给定时间内的 Postgres SUM

Posted 2023-02-16

【中文标题】给定时间内的 Postgres SUM

【英文标题】：Postgres SUM over given time

【发布时间】：2021-03-02 01:02:27

【问题描述】：

我正在尝试使用 Postgres 在给定时间内对谷歌分析数据求和。我不太确定如何解决分组问题。您在下面看到的查询每天返回一列。我想总结给定日期的所有点击次数。因此，查询应该为每个广告系列返回 1 行，每行有一列表示点击的总和。

SELECT
    sa.id AS salesforce_account_id,
    acp.campaignid,
    acp.campaignname,
    date,     
    SUM(clicks) as clicks
FROM
    adwords_campaign_performance acp
INNER JOIN salesforce_account sa ON
    sa.adwords_id = acp.adwords_customerid
WHERE acp.date >= '2020-10-01'
  AND acp.date <= '2020-10-03'
GROUP BY sa.id, acp.campaignid, acp.campaignname, date

我可以编写查询，使其返回整个月份的数字，如下所示：

SELECT
    sa.id AS salesforce_account_id,
    acp.campaignid,
    acp.campaignname,
     date_trunc('month', date) AS MONTH,
        SUM(clicks) as clicks
FROM
    adwords_campaign_performance acp
INNER JOIN salesforce_account sa ON
    sa.adwords_id = acp.adwords_customerid
WHERE MONTH = '2020-10-01 00:00:00'
  AND sa.id = 3148
GROUP BY sa.id, acp.campaignid, acp.campaignname, MONTH

任何帮助将不胜感激。谢谢！

【参考方案1】：

我认为您希望 generate_series() 生成该月的所有日期：

SELECT gs.date, sa.id AS salesforce_account_id, acp.campaignid, acp.campaignname,
       SUM(clicks) as clicks
FROM generate_series('2020-10-01'::date, '2020-10-31'::date, interval '1 day') gs(dte) LEFT JOIN
     adwords_campaign_performance acp
     ON acp.date >= gs.dte AND
        acp.date < gs.dte + INTERVAL '1 DAY' LEFT JOIN 
     salesforce_account sa
     ON sa.adwords_id = acp.adwords_customerid
GROUP BY gs.dte, sa.id, acp.campaignid, acp.campaignname, date

【参考方案2】：

您在下面看到的查询每天返回一列。我想总结给定日期的所有点击次数。因此，查询应该为每个广告系列返回 1 行，每行有一列表示点击的总和。

所以只需从select 和group by 子句中删除date：

SELECT
    sa.id AS salesforce_account_id,
    acp.campaignid,
    acp.campaignname,     
    SUM(clicks) as clicks
FROM adwords_campaign_performance acp
INNER JOIN salesforce_account sa ON sa.adwords_id = acp.adwords_customerid
WHERE acp.date >= '2020-10-01'
  AND acp.date <= '2020-10-03'
GROUP BY sa.id, acp.campaignid, acp.campaignname