I have two tables in my database:
(1) PHRASES:
t_phrase
========
I like
They prefer
...
Somebody else wants
and
(2) PLACES:
n_id t_place
==== =======
1 London
2 Paris
...
N New York
Table PHRASES
has at least as many rows as PLACES
. I need to join these two tables in such a way as to select all place
s with one phrase
for each of them - but phrases need to be randomly distributed across places. The overall places
table isn't too big: maybe, about 3-4 thousand rows, however there will be an additional WHERE
clause on it that will limit the output to about 200 places at most.
Ideally, I'd like this to be in one SQL statement, but so far I haven't been able to get my head around this. Therefore the second option is a stored function returning a row of (int, varchar, varchar)
. For this, I was thinking of something along the lines of:
Somehow this seems to me very inefficient, but I can't come up with anything better.
Can you suggest any better idea? Or, even better, one statement SQL, maybe?
Thanks in advance.
EDIT: Please note that the phrases should NOT be repeated in the resultset. There are always at least as many phrases as there are places.
WITH p AS (
SELECT place, row_number() OVER () AS rn
FROM t_place
WHERE <some condition>
)
, ph AS (
SELECT phrase, row_number() OVER (ORDER BY random()) AS rn
FROM t_phrase
)
SELECT ph.phrase, p.place
FROM p
JOIN ph USING (rn);
It won't get any more random, if you impose a truly random order on both tables, it will only get slower. I impose the random order on phrases, because:
There are always at least as many phrases as there are places.
It needs to be done with the bigger set, lest some non-random part might get cut off. For the smaller set (places), on the other hand, any sequence of numbers without gaps is good, so I pick the fastest way.
My example uses CTEs, but it can be done with subqueries just as well. Both CTE and window functions require PostgreSQL 8.4 or later.
I think the following will work:
select (select phrase from phrases order by random() limit 1),
place
from places
The select within the select should be called for each row, so it should return a different value each time.
If you want just a random arrangement of the phrases and places, you can use windows functions:
select ph.phrase, p.place
from (select place, row_number() over (order by place) as seqnum
from places p
) p join
(select phrase, row_number() over (order by random()) as seqnum
from phrases
) ph
on p.seqnum = ph.seqnum
This orders the places by place (or any field could do). It randomizes the phrases, and joins on the resulting row numbers.
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.