mysql sum join 多表统计求和时不正确的问题

我在做订单统计的时候遇到问题，就是多表求和时发现不正确

我有下面两个表

--
-- 表的结构 `mobile_shops_orders`
--

CREATE TABLE IF NOT EXISTS `mobile_shops_orders` (
`oid` varchar(80) NOT NULL,
`userid` int(10) NOT NULL,
`pid` int(10) NOT NULL DEFAULT '0',
`paytype` tinyint(2) NOT NULL DEFAULT '1',
`cartcount` int(10) NOT NULL DEFAULT '0',
`dprice` float(13,2) NOT NULL DEFAULT '0.00',
`price` float(13,2) NOT NULL DEFAULT '0.00',
`priceCount` float(13,2) NOT NULL,
`state` tinyint(1) NOT NULL DEFAULT '0',
`ip` char(15) NOT NULL DEFAULT '',
`stime` int(10) NOT NULL DEFAULT '0',
`rebateCount` float NOT NULL COMMENT '累计优惠价格',
KEY `stime` (`stime`),
KEY `userid` (`userid`),
KEY `oid` (`oid`)
) ENGINE=MyISAM DEFAULT CHARSET=utf8;

--
-- 转存表中的数据 `mobile_shops_orders`
--

INSERT INTO `mobile_shops_orders` (`oid`, `userid`, `pid`, `paytype`, `cartcount`, `dprice`, `price`, `priceCount`, `state`, `ip`, `stime`, `rebateCount`) VALUES
('2015031710015621508', 1, 1, 1, 8, 10.00, 728.00, 728.00, 0, '127.0.0.1', 1426561489, 10),
('2015031711044980920', 1, 1, 1, 19, 10.00, 855.50, 865.50, 0, '127.0.0.1', 1426581574, 0),
('2015031716393491386', 1, 1, 1, 8, 10.00, 600.00, 610.00, 0, '127.0.0.1', 1426588329, 0),
('2015031809153611521', 2, 1, 1, 6, 10.00, 322.00, 332.00, 0, '127.0.0.1', 1426641353, 0),
('2015031809155386138', 2, 1, 1, 3, 10.00, 30.00, 40.00, 0, '127.0.0.1', 1426641377, 0);

--
-- 数据库: `mobile`
--

-- --------------------------------------------------------

--
-- 表的结构 `mobile_shops_products`
--

CREATE TABLE IF NOT EXISTS `mobile_shops_products` (
`aid` mediumint(8) NOT NULL DEFAULT '0',
`oid` varchar(80) NOT NULL DEFAULT '',
`userid` int(10) NOT NULL,
`title` varchar(80) NOT NULL DEFAULT '',
`price` float(13,2) NOT NULL DEFAULT '0.00',
`buynum` int(10) NOT NULL DEFAULT '9',
KEY `oid` (`oid`),
KEY `userid` (`userid`)
) ENGINE=MyISAM DEFAULT CHARSET=utf8;

--
-- 转存表中的数据 `mobile_shops_products`
--

INSERT INTO `mobile_shops_products` (`aid`, `oid`, `userid`, `title`, `price`, `buynum`) VALUES
(8, '2015031710015621508', 1, '盛盈汇贵妃鸡 光鸡约3.5斤', 139.00, 5),
(9, '2015031710015621508', 1, '紫玉淮山 5斤/份', 11.00, 3),
(14, '2015031711044980920', 1, '紫玉淮山 5斤/份aaaaaaaaa', 12.50, 7),
(8, '2015031711044980920', 1, '盛盈汇贵妃鸡 光鸡约3.5斤', 139.00, 5),
(9, '2015031711044980920', 1, '紫玉淮山 5斤/份', 11.00, 3),
(10, '2015031711044980920', 1, '盛盈汇贵妃鸡 光鸡约', 10.00, 4),
(8, '2015031716393491386', 1, '盛盈汇贵妃鸡 光鸡约3.5斤', 139.00, 4),
(9, '2015031716393491386', 1, '紫玉淮山 5斤/份', 11.00, 4),
(8, '2015031809153611521', 2, '盛盈汇贵妃鸡 光鸡约3.5斤', 139.00, 2),
(9, '2015031809153611521', 2, '紫玉淮山 5斤/份', 11.00, 4),
(10, '2015031809155386138', 2, '盛盈汇贵妃鸡 光鸡约', 10.00, 3);

平时的统计是这样用的， userid 是表示当前用户

SELECT oid , SUM( priceCount ) AS count_price

FROM mobile_shops_orders AS s

WHERE s.userid = '2'

oid	count_price
2015031809153611521	372.00

这里的值372是我们想要的

多表查询时统计时是下面这样的

SELECT s.oid,p.oid  AS poid,  SUM(priceCount) AS count_price
FROM mobile_shops_orders AS s
LEFT JOIN mobile_shops_products AS p ON p.oid = s.oid
WHERE s.userid =  '2'

oid	poid	count_price
2015031809153611521	2015031809153611521	704.00

但是我们发现，这个数据是错误的

于是我想到了用 distinct 但是发现用不了

SELECT s.oid,p.oid  AS poid,  SUM( priceCount) AS count_price,DISTINCT  poid
FROM mobile_shops_orders AS s
LEFT JOIN mobile_shops_products AS p ON p.oid = s.oid
WHERE s.userid =  '2'

<span style="font-family: Arial, Helvetica, sans-serif;">--这个语句是错的</span>

于是又想到 group by

SELECT s.oid,p.oid  AS poid,  SUM( priceCount) AS count_price
FROM mobile_shops_orders AS s
LEFT JOIN mobile_shops_products AS p ON p.oid = s.oid
WHERE s.userid =  '2'
GROUP BY poid

出来的结果是下面这样的

oid	poid	count_price
2015031809153611521	2015031809153611521	664.00
2015031809155386138	2015031809155386138	40.00

发现统计只是分组统计，没达到所有行统计，而且统计的数据不正确

于是我写了下面的语句来查看结果,加多了一个count函数查看

SELECT s.oid,p.oid  AS poid,  SUM( priceCount) AS count_price, COUNT(p.oid) FROM mobile_shops_orders AS s
LEFT JOIN mobile_shops_products AS p ON p.oid = s.oid
WHERE s.userid =  '2'
GROUP BY poid

出来的结果是下面的，一看就知道是什么原因了

oid	poid	count_price	COUNT(p.oid)
2015031809153611521	2015031809153611521	664.00	2
2015031809155386138	2015031809155386138	40.00	1

最后我们的语句应该是下面这个的，join 后面这个是重点，意思是返回不重复的数据

SELECT s.oid, p.oid AS poid, SUM(priceCount) AS count_price 

FROM mobile_shops_orders AS s 

<span style="font-family: Arial, Helvetica, sans-serif;">LEFT JOIN  (select distinct  oid from  mobile_shops_products)   AS  p </span>

<span style="font-family: Arial, Helvetica, sans-serif;">ON p.oid = s.oid </span>

<span style="font-family: Arial, Helvetica, sans-serif;">WHERE s.userid =  '2'</span>

<span style="font-family: Arial, Helvetica, sans-serif;">
</span>

oid	poid	count_price
2015031809153611521	2015031809153611521	372.00

这里有几个要注意的问题，一是，group by 是用于分组统计，如果要所有行统计不能用他

二需要使用子句先排除重复行，否则也会出得统计不正确的问题

还有，看网上说 返回 null 时sum等统计函数是不正确的

oid	poid	count_price	COUNT(p.oid)
2015031809153611521	2015031809153611521	664.00	2
2015031809155386138	2015031809155386138	40.00	1

oid	poid	count_price
2015031809153611521	2015031809153611521	664.00
2015031809155386138	2015031809155386138	40.00

不过我们建表是都应该设置为not null 这样就不用担心这个问题了，而且还能加快访问