如何将这些额外的 job_title 行转置到它们自己的列中？

Question

我使用的软件是

snowflake

我有一张公司员工工作分配表。他们可以同时活跃地从事多项工作（

end_date

中为空）。我想将其合并为 1 行：

EMPLOYEE_ID	START_DATE	END_DATE	工作_标题
1442	7/30/24		导师
1442	7/30/24		导师
1442	6/28/24		教学专家
1442	5/1/24	6/27/24	教学专家
1442	12/16/21	7/29/24	导师
1442	12/16/21		首席导师
1442	12/16/21	7/29/24	导师

如果员工在

end_date

字段中有任何空值，那么我只想检索不同的 job_titles （将前 2 行消除为 1，因为这两个职位在同一开始日期） 1-5 按降序排列，基于

start_date

，如下所示：

EMPLOYEE_ID	职位_标题_1	工作_标题_2	工作_标题_3	工作_标题_4	工作_标题_5
1442	导师	教学专家	首席导师

现在假设该员工目前没有活跃的工作，表格将如下所示：

EMPLOYEE_ID	START_DATE	END_DATE	工作_标题
1442	5/1/24	6/27/24	教学专家
1442	12/16/21	7/29/24	导师
1442	12/16/21	7/29/24	导师

在这种情况下，我希望表格看起来像这样：

EMPLOYEE_ID	职位_标题_1	工作_标题_2	工作_标题_3	工作_标题_4	工作_标题_5
1442	教学专家	导师

这是我正在使用的查询，它可以工作，但它没有按 desc start_date 顺序对 job_title 1-5 列进行排序：

WITH job_position_info_ADP AS (
    SELECT 
        'ADP' AS source, 
        CAST(w.associate_oid AS STRING) AS worker_id,
        CAST(w.id AS STRING) AS Employee_ID,
        TO_CHAR(wah._fivetran_start, 'MM/DD/YY') AS start_date,
        CASE 
            WHEN wah._fivetran_active = TRUE THEN NULL
            ELSE TO_CHAR(wah._fivetran_end, 'MM/DD/YY')
        END AS end_date,
        wah.job_title AS Job_Title,
        ROW_NUMBER() OVER (PARTITION BY CAST(w.id AS STRING) ORDER BY wah._fivetran_start DESC) AS rn
    FROM
        prod_raw.adp_workforce_now.worker w
    JOIN 
        prod_raw.adp_workforce_now.worker_report_to AS wr 
        ON w.id = wr.worker_id
    JOIN 
        prod_raw.adp_workforce_now.work_assignment_history AS wah 
        ON w.id = wah.worker_id
),
recent_jobs_with_null_end AS (
    SELECT 
        Employee_ID,
        Job_Title,
        ROW_NUMBER() OVER (PARTITION BY Employee_ID ORDER BY start_date DESC) AS rn
    FROM 
        job_position_info_ADP
    WHERE 
        end_date IS NULL
),
recent_jobs_all AS (
    SELECT 
        Employee_ID,
        Job_Title,
        ROW_NUMBER() OVER (PARTITION BY Employee_ID ORDER BY start_date DESC) AS rn
    FROM 
        job_position_info_ADP
)
SELECT
    Employee_ID,
    MAX(CASE WHEN rn = 1 THEN Job_Title END) AS Job_Title_1,
    MAX(CASE WHEN rn = 2 THEN Job_Title END) AS Job_Title_2,
    MAX(CASE WHEN rn = 3 THEN Job_Title END) AS Job_Title_3,
    MAX(CASE WHEN rn = 4 THEN Job_Title END) AS Job_Title_4,
    MAX(CASE WHEN rn = 5 THEN Job_Title END) AS Job_Title_5
FROM (
    SELECT * FROM recent_jobs_with_null_end
    UNION ALL
    SELECT * FROM recent_jobs_all
    WHERE Employee_ID NOT IN (SELECT Employee_ID FROM recent_jobs_with_null_end)
) AS combined
WHERE 
    Employee_ID = '1442'
GROUP BY
    Employee_ID;

Answer 1

这是 SQL Server 中的可复制代码，包括架构和数据。

我们的想法是重新组织 work_assignment_history 表以按日期对职位进行排序，而不包括当前职位（END_DATE 为空）。

然后，在job_unique中，删除重复项，并重新计算行号。发现去重后初始行号不一样，而且日期转成字符导致计算新行号变得困难。

最后，应用PIVOT操作将行转换为列。

;WITH job_position_info_ADP AS (
    SELECT 
        CAST(wah.EMPLOYEE_ID AS varchar(10)) AS Employee_ID,
        wah.START_DATE AS start_date,
        wah.job_title AS Job_Title,
        ROW_NUMBER() OVER (PARTITION BY wah.EMPLOYEE_ID ORDER BY wah.START_DATE DESC) AS rn
    FROM
        [development].[dbo].work_assignment_history AS wah
    WHERE 
        wah.END_DATE is not null
),
job_unique as 
(
    select Employee_ID, start_date, Job_Title, max(rn) as rn,
        ROW_NUMBER() OVER (PARTITION BY EMPLOYEE_ID ORDER BY start_date DESC) AS new_rn
    from job_position_info_ADP
    group by Employee_ID, start_date, Job_Title
)
SELECT Employee_ID, [1], [2], [3], [4],[5]
FROM
(
    SELECT Employee_ID, job_title,new_rn
    FROM job_unique
) AS SourceTable
PIVOT
(
    MAX(job_title)
    FOR new_rn IN ([1], [2], [3], [4],[5])
) AS PivotTable;

输出

Employee_ID  1                              2       3       4      5
1442         Especialista en Instrucción    Tutor   NULL    NULL    NULL

我希望它很容易翻译成Snowflake，因为大多数SQL语言都具有语法更改的PIVOT操作。

架构

CREATE TABLE [dbo].[work_assignment_history ](
    [EMPLOYEE_ID] [smallint] NOT NULL,
    [START_DATE] [date] NOT NULL,
    [END_DATE] [date] NULL,
    [JOB_TITLE] [nvarchar](50) NOT NULL
) ON [PRIMARY]
GO

数据

EMPLOYEE_ID;START_DATE;END_DATE;JOB_TITLE
1442;7/30/24;;Tutor
1442;7/30/24;;Tutor
1442;6/28/24;;Especialista en Instrucción
1442;5/1/24;6/27/24;Especialista en Instrucción
1442;12/16/21;7/29/24;Tutor
1442;12/16/21;;Instructor Principal
1442;12/16/21;7/29/24;Tutor

注意 由于某种原因，当将 HTML 数据导入 SQL Server 时，它被翻译成西班牙语。抱歉。