eloraiby / arabtype Goto Github PK

public static char correct(char prev, char next, char ch) {
  if ((ch >= ARABIC_LETTER_START) && (ch <= ARABIC_LETTER_FINAL)) {
    // covert Arabic letter - https://github.com/eloraiby/arabtype/blob/master/arabtype.c
    boolean isLa  = isLamAlef(ch, next);
    boolean isApl = isAlefPrevLam(prev, ch);

    boolean isLapl = isLa | isApl;

    // determine char to return
    if(isLapl) {
      int index = ((isLinkingType(ch) ? 1 : 0) << 1) | (isLinkingType(prev) ? 1 : 0);
      return (char)ARABIC_FORMS_B[next -  ARABIC_LETTER_START][1][index];
    }
    else {
      if (isApl) {
        return ch;  // skip previously processed lam alef
      }
      else {
        int index   = (((isArabicLetter(next) ? 1 : 0) & (isLinkingType(ch) ? 1 : 0)) << 1) | (isLinkingType(prev) ? 1 : 0);
        return (char)ARABIC_FORMS_B[ch -  ARABIC_LETTER_START][0][index];
      }
    }

    // NOTE: compact form of the above...
    // int index = ((((isLapl | isArabicLetter(next)) & isLinkingType(ch)) ? 1 : 0) << 1) | (isLinkingType(prev) ? 1 : 0);
    // int ref = (next * (isLa ? 1 : 0)) + (ch * (isLa ? 0 : 1)) - ARABIC_LETTER_START;
    // return (char)ARABIC_FORMS_B[ref][isLapl ? 1 : 0][index];
  }
  else {
    // not an Arabic letter to be converted!
    return ch;
  }
}

private static final char ARABIC_LETTER_START = 0x0621;
private static final char ARABIC_LETTER_FINAL = 0x064A;

// private static final int ENDING = 1;
private static final int INITIAL = 2;
private static final int MEDIAL = 3;

private static final int UNICODE_LAM = 0x644;

private static boolean isArabicLetter(char cp)    {
  return ( cp >= ARABIC_LETTER_START && cp <=  ARABIC_LETTER_FINAL );
}

private static boolean isLamAlef(char cp, char next)  {
  return cp == UNICODE_LAM && isArabicLetter(next) && ARABIC_FORMS_B[next - ARABIC_LETTER_START][1][INITIAL] != 0;
}

private static boolean isAlefPrevLam(char prev, char cp)  {
  return prev == UNICODE_LAM && isArabicLetter(cp) && ARABIC_FORMS_B[cp - ARABIC_LETTER_START][1][INITIAL] != 0;
}

private static boolean isLinkingType(char cp) {
  return isArabicLetter(cp) && ARABIC_FORMS_B[cp - ARABIC_LETTER_START][0][MEDIAL] != 0;
}

/** Table to convert to Arabic presentation form B. */
private static final int[][][] ARABIC_FORMS_B = {
    { {0xFE80, 0xFE80,      0,      0}, {-1, -1, 0, 0} },    // hamza  (0)
    { {0xFE81, 0xFE82,      0,      0}, {-1, -1, 0xFEF5, 0xFEF6} },  // 2alif madda  (1)
    { {0xFE83, 0xFE84,      0,      0}, {-1, -1, 0xFEF7, 0xFEF8} },  // 2alif hamza  (2)
    { {0xFE85, 0xFE86,      0,      0}, {-1, -1, 0, 0} },    // waw hamza  (3)
    { {0xFE87, 0xFE88,      0,      0}, {-1, -1, 0xFEF9, 0xFEFA} },  // 2alif hamza maksoura  (4)
    { {0xFE89, 0xFE8A, 0xFE8B, 0xFE8C}, {-1, -1, 0, 0} },    // 2alif maqsoura hamza  (5)
    { {0xFE8D, 0xFE8E,      0,      0}, {-1, -1, 0xFEFB, 0xFEFC} },  // 2alif  (6)
    { {0xFE8F, 0xFE90, 0xFE91, 0xFE92}, {-1, -1, 0, 0} },    // ba2    (7)
    { {0xFE93, 0xFE94,      0,      0}, {-1, -1, 0, 0} },    // ta2 marbouta  (8)
    { {0xFE95, 0xFE96, 0xFE97, 0xFE98}, {-1, -1, 0, 0} },    // ta2    (9)
    { {0xFE99, 0xFE9A, 0xFE9B, 0xFE9C}, {-1, -1, 0, 0} },    // tha2    (10)
    { {0xFE9D, 0xFE9E, 0xFE9F, 0xFEA0}, {-1, -1, 0, 0} },    // jim    (11)
    { {0xFEA1, 0xFEA2, 0xFEA3, 0xFEA4}, {-1, -1, 0, 0} },    // 7a2    (12)
    { {0xFEA5, 0xFEA6, 0xFEA7, 0xFEA8}, {-1, -1, 0, 0} },    // kha2    (13)
    { {0xFEA9, 0xFEAA,      0,      0}, {-1, -1, 0, 0} },    // dal    (14)
    { {0xFEAB, 0xFEAC,      0,      0}, {-1, -1, 0, 0} },    // dhal    (15)
    { {0xFEAD, 0xFEAE,      0,      0}, {-1, -1, 0, 0} },    // ra2    (16)
    { {0xFEAF, 0xFEB0,      0,      0}, {-1, -1, 0, 0} },    // zayn    (17)
    { {0xFEB1, 0xFEB2, 0xFEB3, 0xFEB4}, {-1, -1, 0, 0} },    // syn    (18)
    { {0xFEB5, 0xFEB6, 0xFEB7, 0xFEB8}, {-1, -1, 0, 0} },    // shin    (19)
    { {0xFEB9, 0xFEBA, 0xFEBB, 0xFEBC}, {-1, -1, 0, 0} },    // sad    (20)
    { {0xFEBD, 0xFEBE, 0xFEBF, 0xFEC0}, {-1, -1, 0, 0} },    // dad    (21)
    { {0xFEC1, 0xFEC2, 0xFEC3, 0xFEC4}, {-1, -1, 0, 0} },    // tah    (22)
    { {0xFEC5, 0xFEC6, 0xFEC7, 0xFEC8}, {-1, -1, 0, 0} },    // thah    (23)
    { {0xFEC9, 0xFECA, 0xFECB, 0xFECC}, {-1, -1, 0, 0} },    // 3ayn    (24)
    { {0xFECD, 0xFECE, 0xFECF, 0xFED0}, {-1, -1, 0, 0} },    // ghayn  (25)
    { {     0,      0,      0,      0}, {-1, -1, 0, 0} },    //    (26)
    { {     0,      0,      0,      0}, {-1, -1, 0, 0} },    //    (27)
    { {     0,      0,      0,      0}, {-1, -1, 0, 0} },    //    (28)
    { {     0,      0,      0,      0}, {-1, -1, 0, 0} },    //    (29)
    { {     0,      0,      0,      0}, {-1, -1, 0, 0} },    //    (30)
    { {0x0640, 0x0640, 0x0640, 0x0640}, {-1, -1, 0, 0} },    // wasla  (31)
    { {0xFED1, 0xFED2, 0xFED3, 0xFED4}, {-1, -1, 0, 0} },    // fa2    (32)
    { {0xFED5, 0xFED6, 0xFED7, 0xFED8}, {-1, -1, 0, 0} },    // qaf    (33)
    { {0xFED9, 0xFEDA, 0xFEDB, 0xFEDC}, {-1, -1, 0, 0} },    // kaf    (34)
    { {0xFEDD, 0xFEDE, 0xFEDF, 0xFEE0}, {-1, -1, 0, 0} },    // lam    (35)
    { {0xFEE1, 0xFEE2, 0xFEE3, 0xFEE4}, {-1, -1, 0, 0} },    // mim    (36)
    { {0xFEE5, 0xFEE6, 0xFEE7, 0xFEE8}, {-1, -1, 0, 0} },    // noon    (37)
    { {0xFEE9, 0xFEEA, 0xFEEB, 0xFEEC}, {-1, -1, 0, 0} },    // ha2    (38)
    { {0xFEED, 0xFEEE,      0,      0}, {-1, -1, 0, 0} },    // waw    (39)
    { {0xFEFF, 0xFEF0,      0,      0}, {-1, -1, 0, 0} },    // 2alif maksoura  (40)
    { {0xFEF1, 0xFEF2, 0xFEF3, 0xFEF4}, {-1, -1, 0, 0} },    // ya2    (41)
};