ColumnPtr ColumnArray::replicateNumber(const Offsets_t & replicate_offsets) const
{
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);
ColumnPtr res = cloneEmpty();
if (0 == col_size)
return res;
ColumnArray & res_ = typeid_cast<ColumnArray &>(*res);
const typename ColumnVector<T>::Container_t & src_data = typeid_cast<const ColumnVector<T> &>(*data).getData();
const Offsets_t & src_offsets = getOffsets();
typename ColumnVector<T>::Container_t & res_data = typeid_cast<ColumnVector<T> &>(res_.getData()).getData();
Offsets_t & res_offsets = res_.getOffsets();
res_data.reserve(data->size() / col_size * replicate_offsets.back());
res_offsets.reserve(replicate_offsets.back());
Offset_t prev_replicate_offset = 0;
Offset_t prev_data_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;
size_t value_size = src_offsets[i] - prev_data_offset;
for (size_t j = 0; j < size_to_replicate; ++j)
{
current_new_offset += value_size;
res_offsets.push_back(current_new_offset);
res_data.resize(res_data.size() + value_size);
memcpy(&res_data[res_data.size() - value_size], &src_data[prev_data_offset], value_size * sizeof(T));
}
prev_replicate_offset = replicate_offsets[i];
prev_data_offset = src_offsets[i];
}
return res;
}
ColumnPtr ColumnString::replicate(const Offsets_t & replicate_offsets) const
{
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);
std::shared_ptr<ColumnString> res = std::make_shared<ColumnString>();
if (0 == col_size)
return res;
Chars_t & res_chars = res->chars;
Offsets_t & res_offsets = res->offsets;
res_chars.reserve(chars.size() / col_size * replicate_offsets.back());
res_offsets.reserve(replicate_offsets.back());
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;
size_t string_size = offsets[i] - prev_string_offset;
for (size_t j = 0; j < size_to_replicate; ++j)
{
current_new_offset += string_size;
res_offsets.push_back(current_new_offset);
res_chars.resize(res_chars.size() + string_size);
memcpySmallAllowReadWriteOverflow15(
&res_chars[res_chars.size() - string_size], &chars[prev_string_offset], string_size);
}
prev_replicate_offset = replicate_offsets[i];
prev_string_offset = offsets[i];
}
return res;
}
ColumnPtr ColumnArray::replicateConst(const Offsets_t & replicate_offsets) const
{
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);
if (0 == col_size)
return cloneEmpty();
const Offsets_t & src_offsets = getOffsets();
auto res_column_offsets = std::make_shared<ColumnOffsets_t>();
Offsets_t & res_offsets = res_column_offsets->getData();
res_offsets.reserve(replicate_offsets.back());
Offset_t prev_replicate_offset = 0;
Offset_t prev_data_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;
size_t value_size = src_offsets[i] - prev_data_offset;
for (size_t j = 0; j < size_to_replicate; ++j)
{
current_new_offset += value_size;
res_offsets.push_back(current_new_offset);
}
prev_replicate_offset = replicate_offsets[i];
prev_data_offset = src_offsets[i];
}
return std::make_shared<ColumnArray>(getData().cloneResized(current_new_offset), res_column_offsets);
}
ColumnPtr ColumnArray::replicateString(const Offsets_t & replicate_offsets) const
{
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);
ColumnPtr res = cloneEmpty();
if (0 == col_size)
return res;
ColumnArray & res_ = static_cast<ColumnArray &>(*res);
const ColumnString & src_string = typeid_cast<const ColumnString &>(*data);
const ColumnString::Chars_t & src_chars = src_string.getChars();
const Offsets_t & src_string_offsets = src_string.getOffsets();
const Offsets_t & src_offsets = getOffsets();
ColumnString::Chars_t & res_chars = typeid_cast<ColumnString &>(res_.getData()).getChars();
Offsets_t & res_string_offsets = typeid_cast<ColumnString &>(res_.getData()).getOffsets();
Offsets_t & res_offsets = res_.getOffsets();
res_chars.reserve(src_chars.size() / col_size * replicate_offsets.back());
res_string_offsets.reserve(src_string_offsets.size() / col_size * replicate_offsets.back());
res_offsets.reserve(replicate_offsets.back());
Offset_t prev_replicate_offset = 0;
Offset_t prev_src_offset = 0;
Offset_t prev_src_string_offset = 0;
Offset_t current_res_offset = 0;
Offset_t current_res_string_offset = 0;
for (size_t i = 0; i < col_size; ++i)
{
/// Насколько размножить массив.
size_t size_to_replicate = replicate_offsets[i] - prev_replicate_offset;
/// Количество строк в массиве.
size_t value_size = src_offsets[i] - prev_src_offset;
/// Количество символов в строках массива, включая нулевые байты.
size_t sum_chars_size = value_size == 0 ? 0 : (src_string_offsets[prev_src_offset + value_size - 1] - prev_src_string_offset);
for (size_t j = 0; j < size_to_replicate; ++j)
{
current_res_offset += value_size;
res_offsets.push_back(current_res_offset);
size_t prev_src_string_offset_local = prev_src_string_offset;
for (size_t k = 0; k < value_size; ++k)
{
/// Размер одной строки.
size_t chars_size = src_string_offsets[k + prev_src_offset] - prev_src_string_offset_local;
current_res_string_offset += chars_size;
res_string_offsets.push_back(current_res_string_offset);
prev_src_string_offset_local += chars_size;
}
/// Копирование символов массива строк.
res_chars.resize(res_chars.size() + sum_chars_size);
memcpySmallAllowReadWriteOverflow15(
&res_chars[res_chars.size() - sum_chars_size], &src_chars[prev_src_string_offset], sum_chars_size);
}
prev_replicate_offset = replicate_offsets[i];
prev_src_offset = src_offsets[i];
prev_src_string_offset += sum_chars_size;
}
return res;
}
