将 not in 替换为 left join答案

【问题标题】：Replacing not in with left join将 not in 替换为 left join
【发布时间】：2015-07-15 18:01:09
【问题描述】：

我希望更正 mySQL 数据库中左连接的语法。我的查询运行但很慢，我不确定我的左连接是否正确。我需要根据以前的成功选择 consumer_id。为了检查以前的成功，我必须在记录表和映射表上留下连接，以将映射 ID 链接到仅存在于 event_queue 表中的消费者 ID。非常感谢您的帮助。

第一个查询：

SELECT  
CEQ.consumer_ID
, F.client_license_ID
, F.footprint_ID
, F.event_token_ID
, EM.event_ID
, EM.mapping_ID em_ID
, EM.export_value campaign_number
, EM.export_value_2 sequence_number
, EM.export_value_3 campaign_number_2
, EM.export_value_4 sequence_number_2
, EM.export_value_5 ffs_event_id
, EM.export_value_6
, EM.export_value_7
, EM.export_value_8
, EM.export_value_9
, EM.export_value_10

FROM data_transfer.Event_Mappings EM
JOIN data_transfer.Event_Queue CEQ ON CEQ.event_ID = EM.event_ID
JOIN Footprints F ON CEQ.consumer_ID = F.consumer_ID
LEFT JOIN efn_data_transfer.CRM_Records LCR ON LCR.consumer_ID = CEQ.consumer_ID
WHERE EM.data_transfer_ID = 24
AND EM.mode = 'production'
AND EM.active_flag = 1
AND F.sample_flag = 0
AND CEQ.modify_DTM > (SELECT last_transfer FROM (SELECT CAST(DATE_SUB(start,INTERVAL 3 DAY) AS CHAR) last_transfer
FROM data_transfer.DT_Runs DTR
WHERE DTR.data_transfer_ID = 24
AND DTR.result = 1
AND DTR.mode = 'production'
ORDER BY DTR.dt_run_ID DESC
LIMIT 1)as D 
) 
GROUP BY
CEQ.consumer_ID
, F.client_license_ID
, F.footprint_ID
, F.event_token_ID
, EM.event_ID
, EM.mapping_ID 
, EM.export_value 
, EM.export_value_2 
, EM.export_value_3 
, EM.export_value_4 
, EM.export_value_5 
, EM.export_value_6
, EM.export_value_7
, EM.export_value_8
, EM.export_value_9
, EM.export_value_10

ORDER BY F.client_license_ID, CEQ.consumer_ID, EM.mapping_ID, CEQ.modify_DTM;

组合查询：

SELECT DISTINCT CAST(CEQ.consumer_ID AS CHAR) AS consumer_ID_1
FROM data_transfer.Event_Queue CEQ
WHERE 0=0
AND CEQ.consumer_ID NOT IN (SELECT RR.consumer_ID FROM data_transfer.Records RR
  JOIN data_transfer.Event_Mappings EM ON RR.event_mapping_ID = EM.event_mapping_ID
  WHERE 0=0
  AND RR.failure_code = 0
  AND RR.mode = 'production'
  AND RR.`ignore` = 0
  AND RR.response_code = 'SUCCESS'
  AND RR.data_transfer_ID = 24
  AND RR.consumer_ID = ?
  AND EM.event_ID = ?
)
AND CEQ.consumer_ID = ?

尝试的查询：

SELECT  
DISTINCT
EM.event_ID
, F.footprint_ID
, F.event_token_ID
FROM data_transfer.Event_Mappings EM
JOIN data_transfer.Event_Queue CEQ ON CEQ.event_ID = EM.event_ID
JOIN Footprints F ON CEQ.consumer_ID = F.consumer_ID
LEFT JOIN efn_data_transfer.CRM_Records LCR ON LCR.consumer_ID = CEQ.consumer_ID

LEFT JOIN (
SELECT DISTINCT CEQ.consumer_ID AS consumer_ID_1
FROM data_transfer.Event_Queue CEQ, data_transfer.Records RR
LEFT JOIN  data_transfer.Mappings EM ON EM.mapping_id = RR.mapping_id
WHERE RR.consumer_id = CEQ.consumer_id
  AND RR.failure_code = 0
  AND RR.mode = 'production'
  AND RR.`ignore` = 0
  AND RR.response_code = 'SUCCESS'
  AND RR.data_transfer_ID = 24

) AS RECORDS ON CEQ.consumer_id

WHERE EM.data_transfer_ID = 24
AND EM.mode = 'production'
AND EM.active_flag = 1
AND F.sample_flag = 0
AND CEQ.modify_DTM > (SELECT last_transfer FROM (SELECT CAST(DATE_SUB(start,INTERVAL 3 DAY) AS CHAR) last_transfer
FROM data_transfer.DT_Runs DTR
WHERE DTR.data_transfer_ID = 24
AND DTR.result = 1
AND DTR.mode = 'production'
ORDER BY DTR.dt_run_ID DESC
LIMIT 1) AS D 
) 
GROUP BY
CEQ.consumer_ID
, F.client_license_ID
, F.footprint_ID
, F.event_token_ID
, EM.event_ID
, EM.mapping_ID 
, EM.export_value 
, EM.export_value_2 
, EM.export_value_3 
, EM.export_value_4 
, EM.export_value_5 
, EM.export_value_6
, EM.export_value_7
, EM.export_value_8
, EM.export_value_9
, EM.export_value_10

ORDER BY F.client_license_ID, CEQ.consumer_ID, EM.mapping_ID, CEQ.modify_DTM;

【问题讨论】：

尝试消除 where 子句中的依赖子查询，并仅使用连接重写您的查询

标签： mysql left-join

【解决方案1】：

您的子查询不依赖于任何特定的内容，并且可以是它自己的查询，作为“FROM”子句的一部分。由于它仅通过限制返回 1 行，因此您无需将其加入任何内容。对于查询的其余部分，这只会产生一次日期计算值。

对于您的表，我提供以下建议的索引，以帮助针对相应的 WHERE 和 JOIN 标准进行优化。

Table           Index
Event_Mappings  ( data_transfer_id, active_flag, mode, Event_id )
Event_Queue     ( Event_id, consumer_ID, modify_DTM )
Footprints      ( consumer_ID, sample_flag, client_license_ID )
DT_Runs         ( data_transfer_id, result, mode, dt_run_id, start )
Records         ( data_transfer_id, failure_code, mode, ignore, response_code )

SELECT DISTINCT
      CEQ.consumer_ID, 
      F.client_license_ID, 
      F.footprint_ID, 
      F.event_token_ID, 
      EM.event_ID, 
      EM.mapping_ID em_ID, 
      EM.export_value campaign_number, 
      EM.export_value_2 sequence_number, 
      EM.export_value_3 campaign_number_2, 
      EM.export_value_4 sequence_number_2, 
      EM.export_value_5 ffs_event_id, 
      EM.export_value_6, 
      EM.export_value_7, 
      EM.export_value_8, 
      EM.export_value_9, 
      EM.export_value_10
   FROM 
      data_transfer.Event_Mappings EM
         JOIN data_transfer.Event_Queue CEQ 
            ON EM.event_ID = CEQ.event_ID 
            JOIN Footprints F 
               ON CEQ.consumer_ID = F.consumer_ID
              AND F.sample_flag = 0
            LEFT JOIN efn_data_transfer.CRM_Records LCR 
               ON CEQ.consumer_ID = LCR.consumer_ID
            LEFT JOIN ( SELECT DISTINCT 
                              CEQ.consumer_ID AS consumer_ID_1
                           FROM 
                              data_transfer.Records RR
                                 LEFT JOIN data_transfer.Mappings EM 
                                    ON RR.mapping_id = EM.mapping_id 
                                 JOIN data_transfer.Event_Queue CEQ
                                    ON RR.consumer_id = CEQ.consumer_id
                           WHERE 
                                  RR.data_transfer_ID = 24
                              AND RR.failure_code = 0
                              AND RR.mode = 'production'
                              AND RR.`ignore` = 0
                              AND RR.response_code = 'SUCCESS' ) AS RECORDS 
               ON CEQ.consumer_id = RECORDS.consumer_id_1,
      ( SELECT 
              CAST(DATE_SUB(start,INTERVAL 3 DAY) AS CHAR) last_transfer
           FROM 
              data_transfer.DT_Runs DTR
           WHERE 
                  DTR.data_transfer_ID = 24
              AND DTR.result = 1
              AND DTR.mode = 'production'
           ORDER BY 
              DTR.dt_run_ID DESC 
           LIMIT 1 ) as LT
   WHERE 
          EM.data_transfer_ID = 24
      AND EM.mode = 'production'
      AND EM.active_flag = 1
      AND CEQ.modify_DTM > LT.Last_Transfer
      AND RECORDS.consumer_id_1 IS NULL
   ORDER BY 
      F.client_license_ID, 
      CEQ.consumer_ID, 
      EM.mapping_ID, 
      CEQ.modify_DTM;

修改为包含您的 RECORDS 别名左连接...然后，查找 RECORDS“consumer_id_1”列为 NULL（因此 NOT IN）

【讨论】：