Classes written in Java that can translate Chinese characters into pinyin

/***//**
######################################## #####################################
# Describe converting Chinese characters into pinyin
# Date 2006-7-12
# Company flx
# Porject Java
######################################## #####################################
*/
Import java. util. iterator;
Import java. util. linkedhashmap;
Import java. util. Set;

Public class cntospell {
Private Static linkedhashmap spellmap = NULL;

Static {
If (spellmap = NULL ){
Spellmap = new linkedhashmap (400 );
}
Initialize ();
System. Out. println ("Chinese transfer spell done .");
}

Private cntospell (){
}

/***//**
* Obtain the ASCII of a single Chinese character.
* @ Param CN char
* Chinese Characters
* @ Return int
* 0 is returned for an error. Otherwise, ASCII is returned.
*/
Public static int getcnascii (char CN ){
Byte [] bytes = (string. valueof (CN). getbytes ();
If (Bytes = NULL | bytes. length> 2 | bytes. Length <= 0) {// Error
Return 0;
}
If (bytes. Length = 1) {// english character
Return bytes [0];
}
If (bytes. Length = 2) {// Chinese Character
Int hightbyte = 256 + bytes [0];
Int lowbyte = 256 + bytes [1];
Int ASCII = (256 * hightbyte + lowbyte)-256*256;
Return ASCII;
}
Return 0; // Error
}

/***//**
* Returns the full spelling of strings, which are converted from Chinese characters to full spelling. Other characters are not converted.
* @ Param cnstr string
* String
* @ Return string
* Convert the string to a fully-assembled string
*/
Public static string getfullspell (string cnstr ){
If (null = cnstr | "". Equals (cnstr. Trim ())){
Return cnstr;
}

Char [] chars = cnstr. tochararray ();
Stringbuffer retubuf = new stringbuffer ();
For (INT I = 0, Len = chars. length; I <Len; I ++ ){
Int ASCII = getcnascii (chars [I]);
If (ASCII = 0) {// An error occurred while retrieving ASCII
Retubuf. append (chars [I]);
} Else {
String spell = getspellbyascii (ASCII );
If (spell = NULL ){
Retubuf. append (chars [I]);
} Else {
Retubuf. append (spell );
} // End of if spell = NULL
} // End of if ASCII <=-20400
} // End of

Return retubuf. tostring ();
}

/***//**
* Search for the corresponding pinyin in spellmap Based on the ASCII code
* @ Param ASCII int
* ASCII character
* @ Return string
* For pinyin, first determine whether the ASCII value is greater than 0 & <160. If it is a corresponding character,
*
Otherwise, search in spellmap. If no Pinyin is found, null is returned. If no Pinyin is found, Pinyin is returned.
*/
Public static string getspellbyascii (int ascii ){
If (ASCII> 0 & ASCII <160) {// Single Character
Return string. valueof (char) ASCII );
}

If (ASCII <-20319 | ASCII>-10247) {// unknown characters
Return NULL;
}

Set keyset = spellmap. keyset ();
Iterator it = keyset. iterator ();

String spell0 = NULL;
String spell = NULL;

Int asciirang0 =-20319;
Int asciirang;
While (it. hasnext ()){

Spell = (string) it. Next ();
Object valobj = spellmap. Get (spell );
If (valobj instanceof integer ){
Asciirang = (integer) valobj). intvalue ();

If (ASCII> = asciirang0 & ASCII <asciirang) {// interval found
Return (spell0 = NULL )? Spell: spell0;
} Else {
Spell0 = spell;
Asciirang0 = asciirang;
}
}
}

Return NULL;

}

Private Static void initialize (){
Spellput ("A",-20319 );
Spellput ("artificial intelligence",-20317 );
Spellput ("A",-20304 );
Spellput ("Ang",-20295 );
Spellput ("Ao",-20292 );
Spellput ("ba",-20283 );
Spellput ("bai",-20265 );
Spellput ("ban",-20257 );
Spellput ("bang",-20242 );
Spellput ("Bao",-20230 );
Spellput ("Bei",-20051 );
Spellput ("Ben",-1, 20036 );
Spellput ("Beng",-20032 );
Spellput ("bi",-20026 );
Spellput ("Bian",-1, 20002 );
Spellput ("Biao",-19990 );
Spellput ("bie",-19986 );
Spellput ("bin",-19982 );
Spellput ("bing",-19976 );
Spellput ("Bo",-19805 );
Spellput ("bu",-19784 );
Spellput ("ca",-19775 );
Spellput ("Cai",-19774 );
Spellput ("can",-1, 19763 );
Spellput ("Cang",-19756 );
Spellput ("Cao",-19751 );
Spellput ("CE",-19746 );
Spellput ("Ceng",-19741 );
Spellput ("cha",-19739 );
Spellput ("chai",-19728 );
Spellput ("Chan",-19725 );
Spellput ("Chang",-1, 19715 );
Spellput ("Chao",-19540 );
Spellput ("Che",-19531 );
Spellput ("Chen",-19525 );
Spellput ("Cheng",-19515 );
Spellput ("Chi",-19500 );
Spellput ("Chong",-19484 );
Spellput ("Chou",-19479 );
Spellput ("chu",-19467 );
Spellput ("Chuai",-19289 );
Spellput ("Chuan",-19288 );
Spellput ("Chuang",-19281 );
Spellput ("Chui",-19275 );
Spellput ("Chun",-19270 );
Spellput ("Chuo",-19263 );
Spellput ("Ci",-19261 );
Spellput ("Cong",-19249 );
Spellput ("cou",-19243 );
Spellput ("Cu",-19242 );
Spellput ("Cuan",-19238 );
Spellput ("Cui",-19235 );
Spellput ("cun",-19227 );
Spellput ("CuO",-19224 );
Spellput ("da",-19218 );
Spellput ("Dai",-19212 );
Spellput ("dan",-19038 );
Spellput ("dang",-19023 );
Spellput ("Dao",-1, 19018 );
Spellput ("de",-19006 );
Spellput ("Deng",-19003 );
Spellput ("Di",-1, 18996 );
Spellput ("Dian",-18977 );
Spellput ("Diao",-18961 );
Spellput ("die",-1, 18952 );
Spellput ("ding",-18783 );
Spellput ("Diu",-18774 );
Spellput ("Dong",-18773 );
Spellput ("dou",-18763 );
Spellput ("du",-18756 );
Spellput ("Duan",-18741 );
Spellput ("DUI",-18735 );
Spellput ("Dun",-18731 );
Spellput ("duo",-18722 );
Spellput ("e",-18710 );
Spellput ("en",-18697 );
Spellput ("er",-18696 );
Spellput ("Fa",-18526 );
Spellput ("fan",-18518 );
Spellput ("fang",-18501 );
Spellput ("fei",-18490 );
Spellput ("Fen",-18478 );
Spellput ("Feng",-18463 );
Spellput ("FO",-18448 );
Spellput ("fou",-18447 );
Spellput ("Fu",-18446 );
Spellput ("Ga",-18239 );
Spellput ("Gai",-18237 );
Spellput ("Gan",-18231 );
Spellput ("Gang",-18220 );
Spellput ("Gao",-18211 );
Spellput ("Ge",-18201 );
Spellput ("gei",-18184 );
Spellput ("gen",-1, 18183 );
Spellput ("Geng",-18181 );
Spellput ("Gong",-18012 );
Spellput ("gou",-17997 );
Spellput ("Gu",-17988 );
Spellput ("Gua",-17970 );
Spellput ("guai",-17964 );
Spellput ("Guan",-17961 );
Spellput ("Guang",-17950 );
Spellput ("Gui",-17947 );
Spellput ("gun",-17931 );
Spellput ("guo",-17928 );
Spellput ("ha",-17922 );
Spellput ("Hai",-17759 );
Spellput ("Han",-17752 );
Spellput ("hang",-17733 );
Spellput ("hao",-17730 );
Spellput ("he",-17721 );
Spellput ("Hei",-17703 );
Spellput ("success",-17701 );
Spellput ("Heng",-17697 );
Spellput ("Hong",-17692 );
Spellput ("Hou",-17683 );
Spellput ("Hu",-17676 );
Spellput ("Hua",-17496 );
Spellput ("Huai",-17487 );
Spellput ("Huan",-17482 );
Spellput ("Huang",-17468 );
Spellput ("Hui",-17454 );
Spellput ("hun",-17433 );
Spellput ("Huo",-17427 );
Spellput ("Ji",-17417 );
Spellput ("jia",-17202 );
Spellput ("jian",-17185 );
Spellput ("Jiang",-16983 );
Spellput ("Jiao",-16970 );
Spellput ("Jie",-16942 );
Spellput ("Jin",-16915 );
Spellput ("jing",-16733 );
Spellput ("jiong",-16708 );
Spellput ("Jiu",-16706 );
Spellput ("Ju",-16689 );
Spellput ("Juan",-16664 );
Spellput ("Jue",-16657 );
Spellput ("Jun",-16647 );
Spellput ("ka",-16474 );
Spellput ("Kai",-16470 );
Spellput ("kan",-16465 );
Spellput ("Kang",-16459 );
Spellput ("kao",-16452 );
Spellput ("Ke",-16448 );
Spellput ("Ken",-16433 );
Spellput ("Keng",-16429 );
Spellput ("Kong",-16427 );
Spellput ("Kou",-16423 );
Spellput ("ku",-16419 );
Spellput ("Kua",-16412 );
Spellput ("Kuai",-16407 );
Spellput ("Kuan",-16403 );
Spellput ("Kuang",-16401 );
Spellput ("Kui",-16393 );
Spellput ("Kun",-16220 );
Spellput ("Kuo",-16216 );
Spellput ("La",-16212 );
Spellput ("Lai",-16205 );
Spellput ("Lan",-16202 );
Spellput ("Lang",-16187 );
Spellput ("Lao",-16180 );
Spellput ("Le",-16171 );
Spellput ("lei",-16169 );
Spellput ("leng",-16158 );
Spellput ("Li",-16155 );
Spellput ("Lia",-15959 );
Spellput ("Lian",-15958 );
Spellput ("Liang",-15944 );
Spellput ("liao",-15933 );
Spellput ("lie",-1, 15920 );
Spellput ("Lin",-15915 );
Spellput ("Ling",-15903 );
Spellput ("Liu",-15889 );
Spellput ("long",-1, 15878 );
Spellput ("Lou",-15707 );
Spellput ("Lu",-15701 );
Spellput ("LV",-15681 );
Spellput ("Luan",-15667 );
Spellput ("Lue",-15661 );
Spellput ("Lun",-15659 );
Spellput ("Luo",-15652 );
Spellput ("Ma",-15640 );
Spellput ("Mai",-15631 );
Spellput ("man",-1, 15625 );
Spellput ("mang",-15454 );
Spellput ("Mao",-15448 );
Spellput ("me",-15436 );
Spellput ("Mei",-15435 );
Spellput ("man",-15419 );
Spellput ("Meng",-15416 );
Spellput ("mi",-15408 );
Spellput ("Mian",-15394 );
Spellput ("Miao",-15385 );
Spellput ("mini",-15377 );
Spellput ("min",-15375 );
Spellput ("Ming",-15369 );
Spellput ("Miu",-15363 );
Spellput ("Mo",-15362 );
Spellput ("MOU",-15183 );
Spellput ("Mu",-15180 );
Spellput ("Na",-15165 );
Spellput ("Nai",-15158 );
Spellput ("Nan",-15153 );
Spellput ("Nang",-15150 );
Spellput ("Nao",-15149 );
Spellput ("ne",-15144 );
Spellput ("Nei",-15143 );
Spellput ("Nen",-15141 );
Spellput ("Neng",-15140 );
Spellput ("Ni",-15139 );
Spellput ("Nian",-15128 );
Spellput ("Niang",-15121 );
Spellput ("niao",-15119 );
Spellput ("Nie",-15117 );
Spellput ("Nin",-15110 );
Spellput ("Ning",-15109 );
Spellput ("niu",-14941 );
Spellput ("Nong",-14937 );
Spellput ("Nu",-14933 );
Spellput ("NV",-14930 );
Spellput ("Nuan",-14929 );
Spellput ("nue",-14928 );
Spellput ("nuo",-14926 );
Spellput ("O",-14922 );
Spellput ("ou",-14921 );
Spellput ("Pa",-14914 );
Spellput ("Pai",-14908 );
Spellput ("pan",-14902 );
Spellput ("Pang",-14894 );
Spellput ("pao",-14889 );
Spellput ("Pei",-1, 14882 );
Spellput ("pen",-14873 );
Spellput ("peng",-14871 );
Spellput ("Pi",-14857 );
Spellput ("pian",-14678 );
Spellput ("Piao",-14674 );
Spellput ("pie",-14670 );
Spellput ("pin",-14668 );
Spellput ("ping",-14663 );
Spellput ("po",-14654 );
Spellput ("Pu",-14645 );
Spellput ("Qi",-14630 );
Spellput ("QIA",-14594 );
Spellput ("Qian",-14429 );
Spellput ("Qiang",-14407 );
Spellput ("Qiao",-14399 );
Spellput ("Qie",-14384 );
Spellput ("Qin",-14379 );
Spellput ("Qing",-14368 );
Spellput ("Qiong",-14355 );
Spellput ("Qiu",-14353 );
Spellput ("Qu",-14345 );
Spellput ("Quan",-14170 );
Spellput ("que",-14159 );
Speller put ("Qun",-14151 );
Spellput ("ran",-14149 );
Spellput ("rang",-14145 );
Spellput ("Rao",-14140 );
Spellput ("re",-1, 14137 );
Spellput ("Ren",-14135 );
Spellput ("reng",-14125 );
Spellput ("Ri",-14123 );
Spellput ("rong",-14122 );
Spellput ("Rou",-14112 );
Spellput ("Ru",-14109 );
Spellput ("Ruan",-14099 );
Spellput ("Rui",-14097 );
Spellput ("run",-14094 );
Spellput ("RuO",-14092 );
Spellput ("sa",-14090 );
Spellput ("sai",-14087 );
Spellput ("San",-14083 );
Spellput ("sang",-13917 );
Spellput ("Sao",-13914 );
Spellput ("se",-13910 );
Spellput ("Sen",-13907 );
Spellput ("Seng",-13906 );
Spellput ("Sha",-13905 );
Spellput ("Shai",-13896 );
Spellput ("Shan",-13894 );
Spellput ("Shang",-13878 );
Spellput ("shao",-13870 );
Spellput ("she",-13859 );
Spellput ("shen",-13847 );
Spellput ("Sheng",-13831 );
Spellput ("Shi",-13658 );
Spellput ("Shou",-13611 );
Spellput ("Shu",-13601 );
Spellput ("Shua",-13406 );
Spellput ("Shuai",-13404 );
Spellput ("Shuan",-13400 );
Spellput ("Shuang",-13398 );
Spellput ("Shui",-13395 );
Spellput ("Shun",-13391 );
Spellput ("Shuo",-13387 );
Spellput ("Si",-13383 );
Spellput ("song",-13367 );
Spellput ("Sou",-13359 );
Spellput ("Su",-13356 );
Spellput ("Suan",-13343 );
Spellput ("Sui",-13340 );
Spellput ("sun",-13329 );
Spellput ("Suo",-13326 );
Spellput ("ta",-13318 );
Spellput ("Tai",-13147 );
Spellput ("Tan",-13138 );
Spellput ("Tang",-13120 );
Spellput ("Tao",-13107 );
Spellput ("te",-13096 );
Spellput ("Teng",-13095 );
Spellput ("ti",-13091 );
Spellput ("Tian",-13076 );
Spellput ("tiao",-13068 );
Spellput ("Tie",-1, 13063 );
Spellput ("ting",-1, 13060 );
Spellput ("tong",-12888 );
Spellput ("TOU",-12875 );
Spellput ("tu",-12871 );
Spellput ("tuan",-12860 );
Spellput ("Tui",-12858 );
Spellput ("tun",-12852 );
Spellput ("Tuo",-12849 );
Spellput ("wa",-12838 );
Spellput ("Wai",-12831 );
Spellput ("Wan",-12829 );
Spellput ("Wang",-1, 12812 );
Spellput ("wei",-12802 );
Spellput ("Wen",-12607 );
Spellput ("Weng",-12597 );
Spellput ("wo",-12594 );
Spellput ("Wu",-1, 12585 );
Spellput ("Xi",-12556 );
Spellput ("xia",-12359 );
Spellput ("Xian",-12346 );
Spellput ("Xiang",-12320 );
Spellput ("Xiao",-12300 );
Spellput ("xie",-12120 );
Spellput ("Xin",-12099 );
Spellput ("Xing",-12089 );
Spellput ("Xiong",-12074 );
Spellput ("Xiu",-12067 );
Spellput ("Xu",-12058 );
Spellput ("Xuan",-12039 );
Spellput ("Xue",-11867 );
Spellput ("Xun",-11861 );
Spellput ("ya",-11847 );
Spellput ("Yan",-11831 );
Spellput ("yang",-1, 11798 );
Spellput ("Yao",-11781 );
Spellput ("ye",-11604 );
Spellput ("Yi",-11589 );
Spellput ("Yin",-11536 );
Spellput ("ying",-11358 );
Spellput ("yo",-11340 );
Spellput ("Yong",-11339 );
Spellput ("you",-11324 );
Spellput ("yu",-11303 );
Spellput ("Yuan",-11097 );
Spellput ("Yue",-11077 );
Spellput ("Yun",-11067 );
Spellput ("za",-11055 );
Spellput ("Zai",-11052 );
Spellput ("Zan",-11045 );
Spellput ("Zang",-11041 );
Spellput ("Zao",-11038 );
Spellput ("ze",-11024 );
Spellput ("Zei",-11020 );
Spellput ("Zen",-11019 );
Spellput ("Zeng",-11018 );
Spellput ("Zha",-11014 );
Spellput ("Zhai",-10838 );
Spellput ("Zhan",-10832 );
Spellput ("Zhang",-10815 );
Spellput ("Zhao",-10800 );
Spellput ("zhe",-10790 );
Spellput ("Zhen",-10780 );
Spellput ("zheng",-10764 );
Spellput ("Zhi",-10587 );
Spellput ("Zhong",-10544 );
Spellput ("Zhou",-10533 );
Spellput ("zhu",-10519 );
Spellput ("zhua",-10331 );
Spellput ("zhuai",-10329 );
Spellput ("Zhuan",-10328 );
Spellput ("Zhuang",-10322 );
Spellput ("Zhui",-10315 );
Spellput ("Zhun",-10309 );
Spellput ("Zhuo",-10307 );
Spellput ("zi",-10296 );
Spellput ("Zong",-10281 );
Spellput ("zou",-10274 );
Spellput ("Zu",-10270 );
Spellput ("Zuan",-10262 );
Spellput ("Zui",-10260 );
Spellput ("Zun",-10256 );
Spellput ("Zuo",-10254 );
}

Public static void main (string [] ARGs ){
String STR = NULL;
STR = "Xie Hai 101 General Xiyu ";
System. Out. println (STR );
System. Out. println ("spell =" + cntospell. getfullspell (STR ));

STR = "Dancing Claw"., ";
System. Out. println (STR );
System. Out. println ("spell =" + cntospell. getfullspell (STR ));

STR = "La la, shame. ";
System. Out. println (STR );
System. Out. println ("spell =" + cntospell. getfullspell (STR ));

STR = "lard, pig. ";
System. Out. println (STR );
System. Out. println ("spell =" + cntospell. getfullspell (STR ));
}

Private Static void spellput (string spell, int ASCII ){
Spellmap. Put (spell, new INTEGER (ASCII ));
}
}