easiest way in this kind would be write up a small udf.
As Stephen suggested, its just a number so you can do maths to extract year
and month out of the number and then do the comparison.

also 201307 is not a supported date format anywhere as per my knowledge


On Wed, Jul 3, 2013 at 12:55 PM, Jérôme Verdier
<verdier.jerom...@gmail.com>wrote:

> Hi Stephen,
>
> Thanks for your reply.
>
> The problem is that my input date is this : in_co_an_mois (format :
> YYYYMM, integer), for example, this month, we have 201307
>
> and i have to deal with this date : add one month, compare to over date,
> etc...
>
> The problem is that apparently, there is no way to do this, because Hive
> can't deal with this type of data because it's not a date format.
>
> For hive, this is just a number.
>
> Hive can deal with this : 1970-01-01 00:00:00, or this : 2009-03-20, but
> not with this unusual format : 201307.
>
> Thanks.
>
>
>
>
> 2013/7/2 Stephen Sprague <sprag...@gmail.com>
>
>> not sure i fully understand your dilemma.    have you investigated any of
>> the date functions listed here?
>>
>>
>> https://cwiki.apache.org/confluence/display/Hive/LanguageManual+UDF#LanguageManualUDF-DateFunctions
>>
>> seems to me you could pull the year and month from a date.  or if you
>> have an int then do some arithmetic to get the year and month.  eg. year =
>> floor( <your int>/10000) and month = cast( <your int> % 100 as int)  [% ==
>> modulus operator]
>>
>> or am i not even answering your question?
>>
>>
>>
>> On Tue, Jul 2, 2013 at 2:42 AM, Jérôme Verdier <
>> verdier.jerom...@gmail.com> wrote:
>>
>>> Hi,
>>>
>>> i trying to translate some PL/SQL script in HiveQL, and dealing with
>>> unusual date format.
>>>
>>> i added a variable in my hive script : '${hiveconf:in_co_an_mois}' which
>>> is a year/month date format, like this : 201307 (INT format).
>>>
>>> I would like to transform this in date format, because i have to
>>> increment this (add one month/one year).
>>>
>>> Is there a way to do this in hive ?
>>>
>>> Thanks.
>>>
>>>
>>>
>>> --
>>> *Jérôme*
>>>
>>>
>>
>
>


-- 
Nitin Pawar

Reply via email to