ClickHouse/dbms/include/DB/Columns/ColumnVector.h
2017-03-09 07:26:17 +03:00

523 lines
12 KiB
C++

#pragma once
#include <cstring>
#include <cmath>
#include <DB/Common/Exception.h>
#include <DB/Common/Arena.h>
#include <DB/Common/SipHash.h>
#include <DB/IO/WriteBuffer.h>
#include <DB/IO/WriteHelpers.h>
#include <DB/Columns/IColumn.h>
#if __SSE2__
#include <emmintrin.h>
#endif
namespace DB
{
namespace ErrorCodes
{
extern const int PARAMETER_OUT_OF_BOUND;
extern const int SIZES_OF_COLUMNS_DOESNT_MATCH;
}
/** Stuff for comparing numbers.
* Integer values are compared as usual.
* Floating-point numbers are compared this way that NaNs always end up at the end
* (if you don't do this, the sort would not work at all).
*/
template <typename T>
struct CompareHelper
{
static bool less(T a, T b) { return a < b; }
static bool greater(T a, T b) { return a > b; }
/** Compares two numbers. Returns a number less than zero, equal to zero, or greater than zero if a < b, a == b, a > b, respectively.
* If one of the values is NaN, then
* - if nan_direction_hint == -1 - NaN are considered less than all numbers;
* - if nan_direction_hint == 1 - NaN are considered to be larger than all numbers;
* Essentially: nan_direction_hint == -1 says that the comparison is for sorting in descending order.
*/
static int compare(T a, T b, int nan_direction_hint)
{
return a > b ? 1 : (a < b ? -1 : 0);
}
};
template <typename T>
struct FloatCompareHelper
{
static bool less(T a, T b)
{
if (unlikely(std::isnan(b)))
return !std::isnan(a);
return a < b;
}
static bool greater(T a, T b)
{
if (unlikely(std::isnan(b)))
return !std::isnan(a);
return a > b;
}
static int compare(T a, T b, int nan_direction_hint)
{
bool isnan_a = std::isnan(a);
bool isnan_b = std::isnan(b);
if (unlikely(isnan_a || isnan_b))
{
if (isnan_a && isnan_b)
return 0;
return isnan_a
? nan_direction_hint
: -nan_direction_hint;
}
return (T(0) < (a - b)) - ((a - b) < T(0));
}
};
template <> struct CompareHelper<Float32> : public FloatCompareHelper<Float32> {};
template <> struct CompareHelper<Float64> : public FloatCompareHelper<Float64> {};
/** To implement `get64` function.
*/
template <typename T>
inline UInt64 unionCastToUInt64(T x) { return x; }
template <> inline UInt64 unionCastToUInt64(Float64 x)
{
union
{
Float64 src;
UInt64 res;
};
src = x;
return res;
}
template <> inline UInt64 unionCastToUInt64(Float32 x)
{
union
{
Float32 src;
UInt64 res;
};
res = 0;
src = x;
return res;
}
/// To be sure, that this function is zero-cost for non-floating point types.
template <typename T>
inline bool isNaN(T x)
{
return std::is_floating_point<T>::value ? std::isnan(x) : false;
}
template <typename T>
typename std::enable_if<std::is_floating_point<T>::value, T>::type NaNOrZero()
{
return std::numeric_limits<T>::quiet_NaN();
}
template <typename T>
typename std::enable_if<!std::is_floating_point<T>::value, T>::type NaNOrZero()
{
return 0;
}
/** A pattern of columns that use a simple array to store.
*/
template <typename T>
class ColumnVector final : public IColumn
{
private:
using Self = ColumnVector<T>;
public:
using value_type = T;
using Container_t = PaddedPODArray<value_type>;
ColumnVector() {}
ColumnVector(const size_t n) : data{n} {}
ColumnVector(const size_t n, const value_type x) : data{n, x} {}
bool isNumeric() const override { return IsNumber<T>::value; }
bool isFixed() const override { return IsNumber<T>::value; }
size_t sizeOfField() const override { return sizeof(T); }
size_t size() const override
{
return data.size();
}
StringRef getDataAt(size_t n) const override
{
return StringRef(reinterpret_cast<const char *>(&data[n]), sizeof(data[n]));
}
void insertFrom(const IColumn & src, size_t n) override
{
data.push_back(static_cast<const Self &>(src).getData()[n]);
}
void insertData(const char * pos, size_t length) override
{
data.push_back(*reinterpret_cast<const T *>(pos));
}
void insertDefault() override
{
data.push_back(T());
}
void popBack(size_t n) override
{
data.resize_assume_reserved(data.size() - n);
}
StringRef serializeValueIntoArena(size_t n, Arena & arena, char const *& begin) const override
{
auto pos = arena.allocContinue(sizeof(T), begin);
memcpy(pos, &data[n], sizeof(T));
return StringRef(pos, sizeof(T));
}
const char * deserializeAndInsertFromArena(const char * pos) override
{
data.push_back(*reinterpret_cast<const T *>(pos));
return pos + sizeof(T);
}
void updateHashWithValue(size_t n, SipHash & hash) const override
{
hash.update(reinterpret_cast<const char *>(&data[n]), sizeof(T));
}
size_t byteSize() const override
{
return data.size() * sizeof(data[0]);
}
size_t allocatedSize() const override
{
return data.allocated_size() * sizeof(data[0]);
}
void insert(const T value)
{
data.push_back(value);
}
int compareAt(size_t n, size_t m, const IColumn & rhs_, int nan_direction_hint) const override
{
return CompareHelper<T>::compare(data[n], static_cast<const Self &>(rhs_).data[m], nan_direction_hint);
}
struct less
{
const Self & parent;
less(const Self & parent_) : parent(parent_) {}
bool operator()(size_t lhs, size_t rhs) const { return CompareHelper<T>::less(parent.data[lhs], parent.data[rhs]); }
};
struct greater
{
const Self & parent;
greater(const Self & parent_) : parent(parent_) {}
bool operator()(size_t lhs, size_t rhs) const { return CompareHelper<T>::greater(parent.data[lhs], parent.data[rhs]); }
};
void getPermutation(bool reverse, size_t limit, Permutation & res) const override
{
size_t s = data.size();
res.resize(s);
for (size_t i = 0; i < s; ++i)
res[i] = i;
if (limit >= s)
limit = 0;
if (limit)
{
if (reverse)
std::partial_sort(res.begin(), res.begin() + limit, res.end(), greater(*this));
else
std::partial_sort(res.begin(), res.begin() + limit, res.end(), less(*this));
}
else
{
if (reverse)
std::sort(res.begin(), res.end(), greater(*this));
else
std::sort(res.begin(), res.end(), less(*this));
}
}
void reserve(size_t n) override
{
data.reserve(n);
}
std::string getName() const override { return "ColumnVector<" + TypeName<T>::get() + ">"; }
ColumnPtr cloneResized(size_t size) const override
{
ColumnPtr new_col_holder = std::make_shared<Self>();
if (size > 0)
{
auto & new_col = static_cast<Self &>(*new_col_holder);
new_col.data.resize(size);
size_t count = std::min(this->size(), size);
memcpy(&new_col.data[0], &data[0], count * sizeof(data[0]));
if (size > count)
memset(&new_col.data[count], value_type(), size - count);
}
return new_col_holder;
}
Field operator[](size_t n) const override
{
return typename NearestFieldType<T>::Type(data[n]);
}
void get(size_t n, Field & res) const override
{
res = typename NearestFieldType<T>::Type(data[n]);
}
const T & getElement(size_t n) const
{
return data[n];
}
T & getElement(size_t n)
{
return data[n];
}
UInt64 get64(size_t n) const override
{
return unionCastToUInt64(data[n]);
}
void insert(const Field & x) override
{
data.push_back(DB::get<typename NearestFieldType<T>::Type>(x));
}
void insertRangeFrom(const IColumn & src, size_t start, size_t length) override
{
const ColumnVector & src_vec = static_cast<const ColumnVector &>(src);
if (start + length > src_vec.data.size())
throw Exception("Parameters start = "
+ toString(start) + ", length = "
+ toString(length) + " are out of bound in ColumnVector::insertRangeFrom method"
" (data.size() = " + toString(src_vec.data.size()) + ").",
ErrorCodes::PARAMETER_OUT_OF_BOUND);
size_t old_size = data.size();
data.resize(old_size + length);
memcpy(&data[old_size], &src_vec.data[start], length * sizeof(data[0]));
}
ColumnPtr filter(const IColumn::Filter & filt, ssize_t result_size_hint) const override
{
size_t size = data.size();
if (size != filt.size())
throw Exception("Size of filter doesn't match size of column.", ErrorCodes::SIZES_OF_COLUMNS_DOESNT_MATCH);
std::shared_ptr<Self> res = std::make_shared<Self>();
typename Self::Container_t & res_data = res->getData();
if (result_size_hint)
res_data.reserve(result_size_hint > 0 ? result_size_hint : size);
const UInt8 * filt_pos = &filt[0];
const UInt8 * filt_end = filt_pos + size;
const T * data_pos = &data[0];
#if __SSE2__
/** A slightly more optimized version.
* Based on the assumption that often pieces of consecutive values
* completely pass or do not pass the filter completely.
* Therefore, we will optimistically check the parts of `SIMD_BYTES` values.
*/
static constexpr size_t SIMD_BYTES = 16;
const __m128i zero16 = _mm_setzero_si128();
const UInt8 * filt_end_sse = filt_pos + size / SIMD_BYTES * SIMD_BYTES;
while (filt_pos < filt_end_sse)
{
int mask = _mm_movemask_epi8(_mm_cmpgt_epi8(_mm_loadu_si128(reinterpret_cast<const __m128i *>(filt_pos)), zero16));
if (0 == mask)
{
/// Nothing is inserted.
}
else if (0xFFFF == mask)
{
res_data.insert(data_pos, data_pos + SIMD_BYTES);
}
else
{
for (size_t i = 0; i < SIMD_BYTES; ++i)
if (filt_pos[i])
res_data.push_back(data_pos[i]);
}
filt_pos += SIMD_BYTES;
data_pos += SIMD_BYTES;
}
#endif
while (filt_pos < filt_end)
{
if (*filt_pos)
res_data.push_back(*data_pos);
++filt_pos;
++data_pos;
}
return res;
}
ColumnPtr permute(const IColumn::Permutation & perm, size_t limit) const override
{
size_t size = data.size();
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);
std::shared_ptr<Self> res = std::make_shared<Self>(limit);
typename Self::Container_t & res_data = res->getData();
for (size_t i = 0; i < limit; ++i)
res_data[i] = data[perm[i]];
return res;
}
ColumnPtr replicate(const IColumn::Offsets_t & offsets) const override
{
size_t size = data.size();
if (size != offsets.size())
throw Exception("Size of offsets doesn't match size of column.", ErrorCodes::SIZES_OF_COLUMNS_DOESNT_MATCH);
if (0 == size)
return std::make_shared<Self>();
std::shared_ptr<Self> res = std::make_shared<Self>();
typename Self::Container_t & res_data = res->getData();
res_data.reserve(offsets.back());
IColumn::Offset_t prev_offset = 0;
for (size_t i = 0; i < size; ++i)
{
size_t size_to_replicate = offsets[i] - prev_offset;
prev_offset = offsets[i];
for (size_t j = 0; j < size_to_replicate; ++j)
res_data.push_back(data[i]);
}
return res;
}
void getExtremes(Field & min, Field & max) const override
{
size_t size = data.size();
if (size == 0)
{
min = typename NearestFieldType<T>::Type(0);
max = typename NearestFieldType<T>::Type(0);
return;
}
bool has_value = false;
/** Skip all NaNs in extremes calculation.
* If all values are NaNs, then return NaN.
* NOTE: There exist many different NaNs.
* Different NaN could be returned: not bit-exact value as one of NaNs from column.
*/
T cur_min = NaNOrZero<T>();
T cur_max = NaNOrZero<T>();
for (const T x : data)
{
if (isNaN(x))
continue;
if (!has_value)
{
cur_min = x;
cur_max = x;
has_value = true;
continue;
}
if (x < cur_min)
cur_min = x;
if (x > cur_max)
cur_max = x;
}
min = typename NearestFieldType<T>::Type(cur_min);
max = typename NearestFieldType<T>::Type(cur_max);
}
Columns scatter(ColumnIndex num_columns, const Selector & selector) const override
{
return this->scatterImpl<Self>(num_columns, selector);
}
/** More efficient methods of manipulation - to manipulate with data directly. */
Container_t & getData()
{
return data;
}
const Container_t & getData() const
{
return data;
}
protected:
Container_t data;
};
}