左联接使来自不同表的两个计数具有相同的值

Question

I have a very long query with a lot of LEFT JOINS in. The problem is that I have two COUNTS which do not seem to work together. Both the counts will equal the same value even if they're not.

Here's what I have;

SELECT T.ID,
       T.name,
       T.pic,
       T.T_ID,
       COUNT(P.T_ID) AS plays,
       COUNT(L.T_ID) AS likes,
       S.Status,
       G.gig_name,
       G.date_time,
       G.lineup,
       G.price,
       G.ticket,
       E.action,
       E.ID,
       E.timestamp,
       E.E_ID
  FROM events E
       LEFT JOIN TRACKS T
       ON T.ID = E.ID AND E.action = 'has uploaded a track.' AND E.E_ID = T.T_ID
       LEFT JOIN STATUS S
       ON S.ID = E.ID AND E.action = 'has some news.' AND E.E_ID = S.S_ID
       LEFT JOIN GIGS G
       ON G.ID = E.ID AND E.action = 'has posted a gig.' AND E.E_ID = G.G_ID
       LEFT JOIN track_plays P
       ON P.A_ID = E.ID AND E.action = 'has uploaded a track.' AND E.E_ID = P.T_ID
       LEFT JOIN track_likes L
       ON L.ID = E.E_ID AND E.action = 'has uploaded a track.'
 WHERE E.ID = '3'
 GROUP BY E.E_ID
 ORDER BY E.timestamp DESC LIMIT 15

I won't explain all the query but I think you'll get the gist of it. The JOINS in question are the last two. If COUNT(P.T_ID) = 100 COUNT(L.T_ID) will also = 100 .

Thanks in advance!

Answer 1

You should try using COUNT(DISTINCT)

SELECT T.ID,
       T.name,
       T.pic,
       T.T_ID,
       COUNT(DISTINCT P.T_ID) AS plays,
       COUNT(DISTINCT L.T_ID) AS likes,
       S.Status,
       G.gig_name,
       G.date_time,
       G.lineup,
       G.price,
       G.ticket,
       E.action,
       E.ID,
       E.timestamp,
       E.E_ID
  FROM events E
       LEFT JOIN TRACKS T
       ON T.ID = E.ID AND E.action = 'has uploaded a track.' AND E.E_ID = T.T_ID
       LEFT JOIN STATUS S
       ON S.ID = E.ID AND E.action = 'has some news.' AND E.E_ID = S.S_ID
       LEFT JOIN GIGS G
       ON G.ID = E.ID AND E.action = 'has posted a gig.' AND E.E_ID = G.G_ID
       LEFT JOIN track_plays P
       ON P.A_ID = E.ID AND E.action = 'has uploaded a track.' AND E.E_ID = P.T_ID
       LEFT JOIN track_likes L
       ON L.ID = E.E_ID AND E.action = 'has uploaded a track.'
 WHERE E.ID = '3'
 GROUP BY E.E_ID
 ORDER BY E.timestamp DESC LIMIT 15

Answer 2

You might try "rephrasing" those counts as correlated subqueries:

SELECT T.ID,
       T.name,
       T.pic,
       T.T_ID,
       (SELECT COUNT(*) FROM track_plays WHERE A_ID = E.ID AND E.action = 'has uploaded a track.' AND E.E_ID = T_ID) AS plays,
       (SELECT COUNT(*) FROM track_likes WHERE ID = E.E_ID AND E.action = 'has uploaded a track.') AS likes,
       S.Status,
       G.gig_name,
       G.date_time,
       G.lineup,
       G.price,
       G.ticket,
       E.action,
       E.ID,
       E.timestamp,
       E.E_ID
  FROM events E
       LEFT JOIN TRACKS T
       ON T.ID = E.ID AND E.action = 'has uploaded a track.' AND E.E_ID = T.T_ID
       LEFT JOIN STATUS S
       ON S.ID = E.ID AND E.action = 'has some news.' AND E.E_ID = S.S_ID
       LEFT JOIN GIGS G
       ON G.ID = E.ID AND E.action = 'has posted a gig.' AND E.E_ID = G.G_ID
 WHERE E.ID = '3'
 GROUP BY E.E_ID
 ORDER BY E.timestamp DESC LIMIT 15

I find this to be simpler to understand/maintain and often faster.

Answer 3

Both COUNTs have no WHERE clause and appear in the same table, namely the result table. Therefore both report the total number of rows of the resulting table.

Answer 4

May be i am wrong, but you just count # of thevresult rows, isn't it?

May be you need to change count to be like

select count(P_ID) from track_plays where p_id = E.ID