Package org.apache.cassandra.index.internal.composites

Class PartitionKeyIndex

  • All Implemented Interfaces:
    Index
    public class PartitionKeyIndex
extends CassandraIndex
    Index on a PARTITION_KEY column definition. This suppose a composite row key: rk = rk_0 ... rk_n The corresponding index entry will be: - index row key will be rk_i (where i == columnDef.componentIndex) - cell name will be: rk ck where rk is the fully partition key and ck the clustering keys of the original cell names (thus excluding the last column name as we want to refer to the whole CQL3 row, not just the cell itself) Note that contrarily to other type of index, we repeat the indexed value in the index cell name (we use the whole partition key). The reason is that we want to order the index cell name by partitioner first, and skipping a part of the row key would change the order.

    • Method Detail

      • getIndexedValue

        public java.nio.ByteBuffer getIndexedValue​(java.nio.ByteBuffer partitionKey,
                                           Clustering<?> clustering,
                                           CellPath path,
                                           java.nio.ByteBuffer cellValue)
        Description copied from class: CassandraIndex
        Extract the value to be inserted into the index from the components of the base data
        Specified by:
        getIndexedValue in class CassandraIndex
        Parameters:
        partitionKey - from the primary data
        clustering - from the primary data
        path - from the primary data
        cellValue - from the primary data
        Returns:
        a ByteBuffer containing the value to be inserted in the index. This will be used to make the partition key in the index table

      • buildIndexClusteringPrefix

        public <T> CBuilder buildIndexClusteringPrefix​(java.nio.ByteBuffer partitionKey,
                                               ClusteringPrefix<T> prefix,
                                               CellPath path)
        Description copied from class: CassandraIndex
        Used to construct an the clustering for an entry in the index table based on values from the base data. The clustering columns in the index table encode the values required to retrieve the correct data from the base table and varies depending on the kind of the indexed column. See indexCfsMetadata for more details Used whenever a row in the index table is written or deleted.
        Specified by:
        buildIndexClusteringPrefix in class CassandraIndex
        Parameters:
        partitionKey - from the base data being indexed
        prefix - from the base data being indexed
        path - from the base data being indexed
        Returns:
        a clustering prefix to be used to insert into the index table

      • decodeEntry

        public IndexEntry decodeEntry​(DecoratedKey indexedValue,
                              Row indexEntry)
        Description copied from class: CassandraIndex
        Used at search time to convert a row in the index table into a simple struct containing the values required to retrieve the corresponding row from the base table.
        Specified by:
        decodeEntry in class CassandraIndex
        Parameters:
        indexedValue - the partition key of the indexed table (i.e. the value that was indexed)
        indexEntry - a row from the index table
        Returns:

      • isStale

        public boolean isStale​(Row data,
                       java.nio.ByteBuffer indexValue,
                       long nowInSec)
        Description copied from class: CassandraIndex
        Check whether a value retrieved from an index is still valid by comparing it to current row from the base table. Used at read time to identify out of date index entries so that they can be excluded from search results and repaired
        Specified by:
        isStale in class CassandraIndex
        Parameters:
        data - the current row from the primary data table
        indexValue - the value we retrieved from the index
        Returns:
        true if the index is out of date and the entry should be dropped