Hi Shane,
Mmm did you try that ? Unfortunately PigLatin doesn't allow a query like '
A = ORDER Z BY $1 ; '. In fact , according with the docs it's necessary to
do :
A = FOREACH Z {
X = ORDER 'to_order' BY ....;
};
where 'to_order' has to be a bag (if i'm not wrong), and in my case there's
no bag to order.
In previous attempts i tried to group just by 'user' to later on use order
over the groupped bag :
(12 , {(12,145),(12,145),(12,133),(12,145)})
(13 , {(13,192),(13,164),(13,192)})
but i don't achieve to make a nested group in the already groupped bag , in
order to COUNT and finally order.
By the way , my SQL query had to say ORDER BY user,item,COUNT(*) DESC ,
but anyway i don't know how to solve both options.
I will keep trying ..
Thanks.
Eric.
On 8/10/07, Shane Butler <[EMAIL PROTECTED]> wrote:
>
> Hi Eric,
>
> I think it will be something like A = ORDER Z BY .... (see pg 1 of
> PigLatinAdvanced.pdf in the docs).
>
> Cheers,
> Shane
>
> On 8/10/07, Eric Palacios <[EMAIL PROTECTED]> wrote:
> > Hi all,
> >
> > I'm trying to translate this simple query to PigLatin but i remain stuck
> in
> > the ordering.
> >
> > Given this table :
> >
> > (user,item)
> > 12 145
> > 13 192
> > 12 145
> > 12 133
> > 13 164
> > 13 192
> > 12 145
> >
> > i want to run this :
> >
> > SELECT user,item,COUNT(*)
> > FROM table
> > GROUP BY user,item
> > ORDER BY COUNT(*) DESC
> >
> >
> >
> > I tried this :
> >
> > X = LOAD 'table.txt';
> > Y = GROUP X BY ($0,$1);
> > Z = FOREACH X GENERATE group,COUNT($1);
> >
> > and i get this table:
> >
> > ((12,133),1)
> > ((12,145),3)
> > ((13,164),1)
> > ((13,192),2)
> >
> > but i have no clue how to flatten/group this to order it.
> >
> > I hope this is the right mailing list to talk about Hadoop/Yahoo Pig
> stuff
> >
> > Thanks in advance,
> >
> > Eric.
> >
>
