Reputation: 2837
I'm having a bit of trouble debugging a SQL query and would really appreciate some help.
Here is the query:
SELECT p.id, p.type, p.submission_id,
p.title, p.description, p.date, extract('epoch' FROM p.time) AS time,
podcasts.image_url, podcasts.title AS podcast_title,
COUNT(u1) as upvote_count, u2.id as upvote_id,
episodes.mp3_url, episodes.duration,
COUNT(c) as comment_count
FROM posts AS p LEFT JOIN upvotes AS u1 ON p.id=u1.post_id AND u1.comment_id=-1
LEFT JOIN upvotes AS u2 ON p.id=u2.post_id AND u2.user_id=$1 AND u2.comment_id=-1
LEFT JOIN episodes ON p.submission_id = episodes.id
LEFT JOIN podcasts ON episodes.podcast_id=podcasts.id
LEFT JOIN comments AS c ON c.post_id=p.id
WHERE p.type='podcast' AND p.time IS NOT NULL
GROUP BY(p.id, u2.id, podcasts.image_url, episodes.mp3_url, episodes.duration, podcasts.title);
The unexpected behavior comes from the two COUNT statements. I expect upvote_count to be equivalent to
SELECT COUNT(*) FROM upvotes WHERE upvotes.post_id = (individual post id);
for each individual post and same for comment count (which I expect to return the total number of comments for each post. However, I am getting strange seemingly random results from these queries for those two fields. Can anybody help me diagnose the problem?
Upvotes: 2
Views: 10747
Reputation:
count()
(and all other aggregate functions) ignores null values.
However, COUNT(c)
references the complete row ("record") from the table alias c
But that is is always not null even when all columns of that record are null.
You need to change both count()
calls and pass a column from that table to it, e.g. count(u1.post_id)
and count(c.post_id)
Upvotes: 2