我遇到了与MySQL Invalid use of group function相关的问题。
我正在尝试执行的查询就是这个（我知道这是一个非常大的查询，但是如果我可以执行此操作，那么我将调整并修复性能并轻松阅读内容）：
SELECT
`id_c` AS `sceneId`,
`scene`.`id` AS `sceneName`,
`scene`.`frameCount` AS `frames`,
task.id_mv AS `id_mv`,
CONCAT( '[',GROUP_CONCAT(
CONCAT(
'{',
'"idTk": ', `task`.`id_tk`, ', ',
-- '"tktype": ', `task`.`id_tk_type`, ', ',
-- '"subDepType": ', IF ( `task`.`id_sub_dep_type` = 54, `task`.`id_sub_dep_type`, 'null'), ', ',
-- '"color": "', `task`.`color`, '", ',
'"weight": ', MAX(`task`.`weight`), ', ',
'"dept": ', '"',
CASE `task`.`id_tk_type`
WHEN 43 THEN 'MVB'
WHEN 12 THEN 'LYT'
WHEN 32 THEN
CONCAT_WS(
', ',
IF( `task`.`id_sub_dep_type` != 54, 'ANI', NULL ),
IF( `task`.`id_sub_dep_type` = 54, 'CKD', NULL )
)
WHEN 190 THEN 'DYN'
WHEN 104 THEN 'FIN'
WHEN 191 THEN 'LGT'
END, '"',
-- task.id_tk_type, ', ',
-- task.id_sub_dep_type,
'}'
)
ORDER BY
FIELD( `task`.`id_tk_type`, 43, 12, 32, 190, 104, 191 ),
`task`.`weight` DESC
SEPARATOR
', '
), ']') AS `tkInfo`
FROM
`V_task` AS `task`
JOIN `tk_fct_mcsc` AS `mcsc` ON `task`.`id_tk` = `mcsc`.`id_tk`
JOIN `scene` ON `mcsc`.`id_c` = `scene`.`id_scene`
WHERE
`task`.`id_sub_dep_type` IN (
SELECT
`id_sub_dep_type`
FROM
`tk_dim_sub_dep_type`
WHERE `id_tk_type` IN (
SELECT
`id_tk_type`
FROM
`tk_dim_dep_type`
WHERE
`id_tk_type` IN ( 43, 12, 32, 190, 104, 191 )
)
AND `id_sub_dep_type` != 54
UNION
SELECT
`id_sub_dep_type`
FROM
`tk_dim_sub_dep_type`
WHERE
`id_sub_dep_type` = 54
)
AND `task`.`is_appr` = FALSE
AND `mcsc`.`type` = 'sc'
AND `scene`.`id` != '000'
-- AND `mcsc`.`id_b` = 530
GROUP BY
`scene`.`id`,
`scene`.`id_scene`
-- task.weight
ORDER BY
`scene`.`id`,
`task`.`weight` DESC;
错误，我知道是MAX( task.weight )函数中的GROUP_CONCAT，但我不知道如何避免这种情况。
我实际上正在尝试构建某种类型的sqlfiddle来让你测试这些东西，但我想要的只是另一双眼睛让我知道在哪里看。
谢谢！
编辑＃1 正如kickstart所问，我想要获得的是这样的：
sceneId sceneName frames id_mv tkInfo
200 001 1200 1 "[{"dept": "LGT"}, {"dept": "LGT"}]"
1342 001 45 14 "[{"dept": "LYT"}, {"dept": "LGT"}, {"dept": "LGT"}]"
1335 001 460 14 "[{"dept": "MVB"}, {"dept": "LYT"}, {"dept": "LGT"}, {"dept": "LGT"}]"
1351 001 20 11 "[{"dept": "LYT"}, {"dept": "ANI"}, {"dept": "FIN"}, {"dept": "LGT"}]"
1375 001 63 11 "[{"dept": "LYT"}, {"dept": "CKD"}, {"dept": "ANI"}, {"dept": "DYN"}, {"dept": "FIN"}, {"dept": "FIN"}, {"dept": "FIN"}, {"dept": "LGT"}]"
1382 001 66 11 "[{"dept": "LYT"}, {"dept": "CKD"}, {"dept": "ANI"}, {"dept": "FIN"}]"
但是，如果没有“dept”的重复，并且这样做，我需要MAX权重的部分。
编辑＃2 我有点作弊解决了我的这个问题。
首先，我简化了很多主要查询，删除了case-thing并为每个sceneId插入了一个MAX（权重）。
通过这个简化的查询，我构建了一个View，并且在该视图中我已经应用了case / group_concat的东西..并且所有结果都很好。
[..sqlfiddle将立刻插入这里..]
答案 0 :(得分：1)
进行一些游戏，并且（非常）未经测试，但认为你需要有效地进行两次查询。一旦获得细节，一次获得最大重量。你可以移动一些子选择（进行连接 - 联合将在连接之前处理重复）。
这样的事情： -
SELECT
Sub1.id_c AS sceneId,
Sub1.id AS sceneName,
Sub1.frameCount AS frames,
Sub1.id_mv AS id_mv,
CONCAT( '[',GROUP_CONCAT(
CONCAT(
'{',
'"idTk": ', Sub1.id_tk, ', ',
'"weight": ', Sub2.MaxTaskWeight, ', ',
'"dept": ', '"', Sub1.id_tk_type_decode, '"',
'}'
)
ORDER BY
FIELD( Sub1.id_tk_type, 43, 12, 32, 190, 104, 191 ),
Sub2.MaxTaskWeight DESC
SEPARATOR
', '
), ']') AS tkInfo
FROM
(
SELECT
id_c AS sceneId,
scene.id AS sceneName,
scene.frameCount AS frames,
task.id_mv AS id_mv,
task.id_tk,
CASE task.id_tk_type
WHEN 43 THEN 'MVB'
WHEN 12 THEN 'LYT'
WHEN 32 THEN IF( task.id_sub_dep_type = 54, 'CKD', 'ANI' )
WHEN 190 THEN 'DYN'
WHEN 104 THEN 'FIN'
WHEN 191 THEN 'LGT'
END AS id_tk_type_decode,
task.id_tk_type
FROM V_task AS task
INNER JOIN tk_fct_mcsc AS mcsc ON task.id_tk = mcsc.id_tk
INNER JOIN scene ON mcsc.id_c = scene.id_scene
INNER JOIN
(
SELECT id_sub_dep_type
FROM tk_dim_sub_dep_type
INNER JOIN tk_dim_dep_type
ON tk_dim_sub_dep_type.id_tk_type = tk_dim_dep_type.id_tk_type
WHERE tk_dim_dep_type.id_tk_type IN ( 43, 12, 32, 190, 104, 191 )
AND tk_dim_sub_dep_type.id_sub_dep_type != 54
UNION
SELECT id_sub_dep_type
FROM tk_dim_sub_dep_type
WHERE id_sub_dep_type = 54
) Sub1
ON task.id_sub_dep_type = Sub1.id_sub_dep_type
WHERE task.is_appr = FALSE
AND mcsc.`type` = 'sc'
AND scene.`id` != '000'
)
INNER JOIN
(
SELECT scene.id, scene.id_scene, MAX(task.weight) AS MaxTaskWeight
FROM V_task AS task
INNER JOIN tk_fct_mcsc AS mcsc ON task.id_tk = mcsc.id_tk
INNER JOIN scene ON mcsc.id_c = scene.id_scene
INNER JOIN
(
SELECT id_sub_dep_type
FROM tk_dim_sub_dep_type
INNER JOIN tk_dim_dep_type
ON tk_dim_sub_dep_type.id_tk_type = tk_dim_dep_type.id_tk_type
WHERE tk_dim_dep_type.id_tk_type IN ( 43, 12, 32, 190, 104, 191 )
AND tk_dim_sub_dep_type.id_sub_dep_type != 54
UNION
SELECT id_sub_dep_type
FROM tk_dim_sub_dep_type
WHERE id_sub_dep_type = 54
) Sub1
ON task.id_sub_dep_type = Sub1.id_sub_dep_type
WHERE task.is_appr = FALSE
AND mcsc.`type` = 'sc'
AND scene.`id` != '000'
GROUP BY scene.id, scene.id_scene
) Sub2
ON Sub1.id = Sub2.id
AND Sub1.id_scene = Sub2.id_scene
GROUP BY
Sub1.id,
Sub1.id_scene
ORDER BY
Sub1.id,
MaxTaskWeight DESC;