pivot table 在 postgres 中有多个值列

Question

我需要一个查询来帮助我获取如下所述的数据

我有一个table如下

ID	date	STATUS	TIME09_10	TIME10_11	TIME11_12	TIME12_13
1	2021-09-01	RUN	30	60	45	0
2	2021-09-01	WALK	15	0	o	30
3	2021-09-01	STOP	15	0	15	30

我希望此数据采用以下格式。从上面的 table 中，值列名称必须替换为小时并与日期列连接。我真的很感激有人的帮助。我正在使用 postgres sql 数据库。

date & time	run	walk	stop
2021-09-01 09:00	30	15	15
2021-09-01 10:00	60	0	0
2021-09-01 11:00	45	0	15
2021-09-01 12:00	0	30	30

Answer 1

您可以使用crosstab(source_sql text, category_sql text)功能。您需要安装 tablefunc 扩展：

create extension if not exists tablefunc;

了解扩展程序 in the documentation.

该函数需要三列格式的数据 (row_name, category, value)。在这种情况下，它们是 date+time、status 和 duration.

select 
    date+ '8 hour'::interval+ '1 hour'::interval* i as hour,
    status,
    (array[time09_10, time10_11, time11_12, time12_13])[i] as duration
from my_table
cross join generate_series(1, 4) i


        hour         | status | duration
---------------------+--------+----------
 2021-09-01 09:00:00 | RUN    |       30
 2021-09-01 09:00:00 | WALK   |       15
 2021-09-01 09:00:00 | STOP   |       15
 2021-09-01 10:00:00 | RUN    |       60
 2021-09-01 10:00:00 | WALK   |        0
 2021-09-01 10:00:00 | STOP   |        0
 2021-09-01 11:00:00 | RUN    |       45
 2021-09-01 11:00:00 | WALK   |        0
 2021-09-01 11:00:00 | STOP   |       15
 2021-09-01 12:00:00 | RUN    |        0
 2021-09-01 12:00:00 | WALK   |       30
 2021-09-01 12:00:00 | STOP   |       30
(12 rows)

将查询作为函数的第一个参数传递：

select *
from crosstab(
    $source$
        select 
            date+ '8 hour'::interval+ '1 hour'::interval* i as hour,
            status,
            (array[time09_10, time10_11, time11_12, time12_13])[i] as duration
        from my_table
        cross join generate_series(1, 4) i
    $source$,
    $category$
        values('RUN'), ('STOP'), ('WALK')
    $category$
) as (hour timestamp, run int, stop int, walk int)

        hour         | run | stop | walk
---------------------+-----+------+------
 2021-09-01 09:00:00 |  30 |   15 |   15
 2021-09-01 10:00:00 |  60 |    0 |    0
 2021-09-01 11:00:00 |  45 |   15 |    0
 2021-09-01 12:00:00 |   0 |   30 |   30
(4 rows)

如果您不想使用该扩展程序，还有一个不错的选择。使用 jsonb 函数将第一个查询结果转换为预期输出：

select 
    hour, 
    (activities->>'RUN')::int as run,
    (activities->>'STOP')::int as stop,
    (activities->>'WALK')::int as walk
from (
    select hour, jsonb_object_agg(status, duration) as activities
    from (
        select 
            date+ '8 hour'::interval+ '1 hour'::interval* i as hour,
            status,
            (array[time09_10, time10_11, time11_12, time12_13])[i] as duration
        from my_table
        cross join generate_series(1, 4) i
        ) s
    group by hour
    ) s
order by hour

在Db<>fiddle.

中测试jsonb解决方案

pivot table 在 postgres 中有多个值列

pivot table with multiple value columns in postgres

sql

postgresql

pivot