/*
 奄管: 廃越 稽原切妊奄
 尻姥鯵降: 績庚発, 2004鰍
 壕匂: 切政罫惟
*/

/*
厩嬢税 稽原切 妊奄狛:
薦1舌 妊奄税 奄沙 据帳
  薦1牌 厩嬢税 稽原切 妊奄澗 厩嬢税 妊層 降製狛拭 魚虞 旋澗 依聖 据帳生稽 廃陥.
  薦2牌 稽原切 戚須税 採硲澗 鞠亀系 紫遂馬走 省澗陥.
*/
/*
戚 員拭辞澗 降製戚 焼観 庚切研 奄層生稽 梅製.
託板 降製拭 魚献 妊奄狛聖 蓄亜拝 森舛績.
廃越庚切税 切社研 歳軒馬奄 是背辞 廃越繕杯莫 坪球凪戚走研 紫遂馬心製.
刃失莫 坪球凪戚走拭澗 嬢恐 鋭帳戚 赤澗走 焼送 歳汐馬走 公梅製.
*/

#include "KoreanRomanNotation.h"

///////////////////////////////////////////////////////////////////////
namespace Koreanromannotation
{

const char **Cho_seong_roman = 0;
const char **Jung_seong_roman = 0;
const char **Jong_seong_roman = 0;

///////////////////////////////////////////////////////////////////////
/*廃越引 奄展 庚切亜 嬢酔君遭 庚切伸(str)税 廃越 稽原切妊奄研 姥敗
 by_pronunciation : 降製 奄層生稽 稽原切稽 痕発拝走 食採
 each_char_to_capital : 唖 庚切税 稽原切妊奄 衣引 掻 湛庚切研 企庚切稽 拝走 食採
 space_each_char : 廃越 奄層 唖 庚切 紫戚研 句随走 食採
*/
string_type RomanFromKor(const string_type& str,bool by_pronunciation,bool each_char_to_capital,bool space_each_char)
{
 unsigned char *p = (unsigned char *)ConvertCodePage(str.c_str(),CP_KOREAN_WANSUNG,CP_KOREAN_JOHAB,0);
 char buff[128],*p2;
 TCharInfo prev_char_info,char_info;
 string_type res;
 prev_char_info.Clear();
 if(by_pronunciation)
 {
  Cho_seong_roman = (const char **)Cho_seong_roman_by_pronunciation;
  Jung_seong_roman = (const char **)Jung_seong_roman_by_pronunciation;
  Jong_seong_roman = (const char **)Jong_seong_roman_by_pronunciation;
 }
 else
 {
  Cho_seong_roman = (const char **)Cho_seong_roman_by_char;
  Jung_seong_roman = (const char **)Jung_seong_roman_by_char;
  Jong_seong_roman = (const char **)Jong_seong_roman_by_char;
 }

 for(int i=0; p[i] ;++i)
 {
  //廃越:84~D3  焼什徹据庚切析嬢去:D8~DE  廃切:E0~F9
  if(p[i]>=0x84 && p[i]<=0xD3)
  {
    //是税 if庚拭辞 廃越庚切績聖 溌昔馬心生糠稽 焼掘 敗呪硲窒税 鋼発葵精 牌雌 true.
    //幻鉦 鋼発葵戚 true亜 焼艦虞檎 獄益戚糠稽 郊稽説焼醤 敗.
    RomanFromKorChar(*((KorCodeType*)(p+i)),char_info);
    //降製拭 魚献 稽原切 妊奄澗 製錘 薄雌聖 坦軒背醤 敗.
    if(by_pronunciation) Pronounce(prev_char_info,char_info,res);
    if(i>0)
    {
     //'随肩亀'拭辞人 旭戚  'ぉぉ'昔 井酔 'll'稽 背醤 敗.
     if(!by_pronunciation && prev_char_info.cho_seong!=0x4184 && prev_char_info.jong_seong==0x4984 && char_info.cho_seong==0x419C && char_info.jung_seong!=0x4184)
     //if(!by_pronunciation && prev_char_info.cho_seong!=0x4184 && prev_char_info.jong_seong_roman=="l" && char_info.cho_seong_roman=="r" && char_info.jung_seong!=0x4184)
     {
      res[res.size()-1]='l';
      char_info.cho_seong_roman = "l";
     }
     if(space_each_char) res += " ";
    }
    if(each_char_to_capital)
    {
     if(char_info.cho_seong_roman.size()!=0) char_info.cho_seong_roman[0] -= ('a'-'A');
     else if(char_info.jung_seong_roman.size()!=0) char_info.jung_seong_roman[0] -= ('a'-'A');
    }
    res += char_info.cho_seong_roman + char_info.jung_seong_roman + char_info.jong_seong_roman;
    prev_char_info = char_info;
    ++i;
  }

  else
  {
   //ASCII 坪球税 井酔
   if(p[i]<128)
   {
     res += p[i];
   }
   //析嬢庚切, 廃切, 据庚切 去, 陥獣 刃失莫生稽 痕発
   else
   {
     buff[0]=(char)p[i];
     buff[1]=(char)p[i+1];
     buff[2]=0;
     p2 = ConvertCodePage(buff,CP_KOREAN_JOHAB,CP_KOREAN_WANSUNG,0);
     res += p2;
     delete[] p2;
     ++i;
   }
   prev_char_info.Clear();
  }
 }//for(int i=0;
 delete[] p;
 return res;
}
 
///////////////////////////////////////////////////////////////////////
//廃越 廃 庚切税 稽原切妊奄研 姥敗
bool RomanFromKorChar(KorCodeType kor_code,TCharInfo& char_info)
{
 int idx1,idx2,idx3;
 char_info.kor_code = kor_code;
 char_info.Parse();
 idx1 = GetRomanCodeIndexFromChoseong(char_info.cho_seong);
 if(idx1>=0) char_info.cho_seong_roman = Cho_seong_roman[idx1];
 else return false;
 idx2 = GetRomanCodeIndexFromJungseong(char_info.jung_seong);
 if(idx2>=0) char_info.jung_seong_roman = Jung_seong_roman[idx2];
 else return false;
 idx3 = GetRomanCodeIndexFromJongseong(char_info.jong_seong);
 if(idx3>=0) char_info.jong_seong_roman = Jong_seong_roman[idx3];
 else return false;
 //段失生稽幻 戚欠嬢遭 庚切 'し' 坦軒
 //戚 井酔 'ng'稽 坦軒馬延 梅走幻 限澗 依昔走 税庚戚 姶
 if(idx1==12&&char_info.cho_seong_roman.empty()&&idx2==0&&idx3==0) char_info.cho_seong_roman="ng";
 return true;
}

///////////////////////////////////////////////////////////////////////
//繕杯莫 坪球凪戚走拭辞 段失精 0x4184拭辞 獣拙馬食 唖 坪球娃税 娃維戚 4績(0x4184澗 段失 蒸製績).  => 0x4184~0x41D0
int GetRomanCodeIndexFromChoseong(KorCodeType choseong_code)
{
 int idx = (choseong_code-0x4184);
 if(choseong_code>=0x4184&&choseong_code<=0x41D0 && idx%4==0) return (idx/4);
 else return -1;
}

///////////////////////////////////////////////////////////////////////
//繕杯莫 坪球凪戚走拭辞 掻失精 0x4184拭辞 獣拙馬食 坪球税 砧 郊戚闘研 蝕穿獣佃兜紹聖 凶 唖 坪球娃税 娃維戚 0x20績(0x4184澗 掻失 蒸製績). => 0x4184~0xA187
int GetRomanCodeIndexFromJungseong(KorCodeType jungseong_code)
{
 KorCodeType inverse_code = (jungseong_code>>8) | (jungseong_code<<8);
 int idx = (inverse_code-0x8441); 
 if(inverse_code>=0x8441&&inverse_code<=0x87A1 && idx%0x20==0) return (idx/0x0020);
 else return -1;
}

///////////////////////////////////////////////////////////////////////
//繕杯莫 坪球凪戚走拭辞 曽失精 0x4184拭辞 獣拙馬食 坪球税 砧 郊戚闘研 蝕穿獣佃兜紹聖 凶 唖 坪球娃税 娃維戚 1績(0x4184澗 曽失 蒸製績). => 0x4184~0x5D84
int GetRomanCodeIndexFromJongseong(KorCodeType jongseong_code)
{
 KorCodeType inverse_code = (jongseong_code>>8) | (jongseong_code<<8);
 if(inverse_code>=0x8441&&inverse_code<=0x845D) return ((inverse_code-0x8441));
 else return -1;
}

///////////////////////////////////////////////////////////////////////
//段/掻/曽失 掻 嬢恐 依昔走 毒舘
PhonemeType WhatPhoneme(KorCodeType phoneme)
{
 KorCodeType inverse_code = (phoneme>>8) | (phoneme<<8);
 //曽失
 if(inverse_code>=0x8441 && inverse_code<=0x845D) return enpt_jong_seong;
 //掻失
 else if(inverse_code>=0x8441 && inverse_code<=0x87A1 &&((inverse_code-0x8441)%0x20)==0) enpt_jung_seong;
 //段失
 else if(phoneme>=0x4184 && phoneme<=0x41D0 && (phoneme-0x4184)%4==0) return enpt_cho_seong;
 //廃越戚走幻 製社亜 焼還
 else if((phoneme&0x00FF)>=0x84 && (phoneme&0x00FF)<=0xD3) return enpt_char;
 //廃越戚 焼還
 return enpt_invalid;
}

///////////////////////////////////////////////////////////////////////
//廃越 降製拭 魚献 稽原切妊奄 姥薄
void Pronounce(TCharInfo& prev_char_info,TCharInfo& char_info,string_type& res)
{
//耕刃税 引薦
}

///////////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////////
char* ConvertCodePage(const char *buff,UINT nCPFrom,UINT nCPTo, DWORD dwFlags)
{
 wchar_t *wcs=mbcs2wcs(buff,nCPFrom,dwFlags);
 if(!wcs) return NULL;
 char *mbcs=wcs2mbcs(wcs,nCPTo,0);
 delete[] wcs;
 return mbcs;
}

///////////////////////////////////////////////////////////////////////
wchar_t* mbcs2wcs(const char *mbcs, UINT cp, DWORD dwFlags)
{
 if(!mbcs) return NULL;
 if(!mbcs[0]) return NULL;
 //int mbcsLen = strlen(mbcs);
 int wcsLen = MultiByteToWideChar(cp, dwFlags, mbcs, -1, NULL, 0 );
 if(wcsLen==0) return NULL;
 wchar_t* wcs = new wchar_t[wcsLen+1];
 if(!wcs) return NULL;
 if(MultiByteToWideChar(cp, 0, mbcs, -1, wcs, wcsLen)==0) wcs[0]=0;
 else wcs[wcsLen]=0;	// null-terminate
 return wcs;
}

///////////////////////////////////////////////////////////////////////
char* wcs2mbcs(const wchar_t *wcs, UINT cp, DWORD dwFlags)
{
 if(!wcs) return NULL;
 if(!wcs[0]) return NULL;
 //int wcsLen = wcslen(wcs);
 int mbcsLen = WideCharToMultiByte(cp, 0, wcs, -1, NULL, 0, NULL, NULL);
 if(mbcsLen==0) return NULL;
 char* mbcs = new char[mbcsLen+1];
 if(!mbcs) return NULL;
 if(WideCharToMultiByte(cp, 0, wcs, -1, mbcs, mbcsLen, NULL, NULL)==0) mbcs[0]=0;
 else mbcs[mbcsLen]=0;	// null-terminate
 return mbcs;
}
///////////////////////////////////////////////////////////////////////


const char* const Cho_seong_roman_by_char[]={
   ""
  ,"g" //ぁ
  ,"kk" //あ
  ,"n" //い
  ,"d" //ぇ
  ,"tt" //え
  ,"r" //ぉ  //舘 '随肩亀'拭辞人 旭戚 ぉぉ精 ll稽 妊獣背醤 敗.
  ,"m" //け
  ,"b" //げ
  ,"pp" //こ
  ,"s" //さ
  ,"ss" //ざ
  ,"" //し
  ,"j" //じ
  ,"jj" //す
  ,"ch" //ず
  ,"k" //せ
  ,"t" //ぜ
  ,"p" //そ
  ,"h" //ぞ
  ,0
 };

const char* const Jung_seong_roman_by_char[]={
   ""
  ,"a" //た
  ,"ae" //だ
  ,"ya" //ち
  ,"yae" //ぢ
  ,"eo" //っ
  ,"?" //?
  ,"?" //?
  ,"e" //つ
  ,"yeo" //づ
  ,"ye" //て
  ,"o" //で
  ,"wa" //と
  ,"wae" //ど
  ,"?" //?
  ,"?" //?
  ,"oe" //な
  ,"yo"  //に
  ,"u"  //ぬ
  ,"wo" //ね
  ,"we" //の
  ,"wi" //は
  ,"?" //?
  ,"?" //?
  ,"yu" //ば
  ,"eu"  //ぱ
  ,"ui" //ひ
  ,"i"  //び
  ,0
};
  
const char* const Jong_seong_roman_by_char[]={
   ""
  ,"k" //ぁ
  ,"kk" //あ
  ,"ks" //ぃ
  ,"n" //い
  ,"nj" //ぅ
  ,"nh" //う
  ,"t" //ぇ
  ,"l" //ぉ
  ,"lk" //お
  ,"lm" //か
  ,"lp" //が
  ,"ls" //き
  ,"lt" //ぎ
  ,"lp" //く
  ,"lh" //ぐ
  ,"m" //け
  ,"?" //?
  ,"p" //げ
  ,"ps" //ご
  ,"s" //さ
  ,"ss" //ざ
  ,"ng" //し
  ,"j" //じ
  ,"ch" //ず
  ,"k" //せ
  ,"t" //ぜ
  ,"p" //そ
  ,"h" //ぞ
  ,0
};

//焼掘 妊 3鯵澗 呪舛背醤 拝 員戚 赤壱 薄仙 益 鎧遂聖 紫遂馬走 省製
const char* const Cho_seong_roman_by_pronunciation[]={
   ""
  ,"g" //ぁ
  ,"kk" //あ
  ,"n" //い
  ,"d" //ぇ
  ,"tt" //え
  ,"r" //ぉ  //舘 '随肩亀'拭辞人 旭戚 ぉぉ精 ll稽 妊獣背醤 敗.
  ,"m" //け
  ,"b" //げ
  ,"pp" //こ
  ,"s" //さ
  ,"ss" //ざ
  ,"" //し
  ,"j" //じ
  ,"jj" //す
  ,"ch" //ず
  ,"k" //せ
  ,"t" //ぜ
  ,"p" //そ
  ,"h" //ぞ
  ,0
};
const char* const Jung_seong_roman_by_pronunciation[]={
   ""
  ,"a" //た
  ,"ae" //だ
  ,"ya" //ち
  ,"yae" //ぢ
  ,"eo" //っ
  ,"?" //?
  ,"?" //?
  ,"e" //つ
  ,"yeo" //づ
  ,"ye" //て
  ,"o" //で
  ,"wa" //と
  ,"wae" //ど
  ,"?" //?
  ,"?" //?
  ,"oe" //な
  ,"yo"  //に
  ,"u"  //ぬ
  ,"wo" //ね
  ,"we" //の
  ,"wi" //は
  ,"?" //?
  ,"?" //?
  ,"yu" //ば
  ,"eu"  //ぱ
  ,"ui" //ひ
  ,"i"  //び
  ,0
};
const char* const Jong_seong_roman_by_pronunciation[]={
   ""
  ,"k" //ぁ
  ,"kk" //あ
  ,"ks" //ぃ
  ,"n" //い
  ,"nj" //ぅ
  ,"nh" //う
  ,"t" //ぇ
  ,"l" //ぉ
  ,"lk" //お
  ,"lm" //か
  ,"lp" //が
  ,"ls" //き
  ,"lt" //ぎ
  ,"lp" //く
  ,"lh" //ぐ
  ,"m" //け
  ,"?" //?
  ,"p" //げ
  ,"ps" //ご
  ,"s" //さ
  ,"ss" //ざ
  ,"ng" //し
  ,"j" //じ
  ,"ch" //ず
  ,"k" //せ
  ,"t" //ぜ
  ,"p" //そ
  ,"h" //ぞ
  ,0
};

}//namespace Koreanromannotation