我有一个包含几个字段的表格,但我只想计算state
字段所在的每日条目数。 '被拒绝'。字段createdAt
是带有时区的时间戳。
id | createdAt | state
------------------------------------------
1 | 2018-01-17 13:53:17.303+00 | rejected
2 | 2017-10-19 20:39:18.232+00 | approved
所以我想计算两个时间戳之间每天的行数。没有任何条目的天数应返回0或NULL。 我创建的查询返回正确的时间序列,但不知何故,我的连接或计数似乎是奇怪的。它返回了数千个条目,但通常只有0到20之间。(整个表少于3800个条目)
SELECT serie.day AS time, count(t."createdAt") AS created
FROM (
SELECT generate_series(
to_timestamp(1518248353)::date,
to_timestamp(1521469246)::date,
interval '1 day'
) AS day
FROM users t
) serie
LEFT JOIN users t ON t."createdAt"::date = serie.day::date
WHERE state ILIKE 'rejected'
GROUP BY serie.day
ORDER BY time;
错误的结果:
2018-02-10 00:00:00+00 11619
2018-02-11 00:00:00+00 23238
2018-02-12 00:00:00+00 27111
2018-02-13 00:00:00+00 19365
2018-02-14 00:00:00+00 15492
2018-02-15 00:00:00+00 34857
2018-02-16 00:00:00+00 34857
2018-02-17 00:00:00+00 15492
2018-02-18 00:00:00+00 46476
2018-02-19 00:00:00+00 15492
2018-02-21 00:00:00+00 27111
2018-02-22 00:00:00+00 19365
2018-02-23 00:00:00+00 11619
2018-02-24 00:00:00+00 3873
2018-02-25 00:00:00+00 3873
...
预期结果如下:
2018-02-10 00:00:00+00 3
2018-02-11 00:00:00+00 9
2018-02-12 00:00:00+00 0
2018-02-13 00:00:00+00 19
2018-02-14 00:00:00+00 15
2018-02-15 00:00:00+00 3
2018-02-16 00:00:00+00 7
...
答案 0 :(得分:1)
为什么要将生成的系列加入子查询中的用户表?我认为这是你的问题。通常,当您使用generate_series()
时,您只需提供开始日期时间,结束日期时间和间隔,然后将该系列生成为表格。
请改为尝试:
SELECT
serie.day AS time,
COUNT(t."createdAt") AS created
FROM (
SELECT date_series::date AS day
FROM generate_series(
to_timestamp(1518248353)::date,
to_timestamp(1521469246)::date,
'1 day'
) AS date_series
) AS serie
LEFT JOIN users t ON t."createdAt"::date = serie.day::date
AND state ILIKE 'rejected'
GROUP BY serie.day
ORDER BY time;