使用MAX和GROUP BY与公用表表达式（CTE）Postgresql

Question

我尝试使用WITH子句获取数据透视表，但我仍然坚持如何返回多个或重复行使用MAX()。

这是我的原始查询：

SELECT b.detail_id, a.sampling_date, a.sampling_area, 
        b.sampling_point, b.sampling_type,
        b.ha_tpc, b.ha_entero, b.ha_ecoli, b.ha_salmonella
FROM tbl_header a
    JOIN tbl_detail b ON a.headerid = b.headerid
WHERE 
    a.sampling_date = '2016-12-09' AND 
    a.sampling_area = 'CMP' AND 
    (b.sampling_point ~* '.*(flex).*' OR b.sampling_point ~* '.*(HPM).*' OR b.sampling_point ~* '.*(fr).*') AND 
    (b.sampling_type = 'Personil' OR b.sampling_type = 'Equipment')

以下是结果：

----------------------------------------------------------------------------------------------------------------------------------------------------------------------------
detail_id  |   sampling_date  |   sampling_area    |       sampling_point           |   sampling_type   |    ha_tpc   |   ha_entero  |  ha_ecoli      |  ha_salmonella  |
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------
24243      |    2016-12-09    |     CMP            |    BOIL013 (OPERATOR ENFLEX)   |    Personil       |     500     |    50        |     Abs        |     Abs            |
24289      |    2016-12-09    |     CMP            |    MP115 (OPR ENFLEX)          |    Personil       |     300     |    50        |     Abs        |     Abs            |
24284      |    2016-12-09    |     CMP            |    WT033 (FR)                  |    Personil       |     250     |    50        |     Abs        |     Abs            |

使用这些记录，我尝试创建如下所示的数据透视查询：

WITH tmp_date AS (
    SELECT sampling_date.sampling_date::date AS sampling_date
        FROM generate_series(
            (( 
                SELECT min(tbl_header.sampling_date) AS min
                FROM tbl_header
            ))::timestamp with time zone, 
            (( 
                SELECT max(tbl_header.sampling_date) AS max
                FROM tbl_header
            ))::timestamp with time zone, '1 day'::interval
        ) sampling_date(sampling_date)
), 

tmp_detail AS (
    SELECT a.sampling_date, a.sampling_area, 
           b.detail_id, b.sampling_point, b.sampling_type,
           b.ha_tpc, b.ha_entero, b.ha_ecoli, b.ha_salmonella
    FROM tbl_header a
        JOIN tbl_detail b ON a.headerid = b.headerid
), 

resulttable AS (
    SELECT tmp_date.sampling_date, tmp_detail.sampling_point, tmp_detail.sampling_type, 

        -- Case of Filling Room
        CASE
            WHEN tmp_detail.sampling_point ~* '.*(fr).*' AND tmp_detail.sampling_type = 'Personil' THEN tmp_detail.sampling_point
            ELSE NULL
        END AS fr_name, 
        CASE
            WHEN tmp_detail.sampling_point ~* '.*(fr).*' AND tmp_detail.sampling_type = 'Personil' THEN tmp_detail.ha_tpc
            ELSE NULL
        END AS fr_tpc, 
        CASE
            WHEN tmp_detail.sampling_point ~* '.*(fr).*' AND tmp_detail.sampling_type = 'Personil' THEN tmp_detail.ha_entero
            ELSE NULL
        END AS fr_entero, 
        CASE
            WHEN tmp_detail.sampling_point ~* '.*(fr).*' AND tmp_detail.sampling_type = 'Personil' THEN tmp_detail.ha_ecoli
            ELSE NULL
        END AS fr_ecoli, 
        CASE
            WHEN tmp_detail.sampling_point ~* '.*(fr).*' AND tmp_detail.sampling_type = 'Personil' THEN tmp_detail.ha_salmonella
            ELSE NULL
        END AS fr_salmo, 

        -- Case of Hopper Auger Filling 
        CASE
            WHEN (tmp_detail.sampling_point ~* '.*(flex).*' OR tmp_detail.sampling_point ~* '.*(HPM).*') AND tmp_detail.sampling_type = 'Personil' THEN tmp_detail.sampling_point
            ELSE NULL
        END AS hpm_name, 
        CASE
            WHEN (tmp_detail.sampling_point ~* '.*(flex).*' OR tmp_detail.sampling_point ~* '.*(HPM).*') AND tmp_detail.sampling_type = 'Personil' THEN tmp_detail.ha_tpc
            ELSE NULL
        END AS hpm_tpc, 
        CASE
            WHEN (tmp_detail.sampling_point ~* '.*(flex).*' OR tmp_detail.sampling_point ~* '.*(HPM).*') AND tmp_detail.sampling_type = 'Personil' THEN tmp_detail.ha_entero
            ELSE NULL
        END AS hpm_entero, 
        CASE
            WHEN (tmp_detail.sampling_point ~* '.*(flex).*' OR tmp_detail.sampling_point ~* '.*(HPM).*') AND tmp_detail.sampling_type = 'Personil' THEN tmp_detail.ha_ecoli
            ELSE NULL
        END AS hpm_ecoli, 
        CASE
            WHEN (tmp_detail.sampling_point ~* '.*(flex).*' OR tmp_detail.sampling_point ~* '.*(HPM).*') AND tmp_detail.sampling_type = 'Personil' THEN tmp_detail.ha_salmonella
            ELSE NULL
        END AS hpm_salmo

    FROM tmp_date
        FULL JOIN tmp_detail ON tmp_date.sampling_date = tmp_detail.sampling_date
    WHERE 
        tmp_detail.sampling_area = 'CMP' AND 
        tmp_detail.sampling_type = 'Personil' AND 
        (
            tmp_detail.sampling_point ~* ANY ( VALUES  ('(fr)'), ('(flex)'), ('(HPM)') )
        )
)

SELECT 
    resulttable.sampling_date, 
    max(resulttable.sampling_type) AS sampling_type, 

    max(resulttable.fr_name) AS fr_name, 
    max(resulttable.fr_tpc) AS fr_tpc, 
    max(resulttable.fr_entero) AS fr_entero, 
    max(resulttable.fr_ecoli) AS fr_ecoli, 
    max(resulttable.fr_salmo) AS fr_salmonella, 

    max(resulttable.hpm_name) AS hopper_name, 
    max(resulttable.hpm_tpc) AS hopper_tpc, 
    max(resulttable.hpm_entero) AS hopper_entero, 
    max(resulttable.hpm_ecoli) AS hopper_ecoli, 
    max(resulttable.hpm_salmo) AS hopper_salmonella 

FROM resulttable
    GROUP BY resulttable.sampling_date ORDER BY resulttable.sampling_date;

我不知道为什么这些查询只返回1行，是因为我使用了GROUP BY？：

---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
sampling_date   | sampling_type  |         fr_name                | fr_tpc  |   fr_entero  |   fr_ecoli |   fr_salmonella  |         hopper_name          |   hopper_tpc  |   hopper_entero   |  hopper_ecoli  |   hopper_salmonella  |
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
2016-12-09     |     Personil    |        WT033 (FR)               |   250   |   50         |     Abs    |      Abs         |  BOIL013 (OPERATOR ENFLEX)  |     500       |     50            |     Abs        |      Abs             |

我实际上想得到如下结果：

                                 =====================         Case of Filling Room            =========================== | =====================            Case of Hopper Auger Filling                 ============================
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
sampling_date   | sampling_type  |         fr_name                | fr_tpc  |   fr_entero  |   fr_ecoli |   fr_salmonella  |         hopper_name          |   hopper_tpc  |   hopper_entero   |  hopper_ecoli  |   hopper_salmonella  |
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
2016-12-09     |     Personil    |        WT033 (FR)              |   250   |   50         |     Abs    |      Abs         |  BOIL013 (OPERATOR ENFLEX)   |     500       |     50            |     Abs        |      Abs             |
2016-12-09     |     Personil    |                                |         |              |            |                  |  MP115 (OPR ENFLEX)          |     300       |     50            |     Abs        |      Abs             |

任何帮助将不胜感激，谢谢。

Answer 1

您的问题是您的GROUP BY子句指定的标准不够具体，无法满足您的需求。由于您只有一个采样日期，因此只能输出一行。

阅读您的输出，我想您想要将hopper_name添加到GROUP BY条款中。但是，你并不是100％清楚自己要做什么，所以理解这个条款和其他替代方案可能很重要。

没有分组的GROUP BY

目前您正在做的事情。它基本上说，对于在此标准中唯一的每一行，返回一行。这可能就是你想要的。因此，请确保您在此处有GROUP BY完全符合您的要求。

替代方案：使用汇总和多维数据集进行分组设置

要注意的一个复杂解决方案是，PostgreSQL允许您根据条件设置多个分组集并进行汇总。如果ROLLUP无法让您到达目的地，那么可能想要查看CUBE或GROUP BY。但是，这会使检索数据的过程变得复杂，因为您希望获得分组列的位图。但是，由于您正在尝试使用数据透视表，因此可能需要注意进一步的步骤（以及指向其他步骤）。