ccore/src/Unicode.c

#include "core/Unicode.h"

UTF8 convertUnicodeToUTF8(u32 c) {
    UTF8 u = {0};
    if(c >= 0x10000) {
        u.length = 4;
        u.data[0] = 0b1111'0000 | ((c >> 18) & 0b0000'0111);
        u.data[1] = 0b1000'0000 | ((c >> 12) & 0b0011'1111);
        u.data[2] = 0b1000'0000 | ((c >> 6) & 0b0011'1111);
        u.data[3] = 0b1000'0000 | (c & 0b0011'1111);
    } else if(c >= 0x800) {
        u.length = 3;
        u.data[0] = 0b1110'0000 | ((c >> 12) & 0b0000'1111);
        u.data[1] = 0b1000'0000 | ((c >> 6) & 0b0011'1111);
        u.data[2] = 0b1000'0000 | (c & 0b0011'1111);
    } else if(c >= 0x80) {
        u.length = 2;
        u.data[0] = 0b1100'0000 | ((c >> 6) & 0b0001'1111);
        u.data[1] = 0b1000'0000 | (c & 0b0011'1111);
    } else {
        u.length = 1;
        u.data[0] = c & 0b0111'1111;
    }
    return u;
}

u32 convertUTF8toUnicode(UTF8 c) {
    if(c.length == 4) {
        return ((c.data[0] & 0b0000'0111u) << 18) |
               ((c.data[1] & 0b0011'1111u) << 12) |
               ((c.data[2] & 0b0011'1111u) << 6) | (c.data[3] & 0b0011'1111u);
    } else if(c.length == 3) {
        return ((c.data[0] & 0b0000'1111u) << 12) |
               ((c.data[1] & 0b0011'1111u) << 6) | (c.data[2] & 0b0011'1111u);
    } else if(c.length == 2) {
        return ((c.data[0] & 0b0001'1111u) << 6) | (c.data[1] & 0b0011'1111u);
    }
    return c.data[0];
}

bool isUTF8Remainder(u8 c) {
    return (c & 0b1100'0000) == 0b1000'0000;
}

u32 getUTF8Length(u8 c) {
    if((c & 0b1111'1000) == 0b1111'0000) {
        return 4;
    } else if((c & 0b1111'0000) == 0b1110'0000) {
        return 3;
    } else if((c & 0b1110'0000) == 0b1100'0000) {
        return 2;
    }
    return 1;
}