网站开发日记(14)-MYSQL子查询和嵌套查询优化

qazxsw1

查询游戏历史成绩最高分前100
SELECT ps.* FROM cdb_playsgame ps WHERE ps.credits=(select MAX(credits)
FROM cdb_playsgame ps1
where ps.uid=ps1.uid AND ps.gametag=ps1.gametag) AND ps.gametag='yeti3'
GROUP BY ps.uid order by ps.credits desc LIMIT 100;


SELECT ps.*
FROM cdb_playsgame ps,(select ps1.uid, ps1.gametag, MAX(credits) as credits
FROM cdb_playsgame ps1 group by uid,gametag) t
WHERE ps.credits=t.credits AND ps.uid=t.uid AND ps.gametag=t.gametag AND ps.gametag='yeti3'
GROUP BY ps.uid order by ps.credits desc LIMIT 100;执行时间仅为0.22秒，比原来的25秒提高了10000倍

查询当天游戏最好成绩
SELECT ps. * , mf. * , m.username
FROM cdb_playsgame ps
LEFT JOIN cdb_memberfields mf ON mf.uid = ps.uid
LEFT JOIN cdb_members m ON m.uid = ps.uid
WHERE ps.gametag = 'chuansj'
AND FROM_UNIXTIME( ps.dateline, '%Y%m%d' ) = '20081008'
AND ps.credits = (
SELECT MAX( ps1.credits )
FROM cdb_playsgame ps1
WHERE ps.uid = ps1.uid
AND ps1.gametag = 'chuansj'
AND FROM_UNIXTIME( ps1.dateline, '%Y%m%d' ) = '20081008' )
GROUP BY ps.uid
ORDER BY credits DESC
LIMIT 0 , 50

像查询里
AND ps.credits=(SELECT MAX(ps1.credits)
FROM {$tablepre}playsgame ps1 where ps.uid=ps1.uid AND ps1.gametag = '$game'  
AND FROM_UNIXTIME(ps1.dateline, '%Y%m%d') = '$todaytime'  )
特别消耗时间
另外，像：FROM_UNIXTIME(ps1.dateline, '%Y%m%d') = '$todaytime' 这样的语句会导致索引无效，因为对每个dataline的值都需要用函数计算一遍，需要调整为：AND ps1.dateline >= UNIX_TIMESTAMP('$todaytime')  

//更改后
SELECT ps. * , mf. * , m.username
FROM cdb_playsgame ps, cdb_memberfields mf, cdb_members m, (
SELECT ps1.uid, MAX( ps1.credits ) AS credits
FROM cdb_playsgame ps1
WHERE ps1.gametag = 'chuansj'
AND ps1.dateline >= UNIX_TIMESTAMP( '20081008' )
GROUP BY ps1.uid
) AS t
WHERE mf.uid = ps.uid
AND m.uid = ps.uid
AND ps.gametag = 'chuansj'
AND ps.credits = t.credits
AND ps.uid = t.uid
GROUP BY ps.uid
ORDER BY credits DESC
LIMIT 0 , 50

对于每个球员，找出球员号码，名字以及他所引起的罚款的号码，但只是针对那些至少有两次罚款的球员。
更紧凑的查询，在FROM子句中放置一个子查询。
SELECT PLAYERNO,NAME,NUMBER
FROM (SELECT PLAYERNO,NAME,
(SELECT COUNT(*)
FROM PENALTIES
WHERE PENALTIES.PLAYERNO =
PLAYERS.PLAYERNO)
AS NUMBER
FROM PLYERS) AS PN
WHERE NUMBER>=2

FROM子句中的子查询决定了每个球员的号码，名字和罚款的编号。接下来，这个号码变成了中间结果中的一列。然后指定了一个条件（NUMBER>=2）;最后，获取SELECT子句中的列。