正确使用连接

Question

我正在寻找有关如何优化此功能的提示....

SELECT   u.uid,
         u.username,
         u.nid,
         wd.pay
FROM     (users_data AS u
          LEFT JOIN winners_data wd
            ON u.nid = wd.nid
               AND u.uid = wd.uid)
         LEFT JOIN dp_node dn
           ON u.nid = dn.nid
WHERE    u.uid = ".$val."
         AND ((dn.TYPE = 'event'
               AND (SELECT Count(nid)
                    FROM   tournament_event
                    WHERE  nid = u.nid
                           AND type_value IN ('A','B','C')) > 0)
               OR (dn.TYPE = 'new_event'
                   AND (SELECT Count(nid)
                        FROM   user_tournament_event
                        WHERE  nid = u.nid
                               AND type_0_value IN ('Y','X')) > 0))
ORDER BY nid ASC

实际上 $val 只不过是来自以下查询的 uid，作为我循环的一部分，一次一个。

SELECT DISTINCT(dump.uid) FROM leader_Jdump AS 转储

有没有办法将此部分也添加到上述查询中？我想如果我能在 MySQL 级别做到这一点会更快。

Answer 1

试试这个：

SELECT  u.uid, u.username, u.nid, wd.pay
FROM    users_data AS u
LEFT JOIN
        winners_data wd
ON      (wd.nid, wd.uid) = (u.nid, u.uid)
JOIN    dp_node dn 
ON      dn.nid = u.nid
        AND dn.type IN ('event', 'new_event')
WHERE   u.uid = ".$val."
        AND EXISTS
        (
        SELECT  NULL
        FROM    tournament_event te
        WHERE   te.nid = u.nid
                AND type_value IN ('A', 'B', 'C')
                AND dn.type = 'event'
        UNION ALL
        SELECT  NULL
        FROM    user_tournament_event te
        WHERE   te.nid = u.nid
                AND type_0_value IN ('X', 'Y')
                AND dn.type = 'new_event'

        )
ORDER BY
        u.nid ASC

我从dp_node 中删除了OUTER JOIN，因为您的原始查询在WHERE 子句中的dp_node 字段上需要一个非NULL 条件，所以LEFT JOIN 在这里没有用。

创建以下索引：

users_data (uid, nid)
winners_data (uid, nid)
dp_node (nid, event_type)
tournament_event (nid, type_value)
user_tournament_event (nid, type_0_value)

Answer 2

我见过这种风格（单行末尾的所有右括号），我想知道人们如何理解它。之前有人要求重新格式化版本：

SELECT    u.uid,
          u.username,
          u.nid,
          wd.pay
FROM      users_data u
          LEFT JOIN winners_data wd
               ON u.nid = wd.nid
               AND u.uid = wd.uid
          LEFT JOIN dp_node dn
               ON u.nid = dn.nid
WHERE     u.uid = ".$val."
          AND (
                (
                    dn.TYPE = 'event'
                    AND (
                        SELECT Count(nid)
                        FROM   tournament_event
                        WHERE  nid = u.nid
                               AND type_value IN ('A','B','C')
                    ) > 0
                )
                OR (
                    dn.TYPE = 'new_event'
                    AND (
                        SELECT Count(nid)
                        FROM   user_tournament_event
                        WHERE  nid = u.nid
                               AND type_0_value IN ('Y','X')
                    ) > 0
                )
          )
ORDER BY  nid ASC

Answer 3

这可能是等效的，但更容易阅读：

WITH NIDs(type_value,nid) AS (
  SELECT 'event', nid FROM tournament_event
  WHERE type_value IN ('A','B','C')
  UNION ALL
  SELECT 'new_event', nid FROM user_tournament_event
  WHERE type_value IN ('X','Y')
)

SELECT
  u.uid,
  u.username,
  u.nid,
  wd.pay
FROM users_data AS u JOIN NIDs AS n
ON n.nid = u.nid
AND n.type_value = u."type"
LEFT JOIN winners_data AS wd
ON u.nid=wd.nid AND u.uid=wd.uid
LEFT JOIN dp_node AS dn 
ON u.nid = dn.nid
WHERE u.uid=".$val."
ORDER BY nid ASC

如果“什么样的事件”是一个重要的问题，您最好使用一张表格，您可以通过查看一两列来做出决定。您似乎有两个保存此信息的表（tournament_event 和 user_tournament_event）。例如，您可以将信息存储在此查询的 NID 表达式中。

添加于 2009 年 8 月 22 日：

这是相同的查询，CTE 表示为派生表：

SELECT
  u.uid,
  u.username,
  u.nid,
  wd.pay
FROM users_data AS u JOIN (
  SELECT 'event', nid FROM tournament_event
  WHERE type_value IN ('A','B','C')
  UNION ALL
  SELECT 'new_event', nid FROM user_tournament_event
  WHERE type_value IN ('X','Y')
) AS n(type_value,nid)
ON n.nid = u.nid
AND n.type_value = u."type"
LEFT JOIN winners_data AS wd
ON u.nid=wd.nid AND u.uid=wd.uid
LEFT JOIN dp_node AS dn 
ON u.nid = dn.nid
WHERE u.uid=".$val."
ORDER BY nid ASC