I created a parquet file with several columns from spark 2.1.0
On of this column is a map type created based of a dictionary.
I was expecting that i could query this fields exactly in the same way I
query json data from drill.
The problem the field is read by drill in this format
{"key_value":[{"key":"result","value":"win"},{"key":"game_mode","value":"win_race"},{"key":"air_time","value":"2"},{"key":"race_time","value":"45"}]}
when the original was
{'game_mode':'win_race', 'result':'win', 'race_time':'45', 'air_time': '2'}
Any idea to solve this problem?
--
*Abel Castellanos Carrazana*
Data Scientist/Engineer
Barcelona Data Team
Gameloft
