分组、计数和位置

Question

我被难住了，摸不着头脑。它一定很简单，但我没有看到它。

假设我有四个表：

video = id
hastag = id, tag_id, video_id
hasteam = id, team_id, video_id
hasidol = id, idol_id, video_id

此数据集（仅作为示例）：

video = (1), (2), (3)
hastag = (1, 1, 1), (2, 1, 2), (3, 2, 3)
hasteam = (1, 1, 1), (2, 1, 3), (3, 2, 2)
hasidol = (1, 1, 3)

这个查询：

SELECT v.id ,
COUNT(vhtag.id),
COUNT(vhteam.id),
COUNT(vhidol.id)
FROM video v
LEFT JOIN hastag vhtag ON vhtag.video_id = v.id
LEFT JOIN hasteam vhteam ON vhteam.video_id = v.id
LEFT JOIN hasidol vhidol ON vhidol.video_id = v.id

WHERE
v.id <> 1
AND
(
  vhtag.tag_id IN (SELECT htt2.tag_id FROM hastag htt2 WHERE video_id = 1)
  OR
  vhteam.team_id IN (SELECT htt3.team_id FROM hasteam htt3 WHERE video_id = 1)
  OR
  vhidol.idol_id IN (SELECT htt4.idol_id FROM hasidol htt4 WHERE video_id = 1)
)
GROUP BY v.id

它为我提供了与 WHERE 子句不对应的“具有”行数。例如，如果视频行只有一个共同的视频 ID 为 1 的团队和一个完全不相关的标签，当它应该说“共同标签”时，它会给我“共同标签计数：1，共同团队计数：1”计数：0（因为它是一个无关标签），普通团队计数：1”。

现在，只要我将查询限制为只有一个“有”表，如下所示：

SELECT v.id ,
COUNT(vhtag.id)
FROM video v
LEFT JOIN hastag vhtag ON vhtag.video_id = v.id

WHERE
v.id <> 1
AND
(
  vhtag.tag_id IN (SELECT htt2.tag_id FROM hastag htt2 WHERE video_id = 1)
)
GROUP BY v.id

然后它确实有效，但问题是当我尝试在查询中放入多个“has”表时。我尝试使用 HAVING，但它无法识别“vhtag.tag_id”列。 我显然在这里做错了什么，任何人都可以帮助我吗？

编辑：

这种作品：

LEFT JOIN hastag vhtag ON vhtag.video_id = v.id AND vhtag.tag_id IN (SELECT htt2.tag_id FROM hastag htt2 WHERE video_id = 1)
LEFT JOIN hasteam vhteam ON vhteam.video_id = v.id AND vhteam.team_id IN (SELECT htt3.team_id FROM hasteam htt3 WHERE video_id = 1)
LEFT JOIN hasidol vhidol ON vhidol.video_id = v.id AND vhidol.idol_id IN (SELECT htt4.idol_id FROM hasidol htt4 WHERE video_id = 1)

我也可以在 Doctrine 中使用它（我很笨，忘记了 WITH）。这是最佳方式吗？

Answer 1

您遇到的问题是您的联接导致每个团队内部出现交叉联接。

最简单的方法是做count distinct：

SELECT v.id , COUNT(distinct vhtag.id), COUNT(distinct vhteam.id),
       COUNT(distinct vhidol.id)
FROM video v LEFT JOIN
     hastag vhtag
     ON vhtag.video_id = v.id
     LEFT JOIN hasteam vhteam
     ON vhteam.video_id = v.id
     LEFT JOIN hasidol vhidol
     ON vhidol.video_id = v.id 

真正的解决方案是将每个值单独聚合，然后将结果连接在一起。

Answer 2

尽量简化查询，例如这样-

SELECT
  v.id,
  COUNT(vhtag.id),
  COUNT(vhteam.id),
  COUNT(vhidol.id)
FROM
  video v
LEFT JOIN (SELECT * FROM hastag WHERE video_id = 1) vhtag
  ON vhtag.video_id = v.id
LEFT JOIN (SELECT * FROM hasteam WHERE video_id = 1) vhteam
  ON vhteam.video_id = v.id
LEFT JOIN (SELECT * FROM hasidol WHERE video_id = 1) vhidol
  ON vhidol.video_id = v.id
WHERE
  v.id <> 1
GROUP BY
  v.id

它给出正确的结果吗？

输出：

+------+-----------------+------------------+------------------+
| id   | COUNT(vhtag.id) | COUNT(vhteam.id) | COUNT(vhidol.id) |
+------+-----------------+------------------+------------------+
|    2 |               0 |                0 |                0 |
|    3 |               0 |                0 |                0 |
+------+-----------------+------------------+------------------+