Question

我有一个表，其中包含match的列表：

id        |  datetime 
 2864049     2019-04-03 00:00:00
 2864050     2019-04-03 00:00:00
 2864051     2019-04-03 00:00:00

基本上，我必须检查它是否存在以及是否需要更新，为此，我这样做了：

SELECT mm.match_id
FROM (SELECT 2864049 as match_id, CAST("2019-04-03 00:00:00" AS DATETIME) AS match_datetime UNION ALL 
      SELECT 2864050 as match_id, CAST("2019-04-03 00:00:00" AS DATETIME) AS match_datetime UNION ALL 
      SELECT 2864051 as match_id, CAST("2019-04-03 00:00:00" AS DATETIME) AS match_datetime
     ) mm LEFT JOIN
     `match` m
     on m.id = mm.match_id AND m.datetime <> mm.match_datetime
WHERE m.id IS NULL

问题在于此查询不应返回任何结果，因为所有三个match都存在，并且datetime字段等于{{ 1}}表。

仅当matches不同或mm不存在时，查询才需要返回结果。

我做错了什么？

Answer 1

将其他日期移至where子句

实时测试：http://sqlfiddle.com/#!9/ee7ffb/1

CREATE TABLE `match`
    (`id` int, `datetime` datetime)
;

INSERT INTO `match`
    (`id`, `datetime`)
VALUES
    (2864049, '2019-04-03 00:00:00'),
    (2864050, '2019-04-03 00:00:00'),
    (2864051, '2019-04-03 00:00:00'),
    (2864052, '2019-04-03 00:00:00'),
    (2864053, '2019-04-03 00:00:00'),    
    (2864054, '2019-04-03 00:00:00')        
;



SELECT m.id, `datetime`
FROM 
`match` `m`
left join  
     (SELECT 2864049 as match_id, CAST("2019-04-03 00:00:00" AS DATETIME) AS match_datetime UNION ALL 
      SELECT 2864050 as match_id, CAST("2019-04-03 00:00:00" AS DATETIME) AS match_datetime UNION ALL 
      SELECT 2864051 as match_id, CAST("2019-04-03 00:00:00" AS DATETIME) AS match_datetime union all
      SELECT 2864052 as match_id, CAST("2019-04-04 00:00:00" AS DATETIME) AS match_datetime union all
      SELECT 2864054 as match_id, CAST("2019-04-03 00:00:00" AS DATETIME) AS match_datetime  
     ) mm 
     on m.id = mm.match_id 
WHERE 
mm.match_id IS NULL -- not existing
or m.datetime <> mm.match_datetime -- move to where clause

输出：

|      id |             datetime |
|---------|----------------------|
| 2864052 | 2019-04-03T00:00:00Z |
| 2864053 | 2019-04-03T00:00:00Z |

2864052出现了，尽管它具有匹配的ID，因为它的日期时间与mm的日期时间不同

2864053出现，因为它不存在于毫米上

Answer 2

尝试不使用强制转换，而是使用STR_TO_DATE转换日期

SELECT mm.match_id
FROM (SELECT 2864049 as match_id, str_to_date("2019-04-03 00:00:00", '%Y-%m-%d %T' ) AS match_datetime 
      UNION ALL 
      SELECT 2864050 , str_to_date("2019-04-03 00:00:00", '%Y-%m-%d %T' )  
      UNION ALL 
      SELECT 2864051 , str_to_date("2019-04-03 00:00:00", '%Y-%m-%d %T' )
     ) mm 
LEFT JOIN  `match` m on m.id = mm.match_id AND m.datetime <> mm.match_datetime
WHERE m.id IS NULL

但是您应该检查一下逻辑，因为您正在寻找不匹配的值的左连接...因此，如果mm表中的行与m表中的行不匹配，则所有这些行都有结果ID为null可能是您需要左连接来匹配行，例如：

SELECT mm.match_id
FROM (SELECT 2864049 as match_id, str_to_date("2019-04-03 00:00:00", '%Y-%m-%d %T' ) AS match_datetime 
      UNION ALL 
      SELECT 2864050 , str_to_date("2019-04-03 00:00:00", '%Y-%m-%d %T' )  
      UNION ALL 
      SELECT 2864051 , str_to_date("2019-04-03 00:00:00", '%Y-%m-%d %T' )
     ) mm 
LEFT JOIN  `match` m on m.id = mm.match_id AND m.datetime =  mm.match_datetime
WHERE m.id IS NULL

无法返回符合条件的记录

2 个答案: