You can specify map types as colname: map[] in your load statement. E.g.: a = load 'mydata' as (x: int, m: map[]);
Note that Pig does not enforce types for the keys and values in the map. The only constraint is that keys should be basic types (int, long, float, double, string). Internally, the value types are treated as bytearray allowing the use of any Pig type as the value type. This allows you to have any type (int, long, float, double, string, tuple, bag, map) as the value type. I hope that answers your question regarding types inside a map in Pig. Let me know if you have further questions/clarifications. Thanks, Santhosh -----Original Message----- From: ANKUR GOEL [mailto:ankur.g...@corp.aol.com] Sent: Thursday, January 15, 2009 6:35 AM To: pig-dev@hadoop.apache.org Cc: pig-u...@hadoop.apache.org Subject: Data types for Map key value pairs Hi All, I have a custom loader that returns a set of fields after reading a log line. One of the fields returned is of type DataType.Map. My question is how can I set the data types for this map's (key, value) pair. In my script I try to generate a record from k,v of this map and get the error java.io.IOException: Unknown type Unknown at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.PigMapBase. map(PigMapBase.java:178) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.PigMapOnly$ Map.map(PigMapOnly.java:65) at org.apache.hadoop.mapred.MapRunner.run(MapRunner.java:47) at org.apache.hadoop.mapred.MapTask.run(MapTask.java:227) at org.apache.hadoop.mapred.TaskTracker$Child.main(TaskTracker.java:2207) here is my script raw = LOAD 'myfile' USING myUdf.MyCustomLoader() ; filtered = FILTER raw BY (ARG_MAP#'key' is not null); entry = FOREACH filtered GENERATE A, B, myUdf.MySplit(ARG_MAP#'key', '|') as FIELDS; // This returns a map with String (key, value) pairs // The MySplit UDF line splits the value in the map which is "|" separated and puts the splits it into another Map and returns it. Each split is keyed by 'field0', 'field1'...'fieldn' where n is the number of splits. result = FOREACH entry GENERATE A, B, FIELDS#'field0' as CLIENT_ID, FIELDS#'field1' as CHANNEL_ID, FIELDS#'field2' as OTHER_ID; // Here another tuple is generated store results into 'location' using PigStorage(); Any help here is appreciated. TIA -Ankur
