Question

I’d like to iterate through Unicode characters, gobbling up all combining characters that follow the initial code point.

This is what I have so far, but it acts really weird for some random Unicode sequences I tried: (for example, when I pass it things like “a̔” (U+0061 LATIN SMALL LETTER A followed by U+0314 COMBINING REVERSED COMMA ABOVE) it sees it as two characters rather than one. Other things, like “e︠” (U+0065 LATIN SMALL LETTER E followed by U+FE20 COMBINING LIGATURE LEFT HALF) are seen as one character)

int COMBINING[] = {
    0x0300, 0x036F,
    0x1DC0, 0x1DFF,
    0x20D0, 0x20FF,
    0xFE20, 0xFE2F,
    0 //sentinel
};

utf8_index_t ut_nextchar(utf8_t source, utf8_index_t curr)
{
    int c = decode_cp(source, &curr);
    int comb = 0;
    if (c == 0)
        return -1;
    while (COMBINING[comb] != 0)
    {
        for (comb = 0; COMBINING[comb] != 0; comb += 2)
        {
            if (c >= COMBINING[comb] && c <= COMBINING[comb + 1])
            {
                c = decode_cp(source, &curr);
                if (c == 0)
                    return -1;
                break;
            }
        }
    }
    return curr;
}

Answer 1

Actually, Unicode characters are mostly 1:1 to Unicode codepoints – what you’re interested in are Unicode grapheme clusters, which correspond to so-called user-perceived characters.

You can find my implementation of the algorithm, including property data, here at bitbucket.

If you’re not interested in the full algorithm, you can use

gc_break_property(c) == GC_BP_Extend

to check for characters with property Grapheme_Extend and

gc_break_property(c) & GC_FLAG_POSTFIX

if you want to include spacing marks as well.