I'm quite new to Pig, so perhaps my input is off base here, but if you input one such record without defining the schema I believe the JsonLoader will define the schema for you, no? If so, just import one such record and 'describe' the variable to see the schema. Well... if it's not that easy, then I'm not the right one to answer, but maybe it gets you something quick. David
-----Original Message----- From: marian.steinb...@gmail.com [mailto:marian.steinb...@gmail.com] On Behalf Of Marian Steinbach Sent: Wednesday, May 22, 2013 4:13 PM To: user@pig.apache.org Subject: [Bulk] pig 0.10.0 JsonLoader and nested list I would like to load a JSON file containing records of the following format: { "area": "ABC", "date_day": 1, "date_hour": 0, ... "energy": [["17-16", 1], ["18-17", 2]] } The "energy" property represents a sparse matrix. It's a list with an arbitrary number of key-value-pairs (minimum 1). The first element (string) is the matrix unit key, the second element is the value. I need both key and value in order to summarize values with matching keys in my pig job. I understand that it should be possible to import this as a bag. Correct? Can anybody tell me how the schema definition passed to the built-in JsonLoader function should look like? Thanks in advance! Marian
