ClickHouse/src/Processors/Transforms/WindowTransform.h

78 lines
2.2 KiB
C++
Raw Normal View History

2020-12-10 22:16:58 +00:00
#pragma once
#include <Processors/ISimpleTransform.h>
2020-12-15 00:36:03 +00:00
#include <Interpreters/AggregateDescription.h>
#include <Common/AlignedBuffer.h>
2020-12-10 22:16:58 +00:00
namespace DB
{
class ExpressionActions;
using ExpressionActionsPtr = std::shared_ptr<ExpressionActions>;
2020-12-15 00:36:03 +00:00
class Arena;
2020-12-18 17:13:28 +00:00
// Runtime data for computing one window function
2020-12-15 00:36:03 +00:00
struct WindowFunctionWorkspace
{
WindowFunctionDescription window_function;
AlignedBuffer aggregate_function_state;
std::vector<size_t> argument_column_indices;
// Argument and result columns. Be careful, they are per-chunk.
2020-12-15 00:36:03 +00:00
std::vector<const IColumn *> argument_columns;
MutableColumnPtr result_column;
2020-12-15 00:36:03 +00:00
};
2020-12-18 17:13:28 +00:00
/*
* Computes several window functions that share the same window. The input must
* be sorted correctly for this window (PARTITION BY, then ORDER BY).
*/
2020-12-10 22:16:58 +00:00
class WindowTransform : public ISimpleTransform
{
public:
WindowTransform(
2020-12-22 01:41:02 +00:00
const Block & input_header_,
const Block & output_header_,
const WindowDescription & window_description_,
const std::vector<WindowFunctionDescription> &
2020-12-22 04:02:39 +00:00
window_function_descriptions);
2020-12-15 00:36:03 +00:00
~WindowTransform() override;
2020-12-10 22:16:58 +00:00
String getName() const override
{
return "WindowTransform";
}
static Block transformHeader(Block header, const ExpressionActionsPtr & expression);
void transform(Chunk & chunk) override;
2020-12-15 00:36:03 +00:00
public:
Block input_header;
WindowDescription window_description;
2020-12-18 17:13:28 +00:00
2020-12-16 12:57:47 +00:00
// Indices of the PARTITION BY columns in block.
2020-12-15 00:36:03 +00:00
std::vector<size_t> partition_by_indices;
2020-12-18 17:13:28 +00:00
2020-12-16 12:57:47 +00:00
// The columns for PARTITION BY and the row in these columns where the
// current partition started. They might be in some of the previous blocks,
// so we have to keep the shared ownership of the columns. We don't keep the
// entire block to save memory, only the needed columns, in the same order
// as the partition_by_indices array.
// Can be empty if there is no PARTITION BY.
// Columns are nullptr when it is the first partition.
std::vector<ColumnPtr> partition_start_columns;
size_t partition_start_row = 0;
2020-12-15 00:36:03 +00:00
2020-12-18 17:13:28 +00:00
// Data for computing the window functions.
2020-12-15 00:36:03 +00:00
std::vector<WindowFunctionWorkspace> workspaces;
std::unique_ptr<Arena> arena;
2020-12-10 22:16:58 +00:00
};
}