2018-03-03 20:44:01 +00:00
|
|
|
#ifndef PARQUET_CURSOR_H
|
|
|
|
#define PARQUET_CURSOR_H
|
|
|
|
|
|
|
|
#include "parquet_table.h"
|
|
|
|
#include "parquet/api/reader.h"
|
|
|
|
|
|
|
|
class ParquetCursor {
|
|
|
|
|
|
|
|
ParquetTable* table;
|
|
|
|
std::unique_ptr<parquet::ParquetFileReader> reader;
|
|
|
|
std::unique_ptr<parquet::RowGroupMetaData> rowGroupMetadata;
|
|
|
|
std::shared_ptr<parquet::RowGroupReader> rowGroup;
|
|
|
|
std::vector<std::shared_ptr<parquet::Scanner>> scanners;
|
|
|
|
std::vector<parquet::Type::type> types;
|
2018-03-04 22:20:28 +00:00
|
|
|
std::vector<parquet::LogicalType::type> logicalTypes;
|
2018-03-03 20:44:01 +00:00
|
|
|
|
|
|
|
std::vector<int> colRows;
|
|
|
|
std::vector<bool> colNulls;
|
|
|
|
std::vector<uintptr_t> colIntValues;
|
|
|
|
std::vector<double> colDoubleValues;
|
|
|
|
std::vector<parquet::ByteArray> colByteArrayValues;
|
|
|
|
|
|
|
|
int rowId;
|
|
|
|
int rowGroupId;
|
2018-03-05 03:29:35 +00:00
|
|
|
int rowGroupStartRowId;
|
2018-03-03 20:44:01 +00:00
|
|
|
int numRows;
|
|
|
|
int numRowGroups;
|
|
|
|
int rowsLeftInRowGroup;
|
|
|
|
|
2018-03-04 22:49:19 +00:00
|
|
|
bool nextRowGroup();
|
2018-03-03 20:44:01 +00:00
|
|
|
|
|
|
|
public:
|
|
|
|
ParquetCursor(ParquetTable* table);
|
|
|
|
int getRowId();
|
|
|
|
void next();
|
2018-03-05 02:05:26 +00:00
|
|
|
void close();
|
|
|
|
void reset();
|
2018-03-03 20:44:01 +00:00
|
|
|
bool eof();
|
|
|
|
|
|
|
|
void ensureColumn(int col);
|
|
|
|
bool isNull(int col);
|
2018-03-04 01:00:50 +00:00
|
|
|
parquet::Type::type getPhysicalType(int col);
|
2018-03-04 22:20:28 +00:00
|
|
|
parquet::LogicalType::type getLogicalType(int col);
|
2018-03-04 01:00:50 +00:00
|
|
|
|
|
|
|
int getInt32(int col);
|
|
|
|
long getInt64(int col);
|
2018-03-03 20:44:01 +00:00
|
|
|
double getDouble(int col);
|
|
|
|
parquet::ByteArray* getByteArray(int col);
|
|
|
|
};
|
|
|
|
|
|
|
|
#endif
|
|
|
|
|