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];
}
}
SetVariants::Type SetVariants::chooseMethod(const ConstColumnPlainPtrs & key_columns, Sizes & key_sizes)
{
size_t keys_size = key_columns.size();
bool all_fixed = true;
size_t keys_bytes = 0;
key_sizes.resize(keys_size);
for (size_t j = 0; j < keys_size; ++j)
{
if (!key_columns[j]->isFixed())
{
all_fixed = false;
break;
}
key_sizes[j] = key_columns[j]->sizeOfField();
keys_bytes += key_sizes[j];
}
/// Если есть один числовой ключ, который помещается в 64 бита
if (keys_size == 1 && key_columns[0]->isNumeric())
{
size_t size_of_field = key_columns[0]->sizeOfField();
if (size_of_field == 1)
return SetVariants::Type::key8;
if (size_of_field == 2)
return SetVariants::Type::key16;
if (size_of_field == 4)
return SetVariants::Type::key32;
if (size_of_field == 8)
return SetVariants::Type::key64;
throw Exception("Logical error: numeric column has sizeOfField not in 1, 2, 4, 8.", ErrorCodes::LOGICAL_ERROR);
}
/// Если ключи помещаются в N бит, будем использовать хэш-таблицу по упакованным в N-бит ключам
if (all_fixed && keys_bytes <= 16)
return SetVariants::Type::keys128;
if (all_fixed && keys_bytes <= 32)
return SetVariants::Type::keys256;
/// If there is single string key, use hash table of it's values.
if (keys_size == 1 && (typeid_cast<const ColumnString *>(key_columns[0]) || typeid_cast<const ColumnConstString *>(key_columns[0])))
return SetVariants::Type::key_string;
if (keys_size == 1 && typeid_cast<const ColumnFixedString *>(key_columns[0]))
return SetVariants::Type::key_fixed_string;
/// Otherwise, will use set of cryptographic hashes of unambiguously serialized values.
return SetVariants::Type::hashed;
}
void ExternalQueryBuilder::composeKeyTuple(const ConstColumnPlainPtrs & key_columns, const std::size_t row, WriteBuffer & out) const
{
writeString("(", out);
const auto keys_size = key_columns.size();
auto first = true;
for (const auto i : ext::range(0, keys_size))
{
if (!first)
writeString(", ", out);
first = false;
(*dict_struct.key)[i].type->serializeTextQuoted(*key_columns[i], row, out);
}
writeString(")", out);
}
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 ExternalQueryBuilder::composeKeyCondition(const ConstColumnPlainPtrs & key_columns, const std::size_t row, WriteBuffer & out) const
{
writeString("(", out);
const auto keys_size = key_columns.size();
auto first = true;
for (const auto i : ext::range(0, keys_size))
{
if (!first)
writeString(" AND ", out);
first = false;
const auto & key_description = (*dict_struct.key)[i];
/// key_i=value_i
writeString(key_description.name, out);
writeString("=", out);
key_description.type->serializeTextQuoted(*key_columns[i], row, out);
}
writeString(")", out);
}
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);
}
}