Ming Ma created HIVE-7036:
-----------------------------
Summary: get_json_object bug when extract list of list with index
Key: HIVE-7036
URL: https://issues.apache.org/jira/browse/HIVE-7036
Project: Hive
Issue Type: New Feature
Components: UDF
Affects Versions: 0.12.0
Environment: all
Reporter: Ming Ma
Priority: Minor
https://github.com/apache/hive/blob/trunk/ql/src/java/org/apache/hadoop/hive/ql/udf/UDFJson.java#L250
this line should be out of the for-loop
For example
json = '{"h":[1, [2, 3], {"i": 0}, [{"p": 11}, {"p": 12}, {"pp": 13}]}'
get_json_object(json, '$.h[*][0]') should return back the first node(if exists)
of every childrenof '$.h'
which specifically should be
[2,{"p":11}]
but hive returns only
2
because when hive pick the node '2' out, the tmp_jsonList will change to a list
only contains one node '2':
[2]
then it was assigned to variable jsonList, in the next loop, value of i would
be 2 which is greater than the size(always 1) of jsonList, then the loop broke
out.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
