Test UTF-8 decoder using PEXT instruction

pext_utf8_decode.cpp

#include <stdlib.h>

unsigned pext_utf8_decode(unsigned char*& buf) {
    unsigned next4;
    __builtin_memcpy(&next4, buf, 4);
    next4 = __builtin_bswap32(next4);
    if (__builtin_expect(!!(next4 >> 31), 0)) {
        // multi-byte handling
        unsigned r;
        if (((next4 >> 16) & 0b11100000'11000000) == 0b11000000'10000000) {
            __asm__("pext %1, %2, %0"
                    : "=r"(r)
                    : "r"(0b00011111'00111111'00000000'00000000u), "r"(next4));
            buf += 2;
            return r;
        } else if (((next4 >> 8) & 0b11110000'11000000'11000000) ==
                   0b11100000'10000000'10000000) {
            __asm__("pext %1, %2, %0"
                    : "=r"(r)
                    : "r"(0b00001111'00111111'00111111'00000000u), "r"(next4));
            buf += 3;
            return r;
        } else if ((next4 & 0b11111000'11000000'11000000'11000000u) ==
                   0b11110000'10000000'10000000'10000000u) {
            __asm__("pext %1, %2, %0"
                    : "=r"(r)
                    : "r"(0b00000111'00111111'00111111'00111111u), "r"(next4));
            buf += 4;
            return r;
        } else {
            abort();
            __builtin_unreachable();
        }
    } else {
        buf++;
        return next4 >> 24;
    }
}

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

int main() {
    {
        unsigned char b1[] = {0xe2, 0x82, 0xac, 0};
        unsigned char* buf = b1;
        assert(pext_utf8_decode(buf) == 0x20ac);
        assert(buf == b1 + 3);
    }
    {
        unsigned char b2[] = {'g', 0, 0, 0};
        unsigned char* buf = b2;
        assert(pext_utf8_decode(buf) == (unsigned) 'g');
        assert(buf == b2 + 1);
    }
    {
      unsigned char b3[] = {0xc2, 0xa2, 0, 0};
        unsigned char* buf = b3;
        assert(pext_utf8_decode(buf) == 0xa2);
        assert(buf == b3 + 2);
    }
    {
      unsigned char b4[] = {0xe0, 0xa4, 0xb9, 0};
      unsigned char* buf = b4;
      assert(pext_utf8_decode(buf) == 0x939);
      assert(buf == b4 + 3);
    }
    {
      unsigned char b5[] = {0xe2, 0x82, 0xac, 0};
      unsigned char* buf = b5;
      assert(pext_utf8_decode(buf) == 0x20ac);
      assert(buf == b5 + 3);
    }
    {
      unsigned char b6[] = {0xf0, 0x90, 0x8d, 0x88, 0};
      unsigned char* buf = b6;
      assert(pext_utf8_decode(buf) == 0x10348);
      assert(buf == b6 + 4);
    }
}