Package org.apache.cassandra.db.rows

Interface Row

All Superinterfaces:
Clusterable, IMeasurableMemory, Iterable<ColumnData>, Unfiltered
All Known Implementing Classes:
AbstractRow, BTreeRow, RowWithSource
public interface Row extends Unfiltered, Iterable<ColumnData>, IMeasurableMemory
Storage engine representation of a row. A row mainly contains the following informations: 1) Its Clustering, which holds the values for the clustering columns identifying the row. 2) Its row level informations: the primary key liveness infos and the row deletion (see primaryKeyLivenessInfo() and deletion() for more details). 3) Data for the columns it contains, or in other words, it's a (sorted) collection of ColumnData. Also note that as for every other storage engine object, a Row object cannot shadow it's own data. For instance, a Row cannot contains a cell that is deleted by its own row deletion.

  • Method Details

    • clustering

      Clustering<?> clustering()
      The clustering values for this row.
      Specified by:
      clustering in interface Clusterable
      Specified by:
      clustering in interface Unfiltered

    • columns

      Collection<ColumnMetadata> columns()
      An in-natural-order collection of the columns for which data (incl. simple tombstones) is present in this row.

    • columnCount

      int columnCount()
      The number of columns for which data (incl. simple tombstones) is present in this row.

    • deletion

      Row.Deletion deletion()
      The row deletion. This correspond to the last row deletion done on this row.
      Returns:
      the row deletion.

    • primaryKeyLivenessInfo

      LivenessInfo primaryKeyLivenessInfo()
      Liveness information for the primary key columns of this row.

      As a row is uniquely identified by its primary key, all its primary key columns share the same LivenessInfo. This liveness information is what allows us to distinguish between a dead row (it has no live cells and its primary key liveness info is empty) and a live row but where all non PK columns are null (it has no live cells, but its primary key liveness is not empty). Please note that the liveness info (including it's eventually ttl/local deletion time) only apply to the primary key columns and has no impact on the row content.

      Note in particular that a row may have live cells but no PK liveness info, because the primary key liveness informations are only set on INSERT (which makes sense in itself, see #6782) but live cells can be added through UPDATE even if the row wasn't pre-existing (which users are encouraged not to do, but we can't validate).

    • isStatic

      boolean isStatic()
      Whether the row correspond to a static row or not.
      Returns:
      whether the row correspond to a static row or not.

    • isEmpty

      boolean isEmpty()
      Whether the row has no information whatsoever. This means no PK liveness info, no row deletion, no cells and no complex deletion info.
      Specified by:
      isEmpty in interface Unfiltered
      Returns:
      true if the row has no data, false otherwise.

    • hasLiveData

      boolean hasLiveData(long nowInSec, boolean enforceStrictLiveness)
      Whether the row has some live information (i.e. it's not just deletion informations).
      Parameters:
      nowInSec - the current time to decide what is deleted and what isn't
      enforceStrictLiveness - whether the row should be purged if there is no PK liveness info, normally retrieved from TableMetadata.enforceStrictLiveness()
      Returns:
      true if there is some live information

    • getCell

      Cell<?> getCell(ColumnMetadata c)
      Returns a cell for a simple column.
      Parameters:
      c - the simple column for which to fetch the cell.
      Returns:
      the corresponding cell or null if the row has no such cell.

    • getCell

      Cell<?> getCell(ColumnMetadata c, CellPath path)
      Return a cell for a given complex column and cell path.
      Parameters:
      c - the complex column for which to fetch the cell.
      path - the cell path for which to fetch the cell.
      Returns:
      the corresponding cell or null if the row has no such cell.

    • getComplexColumnData

      ComplexColumnData getComplexColumnData(ColumnMetadata c)
      The data for a complex column.

      The returned object groups all the cells for the column, as well as it's complex deletion (if relevant).

      Parameters:
      c - the complex column for which to return the complex data.
      Returns:
      the data for c or null if the row has no data for this column.

    • getColumnData

      ColumnData getColumnData(ColumnMetadata c)
      Returns the ColumnData for the specified column.
      Parameters:
      c - the column for which to fetch the data.
      Returns:
      the data for the column or null if the row has no data for this column.

    • cells

      Iterable<Cell<?>> cells()
      An iterable over the cells of this row.

      The iterable guarantees that cells are returned in order of Cell.comparator.

      Returns:
      an iterable over the cells of this row.

    • columnData

      Collection<ColumnData> columnData()
      A collection of the ColumnData representation of this row, for columns with some data (possibly not live) present

      The data is returned in column order.

      Returns:
      a Collection of the non-empty ColumnData for this row.

    • cellsInLegacyOrder

      Iterable<Cell<?>> cellsInLegacyOrder(TableMetadata metadata, boolean reversed)
      An iterable over the cells of this row that return cells in "legacy order".

      In 3.0+, columns are sorted so that all simple columns are before all complex columns. Previously however, the cells where just sorted by the column name. This iterator return cells in that legacy order. It's only ever meaningful for backward/thrift compatibility code.

      Parameters:
      metadata - the table this is a row of.
      reversed - if cells should returned in reverse order.
      Returns:
      an iterable over the cells of this row in "legacy order".

    • hasComplexDeletion

      boolean hasComplexDeletion()
      Whether the row stores any (non-live) complex deletion for any complex column.

    • hasComplex

      boolean hasComplex()
      Whether the row stores any (non-RT) data for any complex column.

    • hasDeletion

      boolean hasDeletion(long nowInSec)
      Whether the row has any deletion info (row deletion, cell tombstone, expired cell or complex deletion).
      Parameters:
      nowInSec - the current time in seconds to decid if a cell is expired.

    • searchIterator

      SearchIterator<ColumnMetadata,ColumnData> searchIterator()
      An iterator to efficiently search data for a given column.
      Returns:
      a search iterator for the cells of this row.

    • filter

      Row filter(ColumnFilter filter, TableMetadata metadata)
      Returns a copy of this row that: 1) only includes the data for the column included by filter. 2) doesn't include any data that belongs to a dropped column (recorded in metadata).

    • filter

      Row filter(ColumnFilter filter, DeletionTime activeDeletion, boolean setActiveDeletionToRow, TableMetadata metadata)
      Returns a copy of this row that: 1) only includes the data for the column included by filter. 2) doesn't include any data that belongs to a dropped column (recorded in metadata). 3) doesn't include any data that is shadowed/deleted by activeDeletion. 4) uses activeDeletion as row deletion iff setActiveDeletionToRow and activeDeletion supersedes the row deletion.

    • transformAndFilter

      Row transformAndFilter(LivenessInfo info, Row.Deletion deletion, Function<ColumnData,ColumnData> function)
      Requires that function returns either null or ColumnData for the same column. Returns a copy of this row that: 1) function has been applied to the members of 2) doesn't include any null results of function 3) has precisely the provided LivenessInfo and Deletion

    • transformAndFilter

      Row transformAndFilter(Function<ColumnData,ColumnData> function)
      Requires that function returns either null or ColumnData for the same column. Returns a copy of this row that: 1) function has been applied to the members of 2) doesn't include any null results of function

    • clone

      Row clone(Cloner cloner)

    • purge

      Row purge(DeletionPurger purger, long nowInSec, boolean enforceStrictLiveness)
      Returns a copy of this row without any deletion info that should be purged according to purger.
      Parameters:
      purger - the DeletionPurger to use to decide what can be purged.
      nowInSec - the current time to decide what is deleted and what isn't (in the case of expired cells).
      enforceStrictLiveness - whether the row should be purged if there is no PK liveness info, normally retrieved from TableMetadata.enforceStrictLiveness() When enforceStrictLiveness is set, rows with empty PK liveness info and no row deletion are purged. Currently this is only used by views with normal base column as PK column so updates to other base columns do not make the row live when the PK column is not live. See CASSANDRA-11500.
      Returns:
      this row but without any deletion info purged by purger. If the purged row is empty, returns null.

    • withOnlyQueriedData

      Row withOnlyQueriedData(ColumnFilter filter)
      Returns a copy of this row which only include the data queried by filter, excluding anything _fetched_ for internal reasons but not queried by the user (see ColumnFilter for details).
      Parameters:
      filter - the ColumnFilter to use when deciding what is user queried. This should be the filter that was used when querying the row on which this method is called.
      Returns:
      the row but with all data that wasn't queried by the user skipped.

    • purgeDataOlderThan

      Row purgeDataOlderThan(long timestamp, boolean enforceStrictLiveness)

    • markCounterLocalToBeCleared

      Row markCounterLocalToBeCleared()
      Returns a copy of this row where all counter cells have they "local" shard marked for clearing.

    • updateAllTimestamp

      Row updateAllTimestamp(long newTimestamp)
      Returns a copy of this row where all live timestamp have been replaced by newTimestamp and every deletion timestamp by newTimestamp - 1.
      Parameters:
      newTimestamp - the timestamp to use for all live data in the returned row.
      See Also:

    • withRowDeletion

      Row withRowDeletion(DeletionTime deletion)
      Returns a copy of this row with the new deletion as row deletion if it is more recent than the current row deletion.

      WARNING: this method does not check that nothing in the row is shadowed by the provided deletion and if that is the case, the created row will be invalid. It is thus up to the caller to verify that this is not the case and the only reasonable use case of this is probably when the row and the deletion comes from the same UnfilteredRowIterator since that gives use this guarantee.

    • dataSize

      int dataSize()

    • unsharedHeapSizeExcludingData

      long unsharedHeapSizeExcludingData()

    • toString

      String toString(TableMetadata metadata, boolean fullDetails)
      Specified by:
      toString in interface Unfiltered

    • unsharedHeapSize

      long unsharedHeapSize()
      Specified by:
      unsharedHeapSize in interface IMeasurableMemory
      Returns:
      the amount of on-heap memory retained by the object that might be reclaimed if the object were reclaimed, i.e. it should try to exclude globally cached data where possible, or counting portions of arrays that are referenced by the object but used by other objects only (e.g. slabbed byte-buffers), etc.

    • apply

      void apply(Consumer<ColumnData> function)
      Apply a function to every column in a row

    • apply

      <A> void apply(BiConsumer<A,ColumnData> function, A arg)
      Apply a function to every column in a row

    • accumulate

      long accumulate(LongAccumulator<ColumnData> accumulator, long initialValue)
      Apply an accumulation funtion to every column in a row

    • accumulate

      long accumulate(LongAccumulator<ColumnData> accumulator, Comparator<ColumnData> comparator, ColumnData from, long initialValue)

    • accumulate

      <A> long accumulate(BiLongAccumulator<A,ColumnData> accumulator, A arg, long initialValue)

    • accumulate

      <A> long accumulate(BiLongAccumulator<A,ColumnData> accumulator, A arg, Comparator<ColumnData> comparator, ColumnData from, long initialValue)