Detects whether a string has any invalid utf-8 characters

is_invalid_unicode.c

// Stolen from chatgpt
#include <stdbool.h>
#include <stddef.h>
#include <stdint.h>

bool has_invalid_utf8(const char *s)
{
    const uint8_t *p = (const uint8_t *)s;
    while (*p != '\0') {
        if (*p < 0x80) {
            // 1-byte sequence (ASCII)
            p++;
        } else if ((*p & 0xE0) == 0xC0) {
            // 2-byte sequence
            if ((p[1] & 0xC0) != 0x80 || (p[1] & 0x3E) == 0) {
                return true; // invalid sequence
            }
            p += 2;
        } else if ((*p & 0xF0) == 0xE0) {
            // 3-byte sequence
            if ((p[1] & 0xC0) != 0x80 || (p[2] & 0xC0) != 0x80 ||
                (*p == 0xE0 && (p[1] & 0x1F) == 0)) {
                return true; // invalid sequence
            }
            p += 3;
        } else if ((*p & 0xF8) == 0xF0) {
            // 4-byte sequence
            if ((p[1] & 0xC0) != 0x80 || (p[2] & 0xC0) != 0x80 ||
                (p[3] & 0xC0) != 0x80 || (*p == 0xF0 && (p[1] & 0x0F) == 0)) {
                return true; // invalid sequence
            }
            p += 4;
        } else {
            return true; // invalid start byte
        }
    }
    return false; // no invalid sequence found
}