void DistinctBlockInputStream::buildFilter(
Method & method,
const ConstColumnPlainPtrs & columns,
IColumn::Filter & filter,
size_t rows,
SetVariants & variants) const
{
typename Method::State state;
state.init(columns);
for (size_t i = 0; i < rows; ++i)
{
/// Make a key.
typename Method::Key key = state.getKey(columns, columns.size(), i, key_sizes);
typename Method::Data::iterator it = method.data.find(key);
bool inserted;
method.data.emplace(key, it, inserted);
if (inserted)
method.onNewKey(*it, columns.size(), i, variants.string_pool);
/// Emit the record if there is no such key in the current set yet.
/// Skip it otherwise.
filter[i] = inserted;
}
}
void NO_INLINE Set::executeArrayImpl(
Method & method,
const ConstColumnPlainPtrs & key_columns,
const ColumnArray::Offsets_t & offsets,
ColumnUInt8::Container_t & vec_res,
bool negative,
size_t rows) const
{
typename Method::State state;
state.init(key_columns);
size_t keys_size = key_columns.size();
size_t prev_offset = 0;
/// Для всех строчек
for (size_t i = 0; i < rows; ++i)
{
UInt8 res = 0;
/// Для всех элементов
for (size_t j = prev_offset; j < offsets[i]; ++j)
{
/// Строим ключ
typename Method::Key key = state.getKey(key_columns, keys_size, j, key_sizes);
res |= negative ^ (method.data.end() != method.data.find(key));
if (res)
break;
}
vec_res[i] = res;
prev_offset = offsets[i];
}
}
bool DistinctSortedBlockInputStream::buildFilter(
Method & method,
const ColumnRawPtrs & columns,
const ColumnRawPtrs & clearing_hint_columns,
IColumn::Filter & filter,
size_t rows,
ClearableSetVariants & variants) const
{
typename Method::State state;
state.init(columns);
/// Compare last row of previous block and first row of current block,
/// If rows not equal, we can clear HashSet,
/// If clearing_hint_columns is empty, we CAN'T clear HashSet.
if (!clearing_hint_columns.empty() && !prev_block.clearing_hint_columns.empty()
&& !rowsEqual(clearing_hint_columns, 0, prev_block.clearing_hint_columns, prev_block.block.rows() - 1))
{
method.data.clear();
}
bool has_new_data = false;
for (size_t i = 0; i < rows; ++i)
{
/// Compare i-th row and i-1-th row,
/// If rows are not equal, we can clear HashSet,
/// If clearing_hint_columns is empty, we CAN'T clear HashSet.
if (i > 0 && !clearing_hint_columns.empty() && !rowsEqual(clearing_hint_columns, i, clearing_hint_columns, i - 1))
method.data.clear();
/// Make a key.
typename Method::Key key = state.getKey(columns, columns.size(), i, key_sizes);
typename Method::Data::iterator it = method.data.find(key);
bool inserted;
method.data.emplace(key, it, inserted);
if (inserted)
{
method.onNewKey(*it, columns.size(), variants.string_pool);
has_new_data = true;
}
/// Emit the record if there is no such key in the current set yet.
/// Skip it otherwise.
filter[i] = inserted;
}
return has_new_data;
}
void NO_INLINE Set::executeImpl(
Method & method,
const ConstColumnPlainPtrs & key_columns,
ColumnUInt8::Container_t & vec_res,
bool negative,
size_t rows) const
{
typename Method::State state;
state.init(key_columns);
size_t keys_size = key_columns.size();
/// NOTE Не используется оптимизация для подряд идущих одинаковых значений.
/// Для всех строчек
for (size_t i = 0; i < rows; ++i)
{
/// Строим ключ
typename Method::Key key = state.getKey(key_columns, keys_size, i, key_sizes);
vec_res[i] = negative ^ (method.data.end() != method.data.find(key));
}
}
void NO_INLINE Set::insertFromBlockImpl(
Method & method,
const ConstColumnPlainPtrs & key_columns,
size_t rows,
SetVariants & variants)
{
typename Method::State state;
state.init(key_columns);
size_t keys_size = key_columns.size();
/// Для всех строчек
for (size_t i = 0; i < rows; ++i)
{
/// Строим ключ
typename Method::Key key = state.getKey(key_columns, keys_size, i, key_sizes);
typename Method::Data::iterator it = method.data.find(key);
bool inserted;
method.data.emplace(key, it, inserted);
if (inserted)
method.onNewKey(*it, keys_size, i, variants.string_pool);
}
}