static void
TestPrintUnsignedFast32()
{
PoisonOutput();
sprintf(output, "%" PRIoFAST32, uint_fast32_t(0424242));
MOZ_RELEASE_ASSERT(!strcmp(output, "424242"));
PoisonOutput();
sprintf(output, "%" PRIuFAST32, uint_fast32_t(171717));
MOZ_RELEASE_ASSERT(!strcmp(output, "171717"));
PoisonOutput();
sprintf(output, "%" PRIxFAST32, uint_fast32_t(0x2a2a2a));
MOZ_RELEASE_ASSERT(!strcmp(output, "2a2a2a"));
PoisonOutput();
sprintf(output, "%" PRIXFAST32, uint_fast32_t(0xCDCDCD));
MOZ_RELEASE_ASSERT(!strcmp(output, "CDCDCD"));
}
inline bool Utf8x16<Char16, Traits16>::to_utf16(const char*& in_begin, const char* const in_end, Char16*& out_begin,
Char16* const out_end)
{
typedef std::char_traits<char> traits8;
bool invalid = false;
const char* in = in_begin;
Char16* out = out_begin;
while (in != in_end) {
if (REALM_UNLIKELY(out == out_end)) {
break; // Need space in output buffer
}
REALM_ASSERT(&in[0] >= in_begin && &in[0] < in_end);
uint_fast16_t v1 = uint_fast16_t(traits8::to_int_type(in[0]));
if (REALM_LIKELY(v1 < 0x80)) { // One byte
// UTF-8 layout: 0xxxxxxx
*out++ = Traits16::to_char_type(v1);
in += 1;
continue;
}
if (REALM_UNLIKELY(v1 < 0xC0)) {
invalid = true;
break; // Invalid first byte of UTF-8 sequence
}
if (REALM_LIKELY(v1 < 0xE0)) { // Two bytes
if (REALM_UNLIKELY(in_end - in < 2)) {
invalid = true;
break; // Incomplete UTF-8 sequence
}
REALM_ASSERT(&in[1] >= in_begin && &in[1] < in_end);
uint_fast16_t v2 = uint_fast16_t(traits8::to_int_type(in[1]));
// UTF-8 layout: 110xxxxx 10xxxxxx
if (REALM_UNLIKELY((v2 & 0xC0) != 0x80)) {
invalid = true;
break; // Invalid continuation byte
}
uint_fast16_t v = uint_fast16_t(((v1 & 0x1F) << 6) | ((v2 & 0x3F) << 0));
if (REALM_UNLIKELY(v < 0x80)) {
invalid = true;
break; // Overlong encoding is invalid
}
*out++ = Traits16::to_char_type(v);
in += 2;
continue;
}
if (REALM_LIKELY(v1 < 0xF0)) { // Three bytes
if (REALM_UNLIKELY(in_end - in < 3)) {
invalid = true;
break; // Incomplete UTF-8 sequence
}
REALM_ASSERT(&in[1] >= in_begin && &in[2] < in_end);
uint_fast16_t v2 = uint_fast16_t(traits8::to_int_type(in[1]));
uint_fast16_t v3 = uint_fast16_t(traits8::to_int_type(in[2]));
// UTF-8 layout: 1110xxxx 10xxxxxx 10xxxxxx
if (REALM_UNLIKELY((v2 & 0xC0) != 0x80 || (v3 & 0xC0) != 0x80)) {
invalid = true;
break; // Invalid continuation byte
}
uint_fast16_t v = uint_fast16_t(((v1 & 0x0F) << 12) | ((v2 & 0x3F) << 6) | ((v3 & 0x3F) << 0));
if (REALM_UNLIKELY(v < 0x800)) {
invalid = true;
break; // Overlong encoding is invalid
}
if (REALM_UNLIKELY(0xD800 <= v && v < 0xE000)) {
invalid = true;
break; // Illegal code point range (reserved for UTF-16 surrogate pairs)
}
*out++ = Traits16::to_char_type(v);
in += 3;
continue;
}
if (REALM_UNLIKELY(out + 1 == out_end)) {
break; // Need space in output buffer for surrogate pair
}
if (REALM_LIKELY(v1 < 0xF8)) { // Four bytes
if (REALM_UNLIKELY(in_end - in < 4)) {
invalid = true;
break; // Incomplete UTF-8 sequence
}
uint_fast32_t w1 = uint_fast32_t(v1); // 16 bit -> 32 bit
REALM_ASSERT(&in[1] >= in_begin && &in[3] < in_end);
uint_fast32_t v2 = uint_fast32_t(traits8::to_int_type(in[1])); // 32 bit intended
uint_fast16_t v3 = uint_fast16_t(traits8::to_int_type(in[2])); // 16 bit intended
uint_fast16_t v4 = uint_fast16_t(traits8::to_int_type(in[3])); // 16 bit intended
// UTF-8 layout: 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
if (REALM_UNLIKELY((v2 & 0xC0) != 0x80 || (v3 & 0xC0) != 0x80 || (v4 & 0xC0) != 0x80)) {
invalid = true;
break; // Invalid continuation byte
}
uint_fast32_t v = uint_fast32_t(((w1 & 0x07) << 18) | // Parenthesis is 32 bit partial result
((v2 & 0x3F) << 12) | // Parenthesis is 32 bit partial result
((v3 & 0x3F) << 6) | // Parenthesis is 16 bit partial result
((v4 & 0x3F) << 0)); // Parenthesis is 16 bit partial result
if (REALM_UNLIKELY(v < 0x10000)) {
invalid = true;
break; // Overlong encoding is invalid
}
if (REALM_UNLIKELY(0x110000 <= v)) {
invalid = true;
break; // Code point too big for UTF-16
}
v -= 0x10000l;
*out++ = Traits16::to_char_type(0xD800 + (v / 0x400));
*out++ = Traits16::to_char_type(0xDC00 + (v % 0x400));
in += 4;
continue;
}
// Invalid first byte of UTF-8 sequence, or code point too big for UTF-16
invalid = true;
break;
}
REALM_ASSERT(in >= in_begin && in <= in_end);
REALM_ASSERT(out >= out_begin && out <= out_end);
in_begin = in;
out_begin = out;
return !invalid;
}
inline bool Utf8x16<Char16, Traits16>::to_utf8(const Char16*& in_begin, const Char16* const in_end, char*& out_begin,
char* const out_end)
{
typedef std::char_traits<char> traits8;
typedef typename traits8::int_type traits8_int_type;
bool invalid = false;
const Char16* in = in_begin;
char* out = out_begin;
while (in != in_end) {
REALM_ASSERT(&in[0] >= in_begin && &in[0] < in_end);
uint_fast16_t v1 = uint_fast16_t(Traits16::to_int_type(in[0]));
if (REALM_LIKELY(v1 < 0x80)) {
if (REALM_UNLIKELY(out == out_end)) {
break; // Not enough output buffer space
}
// UTF-8 layout: 0xxxxxxx
REALM_ASSERT(out >= out_begin && out < out_end);
*out++ = traits8::to_char_type(traits8_int_type(v1));
in += 1;
continue;
}
if (REALM_LIKELY(v1 < 0x800)) {
if (REALM_UNLIKELY(out_end - out < 2)) {
break; // Not enough output buffer space
}
// UTF-8 layout: 110xxxxx 10xxxxxx
*out++ = traits8::to_char_type(traits8_int_type(0xC0 + v1 / 0x40));
REALM_ASSERT(out >= out_begin && out < out_end);
*out++ = traits8::to_char_type(traits8_int_type(0x80 + v1 % 0x40));
in += 1;
continue;
}
if (REALM_LIKELY(v1 < 0xD800 || 0xE000 <= v1)) {
if (REALM_UNLIKELY(out_end - out < 3)) {
break; // Not enough output buffer space
}
// UTF-8 layout: 1110xxxx 10xxxxxx 10xxxxxx
REALM_ASSERT(out >= out_begin && out + 2 < out_end);
*out++ = traits8::to_char_type(traits8_int_type(0xE0 + v1 / 0x1000));
*out++ = traits8::to_char_type(traits8_int_type(0x80 + v1 / 0x40 % 0x40));
*out++ = traits8::to_char_type(traits8_int_type(0x80 + v1 % 0x40));
in += 1;
continue;
}
// Surrogate pair
if (REALM_UNLIKELY(out_end - out < 4)) {
break; // Not enough output buffer space
}
if (REALM_UNLIKELY(0xDC00 <= v1)) {
invalid = true;
break; // Invalid first half of surrogate pair
}
if (REALM_UNLIKELY(in + 1 == in_end)) {
invalid = true;
break; // Incomplete surrogate pair
}
REALM_ASSERT(&in[1] >= in_begin && &in[1] < in_end);
uint_fast16_t v2 = uint_fast16_t(Traits16::to_int_type(in[1]));
if (REALM_UNLIKELY(v2 < 0xDC00 || 0xE000 <= v2)) {
invalid = true;
break; // Invalid second half of surrogate pair
}
uint_fast32_t v = 0x10000l + (uint_fast32_t(v1 - 0xD800) * 0x400 + (v2 - 0xDC00));
// UTF-8 layout: 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
REALM_ASSERT(out >= out_begin && out + 3 < out_end);
*out++ = traits8::to_char_type(traits8_int_type(0xF0 + v / 0x40000));
*out++ = traits8::to_char_type(traits8_int_type(0x80 + v / 0x1000 % 0x40));
*out++ = traits8::to_char_type(traits8_int_type(0x80 + v / 0x40 % 0x40));
*out++ = traits8::to_char_type(traits8_int_type(0x80 + v % 0x40));
in += 2;
}
REALM_ASSERT(in >= in_begin && in <= in_end);
REALM_ASSERT(out >= out_begin && out <= out_end);
in_begin = in;
out_begin = out;
return !invalid;
}
uint_fast32_t get_mask() const {
return uint_fast32_t(1) << index;
}
