utf.c

#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <assert.h>

#define UTF_INVALID 0xFFFD

static const char*
utf_dec(unsigned *dst, const char *p, const char *e)
{
    assert(e);
    assert(p);
    assert(dst);

    int n = 0;
    unsigned res = 0;
    switch (*p & 0xf0) {
    case 0xf0:  res = *p & 0x07; n = 3;  break;
    case 0xe0:  res = *p & 0x0f; n = 2;  break;
    case 0xd0:
    case 0xc0:  res = *p & 0x1f; n = 1;  break;
    default:    res = *p & 0xff; n = 0;  break;}
    
    if (p + n >= e) {
        *dst = UTF_INVALID;
        return e;
    }
    while (n--) res = (res << 6) | (*(++p) & 0x3f);
    *dst = res;
    return p + 1;
}
static unsigned
utf_decode(const char **p, const char *e)
{
    assert(p);
    assert(e);

    unsigned rune = 0;
    *p = utf_dec(&rune, *p, e);
    return rune;
}
static const char*
utf_prev(const char *b, const char *s)
{
    assert(b);
    assert(s);

    while (s > b) {
        char c = *(--s);
        if ((((c) & 0xC0) != 0x80))
            return s;
    } return 0;
}
static const char*
utf_at(const char *str, const char *end, int idx)
{
    assert(str);
    assert(idx >= 0);
    unsigned rune = 0u;
    end = !end ? str + strlen(str): end;
    for (int i = 0; i < idx && str < end; ++i)
        str = utf_dec(&rune, str, end);
    return str;
}
static int
utf_len(const char *str, const char *end)
{
    int n = 0;  
    end = !end ? str + strlen(str): end;
    while (str < end && *str) {
        unsigned rune = 0;
        str = utf_dec(&rune, str, end);
        n++;
    } return n;
}
int main(void)
{
#if 0
    const char txt[] = "„Führ“";
    const char *end = txt + sizeof(txt)-1;

    const char *p = txt;
    while (p < end) {
        unsigned rune = 0;
        p = utf_dec(&rune, p,end);
        printf("%u\n", rune);
    }
#endif
    
#if 0
    const char txt[] = "„Führ“";
    const char *end = txt + sizeof(txt)-1;

    unsigned rune = 0;
    const char *p = utf_dec(&rune, utf_at(txt,end,2), end);
    assert(rune == 252);

    const char *x = utf_dec(&rune, utf_at(p,end,4), end);
    assert(rune == UTF_INVALID);
    assert(x == end);
#endif

#if 0   
    const char txt[] = "test";
    const char *end = txt + sizeof(txt)-1;

    const char *p = end;
    while ((p = utf_prev(txt, p))) {
        unsigned rune = 0;
        utf_dec(&rune, p,end);
        printf("%c\n", (char)rune);
    }
#endif

#if 1
    const char txt[] = "„Führ“";
    const char *end = txt + sizeof(txt)-1;

    const char *p = txt;
    while (p < end) {
        unsigned rune = utf_decode(&p,end);
        printf("%u\n", rune);
    }
#endif

    return 0;
}