2011-09-04 00:22:19 +00:00
|
|
|
|
#pragma once
|
2010-03-12 18:25:35 +00:00
|
|
|
|
|
2011-09-04 00:22:19 +00:00
|
|
|
|
#include <string.h>
|
2010-05-20 19:29:04 +00:00
|
|
|
|
|
2013-02-03 18:39:09 +00:00
|
|
|
|
#include <DB/Core/Defines.h>
|
|
|
|
|
|
2013-05-05 15:25:25 +00:00
|
|
|
|
#include <DB/Columns/IColumn.h>
|
2013-05-28 16:56:05 +00:00
|
|
|
|
#include <DB/Common/Collator.h>
|
2013-09-16 05:44:47 +00:00
|
|
|
|
#include <DB/Common/PODArray.h>
|
2010-03-12 18:25:35 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
namespace DB
|
|
|
|
|
{
|
|
|
|
|
|
2010-05-13 16:13:38 +00:00
|
|
|
|
/** Cтолбeц значений типа "строка".
|
|
|
|
|
*/
|
2014-06-04 01:00:09 +00:00
|
|
|
|
class ColumnString final : public IColumn
|
2010-05-13 16:13:38 +00:00
|
|
|
|
{
|
2013-03-05 10:29:26 +00:00
|
|
|
|
public:
|
2013-09-16 05:44:47 +00:00
|
|
|
|
//typedef std::vector<UInt8> Chars_t;
|
|
|
|
|
typedef PODArray<UInt8> Chars_t;
|
2013-03-05 10:29:26 +00:00
|
|
|
|
|
2010-05-20 19:29:04 +00:00
|
|
|
|
private:
|
2013-05-05 15:25:25 +00:00
|
|
|
|
/// По индексу i находится смещение до начала i + 1 -го элемента.
|
|
|
|
|
Offsets_t offsets;
|
2010-05-20 19:29:04 +00:00
|
|
|
|
|
2013-05-05 15:25:25 +00:00
|
|
|
|
/// Байты строк, уложенные подряд. Строки хранятся с завершающим нулевым байтом.
|
|
|
|
|
Chars_t chars;
|
|
|
|
|
|
|
|
|
|
size_t __attribute__((__always_inline__)) offsetAt(size_t i) const { return i == 0 ? 0 : offsets[i - 1]; }
|
|
|
|
|
|
|
|
|
|
/// Размер, включая завершающий нулевой байт.
|
|
|
|
|
size_t __attribute__((__always_inline__)) sizeAt(size_t i) const { return i == 0 ? offsets[0] : (offsets[i] - offsets[i - 1]); }
|
2014-08-20 04:57:03 +00:00
|
|
|
|
|
|
|
|
|
public:
|
2013-05-03 05:23:14 +00:00
|
|
|
|
/** Создать пустой столбец строк */
|
2013-05-05 15:25:25 +00:00
|
|
|
|
ColumnString() {}
|
|
|
|
|
|
|
|
|
|
std::string getName() const { return "ColumnString"; }
|
|
|
|
|
|
|
|
|
|
size_t size() const
|
2010-05-13 16:13:38 +00:00
|
|
|
|
{
|
2013-05-05 15:25:25 +00:00
|
|
|
|
return offsets.size();
|
2010-05-13 16:13:38 +00:00
|
|
|
|
}
|
2010-05-21 19:52:50 +00:00
|
|
|
|
|
2013-05-05 15:25:25 +00:00
|
|
|
|
size_t byteSize() const
|
|
|
|
|
{
|
|
|
|
|
return chars.size() + offsets.size() * sizeof(offsets[0]);
|
|
|
|
|
}
|
2011-08-28 00:31:30 +00:00
|
|
|
|
|
2011-08-09 19:19:00 +00:00
|
|
|
|
ColumnPtr cloneEmpty() const
|
2010-05-21 19:52:50 +00:00
|
|
|
|
{
|
|
|
|
|
return new ColumnString;
|
|
|
|
|
}
|
2013-05-05 15:25:25 +00:00
|
|
|
|
|
2010-05-13 16:13:38 +00:00
|
|
|
|
Field operator[](size_t n) const
|
|
|
|
|
{
|
2013-05-05 15:25:25 +00:00
|
|
|
|
return Field(&chars[offsetAt(n)], sizeAt(n) - 1);
|
2010-05-13 16:13:38 +00:00
|
|
|
|
}
|
2010-05-20 19:29:04 +00:00
|
|
|
|
|
2013-01-07 06:47:15 +00:00
|
|
|
|
void get(size_t n, Field & res) const
|
|
|
|
|
{
|
2013-05-05 15:25:25 +00:00
|
|
|
|
res.assignString(&chars[offsetAt(n)], sizeAt(n) - 1);
|
2013-01-07 06:47:15 +00:00
|
|
|
|
}
|
|
|
|
|
|
2012-10-07 06:30:10 +00:00
|
|
|
|
StringRef getDataAt(size_t n) const
|
2013-04-13 00:56:07 +00:00
|
|
|
|
{
|
2013-05-05 15:25:25 +00:00
|
|
|
|
return StringRef(&chars[offsetAt(n)], sizeAt(n) - 1);
|
2013-04-13 00:56:07 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
StringRef getDataAtWithTerminatingZero(size_t n) const
|
2012-10-07 06:30:10 +00:00
|
|
|
|
{
|
2013-05-05 15:25:25 +00:00
|
|
|
|
return StringRef(&chars[offsetAt(n)], sizeAt(n));
|
2012-10-07 06:30:10 +00:00
|
|
|
|
}
|
|
|
|
|
|
2010-05-20 19:29:04 +00:00
|
|
|
|
void insert(const Field & x)
|
|
|
|
|
{
|
2013-01-07 06:47:15 +00:00
|
|
|
|
const String & s = DB::get<const String &>(x);
|
2013-05-05 15:25:25 +00:00
|
|
|
|
size_t old_size = chars.size();
|
2010-05-20 19:29:04 +00:00
|
|
|
|
size_t size_to_append = s.size() + 1;
|
2014-08-20 04:57:03 +00:00
|
|
|
|
|
2013-05-05 15:25:25 +00:00
|
|
|
|
chars.resize(old_size + size_to_append);
|
|
|
|
|
memcpy(&chars[old_size], s.c_str(), size_to_append);
|
|
|
|
|
offsets.push_back((offsets.size() == 0 ? 0 : offsets.back()) + size_to_append);
|
2010-05-20 19:29:04 +00:00
|
|
|
|
}
|
|
|
|
|
|
2012-12-16 00:52:06 +00:00
|
|
|
|
void insertFrom(const IColumn & src_, size_t n)
|
|
|
|
|
{
|
|
|
|
|
const ColumnString & src = static_cast<const ColumnString &>(src_);
|
2013-05-05 15:25:25 +00:00
|
|
|
|
size_t old_size = chars.size();
|
2012-12-16 00:52:06 +00:00
|
|
|
|
size_t size_to_append = src.sizeAt(n);
|
|
|
|
|
size_t offset = src.offsetAt(n);
|
2014-08-20 04:57:03 +00:00
|
|
|
|
|
2013-05-05 15:25:25 +00:00
|
|
|
|
chars.resize(old_size + size_to_append);
|
|
|
|
|
memcpy(&chars[old_size], &src.chars[offset], size_to_append);
|
|
|
|
|
offsets.push_back((offsets.size() == 0 ? 0 : offsets.back()) + size_to_append);
|
2012-12-16 00:52:06 +00:00
|
|
|
|
}
|
|
|
|
|
|
2013-02-16 20:15:45 +00:00
|
|
|
|
void insertData(const char * pos, size_t length)
|
|
|
|
|
{
|
2013-05-05 15:25:25 +00:00
|
|
|
|
size_t old_size = chars.size();
|
2013-02-16 20:15:45 +00:00
|
|
|
|
|
2013-05-05 15:25:25 +00:00
|
|
|
|
chars.resize(old_size + length + 1);
|
|
|
|
|
memcpy(&chars[old_size], pos, length);
|
|
|
|
|
chars[old_size + length] = 0;
|
|
|
|
|
offsets.push_back((offsets.size() == 0 ? 0 : offsets.back()) + length + 1);
|
2013-04-13 00:56:07 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void insertDataWithTerminatingZero(const char * pos, size_t length)
|
|
|
|
|
{
|
2013-05-05 15:25:25 +00:00
|
|
|
|
size_t old_size = chars.size();
|
2013-04-13 00:56:07 +00:00
|
|
|
|
|
2013-05-05 15:25:25 +00:00
|
|
|
|
chars.resize(old_size + length);
|
|
|
|
|
memcpy(&chars[old_size], pos, length);
|
|
|
|
|
offsets.push_back((offsets.size() == 0 ? 0 : offsets.back()) + length);
|
2013-02-16 20:15:45 +00:00
|
|
|
|
}
|
|
|
|
|
|
2013-05-03 15:10:56 +00:00
|
|
|
|
ColumnPtr cut(size_t start, size_t length) const
|
|
|
|
|
{
|
|
|
|
|
if (length == 0)
|
|
|
|
|
return new ColumnString;
|
|
|
|
|
|
2013-05-05 15:25:25 +00:00
|
|
|
|
if (start + length > offsets.size())
|
2013-05-03 15:10:56 +00:00
|
|
|
|
throw Exception("Parameter out of bound in IColumnString::cut() method.",
|
|
|
|
|
ErrorCodes::PARAMETER_OUT_OF_BOUND);
|
|
|
|
|
|
|
|
|
|
size_t nested_offset = offsetAt(start);
|
2013-05-05 15:25:25 +00:00
|
|
|
|
size_t nested_length = offsets[start + length - 1] - nested_offset;
|
2013-05-03 15:10:56 +00:00
|
|
|
|
|
|
|
|
|
ColumnString * res_ = new ColumnString;
|
|
|
|
|
ColumnPtr res = res_;
|
|
|
|
|
|
2013-05-05 16:18:25 +00:00
|
|
|
|
res_->chars.resize(nested_length);
|
2013-05-05 15:25:25 +00:00
|
|
|
|
memcpy(&res_->chars[0], &chars[nested_offset], nested_length);
|
2014-08-20 04:57:03 +00:00
|
|
|
|
|
2013-05-05 15:25:25 +00:00
|
|
|
|
Offsets_t & res_offsets = res_->offsets;
|
2013-05-03 15:10:56 +00:00
|
|
|
|
|
|
|
|
|
if (start == 0)
|
|
|
|
|
{
|
2013-05-05 15:25:25 +00:00
|
|
|
|
res_offsets.assign(offsets.begin(), offsets.begin() + length);
|
2013-05-03 15:10:56 +00:00
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
res_offsets.resize(length);
|
|
|
|
|
|
|
|
|
|
for (size_t i = 0; i < length; ++i)
|
2013-05-05 15:25:25 +00:00
|
|
|
|
res_offsets[i] = offsets[start + i] - nested_offset;
|
2013-05-03 15:10:56 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return res;
|
|
|
|
|
}
|
|
|
|
|
|
2013-05-03 05:23:14 +00:00
|
|
|
|
ColumnPtr filter(const Filter & filt) const
|
2012-10-07 00:23:18 +00:00
|
|
|
|
{
|
2014-08-25 16:23:32 +00:00
|
|
|
|
const size_t size = offsets.size();
|
2012-10-07 00:23:18 +00:00
|
|
|
|
if (size != filt.size())
|
|
|
|
|
throw Exception("Size of filter doesn't match size of column.", ErrorCodes::SIZES_OF_COLUMNS_DOESNT_MATCH);
|
|
|
|
|
|
|
|
|
|
if (size == 0)
|
2013-05-03 05:23:14 +00:00
|
|
|
|
return new ColumnString;
|
2012-10-07 00:23:18 +00:00
|
|
|
|
|
2014-08-25 16:23:32 +00:00
|
|
|
|
auto res = new ColumnString;
|
2013-05-03 05:23:14 +00:00
|
|
|
|
|
2014-08-25 16:23:32 +00:00
|
|
|
|
Chars_t & res_chars = res->chars;
|
|
|
|
|
Offsets_t & res_offsets = res->offsets;
|
2013-05-05 15:25:25 +00:00
|
|
|
|
res_chars.reserve(chars.size());
|
2013-05-03 05:23:14 +00:00
|
|
|
|
res_offsets.reserve(size);
|
2012-10-07 00:23:18 +00:00
|
|
|
|
|
2014-08-25 16:23:32 +00:00
|
|
|
|
Offset_t current_offset = 0;
|
2012-10-07 00:23:18 +00:00
|
|
|
|
|
2014-08-25 16:23:32 +00:00
|
|
|
|
auto filt_pos = &filt[0];
|
|
|
|
|
const auto filt_end = filt_pos + size;
|
|
|
|
|
const auto filt_end_aligned = filt_pos + size / 16 * 16;
|
|
|
|
|
|
|
|
|
|
auto offsets_pos = &offsets[0];
|
|
|
|
|
const auto offsets_begin = offsets_pos;
|
|
|
|
|
|
|
|
|
|
const auto zero16 = _mm_set1_epi8(0);
|
2014-08-20 04:57:03 +00:00
|
|
|
|
|
2014-08-25 16:23:32 +00:00
|
|
|
|
/// copy string ending at *end_offset_ptr
|
|
|
|
|
const auto copy_string = [&] (const Offset_t * offset_ptr) {
|
|
|
|
|
const auto offset = offset_ptr == offsets_begin ? 0 : offset_ptr[-1];
|
|
|
|
|
const auto size = *offset_ptr - offset;
|
2012-10-07 00:23:18 +00:00
|
|
|
|
|
2014-08-25 16:23:32 +00:00
|
|
|
|
current_offset += size;
|
|
|
|
|
res_offsets.push_back(current_offset);
|
|
|
|
|
|
|
|
|
|
const auto chars_size_old = res_chars.size();
|
|
|
|
|
res_chars.resize(chars_size_old + size);
|
|
|
|
|
memcpy(&res_chars[chars_size_old], &chars[offset], size);
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
while (filt_pos < filt_end_aligned)
|
|
|
|
|
{
|
|
|
|
|
const auto mask = _mm_movemask_epi8(_mm_cmpgt_epi8(
|
|
|
|
|
_mm_loadu_si128(reinterpret_cast<const __m128i *>(filt_pos)),
|
|
|
|
|
zero16));
|
|
|
|
|
|
|
|
|
|
if (mask == 0)
|
|
|
|
|
{
|
|
|
|
|
/// 16 consecutive rows do not pass the filter
|
|
|
|
|
}
|
|
|
|
|
else if (mask == 0xffff)
|
|
|
|
|
{
|
|
|
|
|
/// 16 consecutive rows pass the filter
|
|
|
|
|
const auto first = offsets_pos == offsets_begin;
|
|
|
|
|
|
|
|
|
|
const auto chunk_offset = first ? 0 : offsets_pos[-1];
|
|
|
|
|
const auto chunk_size = offsets_pos[16 - 1] - chunk_offset;
|
|
|
|
|
|
|
|
|
|
const auto offsets_size_old = res_offsets.size();
|
|
|
|
|
res_offsets.resize(offsets_size_old + 16);
|
|
|
|
|
memcpy(&res_offsets[offsets_size_old], offsets_pos, 16 * sizeof(Offset_t));
|
|
|
|
|
|
|
|
|
|
if (!first)
|
|
|
|
|
{
|
|
|
|
|
/// difference between current and actual offset
|
|
|
|
|
const auto diff_offset = chunk_offset - current_offset;
|
|
|
|
|
|
|
|
|
|
if (diff_offset > 0)
|
|
|
|
|
{
|
|
|
|
|
const auto res_offsets_pos = &res_offsets[offsets_size_old];
|
|
|
|
|
|
|
|
|
|
/// adjust offsets
|
|
|
|
|
for (size_t i = 0; i < 16; ++i)
|
|
|
|
|
res_offsets_pos[i] -= diff_offset;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
current_offset += chunk_size;
|
|
|
|
|
|
|
|
|
|
/// copy characters for 16 strings at once
|
|
|
|
|
const auto chars_size_old = res_chars.size();
|
|
|
|
|
res_chars.resize(chars_size_old + chunk_size);
|
|
|
|
|
memcpy(&res_chars[chars_size_old], &chars[chunk_offset], chunk_size);
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
for (size_t i = 0; i < 16; ++i)
|
|
|
|
|
if (filt_pos[i])
|
|
|
|
|
copy_string(offsets_pos + i);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
filt_pos += 16;
|
|
|
|
|
offsets_pos += 16;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
while (filt_pos < filt_end)
|
|
|
|
|
{
|
|
|
|
|
if (*filt_pos)
|
|
|
|
|
copy_string(offsets_pos);
|
2012-10-07 00:23:18 +00:00
|
|
|
|
|
2014-08-25 16:23:32 +00:00
|
|
|
|
++filt_pos;
|
|
|
|
|
++offsets_pos;
|
2012-10-07 00:23:18 +00:00
|
|
|
|
}
|
|
|
|
|
|
2013-05-03 05:23:14 +00:00
|
|
|
|
return res;
|
2012-10-07 00:23:18 +00:00
|
|
|
|
}
|
|
|
|
|
|
2013-09-16 05:44:47 +00:00
|
|
|
|
ColumnPtr permute(const Permutation & perm, size_t limit) const
|
2012-10-07 00:23:18 +00:00
|
|
|
|
{
|
2013-05-05 15:25:25 +00:00
|
|
|
|
size_t size = offsets.size();
|
2012-10-07 00:23:18 +00:00
|
|
|
|
|
2013-09-16 05:44:47 +00:00
|
|
|
|
if (limit == 0)
|
|
|
|
|
limit = size;
|
|
|
|
|
else
|
|
|
|
|
limit = std::min(size, limit);
|
|
|
|
|
|
|
|
|
|
if (perm.size() < limit)
|
|
|
|
|
throw Exception("Size of permutation is less than required.", ErrorCodes::SIZES_OF_COLUMNS_DOESNT_MATCH);
|
|
|
|
|
|
|
|
|
|
if (limit == 0)
|
2013-05-03 05:23:14 +00:00
|
|
|
|
return new ColumnString;
|
|
|
|
|
|
|
|
|
|
ColumnString * res_ = new ColumnString;
|
|
|
|
|
ColumnPtr res = res_;
|
|
|
|
|
|
2013-05-05 15:25:25 +00:00
|
|
|
|
Chars_t & res_chars = res_->chars;
|
|
|
|
|
Offsets_t & res_offsets = res_->offsets;
|
2012-10-07 00:23:18 +00:00
|
|
|
|
|
2013-05-05 15:25:25 +00:00
|
|
|
|
res_chars.resize(chars.size());
|
2013-09-16 05:44:47 +00:00
|
|
|
|
res_offsets.resize(limit);
|
2012-10-07 00:23:18 +00:00
|
|
|
|
|
|
|
|
|
Offset_t current_new_offset = 0;
|
|
|
|
|
|
2013-09-16 05:44:47 +00:00
|
|
|
|
for (size_t i = 0; i < limit; ++i)
|
2012-10-07 00:23:18 +00:00
|
|
|
|
{
|
|
|
|
|
size_t j = perm[i];
|
2013-05-05 15:25:25 +00:00
|
|
|
|
size_t string_offset = j == 0 ? 0 : offsets[j - 1];
|
|
|
|
|
size_t string_size = offsets[j] - string_offset;
|
2012-10-07 00:23:18 +00:00
|
|
|
|
|
2013-05-05 15:25:25 +00:00
|
|
|
|
memcpy(&res_chars[current_new_offset], &chars[string_offset], string_size);
|
2012-10-07 00:23:18 +00:00
|
|
|
|
|
|
|
|
|
current_new_offset += string_size;
|
2013-05-03 05:23:14 +00:00
|
|
|
|
res_offsets[i] = current_new_offset;
|
2012-10-07 00:23:18 +00:00
|
|
|
|
}
|
|
|
|
|
|
2013-09-16 05:44:47 +00:00
|
|
|
|
res_chars.resize(res_offsets[limit - 1]);
|
|
|
|
|
|
2013-05-03 05:23:14 +00:00
|
|
|
|
return res;
|
2012-10-07 00:23:18 +00:00
|
|
|
|
}
|
|
|
|
|
|
2010-05-20 19:29:04 +00:00
|
|
|
|
void insertDefault()
|
|
|
|
|
{
|
2013-05-05 15:25:25 +00:00
|
|
|
|
chars.push_back(0);
|
|
|
|
|
offsets.push_back(offsets.size() == 0 ? 1 : (offsets.back() + 1));
|
2010-05-20 19:29:04 +00:00
|
|
|
|
}
|
2011-09-04 00:22:19 +00:00
|
|
|
|
|
2013-11-01 20:10:43 +00:00
|
|
|
|
int compareAt(size_t n, size_t m, const IColumn & rhs_, int nan_direction_hint) const
|
2011-09-04 00:22:19 +00:00
|
|
|
|
{
|
|
|
|
|
const ColumnString & rhs = static_cast<const ColumnString &>(rhs_);
|
|
|
|
|
|
2011-09-26 11:05:38 +00:00
|
|
|
|
/** Для производительности, строки сравниваются до первого нулевого байта.
|
|
|
|
|
* (если нулевой байт в середине строки, то то, что после него - игнорируется)
|
|
|
|
|
* Замечу, что завершающий нулевой байт всегда есть.
|
|
|
|
|
*/
|
|
|
|
|
return strcmp(
|
2013-05-05 15:25:25 +00:00
|
|
|
|
reinterpret_cast<const char *>(&chars[offsetAt(n)]),
|
|
|
|
|
reinterpret_cast<const char *>(&rhs.chars[rhs.offsetAt(m)]));
|
2011-09-26 11:05:38 +00:00
|
|
|
|
}
|
2014-08-20 04:57:03 +00:00
|
|
|
|
|
2013-05-28 16:56:05 +00:00
|
|
|
|
/// Версия compareAt для locale-sensitive сравнения строк
|
2013-05-30 14:28:27 +00:00
|
|
|
|
int compareAtWithCollation(size_t n, size_t m, const IColumn & rhs_, const Collator & collator) const
|
2013-05-28 16:56:05 +00:00
|
|
|
|
{
|
|
|
|
|
const ColumnString & rhs = static_cast<const ColumnString &>(rhs_);
|
2014-08-20 04:57:03 +00:00
|
|
|
|
|
2013-05-28 16:56:05 +00:00
|
|
|
|
return collator.compare(
|
|
|
|
|
reinterpret_cast<const char *>(&chars[offsetAt(n)]), sizeAt(n),
|
|
|
|
|
reinterpret_cast<const char *>(&rhs.chars[rhs.offsetAt(m)]), rhs.sizeAt(m));
|
|
|
|
|
}
|
2011-09-26 11:05:38 +00:00
|
|
|
|
|
2013-09-16 05:44:47 +00:00
|
|
|
|
template <bool positive>
|
2011-09-26 11:05:38 +00:00
|
|
|
|
struct less
|
|
|
|
|
{
|
|
|
|
|
const ColumnString & parent;
|
|
|
|
|
less(const ColumnString & parent_) : parent(parent_) {}
|
|
|
|
|
bool operator()(size_t lhs, size_t rhs) const
|
|
|
|
|
{
|
2014-05-23 02:18:01 +00:00
|
|
|
|
int res = strcmp(
|
2013-05-05 15:25:25 +00:00
|
|
|
|
reinterpret_cast<const char *>(&parent.chars[parent.offsetAt(lhs)]),
|
2014-05-23 02:18:01 +00:00
|
|
|
|
reinterpret_cast<const char *>(&parent.chars[parent.offsetAt(rhs)]));
|
|
|
|
|
|
|
|
|
|
return positive ? (res < 0) : (res > 0);
|
2011-09-26 11:05:38 +00:00
|
|
|
|
}
|
|
|
|
|
};
|
|
|
|
|
|
2013-12-08 02:29:40 +00:00
|
|
|
|
void getPermutation(bool reverse, size_t limit, Permutation & res) const
|
2011-09-26 11:05:38 +00:00
|
|
|
|
{
|
2013-05-05 15:25:25 +00:00
|
|
|
|
size_t s = offsets.size();
|
2013-12-08 02:29:40 +00:00
|
|
|
|
res.resize(s);
|
2011-09-26 11:05:38 +00:00
|
|
|
|
for (size_t i = 0; i < s; ++i)
|
|
|
|
|
res[i] = i;
|
|
|
|
|
|
2014-05-23 01:32:32 +00:00
|
|
|
|
if (limit >= s)
|
2013-09-16 05:44:47 +00:00
|
|
|
|
limit = 0;
|
|
|
|
|
|
|
|
|
|
if (limit)
|
|
|
|
|
{
|
|
|
|
|
if (reverse)
|
|
|
|
|
std::partial_sort(res.begin(), res.begin() + limit, res.end(), less<false>(*this));
|
|
|
|
|
else
|
|
|
|
|
std::partial_sort(res.begin(), res.begin() + limit, res.end(), less<true>(*this));
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
if (reverse)
|
|
|
|
|
std::sort(res.begin(), res.end(), less<false>(*this));
|
|
|
|
|
else
|
|
|
|
|
std::sort(res.begin(), res.end(), less<true>(*this));
|
|
|
|
|
}
|
2011-09-04 00:22:19 +00:00
|
|
|
|
}
|
2013-09-16 05:44:47 +00:00
|
|
|
|
|
|
|
|
|
template <bool positive>
|
2013-05-28 16:56:05 +00:00
|
|
|
|
struct lessWithCollation
|
|
|
|
|
{
|
|
|
|
|
const ColumnString & parent;
|
|
|
|
|
const Collator & collator;
|
2014-08-20 04:57:03 +00:00
|
|
|
|
|
2013-05-28 16:56:05 +00:00
|
|
|
|
lessWithCollation(const ColumnString & parent_, const Collator & collator_) : parent(parent_), collator(collator_) {}
|
2014-08-20 04:57:03 +00:00
|
|
|
|
|
2013-05-28 16:56:05 +00:00
|
|
|
|
bool operator()(size_t lhs, size_t rhs) const
|
|
|
|
|
{
|
2014-05-23 02:18:01 +00:00
|
|
|
|
int res = collator.compare(
|
2013-05-28 16:56:05 +00:00
|
|
|
|
reinterpret_cast<const char *>(&parent.chars[parent.offsetAt(lhs)]), parent.sizeAt(lhs),
|
2014-05-23 02:18:01 +00:00
|
|
|
|
reinterpret_cast<const char *>(&parent.chars[parent.offsetAt(rhs)]), parent.sizeAt(rhs));
|
|
|
|
|
|
|
|
|
|
return positive ? (res < 0) : (res > 0);
|
2013-05-28 16:56:05 +00:00
|
|
|
|
}
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
/// Сортировка с учетом Collation
|
2013-12-08 02:29:40 +00:00
|
|
|
|
void getPermutationWithCollation(const Collator & collator, bool reverse, size_t limit, Permutation & res) const
|
2013-05-28 16:56:05 +00:00
|
|
|
|
{
|
|
|
|
|
size_t s = offsets.size();
|
2013-12-08 02:29:40 +00:00
|
|
|
|
res.resize(s);
|
2013-05-28 16:56:05 +00:00
|
|
|
|
for (size_t i = 0; i < s; ++i)
|
|
|
|
|
res[i] = i;
|
|
|
|
|
|
2014-05-23 01:32:32 +00:00
|
|
|
|
if (limit >= s)
|
2013-09-16 05:44:47 +00:00
|
|
|
|
limit = 0;
|
|
|
|
|
|
|
|
|
|
if (limit)
|
|
|
|
|
{
|
|
|
|
|
if (reverse)
|
|
|
|
|
std::partial_sort(res.begin(), res.begin() + limit, res.end(), lessWithCollation<false>(*this, collator));
|
|
|
|
|
else
|
|
|
|
|
std::partial_sort(res.begin(), res.begin() + limit, res.end(), lessWithCollation<true>(*this, collator));
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
if (reverse)
|
|
|
|
|
std::sort(res.begin(), res.end(), lessWithCollation<false>(*this, collator));
|
|
|
|
|
else
|
|
|
|
|
std::sort(res.begin(), res.end(), lessWithCollation<true>(*this, collator));
|
|
|
|
|
}
|
2013-05-28 16:56:05 +00:00
|
|
|
|
}
|
2012-08-27 05:13:14 +00:00
|
|
|
|
|
2013-05-03 05:23:14 +00:00
|
|
|
|
ColumnPtr replicate(const Offsets_t & replicate_offsets) const
|
2012-08-27 05:13:14 +00:00
|
|
|
|
{
|
|
|
|
|
size_t col_size = size();
|
|
|
|
|
if (col_size != replicate_offsets.size())
|
|
|
|
|
throw Exception("Size of offsets doesn't match size of column.", ErrorCodes::SIZES_OF_COLUMNS_DOESNT_MATCH);
|
|
|
|
|
|
2013-05-03 05:23:14 +00:00
|
|
|
|
ColumnString * res_ = new ColumnString;
|
|
|
|
|
ColumnPtr res = res_;
|
|
|
|
|
|
2014-08-20 04:57:03 +00:00
|
|
|
|
if (0 == col_size)
|
|
|
|
|
return res;
|
|
|
|
|
|
2013-05-05 15:25:25 +00:00
|
|
|
|
Chars_t & res_chars = res_->chars;
|
|
|
|
|
Offsets_t & res_offsets = res_->offsets;
|
|
|
|
|
res_chars.reserve(chars.size() / col_size * replicate_offsets.back());
|
2013-05-03 05:23:14 +00:00
|
|
|
|
res_offsets.reserve(replicate_offsets.back());
|
2012-08-27 05:13:14 +00:00
|
|
|
|
|
|
|
|
|
Offset_t prev_replicate_offset = 0;
|
|
|
|
|
Offset_t prev_string_offset = 0;
|
|
|
|
|
Offset_t current_new_offset = 0;
|
|
|
|
|
|
|
|
|
|
for (size_t i = 0; i < col_size; ++i)
|
|
|
|
|
{
|
|
|
|
|
size_t size_to_replicate = replicate_offsets[i] - prev_replicate_offset;
|
2013-05-05 15:25:25 +00:00
|
|
|
|
size_t string_size = offsets[i] - prev_string_offset;
|
2012-08-27 05:13:14 +00:00
|
|
|
|
|
|
|
|
|
for (size_t j = 0; j < size_to_replicate; ++j)
|
|
|
|
|
{
|
|
|
|
|
current_new_offset += string_size;
|
2013-05-03 05:23:14 +00:00
|
|
|
|
res_offsets.push_back(current_new_offset);
|
2012-09-19 18:45:01 +00:00
|
|
|
|
|
2013-05-03 05:23:14 +00:00
|
|
|
|
res_chars.resize(res_chars.size() + string_size);
|
2013-05-05 15:25:25 +00:00
|
|
|
|
memcpy(&res_chars[res_chars.size() - string_size], &chars[prev_string_offset], string_size);
|
2012-08-27 05:13:14 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
prev_replicate_offset = replicate_offsets[i];
|
2013-05-05 15:25:25 +00:00
|
|
|
|
prev_string_offset = offsets[i];
|
2012-08-27 05:13:14 +00:00
|
|
|
|
}
|
|
|
|
|
|
2013-05-03 05:23:14 +00:00
|
|
|
|
return res;
|
2012-08-27 05:13:14 +00:00
|
|
|
|
}
|
2013-02-03 18:39:09 +00:00
|
|
|
|
|
|
|
|
|
void reserve(size_t n)
|
|
|
|
|
{
|
2013-05-05 15:25:25 +00:00
|
|
|
|
offsets.reserve(n);
|
|
|
|
|
chars.reserve(n * DBMS_APPROX_STRING_SIZE);
|
2013-02-03 18:39:09 +00:00
|
|
|
|
}
|
2013-05-05 15:25:25 +00:00
|
|
|
|
|
2013-09-06 20:28:22 +00:00
|
|
|
|
void getExtremes(Field & min, Field & max) const
|
|
|
|
|
{
|
|
|
|
|
min = String();
|
|
|
|
|
max = String();
|
|
|
|
|
}
|
|
|
|
|
|
2013-05-05 15:25:25 +00:00
|
|
|
|
|
|
|
|
|
Chars_t & getChars() { return chars; }
|
|
|
|
|
const Chars_t & getChars() const { return chars; }
|
|
|
|
|
|
|
|
|
|
Offsets_t & getOffsets() { return offsets; }
|
|
|
|
|
const Offsets_t & getOffsets() const { return offsets; }
|
2010-05-13 16:13:38 +00:00
|
|
|
|
};
|
2010-03-12 18:25:35 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
}
|