Question: How can I get result types for a Hive query?

[RasterBurn]

.description gives:

[(u'my_schema.col_one', u'BIGINT_TYPE', None, None, None, None, True),
 (u'my_schema.col_two', u'DATE_TYPE', None, None, None, None, True),
 (u'my_schema.col_three',
  u'BIGINT_TYPE',
  None,
  None,
  None,
  None,
  True)]

So the second field gives the types, but I would really like to be able to get at column types I can use in DDL statements. In this case, BIGINT, DATE, and BIGINT.

What's the best way to do this?

I see two options:

1. String manipulation. Strip _TYPE off.
2. Add it to DBAPITypeObject by looking it up in TCLIService.constants.TYPE_NAMES:

for type_id in constants.PRIMITIVE_TYPES:
    name = ttypes.TTypeId._VALUES_TO_NAMES[type_id]
    setattr(sys.modules[__name__], name, DBAPITypeObject([name], constants.TYPE_NAMES[type_id]))

Then add __str__ to DBAPITypeObject which returns a valid column name for use in DDL statements.

[jingw]

To be honest I don't know what DBAPITypeObject is supposed to do (dead code for most people?), so I don't really want to touch it without some kind of example in a popular DB-API implementation.

description currently lacks the type parameters for map/array/union, so it might not be quite enough for you. But adding that feature could be an excuse to break backwards compatibility on description and strip the "_TYPE" suffix.

The fastest path for you is probably to just strip "_TYPE" client side.

[ispmarin]

Maybe the results would be similar to running a describe table?