Guilherme Braccialli created HIVE-11741:
-------------------------------------------

             Summary: Add a new hook to run before query parse/compile
                 Key: HIVE-11741
                 URL: https://issues.apache.org/jira/browse/HIVE-11741
             Project: Hive
          Issue Type: New Feature
          Components: hpl/sql, Parser, SQL
            Reporter: Guilherme Braccialli
            Priority: Minor


It would be nice to allow developers to extend hive query language, making 
possible to use custom wildcards on queries. 

People uses Python or R to iterate over vectors or lists and create SQL 
commands, this could be implemented directly on sql syntax.
For example this python script:
>>> sql = "SELECT state, "
>>> for i in range(10):
...   sql += "   sum(case when type = " + str(i) + " then value end) as 
sum_of_" + str(i) + " ,"
...
>>> sql += " count(1) as  total FROM table"
>>> print(sql)

Could be written directly in extended sql like this:
SELECT state,
%for id = 1 to 10%
   sum(case when type = %id% then value end) as sum_of_%id%,
%end%
, count(1) as total
FROM table
GROUP BY state

This kind of extensibility can be easily added if we add a new hook after 
VariableSubstitution call on Driver.compile method.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to