MySQL 8.3.0
Source Code Documentation
Row_bunch< Column_type > Class Template Reference

Implements the row and column memory management for parse and load operations. More...

#include <bulk_data_service.h>

Public Member Functions

 Row_bunch (size_t n_cols)
 Create a new row bunch. More...
 
size_t get_num_rows () const
 
size_t get_num_cols () const
 
template<typename F >
bool process_columns (size_t row_index, F &&cbk)
 Process all columns, invoking callback for each. More...
 
template<typename F >
bool process_columns_by_offset (size_t row_offset, F &&cbk)
 
size_t get_row_offset (size_t row_index) const
 Get current row offset to access columns. More...
 
size_t get_next_row_offset (size_t &offset) const
 Get next row offset from current row offset. More...
 
Column_type & get_column (size_t row_offset, size_t col_index)
 Get column using row offset and column index. More...
 
const Column_type & read_column (size_t row_offset, size_t col_index) const
 Get constant column for reading using row offset and column index. More...
 
bool set_num_rows (size_t n_rows)
 Set the number of rows. More...
 

Static Public Attributes

static const size_t S_MAX_TOTAL_COLS = 600 * 1024 * 1024
 Limit allocation up to 600M columns. More...
 

Private Attributes

std::vector< Column_type > m_columns
 All the columns. More...
 
size_t m_num_rows {}
 Number of rows. More...
 
size_t m_num_columns {}
 Number of columns in each row. More...
 

Detailed Description

template<typename Column_type>
class Row_bunch< Column_type >

Implements the row and column memory management for parse and load operations.

We try to pre-allocate the memory contiguously as much as we can to maximize the performance.

Template Parameters
Column_typeColumn_text when used in the CSV context, Column_sql when used in the InnoDB context.

Constructor & Destructor Documentation

◆ Row_bunch()

template<typename Column_type >
Row_bunch< Column_type >::Row_bunch ( size_t  n_cols)
inline

Create a new row bunch.

Parameters
[in]n_colsnumber of columns

Member Function Documentation

◆ get_column()

template<typename Column_type >
Column_type & Row_bunch< Column_type >::get_column ( size_t  row_offset,
size_t  col_index 
)
inline

Get column using row offset and column index.

Parameters
[in]row_offsetrow offset in column vector
[in]col_indexindex of the column within row
Returns
column data

◆ get_next_row_offset()

template<typename Column_type >
size_t Row_bunch< Column_type >::get_next_row_offset ( size_t &  offset) const
inline

Get next row offset from current row offset.

Parameters
[in,out]offsetrow offset
Returns
true if there is a next row.

◆ get_num_cols()

template<typename Column_type >
size_t Row_bunch< Column_type >::get_num_cols ( ) const
inline
Returns
return number of columns in each row.

◆ get_num_rows()

template<typename Column_type >
size_t Row_bunch< Column_type >::get_num_rows ( ) const
inline
Returns
return number of rows in the bunch.

◆ get_row_offset()

template<typename Column_type >
size_t Row_bunch< Column_type >::get_row_offset ( size_t  row_index) const
inline

Get current row offset to access columns.

Parameters
[in]row_indexrow index
Returns
row offset in column vector.

◆ process_columns()

template<typename Column_type >
template<typename F >
bool Row_bunch< Column_type >::process_columns ( size_t  row_index,
F &&  cbk 
)
inline

Process all columns, invoking callback for each.

Parameters
[in]row_indexindex of the row
[in]cbkcallback function
Returns
true if successful

◆ process_columns_by_offset()

template<typename Column_type >
template<typename F >
bool Row_bunch< Column_type >::process_columns_by_offset ( size_t  row_offset,
F &&  cbk 
)
inline

◆ read_column()

template<typename Column_type >
const Column_type & Row_bunch< Column_type >::read_column ( size_t  row_offset,
size_t  col_index 
) const
inline

Get constant column for reading using row offset and column index.

Parameters
[in]row_offsetrow offset in column vector
[in]col_indexindex of the column within row
Returns
column data

◆ set_num_rows()

template<typename Column_type >
bool Row_bunch< Column_type >::set_num_rows ( size_t  n_rows)
inline

Set the number of rows.

Adjust number of rows base on maximum column storage limit.

Parameters
[in,out]n_rowsnumber of rows
Returns
true if successful, false if too many rows or columns.

Member Data Documentation

◆ m_columns

template<typename Column_type >
std::vector<Column_type> Row_bunch< Column_type >::m_columns
private

All the columns.

◆ m_num_columns

template<typename Column_type >
size_t Row_bunch< Column_type >::m_num_columns {}
private

Number of columns in each row.

◆ m_num_rows

template<typename Column_type >
size_t Row_bunch< Column_type >::m_num_rows {}
private

Number of rows.

◆ S_MAX_TOTAL_COLS

template<typename Column_type >
const size_t Row_bunch< Column_type >::S_MAX_TOTAL_COLS = 600 * 1024 * 1024
static

Limit allocation up to 600M columns.

This number is rounded up from an estimate of the number of columns with the max chunk size (1024M). In the worst case we can have 2 bytes per column so a chunk can contain around 512M columns, and because of rows that spill over chunk boundaries we assume we can append a full additional row (which should have at most 4096 columns). Rounded up to 600M.


The documentation for this class was generated from the following file: