chunk

package
v1.1.0-beta.0...-cc83417 Latest Latest
Warning

This package is not in the latest version of its module.

Go to latest
Published: Dec 16, 2024 License: Apache-2.0 Imports: 36 Imported by: 0

Documentation

Index

Constants

View Source
const (
	InitialCapacity = 32
	ZeroCapacity    = 0
)

Capacity constants.

View Source
const RowPtrSize = int(unsafe.Sizeof(RowPtr{}))

RowPtrSize shows the size of RowPtr

View Source
const RowSize = int64(unsafe.Sizeof(Row{}))

RowSize is the size of `Row{}`

View Source
const SignalCheckpointForSort uint = 10240

SignalCheckpointForSort indicates the times of row comparation that a signal detection will be triggered.

View Source
const VarElemLen = -1

VarElemLen indicates this Column is a variable length Column.

Variables

View Source
var ErrCannotAddBecauseSorted = errors.New("can not add because sorted")

ErrCannotAddBecauseSorted indicate that the SortPartition is sorted and prohibit inserting data.

View Source
var MaxCachedLen = 16 * 1024

MaxCachedLen Maximum cacheable length

Functions

func AppendCellFromRawData

func AppendCellFromRawData(dst *Column, rowData unsafe.Pointer, currentOffset int) int

AppendCellFromRawData appends the cell from raw data

func Compare

func Compare(row Row, colIdx int, ad *types.Datum) int

Compare compares the value with ad. We assume that the collation information of the column is the same with the datum.

func CopyExpectedRowsWithRowIDFunc

func CopyExpectedRowsWithRowIDFunc(dstCol *Column, srcCol *Column, selected []bool, expectedResult bool, start int, end int, rowIDFunc func(int) int)

CopyExpectedRowsWithRowIDFunc copies the expected rows in srcCol to dstCol

func CopyRows

func CopyRows(dstCol *Column, srcCol *Column, selected []int)

CopyRows copies all rows in srcCol to dstCol

func CopySelectedJoinRowsDirect

func CopySelectedJoinRowsDirect(src *Chunk, selected []bool, dst *Chunk) (bool, error)

CopySelectedJoinRowsDirect directly copies the selected joined rows from the source Chunk to the destination Chunk. Return true if at least one joined row was selected.

func CopySelectedJoinRowsWithSameOuterRows

func CopySelectedJoinRowsWithSameOuterRows(src *Chunk, innerColOffset, innerColLen, outerColOffset, outerColLen int, selected []bool, dst *Chunk) (bool, error)

CopySelectedJoinRowsWithSameOuterRows copies the selected joined rows from the source Chunk to the destination Chunk. Return true if at least one joined row was selected.

NOTE: All the outer rows in the source Chunk should be the same.

func CopySelectedRows

func CopySelectedRows(dstCol *Column, srcCol *Column, selected []bool)

CopySelectedRows copies the selected rows in srcCol to dstCol

func CopySelectedRowsWithRowIDFunc

func CopySelectedRowsWithRowIDFunc(dstCol *Column, srcCol *Column, selected []bool, start int, end int, rowIDFunc func(int) int)

CopySelectedRowsWithRowIDFunc copies the selected rows in srcCol to dstCol

func EstimateTypeWidth

func EstimateTypeWidth(colType *types.FieldType) int

EstimateTypeWidth estimates the average width of values of the type. This is used by the planner, which doesn't require absolutely correct results; it's OK (and expected) to guess if we don't know for sure.

mostly study from https://github.com/postgres/postgres/blob/REL_12_STABLE/src/backend/utils/cache/lsyscache.c#L2356

func GetFixedLen

func GetFixedLen(colType *types.FieldType) int

GetFixedLen get the memory size of a fixed-length type. if colType is not fixed-length, it returns VarElemLen, aka -1.

func InitChunkAllocSize

func InitChunkAllocSize(setMaxFreeChunks, setMaxFreeColumns uint32)

InitChunkAllocSize init the maximum cache size

func NewAllocator

func NewAllocator() *allocator

NewAllocator creates an Allocator.

func NewIterator4RowContainer

func NewIterator4RowContainer(c *RowContainer) *iterator4RowContainer

NewIterator4RowContainer create a new iterator for RowContainer

func NewRowContainerReader

func NewRowContainerReader(rc *RowContainer) *rowContainerReader

NewRowContainerReader creates a forward only iterator for row container

Types

type Allocator

type Allocator interface {
	Alloc(fields []*types.FieldType, capacity, maxChunkSize int) *Chunk
	CheckReuseAllocSize() bool
	Reset()
}

Allocator is an interface defined to reduce object allocation. The typical usage is to call Reset() to recycle objects into a pool, and Alloc() allocates from the pool.

func NewEmptyAllocator

func NewEmptyAllocator() Allocator

NewEmptyAllocator creates an empty pool, which will always call `chunk.New` to create a new chunk

func NewReuseHookAllocator

func NewReuseHookAllocator(alloc Allocator, f func()) Allocator

NewReuseHookAllocator creates an allocator, which will call the function `f` when the first reused chunk is allocated.

func NewSyncAllocator

func NewSyncAllocator(alloc Allocator) Allocator

NewSyncAllocator creates the synchronized version of the `alloc`

type Chunk

type Chunk struct {
	// contains filtered or unexported fields
}

Chunk stores multiple rows of data in columns. Columns are in Apache Arrow format. See https://arrow.apache.org/docs/format/Columnar.html#physical-memory-layout. Apache Arrow is not used directly because we want to access MySQL types without decoding.

Values are appended in compact format and can be directly accessed without decoding. When the chunk is done processing, we can reuse the allocated memory by resetting it.

All Chunk's API should not do the validation work, and the user should ensure it is used correctly.

func New

func New(fields []*types.FieldType, capacity, maxChunkSize int) *Chunk

New creates a new chunk.

cap: the limit for the max number of rows.
maxChunkSize: the max limit for the number of rows.

func NewChunkFromPoolWithCapacity

func NewChunkFromPoolWithCapacity(fields []*types.FieldType, initCap int) *Chunk

NewChunkFromPoolWithCapacity creates a new chunk with field types and capacity from the pool.

func NewChunkWithCapacity

func NewChunkWithCapacity(fields []*types.FieldType, capacity int) *Chunk

NewChunkWithCapacity creates a new chunk with field types and capacity.

func NewEmptyChunk

func NewEmptyChunk(fields []*types.FieldType) *Chunk

NewEmptyChunk creates an empty chunk

func Renew

func Renew(chk *Chunk, maxChunkSize int) *Chunk

Renew creates a new Chunk based on an existing Chunk. The newly created Chunk has the same data schema with the old Chunk. The capacity of the new Chunk might be doubled based on the capacity of the old Chunk and the maxChunkSize.

chk: old chunk(often used in previous call).
maxChunkSize: the limit for the max number of rows.

func (*Chunk) Append

func (c *Chunk) Append(other *Chunk, begin, end int)

Append appends rows in [begin, end) in another Chunk to a Chunk.

func (*Chunk) AppendBytes

func (c *Chunk) AppendBytes(colIdx int, b []byte)

AppendBytes appends a bytes value to the chunk.

func (*Chunk) AppendDatum

func (c *Chunk) AppendDatum(colIdx int, d *types.Datum)

AppendDatum appends a datum into the chunk.

func (*Chunk) AppendDuration

func (c *Chunk) AppendDuration(colIdx int, dur types.Duration)

AppendDuration appends a Duration value to the chunk. Fsp is ignored.

func (*Chunk) AppendEnum

func (c *Chunk) AppendEnum(colIdx int, enum types.Enum)

AppendEnum appends an Enum value to the chunk.

func (*Chunk) AppendFloat32

func (c *Chunk) AppendFloat32(colIdx int, f float32)

AppendFloat32 appends a float32 value to the chunk.

func (*Chunk) AppendFloat64

func (c *Chunk) AppendFloat64(colIdx int, f float64)

AppendFloat64 appends a float64 value to the chunk.

func (*Chunk) AppendInt64

func (c *Chunk) AppendInt64(colIdx int, i int64)

AppendInt64 appends a int64 value to the chunk.

func (*Chunk) AppendJSON

func (c *Chunk) AppendJSON(colIdx int, j types.BinaryJSON)

AppendJSON appends a JSON value to the chunk.

func (*Chunk) AppendMyDecimal

func (c *Chunk) AppendMyDecimal(colIdx int, dec *types.MyDecimal)

AppendMyDecimal appends a MyDecimal value to the chunk.

func (*Chunk) AppendNull

func (c *Chunk) AppendNull(colIdx int)

AppendNull appends a null value to the chunk.

func (*Chunk) AppendPartialRow

func (c *Chunk) AppendPartialRow(colOff int, row Row)

AppendPartialRow appends a row to the chunk, starting from colOff.

func (*Chunk) AppendPartialRowByColIdxs

func (c *Chunk) AppendPartialRowByColIdxs(colOff int, row Row, colIdxs []int) (wide int)

AppendPartialRowByColIdxs appends a row to the chunk starting from colOff, using the row's columns specified by colIdxs. 1. every columns are used if colIdxs is nil. 2. no columns are used if colIdxs is not nil but the size of colIdxs is 0.

func (*Chunk) AppendPartialRows

func (c *Chunk) AppendPartialRows(colOff int, rows []Row)

AppendPartialRows appends multiple rows to the chunk.

func (*Chunk) AppendRow

func (c *Chunk) AppendRow(row Row)

AppendRow appends a row to the chunk.

func (*Chunk) AppendRowByColIdxs

func (c *Chunk) AppendRowByColIdxs(row Row, colIdxs []int) (wide int)

AppendRowByColIdxs appends a row to the chunk, using the row's columns specified by colIdxs. 1. every columns are used if colIdxs is nil. 2. no columns are used if colIdxs is not nil but the size of colIdxs is 0.

func (*Chunk) AppendRows

func (c *Chunk) AppendRows(rows []Row)

AppendRows appends multiple rows to the chunk.

func (*Chunk) AppendSet

func (c *Chunk) AppendSet(colIdx int, set types.Set)

AppendSet appends a Set value to the chunk.

func (*Chunk) AppendString

func (c *Chunk) AppendString(colIdx int, str string)

AppendString appends a string value to the chunk.

func (*Chunk) AppendTime

func (c *Chunk) AppendTime(colIdx int, t types.Time)

AppendTime appends a Time value to the chunk.

func (*Chunk) AppendUint64

func (c *Chunk) AppendUint64(colIdx int, u uint64)

AppendUint64 appends a uint64 value to the chunk.

func (*Chunk) AppendVectorFloat32

func (c *Chunk) AppendVectorFloat32(colIdx int, v types.VectorFloat32)

AppendVectorFloat32 appends a VectorFloat32 value to the chunk.

func (*Chunk) Capacity

func (c *Chunk) Capacity() int

Capacity returns the capacity of the Chunk.

func (*Chunk) CloneEmpty

func (c *Chunk) CloneEmpty(maxCapacity int) *Chunk

CloneEmpty returns an empty chunk that has the same schema with current chunk

func (*Chunk) Column

func (c *Chunk) Column(colIdx int) *Column

Column returns the specific column.

func (*Chunk) CopyConstruct

func (c *Chunk) CopyConstruct() *Chunk

CopyConstruct creates a new chunk and copies this chunk's data into it.

func (*Chunk) CopyConstructSel

func (c *Chunk) CopyConstructSel() *Chunk

CopyConstructSel is just like CopyConstruct, but ignore the rows that was not selected.

func (*Chunk) Destroy

func (c *Chunk) Destroy(initCap int, fields []*types.FieldType)

Destroy is to destroy the Chunk and put Chunk into the pool

func (*Chunk) GetNumVirtualRows

func (c *Chunk) GetNumVirtualRows() int

GetNumVirtualRows return c.numVirtualRows, used only in test

func (*Chunk) GetRow

func (c *Chunk) GetRow(idx int) Row

GetRow gets the Row in the chunk with the row index.

func (*Chunk) GrowAndReset

func (c *Chunk) GrowAndReset(maxChunkSize int)

GrowAndReset resets the Chunk and doubles the capacity of the Chunk. The doubled capacity should not be larger than maxChunkSize.

func (*Chunk) IsFull

func (c *Chunk) IsFull() bool

IsFull returns if this chunk is considered full.

func (*Chunk) IsInCompleteChunk

func (c *Chunk) IsInCompleteChunk() bool

IsInCompleteChunk returns true if this chunk is inCompleteChunk, used only in test

func (*Chunk) LowerBound

func (c *Chunk) LowerBound(colIdx int, d *types.Datum) (index int, match bool)

LowerBound searches on the non-decreasing Column colIdx, returns the smallest index i such that the value at row i is not less than `d`.

func (*Chunk) MakeRef

func (c *Chunk) MakeRef(srcColIdx, dstColIdx int)

MakeRef makes Column in "dstColIdx" reference to Column in "srcColIdx".

func (*Chunk) MakeRefTo

func (c *Chunk) MakeRefTo(dstColIdx int, src *Chunk, srcColIdx int) error

MakeRefTo copies columns `src.columns[srcColIdx]` to `c.columns[dstColIdx]`.

func (*Chunk) MemoryUsage

func (c *Chunk) MemoryUsage() (sum int64)

MemoryUsage returns the total memory usage of a Chunk in bytes. We ignore the size of Column.length and Column.nullCount since they have little effect of the total memory usage.

func (*Chunk) NumCols

func (c *Chunk) NumCols() int

NumCols returns the number of columns in the chunk.

func (*Chunk) NumRows

func (c *Chunk) NumRows() int

NumRows returns the number of rows in the chunk.

func (*Chunk) Prune

func (c *Chunk) Prune(usedColIdxs []int) *Chunk

Prune creates a new Chunk according to `c` and prunes the columns whose index is not in `usedColIdxs`

func (*Chunk) Reconstruct

func (c *Chunk) Reconstruct()

Reconstruct removes all filtered rows in this Chunk.

func (*Chunk) RequiredRows

func (c *Chunk) RequiredRows() int

RequiredRows returns how many rows is considered full.

func (*Chunk) Reset

func (c *Chunk) Reset()

Reset resets the chunk, so the memory it allocated can be reused. Make sure all the data in the chunk is not used anymore before you reuse this chunk.

func (*Chunk) Sel

func (c *Chunk) Sel() []int

Sel returns Sel of this Chunk.

func (*Chunk) SetCol

func (c *Chunk) SetCol(colIdx int, col *Column) *Column

SetCol sets the colIdx Column to col and returns the old Column.

func (*Chunk) SetInCompleteChunk

func (c *Chunk) SetInCompleteChunk(isInCompleteChunk bool)

SetInCompleteChunk will set c.inCompleteChunk, used in join

func (*Chunk) SetNumVirtualRows

func (c *Chunk) SetNumVirtualRows(numVirtualRows int)

SetNumVirtualRows sets the virtual row number for a Chunk. It should only be used when there exists no Column in the Chunk.

func (*Chunk) SetRequiredRows

func (c *Chunk) SetRequiredRows(requiredRows, maxChunkSize int) *Chunk

SetRequiredRows sets the number of required rows.

func (*Chunk) SetSel

func (c *Chunk) SetSel(sel []int)

SetSel sets a Sel for this Chunk.

func (*Chunk) SwapColumns

func (c *Chunk) SwapColumns(other *Chunk)

SwapColumns swaps columns with another Chunk.

func (*Chunk) ToString

func (c *Chunk) ToString(ft []*types.FieldType) string

ToString returns all the values in a chunk.

func (*Chunk) TruncateTo

func (c *Chunk) TruncateTo(numRows int)

TruncateTo truncates rows from tail to head in a Chunk to "numRows" rows.

func (*Chunk) UpperBound

func (c *Chunk) UpperBound(colIdx int, d *types.Datum) int

UpperBound searches on the non-decreasing Column colIdx, returns the smallest index i such that the value at row i is larger than `d`.

type Codec

type Codec struct {
	// contains filtered or unexported fields
}

Codec is used to: 1. encode a Chunk to a byte slice. 2. decode a Chunk from a byte slice.

func NewCodec

func NewCodec(colTypes []*types.FieldType) *Codec

NewCodec creates a new Codec object for encode or decode a Chunk.

func (*Codec) Decode

func (c *Codec) Decode(buffer []byte) (*Chunk, []byte)

Decode decodes a Chunk from a byte slice, return the remained unused bytes.

func (*Codec) DecodeToChunk

func (c *Codec) DecodeToChunk(buffer []byte, chk *Chunk) (remained []byte)

DecodeToChunk decodes a Chunk from a byte slice, return the remained unused bytes.

func (*Codec) Encode

func (c *Codec) Encode(chk *Chunk) []byte

Encode encodes a Chunk to a byte slice.

type Column

type Column struct {
	// contains filtered or unexported fields
}

Column stores one column of data in Apache Arrow format. See https://arrow.apache.org/docs/format/Columnar.html#format-columnar

func NewColumn

func NewColumn(ft *types.FieldType, capacity int) *Column

NewColumn creates a new column with the specific type and capacity.

func NewEmptyColumn

func NewEmptyColumn(ft *types.FieldType) *Column

NewEmptyColumn creates a new column with nothing.

func (*Column) AppendBytes

func (c *Column) AppendBytes(b []byte)

AppendBytes appends a byte slice into this Column.

func (*Column) AppendCellNTimes

func (c *Column) AppendCellNTimes(src *Column, pos, times int)

AppendCellNTimes append the pos-th Cell in source column to target column N times

func (*Column) AppendDuration

func (c *Column) AppendDuration(dur types.Duration)

AppendDuration appends a duration value into this Column. Fsp is ignored.

func (*Column) AppendEnum

func (c *Column) AppendEnum(enum types.Enum)

AppendEnum appends a Enum value into this Column.

func (*Column) AppendFloat32

func (c *Column) AppendFloat32(f float32)

AppendFloat32 appends a float32 value into this Column.

func (*Column) AppendFloat64

func (c *Column) AppendFloat64(f float64)

AppendFloat64 appends a float64 value into this Column.

func (*Column) AppendInt64

func (c *Column) AppendInt64(i int64)

AppendInt64 appends an int64 value into this Column.

func (*Column) AppendJSON

func (c *Column) AppendJSON(j types.BinaryJSON)

AppendJSON appends a BinaryJSON value into this Column.

func (*Column) AppendMyDecimal

func (c *Column) AppendMyDecimal(dec *types.MyDecimal)

AppendMyDecimal appends a MyDecimal value into this Column.

func (*Column) AppendNNulls

func (c *Column) AppendNNulls(n int)

AppendNNulls append n nulls to the column

func (*Column) AppendNull

func (c *Column) AppendNull()

AppendNull appends a null value into this Column.

func (*Column) AppendNullBitmap

func (c *Column) AppendNullBitmap(notNull bool)

AppendNullBitmap append a null/notnull value to the column's null map

func (*Column) AppendSet

func (c *Column) AppendSet(set types.Set)

AppendSet appends a Set value into this Column.

func (*Column) AppendString

func (c *Column) AppendString(str string)

AppendString appends a string value into this Column.

func (*Column) AppendTime

func (c *Column) AppendTime(t types.Time)

AppendTime appends a time value into this Column.

func (*Column) AppendUint64

func (c *Column) AppendUint64(u uint64)

AppendUint64 appends a uint64 value into this Column.

func (*Column) AppendVectorFloat32

func (c *Column) AppendVectorFloat32(v types.VectorFloat32)

AppendVectorFloat32 appends a VectorFloat32 value into this Column.

func (*Column) ContainsVeryLargeElement

func (c *Column) ContainsVeryLargeElement() bool

ContainsVeryLargeElement checks if the column contains element whose length is greater than math.MaxUint32.

func (*Column) CopyConstruct

func (c *Column) CopyConstruct(dst *Column) *Column

CopyConstruct copies this Column to dst. If dst is nil, it creates a new Column and returns it.

func (*Column) CopyReconstruct

func (c *Column) CopyReconstruct(sel []int, dst *Column) *Column

CopyReconstruct copies this Column to dst and removes unselected rows. If dst is nil, it creates a new Column and returns it.

func (*Column) Decimals

func (c *Column) Decimals() []types.MyDecimal

Decimals returns a MyDecimal slice stored in this Column.

func (*Column) DestroyDataForTest

func (c *Column) DestroyDataForTest()

DestroyDataForTest destroies data in the column so that we can test if data in column are deeply copied.

func (*Column) Float32s

func (c *Column) Float32s() []float32

Float32s returns a float32 slice stored in this Column.

func (*Column) Float64s

func (c *Column) Float64s() []float64

Float64s returns a float64 slice stored in this Column.

func (*Column) GetBytes

func (c *Column) GetBytes(rowID int) []byte

GetBytes returns the byte slice in the specific row.

func (*Column) GetDecimal

func (c *Column) GetDecimal(rowID int) *types.MyDecimal

GetDecimal returns the decimal in the specific row.

func (*Column) GetDuration

func (c *Column) GetDuration(rowID int, fillFsp int) types.Duration

GetDuration returns the Duration in the specific row.

func (*Column) GetEnum

func (c *Column) GetEnum(rowID int) types.Enum

GetEnum returns the Enum in the specific row.

func (*Column) GetFloat32

func (c *Column) GetFloat32(rowID int) float32

GetFloat32 returns the float32 in the specific row.

func (*Column) GetFloat64

func (c *Column) GetFloat64(rowID int) float64

GetFloat64 returns the float64 in the specific row.

func (*Column) GetInt64

func (c *Column) GetInt64(rowID int) int64

GetInt64 returns the int64 in the specific row.

func (*Column) GetJSON

func (c *Column) GetJSON(rowID int) types.BinaryJSON

GetJSON returns the JSON in the specific row.

func (*Column) GetRaw

func (c *Column) GetRaw(rowID int) []byte

GetRaw returns the underlying raw bytes in the specific row.

func (*Column) GetSet

func (c *Column) GetSet(rowID int) types.Set

GetSet returns the Set in the specific row.

func (*Column) GetString

func (c *Column) GetString(rowID int) string

GetString returns the string in the specific row.

func (*Column) GetTime

func (c *Column) GetTime(rowID int) types.Time

GetTime returns the Time in the specific row.

func (*Column) GetUint64

func (c *Column) GetUint64(rowID int) uint64

GetUint64 returns the uint64 in the specific row.

func (*Column) GetVectorFloat32

func (c *Column) GetVectorFloat32(rowID int) types.VectorFloat32

GetVectorFloat32 returns the VectorFloat32 in the specific row.

func (*Column) GoDurations

func (c *Column) GoDurations() []time.Duration

GoDurations returns a Golang time.Duration slice stored in this Column. Different from the Row.GetDuration method, the argument Fsp is ignored, so the user should handle it outside.

func (*Column) Int64s

func (c *Column) Int64s() []int64

Int64s returns an int64 slice stored in this Column.

func (*Column) IsNull

func (c *Column) IsNull(rowIdx int) bool

IsNull returns if this row is null.

func (*Column) MergeNulls

func (c *Column) MergeNulls(cols ...*Column)

MergeNulls merges these columns' null bitmaps. For a row, if any column of it is null, the result is null. It works like: if col1.IsNull || col2.IsNull || col3.IsNull. The caller should ensure that all these columns have the same length, and data stored in the result column is fixed-length type.

func (*Column) ReserveBytes

func (c *Column) ReserveBytes(n int)

ReserveBytes changes the column capacity to store n bytes elements and set the length to zero.

func (*Column) ReserveEnum

func (c *Column) ReserveEnum(n int)

ReserveEnum changes the column capacity to store n enum elements and set the length to zero.

func (*Column) ReserveJSON

func (c *Column) ReserveJSON(n int)

ReserveJSON changes the column capacity to store n JSON elements and set the length to zero.

func (*Column) ReserveSet

func (c *Column) ReserveSet(n int)

ReserveSet changes the column capacity to store n set elements and set the length to zero.

func (*Column) ReserveString

func (c *Column) ReserveString(n int)

ReserveString changes the column capacity to store n string elements and set the length to zero.

func (*Column) ReserveVectorFloat32

func (c *Column) ReserveVectorFloat32(n int)

ReserveVectorFloat32 changes the column capacity to store n vectorFloat32 elements and set the length to zero.

func (*Column) Reset

func (c *Column) Reset(eType types.EvalType)

Reset resets this Column according to the EvalType. Different from reset, Reset will reset the elemBuf.

func (*Column) ResizeDecimal

func (c *Column) ResizeDecimal(n int, isNull bool)

ResizeDecimal resizes the column so that it contains n decimal elements.

func (*Column) ResizeFloat32

func (c *Column) ResizeFloat32(n int, isNull bool)

ResizeFloat32 resizes the column so that it contains n float32 elements.

func (*Column) ResizeFloat64

func (c *Column) ResizeFloat64(n int, isNull bool)

ResizeFloat64 resizes the column so that it contains n float64 elements.

func (*Column) ResizeGoDuration

func (c *Column) ResizeGoDuration(n int, isNull bool)

ResizeGoDuration resizes the column so that it contains n duration elements.

func (*Column) ResizeInt64

func (c *Column) ResizeInt64(n int, isNull bool)

ResizeInt64 resizes the column so that it contains n int64 elements.

func (*Column) ResizeTime

func (c *Column) ResizeTime(n int, isNull bool)

ResizeTime resizes the column so that it contains n Time elements.

func (*Column) ResizeUint64

func (c *Column) ResizeUint64(n int, isNull bool)

ResizeUint64 resizes the column so that it contains n uint64 elements.

func (*Column) Rows

func (c *Column) Rows() int

Rows returns the row number in current column

func (*Column) SetNull

func (c *Column) SetNull(rowIdx int, isNull bool)

SetNull sets the rowIdx to null.

func (*Column) SetNulls

func (c *Column) SetNulls(begin, end int, isNull bool)

SetNulls sets rows in [begin, end) to null.

func (*Column) SetRaw

func (c *Column) SetRaw(rowID int, bs []byte)

SetRaw sets the raw bytes for the rowIdx-th element. NOTE: Two conditions must be satisfied before calling this function: 1. The column should be stored with variable-length elements. 2. The length of the new element should be exactly the same as the old one.

func (*Column) Times

func (c *Column) Times() []types.Time

Times returns a Time slice stored in this Column.

func (*Column) Uint64s

func (c *Column) Uint64s() []uint64

Uint64s returns a uint64 slice stored in this Column.

type ColumnAllocator

type ColumnAllocator interface {
	NewColumn(ft *types.FieldType, count int) *Column
}

ColumnAllocator defines an allocator for Column.

type ColumnSwapHelper

type ColumnSwapHelper struct {
	// InputIdxToOutputIdxes maps the input column index to the output column indexes.
	InputIdxToOutputIdxes map[int][]int
	// contains filtered or unexported fields
}

ColumnSwapHelper is used to help swap columns in a chunk.

func NewColumnSwapHelper

func NewColumnSwapHelper(usedColumnIndex []int) *ColumnSwapHelper

NewColumnSwapHelper creates a new ColumnSwapHelper.

func (*ColumnSwapHelper) SwapColumns

func (helper *ColumnSwapHelper) SwapColumns(input, output *Chunk) error

SwapColumns evaluates "Column" expressions. it will change the content of the input Chunk.

type CompareFunc

type CompareFunc = func(l Row, lCol int, r Row, rCol int) int

CompareFunc is a function to compare the two values in Row, the two columns must have the same type.

func GetCompareFunc

func GetCompareFunc(tp *types.FieldType) CompareFunc

GetCompareFunc gets a compare function for the field type.

type DataInDiskByChunks

type DataInDiskByChunks struct {
	// contains filtered or unexported fields
}

DataInDiskByChunks represents some data stored in temporary disk. They can only be restored by chunks.

func NewDataInDiskByChunks

func NewDataInDiskByChunks(fieldTypes []*types.FieldType) *DataInDiskByChunks

NewDataInDiskByChunks creates a new DataInDiskByChunks with field types.

func (*DataInDiskByChunks) Add

func (d *DataInDiskByChunks) Add(chk *Chunk) (err error)

Add adds a chunk to the DataInDiskByChunks. Caller must make sure the input chk has the same field types. Warning: Do not concurrently call this function.

func (*DataInDiskByChunks) Close

func (d *DataInDiskByChunks) Close()

Close releases the disk resource.

func (*DataInDiskByChunks) FillChunk

func (d *DataInDiskByChunks) FillChunk(srcChkIdx int, destChk *Chunk) error

FillChunk fills a Chunk from the DataInDiskByChunks by chkIdx.

func (*DataInDiskByChunks) GetChunk

func (d *DataInDiskByChunks) GetChunk(chkIdx int) (*Chunk, error)

GetChunk gets a Chunk from the DataInDiskByChunks by chkIdx.

func (*DataInDiskByChunks) GetDiskTracker

func (d *DataInDiskByChunks) GetDiskTracker() *disk.Tracker

GetDiskTracker returns the memory tracker of this List.

func (*DataInDiskByChunks) NumChunks

func (d *DataInDiskByChunks) NumChunks() int

NumChunks returns total spilled chunk number

func (*DataInDiskByChunks) NumRows

func (d *DataInDiskByChunks) NumRows() int64

NumRows returns total spilled row number

type DataInDiskByRows

type DataInDiskByRows struct {
	// contains filtered or unexported fields
}

DataInDiskByRows represents some data stored in temporary disk. These data are stored in row format, so they can only be restored by rows.

func NewDataInDiskByRows

func NewDataInDiskByRows(fieldTypes []*types.FieldType) *DataInDiskByRows

NewDataInDiskByRows creates a new DataInDiskByRows with field types.

func (*DataInDiskByRows) Add

func (l *DataInDiskByRows) Add(chk *Chunk) (err error)

Add adds a chunk to the DataInDiskByRows. Caller must make sure the input chk is not empty and not used any more and has the same field types. Warning: Do not use Add concurrently.

func (*DataInDiskByRows) Close

func (l *DataInDiskByRows) Close() error

Close releases the disk resource.

func (*DataInDiskByRows) GetChunk

func (l *DataInDiskByRows) GetChunk(chkIdx int) (*Chunk, error)

GetChunk gets a Chunk from the DataInDiskByRows by chkIdx.

func (*DataInDiskByRows) GetDiskTracker

func (l *DataInDiskByRows) GetDiskTracker() *disk.Tracker

GetDiskTracker returns the memory tracker of this List.

func (*DataInDiskByRows) GetRow

func (l *DataInDiskByRows) GetRow(ptr RowPtr) (row Row, err error)

GetRow gets a Row from the DataInDiskByRows by RowPtr.

func (*DataInDiskByRows) GetRowAndAppendToChunk

func (l *DataInDiskByRows) GetRowAndAppendToChunk(ptr RowPtr, chk *Chunk) (row Row, _ *Chunk, err error)

GetRowAndAppendToChunk gets a Row from the DataInDiskByRows by RowPtr. Return the Row and the Ref Chunk.

func (*DataInDiskByRows) Len

func (l *DataInDiskByRows) Len() int

Len returns the number of rows in DataInDiskByRows

func (*DataInDiskByRows) NumChunks

func (l *DataInDiskByRows) NumChunks() int

NumChunks returns the number of chunks in the DataInDiskByRows.

func (*DataInDiskByRows) NumRowsOfChunk

func (l *DataInDiskByRows) NumRowsOfChunk(chkID int) int

NumRowsOfChunk returns the number of rows of a chunk in the DataInDiskByRows.

type Decoder

type Decoder struct {
	// contains filtered or unexported fields
}

Decoder decodes the data returned from the coprocessor and stores the result in Chunk. How Decoder works:

  1. Initialization phase: Decode a whole input byte slice to Decoder.intermChk(intermediate chunk) using Codec.Decode. intermChk is introduced to simplify the implementation of decode phase. This phase uses pointer operations with less CPU and memory cost.
  2. Decode phase: 2.1 Set the number of rows to be decoded to a value that is a multiple of 8 and greater than `chk.RequiredRows() - chk.NumRows()`. This reduces the overhead of copying the srcCol.nullBitMap into destCol.nullBitMap. 2.2 Append srcCol.offsets to destCol.offsets when the elements is of var-length type. And further adjust the offsets according to descCol.offsets[destCol.length]-srcCol.offsets[0]. 2.3 Append srcCol.nullBitMap to destCol.nullBitMap.
  3. Go to step 1 when the input byte slice is consumed.

func NewDecoder

func NewDecoder(chk *Chunk, colTypes []*types.FieldType) *Decoder

NewDecoder creates a new Decoder object for decode a Chunk.

func (*Decoder) Decode

func (c *Decoder) Decode(chk *Chunk)

Decode decodes multiple rows of Decoder.intermChk and stores the result in chk.

func (*Decoder) IsFinished

func (c *Decoder) IsFinished() bool

IsFinished indicates whether Decoder.intermChk has been dried up.

func (*Decoder) RemainedRows

func (c *Decoder) RemainedRows() int

RemainedRows indicates Decoder.intermChk has remained rows.

func (*Decoder) Reset

func (c *Decoder) Reset(data []byte)

Reset decodes data and store the result in Decoder.intermChk. This decode phase uses pointer operations with less CPU and memory costs.

func (*Decoder) ReuseIntermChk

func (c *Decoder) ReuseIntermChk(chk *Chunk)

ReuseIntermChk swaps `Decoder.intermChk` with `chk` directly when `Decoder.intermChk.NumRows()` is no less than `chk.requiredRows * factor` where `factor` is 0.8 now. This can avoid the overhead of appending the data from `Decoder.intermChk` to `chk`. Moreover, the column.offsets needs to be further adjusted according to column.offset[0].

type DefaultColumnAllocator

type DefaultColumnAllocator struct{}

DefaultColumnAllocator is the default implementation of ColumnAllocator.

func (DefaultColumnAllocator) NewColumn

func (DefaultColumnAllocator) NewColumn(ft *types.FieldType, capacity int) *Column

NewColumn implements the ColumnAllocator interface.

type Iterator

type Iterator interface {
	// Begin resets the cursor of the iterator and returns the first Row.
	Begin() Row

	// Next returns the next Row.
	Next() Row

	// End returns the invalid end Row.
	End() Row

	// Len returns the length.
	Len() int

	// Current returns the current Row.
	Current() Row

	// ReachEnd reaches the end of iterator.
	ReachEnd()

	// Error returns none-nil error if anything wrong happens during the iteration.
	Error() error
}

Iterator is used to iterate a number of rows.

for row := it.Begin(); row != it.End(); row = it.Next() {
    ...
}

func NewIterator4List

func NewIterator4List(li *List) Iterator

NewIterator4List returns a Iterator for List.

func NewIterator4RowPtr

func NewIterator4RowPtr(li *List, ptrs []RowPtr) Iterator

NewIterator4RowPtr returns a Iterator for RowPtrs.

func NewMultiIterator

func NewMultiIterator(iters ...Iterator) Iterator

NewMultiIterator creates a new multiIterator

type Iterator4Chunk

type Iterator4Chunk struct {
	// contains filtered or unexported fields
}

Iterator4Chunk is used to iterate rows inside a chunk.

func NewIterator4Chunk

func NewIterator4Chunk(chk *Chunk) *Iterator4Chunk

NewIterator4Chunk returns a iterator for Chunk.

func (*Iterator4Chunk) Begin

func (it *Iterator4Chunk) Begin() Row

Begin implements the Iterator interface.

func (*Iterator4Chunk) Current

func (it *Iterator4Chunk) Current() Row

Current implements the Iterator interface.

func (*Iterator4Chunk) End

func (*Iterator4Chunk) End() Row

End implements the Iterator interface.

func (*Iterator4Chunk) Error

func (*Iterator4Chunk) Error() error

Error returns none-nil error if anything wrong happens during the iteration.

func (*Iterator4Chunk) GetChunk

func (it *Iterator4Chunk) GetChunk() *Chunk

GetChunk returns the chunk stored in the Iterator4Chunk

func (*Iterator4Chunk) Len

func (it *Iterator4Chunk) Len() int

Len implements the Iterator interface

func (*Iterator4Chunk) Next

func (it *Iterator4Chunk) Next() Row

Next implements the Iterator interface.

func (*Iterator4Chunk) ReachEnd

func (it *Iterator4Chunk) ReachEnd()

ReachEnd implements the Iterator interface.

func (*Iterator4Chunk) ResetChunk

func (it *Iterator4Chunk) ResetChunk(chk *Chunk)

ResetChunk reset the chunk this iter holds

type Iterator4Slice

type Iterator4Slice struct {
	// contains filtered or unexported fields
}

Iterator4Slice is used to iterate rows inside a slice.

func NewIterator4Slice

func NewIterator4Slice(rows []Row) *Iterator4Slice

NewIterator4Slice returns a *Iterator4Slice for Row slice.

func (*Iterator4Slice) Begin

func (it *Iterator4Slice) Begin() Row

Begin implements the Iterator interface.

func (*Iterator4Slice) Current

func (it *Iterator4Slice) Current() Row

Current implements the Iterator interface.

func (*Iterator4Slice) End

func (*Iterator4Slice) End() Row

End implements the Iterator interface.

func (*Iterator4Slice) Error

func (*Iterator4Slice) Error() error

Error returns none-nil error if anything wrong happens during the iteration.

func (*Iterator4Slice) Len

func (it *Iterator4Slice) Len() int

Len implements the Iterator interface.

func (*Iterator4Slice) Next

func (it *Iterator4Slice) Next() Row

Next implements the Iterator interface.

func (*Iterator4Slice) ReachEnd

func (it *Iterator4Slice) ReachEnd()

ReachEnd implements the Iterator interface.

func (*Iterator4Slice) Reset

func (it *Iterator4Slice) Reset(rows []Row)

Reset iterator.rows and cursor.

type List

type List struct {
	// contains filtered or unexported fields
}

List holds a slice of chunks, use to append rows with max chunk size properly handled.

func NewList

func NewList(fieldTypes []*types.FieldType, initChunkSize, maxChunkSize int) *List

NewList creates a new List with field types, init chunk size and max chunk size.

func NewListWithMemTracker

func NewListWithMemTracker(fieldTypes []*types.FieldType, initChunkSize, maxChunkSize int, tracker *memory.Tracker) *List

NewListWithMemTracker creates a new List with field types, init chunk size, max chunk size and memory tracker.

func (*List) Add

func (l *List) Add(chk *Chunk)

Add adds a chunk to the List, the chunk may be modified later by the list. Caller must make sure the input chk is not empty and not used any more and has the same field types.

func (*List) AppendRow

func (l *List) AppendRow(row Row) RowPtr

AppendRow appends a row to the List, the row is copied to the List.

func (*List) Clear

func (l *List) Clear()

Clear triggers GC for all the allocated chunks and reset the list

func (*List) FieldTypes

func (l *List) FieldTypes() []*types.FieldType

FieldTypes returns the fieldTypes of the list

func (*List) GetChunk

func (l *List) GetChunk(chkIdx int) *Chunk

GetChunk gets the Chunk by ChkIdx.

func (*List) GetMemTracker

func (l *List) GetMemTracker() *memory.Tracker

GetMemTracker returns the memory tracker of this List.

func (*List) GetRow

func (l *List) GetRow(ptr RowPtr) Row

GetRow gets a Row from the list by RowPtr.

func (*List) Len

func (l *List) Len() int

Len returns the length of the List.

func (*List) NumChunks

func (l *List) NumChunks() int

NumChunks returns the number of chunks in the List.

func (*List) NumRowsOfChunk

func (l *List) NumRowsOfChunk(chkID int) int

NumRowsOfChunk returns the number of rows of a chunk in the DataInDiskByRows.

func (*List) Reset

func (l *List) Reset()

Reset resets the List.

func (*List) Walk

func (l *List) Walk(walkFunc ListWalkFunc) error

Walk iterate the list and call walkFunc for each row.

type ListWalkFunc

type ListWalkFunc = func(row Row) error

ListWalkFunc is used to walk the list. If error is returned, it will stop walking.

type MutRow

type MutRow Row

MutRow represents a mutable Row. The underlying columns only contains one row and not exposed to the user.

func MutRowFromDatums

func MutRowFromDatums(datums []types.Datum) MutRow

MutRowFromDatums creates a MutRow from a datum slice.

func MutRowFromTypes

func MutRowFromTypes(types []*types.FieldType) MutRow

MutRowFromTypes creates a MutRow from a FieldType slice, each Column is initialized to zero value.

func MutRowFromValues

func MutRowFromValues(vals ...any) MutRow

MutRowFromValues creates a MutRow from a interface slice.

func (MutRow) Clone

func (mr MutRow) Clone() MutRow

Clone deep clone a MutRow.

func (MutRow) Len

func (mr MutRow) Len() int

Len returns the number of columns.

func (MutRow) SetDatum

func (mr MutRow) SetDatum(colIdx int, d types.Datum)

SetDatum sets the MutRow with colIdx and datum.

func (MutRow) SetDatums

func (mr MutRow) SetDatums(datums ...types.Datum)

SetDatums sets the MutRow with datum slice.

func (MutRow) SetRow

func (mr MutRow) SetRow(row Row)

SetRow sets the MutRow with Row.

func (MutRow) SetValue

func (mr MutRow) SetValue(colIdx int, val any)

SetValue sets the MutRow with colIdx and value.

func (MutRow) SetValues

func (mr MutRow) SetValues(vals ...any)

SetValues sets the MutRow with values.

func (MutRow) ShallowCopyPartialRow

func (mr MutRow) ShallowCopyPartialRow(colIdx int, row Row)

ShallowCopyPartialRow shallow copies the data of `row` to MutRow.

func (MutRow) ToRow

func (mr MutRow) ToRow() Row

ToRow converts the MutRow to Row, so it can be used to read data.

type Pool

type Pool struct {
	// contains filtered or unexported fields
}

Pool is the Column pool. NOTE: Pool is non-copyable.

func NewPool

func NewPool(initCap int) *Pool

NewPool creates a new Pool.

func (*Pool) GetChunk

func (p *Pool) GetChunk(fields []*types.FieldType) *Chunk

GetChunk gets a Chunk from the Pool.

func (*Pool) PutChunk

func (p *Pool) PutChunk(fields []*types.FieldType, chk *Chunk)

PutChunk puts a Chunk back to the Pool.

type ReaderWithCache

type ReaderWithCache struct {
	// contains filtered or unexported fields
}

ReaderWithCache helps to read data that has not be flushed to underlying layer. By using ReaderWithCache, user can still write data into DataInDiskByRows even after reading.

func NewReaderWithCache

func NewReaderWithCache(r io.ReaderAt, cache []byte, cacheOff int64) *ReaderWithCache

NewReaderWithCache returns a ReaderWithCache.

func (*ReaderWithCache) ReadAt

func (r *ReaderWithCache) ReadAt(p []byte, off int64) (readCnt int, err error)

ReadAt implements the ReadAt interface.

type Row

type Row struct {
	// contains filtered or unexported fields
}

Row represents a row of data, can be used to access values.

func (Row) Chunk

func (r Row) Chunk() *Chunk

Chunk returns the Chunk which the row belongs to.

func (Row) CopyConstruct

func (r Row) CopyConstruct() Row

CopyConstruct creates a new row and copies this row's data into it.

func (Row) DatumWithBuffer

func (r Row) DatumWithBuffer(colIdx int, tp *types.FieldType, d *types.Datum)

DatumWithBuffer gets datum using the buffer d.

func (Row) GetBytes

func (r Row) GetBytes(colIdx int) []byte

GetBytes returns the bytes value with the colIdx.

func (Row) GetDatum

func (r Row) GetDatum(colIdx int, tp *types.FieldType) types.Datum

GetDatum implements the chunk.Row interface.

func (Row) GetDatumRow

func (r Row) GetDatumRow(fields []*types.FieldType) []types.Datum

GetDatumRow converts chunk.Row to types.DatumRow. Keep in mind that GetDatumRow has a reference to r.c, which is a chunk, this function works only if the underlying chunk is valid or unchanged.

func (Row) GetDatumRowWithBuffer

func (r Row) GetDatumRowWithBuffer(fields []*types.FieldType, datumRow []types.Datum) []types.Datum

GetDatumRowWithBuffer gets datum using the buffer datumRow.

func (Row) GetDuration

func (r Row) GetDuration(colIdx int, fillFsp int) types.Duration

GetDuration returns the Duration value with the colIdx.

func (Row) GetEnum

func (r Row) GetEnum(colIdx int) types.Enum

GetEnum returns the Enum value with the colIdx.

func (Row) GetFloat32

func (r Row) GetFloat32(colIdx int) float32

GetFloat32 returns the float32 value with the colIdx.

func (Row) GetFloat64

func (r Row) GetFloat64(colIdx int) float64

GetFloat64 returns the float64 value with the colIdx.

func (Row) GetInt64

func (r Row) GetInt64(colIdx int) int64

GetInt64 returns the int64 value with the colIdx.

func (Row) GetJSON

func (r Row) GetJSON(colIdx int) types.BinaryJSON

GetJSON returns the JSON value with the colIdx.

func (Row) GetMyDecimal

func (r Row) GetMyDecimal(colIdx int) *types.MyDecimal

GetMyDecimal returns the MyDecimal value with the colIdx.

func (Row) GetRaw

func (r Row) GetRaw(colIdx int) []byte

GetRaw returns the underlying raw bytes with the colIdx.

func (Row) GetSet

func (r Row) GetSet(colIdx int) types.Set

GetSet returns the Set value with the colIdx.

func (Row) GetString

func (r Row) GetString(colIdx int) string

GetString returns the string value with the colIdx.

func (Row) GetTime

func (r Row) GetTime(colIdx int) types.Time

GetTime returns the Time value with the colIdx.

func (Row) GetUint64

func (r Row) GetUint64(colIdx int) uint64

GetUint64 returns the uint64 value with the colIdx.

func (Row) GetVectorFloat32

func (r Row) GetVectorFloat32(colIdx int) types.VectorFloat32

GetVectorFloat32 returns the VectorFloat32 value with the colIdx.

func (Row) Idx

func (r Row) Idx() int

Idx returns the row index of Chunk.

func (Row) IsEmpty

func (r Row) IsEmpty() bool

IsEmpty returns true if the Row is empty.

func (Row) IsNull

func (r Row) IsNull(colIdx int) bool

IsNull returns if the datum in the chunk.Row is null.

func (Row) Len

func (r Row) Len() int

Len returns the number of values in the row.

func (Row) ToString

func (r Row) ToString(ft []*types.FieldType) string

ToString returns all the values in a row.

type RowContainer

type RowContainer struct {
	// contains filtered or unexported fields
}

RowContainer provides a place for many rows, so many that we might want to spill them into disk. nolint:structcheck

func NewRowContainer

func NewRowContainer(fieldType []*types.FieldType, chunkSize int) *RowContainer

NewRowContainer creates a new RowContainer in memory.

func (*RowContainer) ActionSpill

func (c *RowContainer) ActionSpill() *SpillDiskAction

ActionSpill returns a SpillDiskAction for spilling over to disk.

func (*RowContainer) ActionSpillForTest

func (c *RowContainer) ActionSpillForTest() *SpillDiskAction

ActionSpillForTest returns a SpillDiskAction for spilling over to disk for test.

func (*RowContainer) Add

func (c *RowContainer) Add(chk *Chunk) (err error)

Add appends a chunk into the RowContainer.

func (*RowContainer) AllocChunk

func (c *RowContainer) AllocChunk() (chk *Chunk)

AllocChunk allocates a new chunk from RowContainer.

func (*RowContainer) AlreadySpilledSafeForTest

func (c *RowContainer) AlreadySpilledSafeForTest() bool

AlreadySpilledSafeForTest indicates that records have spilled out into disk. It's thread-safe. The function is only used for test.

func (*RowContainer) Close

func (c *RowContainer) Close() (err error)

Close close the RowContainer

func (*RowContainer) GetChunk

func (c *RowContainer) GetChunk(chkIdx int) (*Chunk, error)

GetChunk returns chkIdx th chunk of in memory records.

func (*RowContainer) GetDiskTracker

func (c *RowContainer) GetDiskTracker() *disk.Tracker

GetDiskTracker returns the underlying disk usage tracker in recordsInDisk.

func (*RowContainer) GetMemTracker

func (c *RowContainer) GetMemTracker() *memory.Tracker

GetMemTracker returns the memory tracker in records, panics if the RowContainer has already spilled.

func (*RowContainer) GetRow

func (c *RowContainer) GetRow(ptr RowPtr) (row Row, err error)

GetRow returns the row the ptr pointed to.

func (*RowContainer) GetRowAndAlwaysAppendToChunk

func (c *RowContainer) GetRowAndAlwaysAppendToChunk(ptr RowPtr, chk *Chunk) (row Row, _ *Chunk, err error)

GetRowAndAlwaysAppendToChunk gets a Row from the RowContainer by RowPtr. Unlike `GetRowAndAppendToChunkIfInDisk`, this function always appends the row to the chunk, without considering whether it has spilled. It'll return `nil` chunk if it returns an error, or the chunk will be the same with the argument.

func (*RowContainer) GetRowAndAppendToChunkIfInDisk

func (c *RowContainer) GetRowAndAppendToChunkIfInDisk(ptr RowPtr, chk *Chunk) (row Row, _ *Chunk, err error)

GetRowAndAppendToChunkIfInDisk gets a Row from the RowContainer by RowPtr. If the container has spilled, the row will be appended to the chunk. It'll return `nil` chunk if the container hasn't spilled, or it returns an error.

func (*RowContainer) NumChunks

func (c *RowContainer) NumChunks() int

NumChunks returns the number of chunks in the container.

func (*RowContainer) NumRow

func (c *RowContainer) NumRow() int

NumRow returns the number of rows in the container

func (*RowContainer) NumRowsOfChunk

func (c *RowContainer) NumRowsOfChunk(chkID int) int

NumRowsOfChunk returns the number of rows of a chunk in the DataInDiskByRows.

func (*RowContainer) Reset

func (c *RowContainer) Reset() error

Reset resets RowContainer.

func (*RowContainer) ShallowCopyWithNewMutex

func (c *RowContainer) ShallowCopyWithNewMutex() *RowContainer

ShallowCopyWithNewMutex shallow clones a RowContainer. The new RowContainer shares the same underlying data with the old one but holds an individual rLock.

func (*RowContainer) SpillToDisk

func (c *RowContainer) SpillToDisk()

SpillToDisk spills data to disk. This function may be called in parallel.

type RowContainerReader

type RowContainerReader interface {
	// Next returns the next Row.
	Next() Row

	// Current returns the current Row.
	Current() Row

	// End returns the invalid end Row.
	End() Row

	// Error returns none-nil error if anything wrong happens during the iteration.
	Error() error

	// Close closes the dumper
	Close()
}

RowContainerReader is a forward-only iterator for the row container. It provides an interface similar to other iterators, but it doesn't provide `ReachEnd` function and requires manually closing to release goroutine.

It's recommended to use the following pattern to use it:

for iter := NewRowContainerReader(rc); iter.Current() != iter.End(); iter.Next() {
	...
}
iter.Close()
if iter.Error() != nil {
}

type RowPtr

type RowPtr struct {
	ChkIdx uint32
	RowIdx uint32
}

RowPtr is used to get a row from a list. It is only valid for the list that returns it.

type SortAndSpillDiskAction

type SortAndSpillDiskAction struct {
	// contains filtered or unexported fields
}

SortAndSpillDiskAction implements memory.ActionOnExceed for chunk.List. If the memory quota of a query is exceeded, SortAndSpillDiskAction.Action is triggered.

func (*SortAndSpillDiskAction) Action

func (a *SortAndSpillDiskAction) Action(t *memory.Tracker)

Action sends a signal to trigger sortAndSpillToDisk method of RowContainer and if it is already triggered before, call its fallbackAction.

func (SortAndSpillDiskAction) GetPriority

func (SortAndSpillDiskAction) GetPriority() int64

GetPriority get the priority of the Action.

func (SortAndSpillDiskAction) Reset

func (a SortAndSpillDiskAction) Reset()

Reset resets the status for SpillDiskAction.

func (*SortAndSpillDiskAction) WaitForTest

func (a *SortAndSpillDiskAction) WaitForTest()

WaitForTest waits all goroutine have gone.

type SortedRowContainer

type SortedRowContainer struct {
	*RowContainer

	ByItemsDesc []bool
	// contains filtered or unexported fields
}

SortedRowContainer provides a place for many rows, so many that we might want to sort and spill them into disk.

func NewSortedRowContainer

func NewSortedRowContainer(fieldType []*types.FieldType, chunkSize int, byItemsDesc []bool,
	keyColumns []int, keyCmpFuncs []CompareFunc) *SortedRowContainer

NewSortedRowContainer creates a new SortedRowContainer in memory.

func (*SortedRowContainer) ActionSpill

func (c *SortedRowContainer) ActionSpill() *SortAndSpillDiskAction

ActionSpill returns a SortAndSpillDiskAction for sorting and spilling over to disk.

func (*SortedRowContainer) ActionSpillForTest

func (c *SortedRowContainer) ActionSpillForTest() *SortAndSpillDiskAction

ActionSpillForTest returns a SortAndSpillDiskAction for sorting and spilling over to disk for test.

func (*SortedRowContainer) Add

func (c *SortedRowContainer) Add(chk *Chunk) (err error)

Add appends a chunk into the SortedRowContainer.

func (*SortedRowContainer) Close

func (c *SortedRowContainer) Close() error

Close close the SortedRowContainer

func (*SortedRowContainer) GetMemTracker

func (c *SortedRowContainer) GetMemTracker() *memory.Tracker

GetMemTracker return the memory tracker for the sortedRowContainer

func (*SortedRowContainer) GetSortedRow

func (c *SortedRowContainer) GetSortedRow(idx int) (Row, error)

GetSortedRow returns the row the idx pointed to.

func (*SortedRowContainer) GetSortedRowAndAlwaysAppendToChunk

func (c *SortedRowContainer) GetSortedRowAndAlwaysAppendToChunk(idx int, chk *Chunk) (Row, *Chunk, error)

GetSortedRowAndAlwaysAppendToChunk returns the row the idx pointed to.

func (*SortedRowContainer) Sort

func (c *SortedRowContainer) Sort() (ret error)

Sort inits pointers and sorts the records.

func (*SortedRowContainer) SpillToDisk

func (c *SortedRowContainer) SpillToDisk()

SpillToDisk spills data to disk. This function may be called in parallel.

type SpillDiskAction

type SpillDiskAction struct {
	// contains filtered or unexported fields
}

SpillDiskAction implements memory.ActionOnExceed for chunk.List. If the memory quota of a query is exceeded, SpillDiskAction.Action is triggered.

func (*SpillDiskAction) Action

func (a *SpillDiskAction) Action(t *memory.Tracker)

Action sends a signal to trigger spillToDisk method of RowContainer and if it is already triggered before, call its fallbackAction.

func (SpillDiskAction) GetPriority

func (SpillDiskAction) GetPriority() int64

GetPriority get the priority of the Action.

func (SpillDiskAction) Reset

func (a SpillDiskAction) Reset()

Reset resets the status for SpillDiskAction.

func (SpillDiskAction) WaitForTest

func (a SpillDiskAction) WaitForTest()

WaitForTest waits all goroutine have gone.

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL