Field Detail

• GROUP_COLUMNS

  public static final String GROUP_COLUMNS

  JobConf parameter to specify the columns used to produce the key passed to collect from the map phase

  See Also:

  Constant Field Values

• columns

  protected byte[][] columns

Constructor Detail

• GroupingTableMap

  public GroupingTableMap()

Method Detail

• initJob

  public static void initJob(String table,
             String columns,
             String groupColumns,
             Class<? extends TableMap> mapper,
             org.apache.hadoop.mapred.JobConf job)

  Use this before submitting a TableMap job. It will appropriately set up the JobConf.

  Parameters:

  table - table to be processed

  columns - space separated list of columns to fetch

  groupColumns - space separated list of columns used to form the key used in collect

  mapper - map class

  job - job configuration object

• configure

  public void configure(org.apache.hadoop.mapred.JobConf job)

  Specified by:

  configure in interface org.apache.hadoop.mapred.JobConfigurable

  Overrides:

  configure in class org.apache.hadoop.mapred.MapReduceBase

• map

  public void map(ImmutableBytesWritable key,
         Result value,
         org.apache.hadoop.mapred.OutputCollector<ImmutableBytesWritable,Result> output,
         org.apache.hadoop.mapred.Reporter reporter)
           throws IOException

  Extract the grouping columns from value to construct a new key. Pass the new key and value to reduce. If any of the grouping columns are not found in the value, the record is skipped.

  Specified by:

  map in interface org.apache.hadoop.mapred.Mapper<ImmutableBytesWritable,Result,ImmutableBytesWritable,Result>

  Parameters:

  key -

  value -

  output -

  reporter -

  Throws:

  IOException

• extractKeyValues

  protected byte[][] extractKeyValues(Result r)

  Extract columns values from the current record. This method returns null if any of the columns are not found. Override this method if you want to deal with nulls differently.

  Parameters:

  r -

  Returns:

  array of byte values

• createGroupKey

  protected ImmutableBytesWritable createGroupKey(byte[][] vals)

  Create a key by concatenating multiple column values. Override this function in order to produce different types of keys.

  Parameters:

  vals -

  Returns:

  key generated by concatenating multiple column values