ClickHouse/dbms/src/Functions/flatten.cpp

127 lines
4.0 KiB
C++
Raw Normal View History

2019-02-14 14:56:11 +00:00
#include <Functions/IFunction.h>
#include <Functions/FunctionFactory.h>
2019-02-15 05:47:49 +00:00
#include <Functions/FunctionHelpers.h>
#include <DataTypes/DataTypeArray.h>
#include <Columns/ColumnArray.h>
2019-02-14 14:56:11 +00:00
namespace DB
{
namespace ErrorCodes
{
extern const int ILLEGAL_COLUMN;
}
2019-02-15 05:47:49 +00:00
/// flatten([[1, 2, 3], [4, 5]]) = [1, 2, 3, 4, 5] - flatten array.
class FunctionFlatten : public IFunction
{
public:
static constexpr auto name = "flatten";
2019-02-14 14:56:11 +00:00
2019-03-01 20:07:58 +00:00
static FunctionPtr create(const Context &) { return std::make_shared<FunctionFlatten>(); }
2019-02-14 14:56:11 +00:00
2019-02-15 05:47:49 +00:00
size_t getNumberOfArguments() const override { return 1; }
2019-03-01 20:07:58 +00:00
bool useDefaultImplementationForConstants() const override { return true; }
2019-02-14 14:56:11 +00:00
2019-02-15 05:47:49 +00:00
DataTypePtr getReturnTypeImpl(const DataTypes & arguments) const override
{
if (!isArray(arguments[0]))
throw Exception("Illegal type " + arguments[0]->getName() +
" of argument of function " + getName() +
", expected Array", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT);
2019-02-14 14:56:11 +00:00
2019-02-15 05:47:49 +00:00
DataTypePtr nested_type = arguments[0];
while (isArray(nested_type))
nested_type = checkAndGetDataType<DataTypeArray>(nested_type.get())->getNestedType();
2019-02-14 14:56:11 +00:00
2019-02-15 05:47:49 +00:00
return std::make_shared<DataTypeArray>(nested_type);
}
2019-02-14 14:56:11 +00:00
2019-02-15 05:47:49 +00:00
void executeImpl(Block & block, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) override
{
2019-03-01 20:07:58 +00:00
/** We create an array column with array elements as the most deep elements of nested arrays,
* and construct offsets by selecting elements of most deep offsets by values of ancestor offsets.
*
2019-03-01 20:16:56 +00:00
Example 1:
2019-03-01 20:07:58 +00:00
Source column: Array(Array(UInt8)):
Row 1: [[1, 2, 3], [4, 5]], Row 2: [[6], [7, 8]]
data: [1, 2, 3], [4, 5], [6], [7, 8]
offsets: 2, 4
data.data: 1 2 3 4 5 6 7 8
data.offsets: 3 5 6 8
Result column: Array(UInt8):
Row 1: [1, 2, 3, 4, 5], Row 2: [6, 7, 8]
data: 1 2 3 4 5 6 7 8
offsets: 5 8
Result offsets are selected from the most deep (data.offsets) by previous deep (offsets) (and values are decremented by one):
3 5 6 8
^ ^
Example 2:
Source column: Array(Array(Array(UInt8))):
Row 1: [[], [[1], [], [2, 3]]], Row 2: [[[4]]]
most deep data: 1 2 3 4
offsets1: 2 3
offsets2: 0 3 4
- ^ ^ - select by prev offsets
offsets3: 1 1 3 4
- ^ ^ - select by prev offsets
result offsets: 3, 4
result: Row 1: [1, 2, 3], Row2: [4]
*/
const ColumnArray * src_col = checkAndGetColumn<ColumnArray>(block.getByPosition(arguments[0]).column.get());
if (!src_col)
throw Exception("Illegal column " + block.getByPosition(arguments[0]).column->getName() + " in argument of function 'flatten'",
ErrorCodes::ILLEGAL_COLUMN);
const IColumn::Offsets & src_offsets = src_col->getOffsets();
ColumnArray::ColumnOffsets::MutablePtr result_offsets_column;
const IColumn::Offsets * prev_offsets = &src_offsets;
const IColumn * prev_data = &src_col->getData();
while (const ColumnArray * next_col = checkAndGetColumn<ColumnArray>(prev_data))
2019-02-14 14:56:11 +00:00
{
2019-03-01 20:07:58 +00:00
if (!result_offsets_column)
result_offsets_column = ColumnArray::ColumnOffsets::create(input_rows_count);
IColumn::Offsets & result_offsets = result_offsets_column->getData();
const IColumn::Offsets * next_offsets = &next_col->getOffsets();
for (size_t i = 0; i < input_rows_count; ++i)
2019-03-01 20:17:30 +00:00
result_offsets[i] = (*next_offsets)[(*prev_offsets)[i] - 1]; /// -1 array subscript is Ok, see PaddedPODArray
2019-03-01 20:07:58 +00:00
prev_offsets = &result_offsets;
prev_data = &next_col->getData();
2019-02-14 14:56:11 +00:00
}
2019-03-01 20:07:58 +00:00
block.getByPosition(result).column = ColumnArray::create(
prev_data->getPtr(),
result_offsets_column ? std::move(result_offsets_column) : src_col->getOffsetsPtr());
2019-02-15 05:47:49 +00:00
}
2019-02-14 14:56:11 +00:00
2019-02-15 05:47:49 +00:00
private:
String getName() const override
{
return name;
}
};
void registerFunctionFlatten(FunctionFactory & factory)
{
factory.registerFunction<FunctionFlatten>();
}
2019-02-14 14:56:11 +00:00
}