ClickHouse/dbms/src/Interpreters/Set.cpp

376 lines
12 KiB
C++
Raw Normal View History

2012-08-23 23:49:28 +00:00
#include <iomanip>
2012-08-23 20:22:44 +00:00
#include <DB/Core/Field.h>
#include <DB/Columns/ColumnString.h>
#include <DB/Columns/ColumnFixedString.h>
#include <DB/Columns/ColumnsNumber.h>
2012-08-23 23:49:28 +00:00
#include <DB/DataStreams/IProfilingBlockInputStream.h>
2012-08-24 19:42:03 +00:00
#include <DB/DataStreams/OneBlockInputStream.h>
#include <DB/Parsers/ASTExpressionList.h>
#include <DB/Parsers/ASTFunction.h>
#include <DB/Parsers/ASTLiteral.h>
2012-08-23 23:49:28 +00:00
2012-08-23 20:22:44 +00:00
#include <DB/Interpreters/Set.h>
namespace DB
{
Set::Type Set::chooseMethod(const ConstColumnPlainPtrs & key_columns, bool & keys_fit_128_bits, Sizes & key_sizes)
2012-08-23 20:22:44 +00:00
{
size_t keys_size = key_columns.size();
keys_fit_128_bits = true;
size_t keys_bytes = 0;
key_sizes.resize(keys_size);
for (size_t j = 0; j < keys_size; ++j)
{
if (!key_columns[j]->isNumeric())
{
keys_fit_128_bits = false;
break;
}
key_sizes[j] = key_columns[j]->sizeOfField();
keys_bytes += key_sizes[j];
}
if (keys_bytes > 16)
keys_fit_128_bits = false;
/// Если есть один ключ, который помещается в 64 бита, и это не число с плавающей запятой
if (keys_size == 1 && key_columns[0]->isNumeric()
&& !dynamic_cast<const ColumnFloat32 *>(key_columns[0]) && !dynamic_cast<const ColumnFloat64 *>(key_columns[0])
&& !dynamic_cast<const ColumnConstFloat32 *>(key_columns[0]) && !dynamic_cast<const ColumnConstFloat64 *>(key_columns[0]))
2012-08-23 20:22:44 +00:00
return KEY_64;
/// Если есть один строковый ключ, то используем хэш-таблицу с ним
if (keys_size == 1
&& (dynamic_cast<const ColumnString *>(key_columns[0]) || dynamic_cast<const ColumnFixedString *>(key_columns[0])
|| dynamic_cast<const ColumnConstString *>(key_columns[0])))
2012-08-23 20:22:44 +00:00
return KEY_STRING;
/// Если много ключей - будем строить множество хэшей от них
return HASHED;
}
void Set::create(BlockInputStreamPtr stream)
{
LOG_TRACE(log, "Creating set");
2012-08-23 23:49:28 +00:00
Stopwatch watch;
size_t entries = 0;
2012-08-23 20:22:44 +00:00
/// Читаем все данные
while (Block block = stream->read())
{
size_t keys_size = block.columns();
Row key(keys_size);
ConstColumnPlainPtrs key_columns(keys_size);
2012-08-23 22:40:51 +00:00
data_types.resize(keys_size);
2012-08-23 20:22:44 +00:00
/// Запоминаем столбцы, с которыми будем работать
for (size_t i = 0; i < keys_size; ++i)
2012-08-23 22:40:51 +00:00
{
2012-08-23 20:22:44 +00:00
key_columns[i] = block.getByPosition(i).column;
2012-08-23 22:40:51 +00:00
data_types[i] = block.getByPosition(i).type;
}
2012-08-23 20:22:44 +00:00
size_t rows = block.rows();
/// Какую структуру данных для множества использовать?
bool keys_fit_128_bits = false;
Sizes key_sizes;
type = chooseMethod(key_columns, keys_fit_128_bits, key_sizes);
if (type == KEY_64)
{
SetUInt64 & res = key64;
const IColumn & column = *key_columns[0];
2012-08-23 20:22:44 +00:00
/// Для всех строчек
for (size_t i = 0; i < rows; ++i)
{
/// Строим ключ
UInt64 key = get<UInt64>(column[i]);
2012-08-23 20:22:44 +00:00
res.insert(key);
}
2012-08-23 23:49:28 +00:00
entries = res.size();
2012-08-23 20:22:44 +00:00
}
else if (type == KEY_STRING)
{
SetString & res = key_string;
const IColumn & column = *key_columns[0];
2012-08-23 20:22:44 +00:00
if (const ColumnString * column_string = dynamic_cast<const ColumnString *>(&column))
{
const ColumnString::Offsets_t & offsets = column_string->getOffsets();
2012-08-23 20:35:05 +00:00
const ColumnUInt8::Container_t & data = dynamic_cast<const ColumnUInt8 &>(column_string->getData()).getData();
2012-08-23 20:22:44 +00:00
/// Для всех строчек
for (size_t i = 0; i < rows; ++i)
{
/// Строим ключ
StringRef ref(&data[i == 0 ? 0 : offsets[i - 1]], (i == 0 ? offsets[i] : (offsets[i] - offsets[i - 1])) - 1);
SetString::iterator it;
bool inserted;
res.emplace(ref, it, inserted);
if (inserted)
it->data = string_pool.insert(ref.data, ref.size);
}
}
else if (const ColumnFixedString * column_string = dynamic_cast<const ColumnFixedString *>(&column))
{
size_t n = column_string->getN();
const ColumnUInt8::Container_t & data = dynamic_cast<const ColumnUInt8 &>(column_string->getData()).getData();
/// Для всех строчек
for (size_t i = 0; i < rows; ++i)
{
/// Строим ключ
StringRef ref(&data[i * n], n);
SetString::iterator it;
bool inserted;
res.emplace(ref, it, inserted);
if (inserted)
it->data = string_pool.insert(ref.data, ref.size);
}
}
else
throw Exception("Illegal type of column when creating set with string key: " + column.getName(), ErrorCodes::ILLEGAL_COLUMN);
2012-08-23 23:49:28 +00:00
entries = res.size();
2012-08-23 20:22:44 +00:00
}
else if (type == HASHED)
{
SetHashed & res = hashed;
/// Для всех строчек
for (size_t i = 0; i < rows; ++i)
2012-08-23 23:49:28 +00:00
res.insert(pack128(i, keys_fit_128_bits, keys_size, key, key_columns, key_sizes));
2012-08-23 20:22:44 +00:00
2012-08-23 23:49:28 +00:00
entries = res.size();
2012-08-23 20:22:44 +00:00
}
else if (type == GENERIC)
{
/// Общий способ
SetGeneric & res = generic;
/// Для всех строчек
for (size_t i = 0; i < rows; ++i)
{
/// Строим ключ
for (size_t j = 0; j < keys_size; ++j)
key_columns[j]->get(i, key[j]);
2012-08-23 20:22:44 +00:00
res.insert(key);
key.resize(keys_size);
2012-08-23 20:22:44 +00:00
}
2012-08-23 23:49:28 +00:00
entries = res.size();
2012-08-23 20:22:44 +00:00
}
else
throw Exception("Unknown set variant.", ErrorCodes::UNKNOWN_SET_DATA_VARIANT);
}
2012-08-23 23:49:28 +00:00
logProfileInfo(watch, *stream, entries);
}
void Set::logProfileInfo(Stopwatch & watch, IBlockInputStream & in, size_t entries)
{
/// Выведем информацию о том, сколько считано строк и байт.
size_t rows = 0;
size_t bytes = 0;
in.getLeafRowsBytes(rows, bytes);
size_t head_rows = 0;
if (IProfilingBlockInputStream * profiling_in = dynamic_cast<IProfilingBlockInputStream *>(&in))
head_rows = profiling_in->getInfo().rows;
if (rows != 0)
{
LOG_DEBUG(log, std::fixed << std::setprecision(3)
<< "Created set with " << entries << " entries from " << head_rows << " rows."
<< " Read " << rows << " rows, " << bytes / 1048576.0 << " MiB in " << watch.elapsedSeconds() << " sec., "
<< static_cast<size_t>(rows / watch.elapsedSeconds()) << " rows/sec., " << bytes / 1048576.0 / watch.elapsedSeconds() << " MiB/sec.");
}
2012-08-23 20:22:44 +00:00
}
2012-08-24 19:42:03 +00:00
void Set::create(DataTypes & types, ASTPtr node)
{
data_types = types;
/// Засунем множество в блок.
Block block;
for (size_t i = 0, size = data_types.size(); i < size; ++i)
{
ColumnWithNameAndType col;
col.type = data_types[i];
col.column = data_types[i]->createColumn();
col.name = "_" + Poco::NumberFormatter::format(i);
block.insert(col);
}
ASTExpressionList & list = dynamic_cast<ASTExpressionList &>(*node);
for (ASTs::iterator it = list.children.begin(); it != list.children.end(); ++it)
{
if (data_types.size() == 1)
{
if (ASTLiteral * lit = dynamic_cast<ASTLiteral *>(&**it))
block.getByPosition(0).column->insert(lit->value);
else
throw Exception("Incorrect element of set. Must be literal.", ErrorCodes::INCORRECT_ELEMENT_OF_SET);
}
else if (ASTFunction * func = dynamic_cast<ASTFunction *>(&**it))
{
if (func->name != "tuple")
throw Exception("Incorrect element of set. Must be tuple.", ErrorCodes::INCORRECT_ELEMENT_OF_SET);
size_t tuple_size = func->arguments->children.size();
if (tuple_size != data_types.size())
throw Exception("Incorrect size of tuple in set.", ErrorCodes::INCORRECT_ELEMENT_OF_SET);
for (size_t j = 0; j < tuple_size; ++j)
{
if (ASTLiteral * lit = dynamic_cast<ASTLiteral *>(&*func->arguments->children[j]))
block.getByPosition(j).column->insert(lit->value);
else
throw Exception("Incorrect element of tuple in set. Must be literal.", ErrorCodes::INCORRECT_ELEMENT_OF_SET);
}
}
else
throw Exception("Incorrect element of set", ErrorCodes::INCORRECT_ELEMENT_OF_SET);
/// NOTE: Потом можно реализовать возможность задавать константные выражения в множествах.
}
create(new OneBlockInputStream(block));
}
2012-08-23 20:22:44 +00:00
void Set::execute(Block & block, const ColumnNumbers & arguments, size_t result, bool negative) const
{
ColumnUInt8 * c_res = new ColumnUInt8;
block.getByPosition(result).column = c_res;
ColumnUInt8::Container_t & vec_res = c_res->getData();
vec_res.resize(block.getByPosition(arguments[0]).column->size());
2012-08-23 20:35:05 +00:00
2012-08-24 20:40:34 +00:00
/// Если множество пусто
if (data_types.empty())
return;
2012-08-23 20:35:05 +00:00
size_t keys_size = arguments.size();
Row key(keys_size);
ConstColumnPlainPtrs key_columns(keys_size);
2012-08-23 20:35:05 +00:00
2012-08-23 22:40:51 +00:00
/// Запоминаем столбцы, с которыми будем работать. Также проверим, что типы данных правильные.
2012-08-23 20:35:05 +00:00
for (size_t i = 0; i < keys_size; ++i)
2012-08-23 22:40:51 +00:00
{
2012-08-23 20:35:05 +00:00
key_columns[i] = block.getByPosition(arguments[i]).column;
2012-08-23 22:40:51 +00:00
if (data_types[i]->getName() != block.getByPosition(arguments[i]).type->getName())
throw Exception("Types in section IN doesn't match.", ErrorCodes::TYPE_MISMATCH);
}
2012-08-23 20:35:05 +00:00
size_t rows = block.rows();
/// Какую структуру данных для множества использовать?
bool keys_fit_128_bits = false;
Sizes key_sizes;
if (type != chooseMethod(key_columns, keys_fit_128_bits, key_sizes))
throw Exception("Incompatible columns in IN section", ErrorCodes::INCOMPATIBLE_COLUMNS);
if (type == KEY_64)
{
2012-08-23 22:27:10 +00:00
const SetUInt64 & set = key64;
const IColumn & column = *key_columns[0];
2012-08-23 20:35:05 +00:00
/// Для всех строчек
for (size_t i = 0; i < rows; ++i)
{
/// Строим ключ
UInt64 key = get<UInt64>(column[i]);
2012-08-23 20:35:05 +00:00
vec_res[i] = negative ^ (set.end() != set.find(key));
}
}
else if (type == KEY_STRING)
{
2012-08-23 22:27:10 +00:00
const SetString & set = key_string;
const IColumn & column = *key_columns[0];
2012-08-23 20:35:05 +00:00
if (const ColumnString * column_string = dynamic_cast<const ColumnString *>(&column))
{
const ColumnString::Offsets_t & offsets = column_string->getOffsets();
const ColumnUInt8::Container_t & data = dynamic_cast<const ColumnUInt8 &>(column_string->getData()).getData();
/// Для всех строчек
for (size_t i = 0; i < rows; ++i)
{
/// Строим ключ
StringRef ref(&data[i == 0 ? 0 : offsets[i - 1]], (i == 0 ? offsets[i] : (offsets[i] - offsets[i - 1])) - 1);
vec_res[i] = negative ^ (set.end() != set.find(ref));
}
}
else if (const ColumnFixedString * column_string = dynamic_cast<const ColumnFixedString *>(&column))
{
size_t n = column_string->getN();
const ColumnUInt8::Container_t & data = dynamic_cast<const ColumnUInt8 &>(column_string->getData()).getData();
/// Для всех строчек
for (size_t i = 0; i < rows; ++i)
{
/// Строим ключ
StringRef ref(&data[i * n], n);
vec_res[i] = negative ^ (set.end() != set.find(ref));
}
}
2012-08-24 19:42:03 +00:00
else if (const ColumnConstString * column_string = dynamic_cast<const ColumnConstString *>(&column))
{
bool res = negative ^ (set.end() != set.find(StringRef(column_string->getData())));
/// Для всех строчек
for (size_t i = 0; i < rows; ++i)
vec_res[i] = res;
}
2012-08-23 20:35:05 +00:00
else
throw Exception("Illegal type of column when creating set with string key: " + column.getName(), ErrorCodes::ILLEGAL_COLUMN);
}
else if (type == HASHED)
{
2012-08-23 22:27:10 +00:00
const SetHashed & set = hashed;
2012-08-23 20:35:05 +00:00
/// Для всех строчек
for (size_t i = 0; i < rows; ++i)
2012-08-23 23:49:28 +00:00
vec_res[i] = negative ^ (set.end() != set.find(pack128(i, keys_fit_128_bits, keys_size, key, key_columns, key_sizes)));
2012-08-23 20:35:05 +00:00
}
else if (type == GENERIC)
{
/// Общий способ
2012-08-23 22:27:10 +00:00
const SetGeneric & set = generic;
2012-08-23 20:35:05 +00:00
/// Для всех строчек
for (size_t i = 0; i < rows; ++i)
{
/// Строим ключ
for (size_t j = 0; j < keys_size; ++j)
key_columns[j]->get(i, key[j]);
2012-08-23 20:35:05 +00:00
vec_res[i] = negative ^ (set.end() != set.find(key));
}
}
else
throw Exception("Unknown set variant.", ErrorCodes::UNKNOWN_SET_DATA_VARIANT);
2012-08-23 20:22:44 +00:00
}
}