hive-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sanjay Subramanian <Sanjay.Subraman...@wizecommerce.com>
Subject Re: Hive UDAF extending UDAF class: iterate or evaluate method
Date Thu, 08 Aug 2013 04:03:08 GMT
Please follow the guidance for UDF provided in the Hive Programming book by Wampler/Capriolo.
That will work for u.

I can say with confidence that their book was mighty helpful to me in my project from start
to production...

And I would recommend go ahead with a way, implement and then fine tune otherwise u will be
in analysis paralysis mode…

We are all on a  path of discovery here ...

Regards

sanjay

From: Ritesh Agrawal <ragrawal@netflix.com<mailto:ragrawal@netflix.com>>
Reply-To: "user@hive.apache.org<mailto:user@hive.apache.org>" <user@hive.apache.org<mailto:user@hive.apache.org>>
Date: Wednesday, August 7, 2013 5:57 PM
To: "user@hive.apache.org<mailto:user@hive.apache.org>" <user@hive.apache.org<mailto:user@hive.apache.org>>
Subject: Re: Hive UDAF extending UDAF class: iterate or evaluate method

Hi Sanjay, Lefty

Thanks for the help but none of above responses directly answering my question (probably I
am not asking clear enough :-(  ).

Below I have two different structure of a UDAF (aggregation function). My question is which
one is the preferred/right approach

http://pastebin.com/QCgd4Hxc  : This version is based on based on what I could understand
from API docs about UDAF class.

http://pastebin.com/Uctamtek : This version is based on the book Hadoop The definitive guide.
Notice the function names for different from the first one.

I hope this clarifies my question.

Thanks
Ritesh





On Wed, Aug 7, 2013 at 5:34 PM, Lefty Leverenz <leftyleverenz@gmail.com<mailto:leftyleverenz@gmail.com>>
wrote:
Sounds like the wikidoc needs some work.  I'm open to suggestions.  If Sanjay's simple UDF
helps, I could put it in the wiki along with any advice you think would help.

Does anyone else have use cases to contribute?

-- Lefty


On Mon, Aug 5, 2013 at 2:45 PM, Sanjay Subramanian <Sanjay.Subramanian@wizecommerce.com<mailto:Sanjay.Subramanian@wizecommerce.com>>
wrote:
Hi Ritesh

To help u get started , I am writing a simple HelloWorld-ish UDF that might help…If it doesn't
please ask for more clarifications...

Good Luck
Thanks

sanjay

********************************************************************************
ToUpperCase.java


package com.sanjaysubramanian.utils.hive.udf;


import org.apache.hadoop.hive.ql.exec.UDF;


public finalclass ToUpperCase extends UDF{

    protected final Log logger = LogFactory.getLog(toUpperCase.class);


publicString evaluate(final String inputString) {

     if (inputString != null){

return inputString.toUpper;

             }

             else {

return inputString;

    }

       }

}

********************************************************************************

Usage in a Hive script

hive -e "

create temporary function toupper  as 'com.sanjaysubramanian.utils.hive.udf.ToUpperCase';

SELECT
      first_name,
      toupper(first_name)
FROM
      company_names
"

***********************************************************************************

From: Ritesh Agrawal <ragrawal@netflix.com<mailto:ragrawal@netflix.com>>
Reply-To: "user@hive.apache.org<mailto:user@hive.apache.org>" <user@hive.apache.org<mailto:user@hive.apache.org>>
Date: Monday, August 5, 2013 9:41 AM
To: "user@hive.apache.org<mailto:user@hive.apache.org>" <user@hive.apache.org<mailto:user@hive.apache.org>>
Subject: Re: Hive UDAF extending UDAF class: iterate or evaluate method

Hi Lefty,

I used the wiki you sent to write my first version of UDAF. However, I found it to be utterly
complex, especially for storing partial results as I am not very familiar with hive API. Then
I found another example of UDAF in the hadoop the definitive guide book and it had much simpler
code but using different method. Instead of using iterate it was using evaluate method and
so I am getting confused.

Ritesh


On Sun, Aug 4, 2013 at 2:18 PM, Lefty Leverenz <leftyleverenz@gmail.com<mailto:leftyleverenz@gmail.com>>
wrote:
You might find this wikidoc useful:  GenericUDAFCaseStudy<https://cwiki.apache.org/confluence/display/Hive/GenericUDAFCaseStudy>.

The O'Reilly book "Programming Hive" also has a section called "User-Defined Aggregate Functions"
in chapter 13 (Functions), pages 172 to 176.

-- Lefty


On Sun, Aug 4, 2013 at 7:12 AM, Ritesh Agrawal <ragrawal@netflix.com<mailto:ragrawal@netflix.com>>
wrote:
Hi all,

I am trying to write a UDAF function. I found an example that shows how to implement a UDAF
in "Hadoop The Definitive Guide" book. However I am little confused. In the book, the author
extends UDAF class and implements init, iterate, terminatePartial,  merge and terminate function.
However looking at the hive docs (http://hive.apache.org/docs/r0.11.0/api/org/apache/hadoop/hive/ql/exec/UDAF.html),
it seems I need to implement init, aggregate, evaluatePartial, aggregatePartial and evaluate
function. Please let me know what are the write functions to implement.

Ritesh



--
Lefty


CONFIDENTIALITY NOTICE
======================
This email message and any attachments are for the exclusive use of the intended recipient(s)
and may contain confidential and privileged information. Any unauthorized review, use, disclosure
or distribution is prohibited. If you are not the intended recipient, please contact the sender
by reply email and destroy all copies of the original message along with any attachments,
from your computer system. If you are the intended recipient, please be advised that the content
of this message is subject to access, review and disclosure by the sender's Email System Administrator.



--
Lefty


CONFIDENTIALITY NOTICE
======================
This email message and any attachments are for the exclusive use of the intended recipient(s)
and may contain confidential and privileged information. Any unauthorized review, use, disclosure
or distribution is prohibited. If you are not the intended recipient, please contact the sender
by reply email and destroy all copies of the original message along with any attachments,
from your computer system. If you are the intended recipient, please be advised that the content
of this message is subject to access, review and disclosure by the sender's Email System Administrator.

Mime
View raw message