should be it

This commit is contained in:
2025-10-24 19:21:19 -05:00
parent a4b23fc57c
commit f09560c7b1
14047 changed files with 3161551 additions and 1 deletions

View File

@@ -0,0 +1,32 @@
//===----------------------------------------------------------------------===//
// DuckDB
//
// decoder/byte_stream_split_decoder.hpp
//
//
//===----------------------------------------------------------------------===//
#pragma once
#include "duckdb.hpp"
#include "parquet_bss_decoder.hpp"
namespace duckdb {
class ColumnReader;
class ByteStreamSplitDecoder {
public:
explicit ByteStreamSplitDecoder(ColumnReader &reader);
public:
void InitializePage();
void Read(uint8_t *defines, idx_t read_count, Vector &result, idx_t result_offset);
void Skip(uint8_t *defines, idx_t skip_count);
private:
ColumnReader &reader;
ResizeableBuffer &decoded_data_buffer;
unique_ptr<BssDecoder> bss_decoder;
};
} // namespace duckdb

View File

@@ -0,0 +1,33 @@
//===----------------------------------------------------------------------===//
// DuckDB
//
// decoder/delta_binary_packed_decoder.hpp
//
//
//===----------------------------------------------------------------------===//
#pragma once
#include "duckdb.hpp"
#include "parquet_dbp_decoder.hpp"
#include "resizable_buffer.hpp"
namespace duckdb {
class ColumnReader;
class DeltaBinaryPackedDecoder {
public:
explicit DeltaBinaryPackedDecoder(ColumnReader &reader);
public:
void InitializePage();
void Read(uint8_t *defines, idx_t read_count, Vector &result, idx_t result_offset);
void Skip(uint8_t *defines, idx_t skip_count);
private:
ColumnReader &reader;
ResizeableBuffer &decoded_data_buffer;
unique_ptr<DbpDecoder> dbp_decoder;
};
} // namespace duckdb

View File

@@ -0,0 +1,38 @@
//===----------------------------------------------------------------------===//
// DuckDB
//
// decoder/delta_byte_array_decoder.hpp
//
//
//===----------------------------------------------------------------------===//
#pragma once
#include "duckdb.hpp"
#include "parquet_dbp_decoder.hpp"
#include "resizable_buffer.hpp"
namespace duckdb {
class ColumnReader;
class DeltaByteArrayDecoder {
public:
explicit DeltaByteArrayDecoder(ColumnReader &reader);
public:
void InitializePage();
void Read(uint8_t *defines, idx_t read_count, Vector &result, idx_t result_offset);
void Skip(uint8_t *defines, idx_t skip_count);
static void ReadDbpData(Allocator &allocator, ResizeableBuffer &buffer, ResizeableBuffer &result_buffer,
idx_t &value_count);
private:
ColumnReader &reader;
unique_ptr<Vector> byte_array_data;
idx_t byte_array_count = 0;
idx_t delta_offset = 0;
};
} // namespace duckdb

View File

@@ -0,0 +1,43 @@
//===----------------------------------------------------------------------===//
// DuckDB
//
// decoder/delta_length_byte_array_decoder.hpp
//
//
//===----------------------------------------------------------------------===//
#pragma once
#include "duckdb.hpp"
#include "parquet_dbp_decoder.hpp"
#include "resizable_buffer.hpp"
namespace duckdb {
class ColumnReader;
class DeltaLengthByteArrayDecoder {
public:
explicit DeltaLengthByteArrayDecoder(ColumnReader &reader);
public:
void InitializePage();
void Read(shared_ptr<ResizeableBuffer> &block, uint8_t *defines, idx_t read_count, Vector &result,
idx_t result_offset);
void Skip(uint8_t *defines, idx_t skip_count);
private:
template <bool HAS_DEFINES>
void ReadInternal(shared_ptr<ResizeableBuffer> &block, uint8_t *defines, idx_t read_count, Vector &result,
idx_t result_offset);
template <bool HAS_DEFINES>
void SkipInternal(uint8_t *defines, idx_t skip_count);
private:
ColumnReader &reader;
ResizeableBuffer &length_buffer;
idx_t byte_array_count = 0;
idx_t length_idx;
};
} // namespace duckdb

View File

@@ -0,0 +1,56 @@
//===----------------------------------------------------------------------===//
// DuckDB
//
// decoder/dictionary_decoder.hpp
//
//
//===----------------------------------------------------------------------===//
#pragma once
#include "duckdb.hpp"
#include "parquet_rle_bp_decoder.hpp"
#include "resizable_buffer.hpp"
namespace duckdb {
class ColumnReader;
struct TableFilterState;
class DictionaryDecoder {
public:
explicit DictionaryDecoder(ColumnReader &reader);
public:
void InitializeDictionary(idx_t dictionary_size, optional_ptr<const TableFilter> filter,
optional_ptr<TableFilterState> filter_state, bool has_defines);
void InitializePage();
idx_t Read(uint8_t *defines, idx_t read_count, Vector &result, idx_t result_offset);
void Skip(uint8_t *defines, idx_t skip_count);
bool CanFilter(const TableFilter &filter, TableFilterState &filter_state);
bool DictionarySupportsFilter(const TableFilter &filter, TableFilterState &filter_state);
void Filter(uint8_t *defines, idx_t read_count, Vector &result, SelectionVector &sel, idx_t &approved_tuple_count);
bool HasFilter() const {
return filter_result.get();
}
bool HasFilteredOutAllValues() const {
return HasFilter() && filter_count == 0;
}
private:
idx_t GetValidValues(uint8_t *defines, idx_t read_count, idx_t result_offset);
void ConvertDictToSelVec(uint32_t *offsets, const SelectionVector &rows, idx_t count);
private:
ColumnReader &reader;
ResizeableBuffer &offset_buffer;
unique_ptr<RleBpDecoder> dict_decoder;
SelectionVector valid_sel;
SelectionVector dictionary_selection_vector;
idx_t dictionary_size;
buffer_ptr<VectorChildBuffer> dictionary;
unsafe_unique_array<bool> filter_result;
idx_t filter_count;
bool can_have_nulls;
};
} // namespace duckdb

View File

@@ -0,0 +1,32 @@
//===----------------------------------------------------------------------===//
// DuckDB
//
// decoder/rle_decoder.hpp
//
//
//===----------------------------------------------------------------------===//
#pragma once
#include "duckdb.hpp"
#include "parquet_rle_bp_decoder.hpp"
namespace duckdb {
class ColumnReader;
class RLEDecoder {
public:
explicit RLEDecoder(ColumnReader &reader);
public:
void InitializePage();
void Read(uint8_t *defines, idx_t read_count, Vector &result, idx_t result_offset);
void Skip(uint8_t *defines, idx_t skip_count);
private:
ColumnReader &reader;
ResizeableBuffer &decoded_data_buffer;
unique_ptr<RleBpDecoder> rle_decoder;
};
} // namespace duckdb