我目前使用以下查询来获取每个用户的详细信息。
SELECT u.*, sums.total_votes, sums.no_of_events
FROM user u
LEFT JOIN
(
SELECT
us.user_uid,
count(ev.event_vote_id) AS total_votes
count(distinct ue.event_uid) AS no_of_events
FROM user_event ue
LEFT JOIN event_vote ev
ON ev.event_uid = ue.event_uid
GROUP BY ue.user_uid
) sums ON sums.user_uid = u.user_uid
但是，我希望也能恢复他们最高投票事件的排名（在所有事件中 - 不仅仅是他们自己的事件）。
用户
| USER_UID | FIRSTNAME | LASTNAME |
1 bob smith
2 rob smithies
3 john clark
EVENT
| GUID | NAME |
101 event1
102 event2
103 event3
USER_EVENT
| USER_EVENT_ID | USER_UID | EVENT_UID |
1001 1 101
1002 2 102
1003 1 103
EVENT_VOTE
| EVENT_VOTE_ID | USER_UID | EVENT_UID |
2001 2 101
2002 3 101
2003 2 103
预期结果
user_uid: 1
firstname: bob
lastname: smith
votes: 3 // 2 for 101, 1 for 103.
no_of_events: 2
bestRank: 1 (1st) // ranked 1st and 2nd but 1st is higher.
user_uid: 2
firstname: rob
lastname: smithies
votes: 0
no_of_events: 1
bestRank: 3 (3rd)
答案 0 :(得分：1)
此查询有3个部分
在演示中，您还可以看到三个查询，以便调试部分结果。
当前输出也是部分结果，以获得您需要添加的愿望结果
WHERE R.event_uid IS NULL
<强> SQL DEMO
SELECT *
FROM ( SELECT u.*, sums.total_votes, sums.no_of_events
FROM user u
JOIN ( SELECT ue.user_uid,
count(ev.event_vote_id) AS total_votes,
count(distinct ue.event_uid) AS no_of_events
FROM user_event ue
LEFT JOIN event_vote ev
ON ev.event_uid = ue.event_uid
GROUP BY ue.user_uid
) as sums
ON u.user_uid = sums.user_uid
) as U
JOIN ( SELECT T.*,
@rank := @rank + 1 as rn,
@dense := if (@votes = votes,
@dense,
if(@votes := votes, @rank, @rank)
) as dense
FROM (
SELECT
e.guid as event_uid,
ue.user_uid,
count(ev.event_uid) AS votes
FROM event e
JOIN user_event ue
ON e.guid = ue.event_uid
LEFT JOIN event_vote ev
ON ev.event_uid = ue.event_uid
GROUP BY e.GUID, ue.user_uid
ORDER BY count(ue.event_uid) DESC
) as T
CROSS JOIN (SELECT @rank := 0, @dense := 0, @votes := 0 ) as vars
ORDER BY votes desc, event_uid
) as Q
ON U.user_uid = Q.user_uid
LEFT JOIN ( SELECT T.*,
@rank2 := @rank2 + 1 as rn,
@dense2 := if (@votes2 = votes,
@dense2,
if(@votes2 := votes, @rank2, @rank2)
) as dense
FROM (
SELECT
e.guid as event_uid,
ue.user_uid,
count(ev.event_uid) AS votes
FROM event e
JOIN user_event ue
ON e.guid = ue.event_uid
LEFT JOIN event_vote ev
ON ev.event_uid = ue.event_uid
GROUP BY e.GUID, ue.user_uid
ORDER BY count(ue.event_uid) DESC
) as T
CROSS JOIN (SELECT @rank2 := 0, @dense2 := 0, @votes2 := 0 ) as vars
ORDER BY votes desc, event_uid
) as R
ON Q.user_uid = R.user_uid
AND Q.rn > R.rn
-- WHERE R.event_uid IS NULL
<强>输出
| USER_UID | FIRSTNAME | LASTNAME | total_votes | no_of_events | event_uid | user_uid | votes | rn | dense | event_uid | user_uid | votes | rn | dense |
|----------|-----------|----------|-------------|--------------|-----------|----------|-------|----|-------|-----------|----------|--------|--------|--------|
| 1 | bob | smith | 3 | 2 | 101 | 1 | 2 | 1 | 1 | (null) | (null) | (null) | (null) | (null) |
| 1 | bob | smith | 3 | 2 | 103 | 1 | 1 | 3 | 2 | 101 | 1 | 2 | 1 | 1 |
| 2 | rob | smithies | 1 | 3 | 102 | 2 | 1 | 2 | 2 | (null) | (null) | (null) | (null) | (null) |
| 2 | rob | smithies | 1 | 3 | 104 | 2 | 0 | 4 | 4 | 102 | 2 | 1 | 2 | 2 |
| 2 | rob | smithies | 1 | 3 | 105 | 2 | 0 | 5 | 4 | 102 | 2 | 1 | 2 | 2 |
| 2 | rob | smithies | 1 | 3 | 105 | 2 | 0 | 5 | 4 | 104 | 2 | 0 | 4 | 4 |