aboutsummaryrefslogtreecommitdiffstats
path: root/contrib/simdutf/src/scalar/utf32_to_utf8/valid_utf32_to_utf8.h
blob: 7a7cdd56848193fc8bb736528865fb529e7e36bb (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
#ifndef SIMDUTF_VALID_UTF32_TO_UTF8_H
#define SIMDUTF_VALID_UTF32_TO_UTF8_H

namespace simdutf {
namespace scalar {
namespace {
namespace utf32_to_utf8 {

#if SIMDUTF_IMPLEMENTATION_FALLBACK || SIMDUTF_IMPLEMENTATION_PPC64
// only used by the fallback and POWER kernel
inline size_t convert_valid(const char32_t *buf, size_t len,
                            char *utf8_output) {
  const uint32_t *data = reinterpret_cast<const uint32_t *>(buf);
  size_t pos = 0;
  char *start{utf8_output};
  while (pos < len) {
    // try to convert the next block of 2 ASCII characters
    if (pos + 2 <=
        len) { // if it is safe to read 8 more bytes, check that they are ascii
      uint64_t v;
      ::memcpy(&v, data + pos, sizeof(uint64_t));
      if ((v & 0xFFFFFF80FFFFFF80) == 0) {
        *utf8_output++ = char(buf[pos]);
        *utf8_output++ = char(buf[pos + 1]);
        pos += 2;
        continue;
      }
    }
    uint32_t word = data[pos];
    if ((word & 0xFFFFFF80) == 0) {
      // will generate one UTF-8 bytes
      *utf8_output++ = char(word);
      pos++;
    } else if ((word & 0xFFFFF800) == 0) {
      // will generate two UTF-8 bytes
      // we have 0b110XXXXX 0b10XXXXXX
      *utf8_output++ = char((word >> 6) | 0b11000000);
      *utf8_output++ = char((word & 0b111111) | 0b10000000);
      pos++;
    } else if ((word & 0xFFFF0000) == 0) {
      // will generate three UTF-8 bytes
      // we have 0b1110XXXX 0b10XXXXXX 0b10XXXXXX
      *utf8_output++ = char((word >> 12) | 0b11100000);
      *utf8_output++ = char(((word >> 6) & 0b111111) | 0b10000000);
      *utf8_output++ = char((word & 0b111111) | 0b10000000);
      pos++;
    } else {
      // will generate four UTF-8 bytes
      // we have 0b11110XXX 0b10XXXXXX 0b10XXXXXX 0b10XXXXXX
      *utf8_output++ = char((word >> 18) | 0b11110000);
      *utf8_output++ = char(((word >> 12) & 0b111111) | 0b10000000);
      *utf8_output++ = char(((word >> 6) & 0b111111) | 0b10000000);
      *utf8_output++ = char((word & 0b111111) | 0b10000000);
      pos++;
    }
  }
  return utf8_output - start;
}
#endif // SIMDUTF_IMPLEMENTATION_FALLBACK || SIMDUTF_IMPLEMENTATION_PPC64

} // namespace utf32_to_utf8
} // unnamed namespace
} // namespace scalar
} // namespace simdutf

#endif