mirror of
https://github.com/ClickHouse/ClickHouse.git
synced 2024-12-03 13:02:00 +00:00
175 lines
5.5 KiB
C++
175 lines
5.5 KiB
C++
#include <DataTypes/DataTypeString.h>
|
|
#include <Functions/FunctionFactory.h>
|
|
#include <Functions/FunctionStringToString.h>
|
|
#include <IO/WriteBufferFromVector.h>
|
|
#include <IO/WriteHelpers.h>
|
|
#include <Poco/UTF8Encoding.h>
|
|
|
|
#include <string_view>
|
|
|
|
#ifdef __SSE2__
|
|
# include <emmintrin.h>
|
|
#endif
|
|
|
|
#if defined(__aarch64__) && defined(__ARM_NEON)
|
|
# include <arm_neon.h>
|
|
# ifdef HAS_RESERVED_IDENTIFIER
|
|
# pragma clang diagnostic ignored "-Wreserved-identifier"
|
|
# endif
|
|
#endif
|
|
|
|
namespace DB
|
|
{
|
|
|
|
namespace ErrorCodes
|
|
{
|
|
extern const int ILLEGAL_COLUMN;
|
|
}
|
|
|
|
extern const UInt8 length_of_utf8_sequence[256];
|
|
|
|
namespace
|
|
{
|
|
|
|
struct ToValidUTF8Impl
|
|
{
|
|
static void toValidUTF8One(const char * begin, const char * end, WriteBuffer & write_buffer)
|
|
{
|
|
static constexpr std::string_view replacement = "\xEF\xBF\xBD";
|
|
|
|
const char * p = begin;
|
|
const char * valid_start = begin;
|
|
|
|
/// The last recorded character was `replacement`.
|
|
bool just_put_replacement = false;
|
|
|
|
auto put_valid = [&write_buffer, &just_put_replacement](const char * data, size_t len)
|
|
{
|
|
if (len == 0)
|
|
return;
|
|
just_put_replacement = false;
|
|
write_buffer.write(data, len);
|
|
};
|
|
|
|
auto put_replacement = [&write_buffer, &just_put_replacement]()
|
|
{
|
|
if (just_put_replacement)
|
|
return;
|
|
just_put_replacement = true;
|
|
write_buffer.write(replacement.data(), replacement.size());
|
|
};
|
|
|
|
while (p < end)
|
|
{
|
|
#ifdef __SSE2__
|
|
/// Fast skip of ASCII
|
|
static constexpr size_t SIMD_BYTES = 16;
|
|
const char * simd_end = p + (end - p) / SIMD_BYTES * SIMD_BYTES;
|
|
|
|
while (p < simd_end && !_mm_movemask_epi8(_mm_loadu_si128(reinterpret_cast<const __m128i *>(p))))
|
|
p += SIMD_BYTES;
|
|
|
|
if (!(p < end))
|
|
break;
|
|
#elif defined(__aarch64__) && defined(__ARM_NEON)
|
|
/// Fast skip of ASCII for aarch64.
|
|
static constexpr size_t SIMD_BYTES = 16;
|
|
const char * simd_end = p + (end - p) / SIMD_BYTES * SIMD_BYTES;
|
|
/// Returns a 64 bit mask of nibbles (4 bits for each byte).
|
|
auto get_nibble_mask = [](uint8x16_t input) -> uint64_t
|
|
{ return vget_lane_u64(vreinterpret_u64_u8(vshrn_n_u16(vreinterpretq_u16_u8(input), 4)), 0); };
|
|
/// Other options include
|
|
/// vmaxvq_u8(input) < 0b10000000;
|
|
/// Used by SIMDJSON, has latency 3 for M1, 6 for everything else
|
|
/// SIMDJSON uses it for 64 byte masks, so it's a little different.
|
|
/// vmaxvq_u32(vandq_u32(input, vdupq_n_u32(0x80808080))) // u32 version has latency 3
|
|
/// shrn version has universally <=3 cycles, on servers 2 cycles.
|
|
while (p < simd_end && get_nibble_mask(vcgeq_u8(vld1q_u8(reinterpret_cast<const uint8_t *>(p)), vdupq_n_u8(0x80))) == 0)
|
|
p += SIMD_BYTES;
|
|
|
|
if (!(p < end))
|
|
break;
|
|
#endif
|
|
|
|
size_t len = length_of_utf8_sequence[static_cast<unsigned char>(*p)];
|
|
|
|
if (len > 4)
|
|
{
|
|
/// Invalid start of sequence. Skip one byte.
|
|
put_valid(valid_start, p - valid_start);
|
|
put_replacement();
|
|
++p;
|
|
valid_start = p;
|
|
}
|
|
else if (p + len > end)
|
|
{
|
|
/// Sequence was not fully written to this buffer.
|
|
break;
|
|
}
|
|
else if (Poco::UTF8Encoding::isLegal(reinterpret_cast<const unsigned char *>(p), len))
|
|
{
|
|
/// Valid sequence.
|
|
p += len;
|
|
}
|
|
else
|
|
{
|
|
/// Invalid sequence. Skip just first byte.
|
|
put_valid(valid_start, p - valid_start);
|
|
put_replacement();
|
|
++p;
|
|
valid_start = p;
|
|
}
|
|
}
|
|
|
|
put_valid(valid_start, p - valid_start);
|
|
|
|
if (p != end)
|
|
put_replacement();
|
|
}
|
|
|
|
static void vector(
|
|
const ColumnString::Chars & data,
|
|
const ColumnString::Offsets & offsets,
|
|
ColumnString::Chars & res_data,
|
|
ColumnString::Offsets & res_offsets)
|
|
{
|
|
const size_t offsets_size = offsets.size();
|
|
/// It can be larger than that, but we believe it is unlikely to happen.
|
|
res_data.resize(data.size());
|
|
res_offsets.resize(offsets_size);
|
|
|
|
size_t prev_offset = 0;
|
|
WriteBufferFromVector<ColumnString::Chars> write_buffer(res_data);
|
|
for (size_t i = 0; i < offsets_size; ++i)
|
|
{
|
|
const char * haystack_data = reinterpret_cast<const char *>(&data[prev_offset]);
|
|
const size_t haystack_size = offsets[i] - prev_offset - 1;
|
|
toValidUTF8One(haystack_data, haystack_data + haystack_size, write_buffer);
|
|
writeChar(0, write_buffer);
|
|
res_offsets[i] = write_buffer.count();
|
|
prev_offset = offsets[i];
|
|
}
|
|
write_buffer.finalize();
|
|
}
|
|
|
|
[[noreturn]] static void vectorFixed(const ColumnString::Chars &, size_t, ColumnString::Chars &)
|
|
{
|
|
throw Exception("Column of type FixedString is not supported by toValidUTF8 function", ErrorCodes::ILLEGAL_COLUMN);
|
|
}
|
|
};
|
|
|
|
struct NameToValidUTF8
|
|
{
|
|
static constexpr auto name = "toValidUTF8";
|
|
};
|
|
using FunctionToValidUTF8 = FunctionStringToString<ToValidUTF8Impl, NameToValidUTF8>;
|
|
|
|
}
|
|
|
|
void registerFunctionToValidUTF8(FunctionFactory & factory)
|
|
{
|
|
factory.registerFunction<FunctionToValidUTF8>();
|
|
}
|
|
|
|
}
|