SQL Query按类别分组

Question

Using SQL Server 2016

I have the following table:

Action Id   Epic    Due Date    First Name
Action 1    Epic 1  27-Feb-19   Tom
Action 2    Epic 2  28-Feb-19   John
Action 3    Epic 3  1-Mar-19    Ana
Action 4    Epic 3  15-Apr-19   Ana

Is it possible to design a query with the following output?

Action/Epic   Due Date  First Name   Type
Epic 1                               Epic
Action 1      27-Feb-19 Tom          Action
Epic 2                               Epic
Action 2      28-Feb-19 John         Action
Epic 3                               Epic
Action 3      1-Mar-19  Ana          Action
Action 4      15-Apr-19 Ana          Action

Answer 1

Yeah, you can do it this way (one of the many ways, I suppose):

Table

create table test (
    actionid varchar(100),
    epic varchar(100),
    duedate varchar(100),
    firstname varchar(100)
);

insert into test values
('Action 1', 'Epic 1', '27-Feb-19', 'Tom'),
('Action 2', 'Epic 2', '28-Feb-19', 'John'),
('Action 3', 'Epic 3', '1-Mar-19', 'Ana'),
('Action 4', 'Epic 3', '15-Apr-19', 'Ana');

Query

with data as (
    select 
        row_number() over(order by actionid) as sr,
        *
    from test
),
compiled as (
    select min(sr) as sr, epic as actionid, '' as epic, '' as duedate, '' as firstname, 'Epic' as type
    from data group by epic
    union all
    select *, 'Action' as type from data
)
select actionid, epic, duedate, firstname, type from compiled order by sr, actionid desc

Result

actionid    epic    duedate   firstname type
Epic 1                                  Epic
Action 1    Epic 1  27-Feb-19   Tom     Action
Epic 2                                  Epic
Action 2    Epic 2  28-Feb-19   John    Action
Epic 3                                  Epic
Action 3    Epic 3  1-Mar-19    Ana     Action
Action 4    Epic 3  15-Apr-19   Ana     Action

Example: https://rextester.com/JIN82148

Explanation

• Give each row a number sequentially after sorting by actionid
• For each row, pull out the Epic. If epic is duplicate, get the minimum serial number
• Get the action records
• Combine the data together
• Sort it by the serial number knowing that some Epic and Action will have the same serial number since they originated from the same row
• After serial number is sorted, sort by Epic so that E comes before A