[jira] [Created] (AIRFLOW-3633) DESCRIBE table via HiverServer2Hook

["Shubham Gupta (JIRA)" <ji...@apache.org>]

Shubham Gupta created AIRFLOW-3633:
--------------------------------------

             Summary: DESCRIBE table via HiverServer2Hook
                 Key: AIRFLOW-3633
                 URL: https://issues.apache.org/jira/browse/AIRFLOW-3633
             Project: Apache Airflow
          Issue Type: Wish
          Components: hive_hooks
    Affects Versions: 1.10.0
            Reporter: Shubham Gupta


It seems a bit unfair that  [HiveServer2Hook.get_records()|https://github.com/apache/airflow/blob/v1-10-stable/airflow/hooks/hive_hooks.py#L855] only allows SELECT (and WITH c.t.e.) queries as of [v1.10|https://github.com/apache/airflow/blob/v1-10-stable/airflow/hooks/hive_hooks.py#L787]

While allowed queries have been [expanded|https://github.com/apache/airflow/blob/master/airflow/hooks/hive_hooks.py#L811] in master but it still isn't an exhaustive set.

 

My use-case involved DESCRIBE queries on tables and while I can use [headers|https://github.com/apache/airflow/blob/v1-10-stable/airflow/hooks/hive_hooks.py#L813] field from [get_results()|https://github.com/apache/airflow/blob/v1-10-stable/airflow/hooks/hive_hooks.py#L808] method, it isn't the same as some extra work is required to process the results from cursor description:
 * Column names are prefixed with table name
 * Data-types (strings) are not same as Hive

 

Is there a specific reason to support only limited set of queries (amongst those that return result)? If no, can we please have DESCRIBE queries too?



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)