMPJLPZJOQLCDHJJYSPRCHN[NNLHLYYQYHWZPTCZGWWMZFFJQQQQYXACLBHKDJXDGMMY"
"DJXZLLSYGXGKJRYWZWYCLZMSSJZLDBYD[FCXYHLXCHYZJQ[[QAGMNYXPFRKSSBJLYXYSYGLNSCMHZWWMNZJJLXXHCHSY[[TTXRYCYXBYHCSMXJSZNPWGPXXTAYBGAJCXLY[DCCWZOCWKCCSBNHCPDYZNFCYYTYCKXKYBSQKKYTQQXFCWCHCYKELZQBSQYJQCCLMTHSYWHMKTLKJLYCXWHEQQHTQH[PQ[QSCFYMNDMGBWHWLGSLLYSDLMLXPTHMJHWLJZYHZJXHTXJLHXRSWLWZJCBXMHZQXSDZP"
"MGFCSGLSXYMJSHXPJXWMYQKSMYPLRTHBXFTPMHYXLCHLHLZYLXGSSSSTCLSLDCLRPBHZHXYYFHB[GDMYCNQQWLQHJJ[YWJZYEJJDHPBLQXTQKWHLCHQXAGTLXLJXMSL[HTZKZJECXJCJNMFBY[SFYWYBJZGNYSDZSQYRSLJPCLPWXSDWEJBJCBCNAYTWGMPAPCLYQPCLZXSBNMSGGFNZJJBZSFZYNDXHPLQKZCZWALSBCCJX[YZGWKYPSGXFZFCDKHJGXDLQFSGDSLQWZKXTMHSBGZMJZRGLYJB"
"PMLMSXLZJQQHZYJCZYDJWBMYKLDDPMJEGXYHYLXHLQYQHKYCWCJMYYXNATJHYCCXZPCQLBZWWYTWBQCMLPMYRJCCCXFPZNZZLJPLXXYZTZLGDLDCKLYRZZGQTGJHHGJLJAXFGFJZSLCFDQZLCLGJDJCSNZLLJPJQDCCLCJXMYZFTSXGCGSBRZXJQQCTZHGYQTJQQLZXJYLYLBCYAMCSTYLPDJBYREGKLZYZHLYSZQLZNWCZCLLWJQJJJKDGJZOLBBZPPGLGHTGZXYGHZMYCNQSYCYHBHGXKAMTX"
"YXNBSKYZZGJZLQJDFCJXDYGJQJJPMGWGJJJPKQSBGBMMCJSSCLPQPDXCDYYKY[CJDDYYGYWRHJRTGZNYQLDKLJSZZGZQZJGDYKSHPZMTLCPWNJAFYZDJCNMWESCYGLBTZCGMSSLLYXQSXSBSJSBBSGGHFJLYPMZJNLYYWDQSHZXTYYWHMZYHYWDBXBTLMSYYYFSXJC[DXXLHJHF[SXZQHFZMZCZTQCXZXRTTDJHNNYZQQMNQDMMG[YDXMJGDHCDYZBFFALLZTDLTFXMXQZDNGWQDBDCZJDXBZGS"
"QQDDJCMBKZFFXMKDMDSYYSZCMLJDSYNSBRSKMKMPCKLGDBQTFZSWTFGGLYPLLJZHGJ[GYPZLTCSMCNBTJBQFKTHBYZGKPBBYMTDSSXTBNPDKLEYCJNYDDYKZDDHQHSDZSCTARLLTKZLGECLLKJLQJAQNBDKKGHPJTZQKSECSHALQFMMGJNLYJBBTMLYZXDCJPLDLPCQDHZYCBZSCZBZMSLJFLKRZJSNFRGJHXPDHYJYBZGDLQCSEZGXLBLGYXTWMABCHECMWYJYZLLJJYHLG[DJLSLYGKDZPZXJ"
"YYZLWCXSZFGWYYDLYHCLJSCMBJHBLYZLYCBLYDPDQYSXQZBYTDKYXJY[CNRJMPDJGKLCLJBCTBJDDBBLBLCZQRPPXJCJLZCSHLTOLJNMDDDLNGKAQHQHJGYKHEZNMSHRP[QQJCHGMFPRXHJGDYCHGHLYRZQLCYQJNZSQTKQJYMSZSWLCFQQQXYFGGYPTQWLMCRNFKKFSYYLQBMQAMMMYXCTPSHCPTXXZZSMPHPSHMCLMLDQFYQXSZYYDYJZZHQPDSZGLSTJBCKBXYQZJSGPSXQZQZRQTBDKYXZK"
"HHGFLBCSMDLDGDZDBLZYYCXNNCSYBZBFGLZZXSWMSCCMQNJQSBDQSJTXXMBLTXZCLZSHZCXRQJGJYLXZFJPHYMZQQYDFQJJLZZNZJCDGZYGCTXMZYSCTLKPHTXHTLBJXJLXSCDQXCBBTJFQZFSLTJBTKQBXXJJLJCHCZDBZJDCZJDCPRNPQCJPFCZLCLZXZDMXMPHJSGZGSZZQLYLWTJPFSYASMCJBTZKYCWMYTCSJJLJCQLWZMALBXYFBPNLSFHTGJWEJJXXGLLJSTGSHJQLZFKCGNNNSZFDEQ"
"FHBSAQTGYLBXMMYGSZLDYDQMJJRGBJTKGDHGKBLQKBDMBYLXWCXYTTYBKMRTJZXQJBHLMHMJJZMQASLDCYXYQDLQCAFYWYXQHZ";


//将unicode编码转换成ANSI编码  
int nSecondSecTable = strlen(ls_SecondSecTable);
int stringlen = wcslen(strName);
char   *buffer = NULL;
buffer = new char[stringlen * sizeof(TCHAR)+1];
ZeroMemory(buffer, stringlen * sizeof(TCHAR)+1);
WideCharToMultiByte(CP_ACP, NULL, strName, stringlen, buffer, stringlen * sizeof(TCHAR)+1, NULL, NULL);
//  
std::string result;
int nHigh, nLow, nCode;
int i, j;
for (i = 0; i < stringlen*sizeof(TCHAR); i++)
{
//找到汉字的机内码  
nHigh = (UCHAR)(buffer[i + 0]);
nLow = (UCHAR)(buffer[i + 1]);


if (nHigh < 0xA1 || nLow < 0xA1)    //非汉字保留  
{
result += buffer[i];
continue;
}
else
{
//一级汉字机内码与区位码的关系  
//机内码高字节 = 区码 + 0XA0;  
//机内码低字节 = 位码 + 0XA0;  
//区位码 = (机内码高字节 - 0XA0) * 100 + 机内码低字节 - 0XA0;  
nCode = (nHigh - 0XA0) * 100 + nLow - 0XA0;
}


if (nCode > 1600 && nCode < 5590)       //一级汉字 16-55区      
{
for (j = 22; j >= 0; j--)
{
if (nCode >= li_SecPosValue[j])
{
result += lc_FirstLetter[j];
i++;
break;
}
}


continue;
}
else  //二级汉字通过查表  56-87区;将二级汉字拼音的首字母放在ls_SecondSecTable表中  
{
i++;
//根据机内码转换成ls_SecondSecTable表的下标  
nCode = (nHigh - 0XA0 - 56) * 94 + nLow - 0XA0 - 1;
if (nCode >= 0 && nCode < nSecondSecTable)
result += ls_SecondSecTable[nCode];
else
{
result += (char)nHigh;
result += (char)nLow;
}
}
}


//将ANSI编码转换成unicode编码  
TCHAR *pwcsTemp = new TCHAR[stringlen + 1];
ZeroMemory(pwcsTemp, (stringlen + 1) * sizeof(TCHAR));


MultiByteToWideChar(CP_ACP, NULL, result.c_str(), result.length(), pwcsTemp, stringlen);
CString str = pwcsTemp;
//  
delete[]pwcsTemp;
pwcsTemp = NULL;


delete[]buffer;
buffer = NULL;


return str;
}