[PIG-4447] Pig Cannot handle nullable values (arrays and records) in avro records

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.13.0
• Fix Version/s: 0.15.0
• Patch Info: Patch Available
• Hadoop Flags: Reviewed
• Assignee: Ratandeep Ratti
• Reporter: Ratandeep Ratti
• Created: 05/Mar/15 16:58
• Updated: 07/Jun/15 03:48
• Resolved: 05/Mar/15 17:46

Description

Here's an example of an avro schema containing nullable values in a map

{
    "name" : "nullableRecordInMap",
    "namespace" : "org.apache.pig.test.builtin",
    "type" : "record",
    "fields" : [
        {"name" : "key", "type" : "string"},
        {"name" : "value", "type" : "int"},
        {
            "name" : "parameters",
            "type": [
                "null",
                {
                    "type": "map",
                    "values": [
                        "null",
                        {
                            "type": "record",
                            "name": "nullable_record",
                            "fields": [
                                {
                                    "name": "id",
                                    "type": [
                                        "null",
                                        "string"
                                    ]
                                }
                            ]
                        }
                    ]
                }
            ]
        }
    ]
}

Here's the corresponding Pig resource schema on running it through org.apache.pig.impl.util.avro.AvroStorageSchemaConversionUtilities

key:chararray,value:int,parameters:[nullable_record:(union:(id:chararray))]

Note that Pig should unpack the underlying schema from the nullable union and the Pig schema should be

key:chararray,value:int,parameters:[nullable_record:(id:chararray)]

There's similar behavior if the nullal map value is of type array

I've created a patch with a few testcases written.

Issue Links

• depends upon: HADOOP-9365 TestHAZKUtil fails on Windows