如何加入一个表只有结果有特定参数

Question

我需要从已完成2个或多个类给下面的表格结构中的所有用户USER_NAME ...

table users 
=============================== 
user_id  | user_name 
=============================== 
1234   completed_two 
1235   completed_cero 
1236   completed_one 
1237   completed_two_no_survey 

table class_registration 
======================================================================== 
reg_id | class_id | class_type  | user_id  | attendance 
======================================================================== 
1   1   stress    1234   1 
2   1   stress    1236   1 
3   2   who_cares    1234   1 
4   1   stress    1235   0 
5   5   nutrition    1236   1 
6   9   who_cares_2   1237   1 
7   10   return_of_the_care 1237   1 

table surveys 
================================================================== 
survey_id | user_id | class_id |  survey_type 
================================================================== 
1    1234   1   'Pre Workshop Survey' 
2    1236   1   'Pre Workshop Survey' 
3    1235   1   'Pre Workshop Survey' 
4    1236   2   'Pre Workshop Survey' 
5    1234   1   'Post Workshop Survey' 
6    1236   2   'Post Workshop Survey' 

在此示例中，USER_ID 1234有一类who_cares，考勤次数设为1，所以这个计数对于1个班级。压力类型有额外的要求，主要是在调查表中必须完成一项后期调查。具有user_id的用户也参加了2个班，但对于压力班而言，缺少调查。 user_id为1235的用户注册了压力班，但未能出席。

结果应该包含用户名称completed_two和completed_two_no_survey，因为这2个用户总共有2个班级参加了即使寿，用户1237没有调查。只有class_type压力或class_type营养才有调查。

我甚至不知道从哪里开始做这样的查询......我确定有一个案例我可以在那里展开，但我不知道如何使它只有压力和根据调查表检查营养。我想是这样的，在某处有一个CASE子句...

SELECT users.user_name FROM class_registration 
JOIN users ON users.user_id = class_registration.user_id 
WHERE class_registration.attendance = 1 
GROUP BY class_registration.user_id, users.user_name 
HAVING COUNT(class_registration.attendance) >= 2 

同样，我需要的是检查调查表后车间的调查，当且仅当class_type是压力或营养。

Answer 1

declare @users table (user_id int, user_name varchar(50)) 
insert into @users values 
(1234,   'completed_two'), 
(1235,   'completed_cero'), 
(1236,   'completed_one'), 
(1237,   'completed_two_no_survey') 

declare @class_registration table 
(reg_id int, class_id int, class_type varchar(50), user_id int, attendance bit) 
insert into @class_registration values 
(1,   1,   'stress',    1234,   1), 
(2,   1,   'stress',    1236,   1), 
(3,   2,   'who_cares',    1234,   1), 
(4,   1,   'stress',    1235,   0), 
(5,   5,   'nutrition',    1236,   1), 
(6,   9,   'who_cares_2',   1237,   1), 
(7,   10,   'return_of_the_care', 1237,   1) 

declare @surveys table 
(survey_id int, user_id int, class_id int, survey_type varchar(50)) 
insert into @surveys values 
(1,    1234,   1,   'Pre Workshop Survey'), 
(2,    1236,   1,   'Pre Workshop Survey'), 
(3,    1235,   1,   'Pre Workshop Survey'), 
(4,    1236,   2,   'Pre Workshop Survey'), 
(5,    1234,   1,   'Post Workshop Survey'), 
(6,    1236,   2,   'Post Workshop Survey') 


select u.user_id, u.user_name 
from @users u 
    inner join 
    ( 
     select c.user_id 
     from @class_registration as c 
     inner join @surveys as s 
      on c.class_id = s.class_id and 
      c.user_id = s.user_id 
     where 
     class_type in ('nutrition', 'stress') and 
     survey_type = 'Post Workshop Survey' and 
     attendance = 1 
     union all 
     select c.user_id 
     from @class_registration as c 
     where 
     class_type not in ('nutrition', 'stress') and 
     attendance = 1 
    ) as cc 
    on u.user_id = cc.user_id 
group by u.user_id, u.user_name 
having count(*) > 1 

Answer 2

希望我正确地解释你的问题 - 你的一些例子是缺少实际的用户ID ...

select u.user_name, 
     cr.class_type, 
     count(*) 
from users     u, 
     class_registration  cr 
where u.user_id = cr.user_id 
and  cr.class_type in ('stress', 'nutrition') 
and  cr.attendance = 1 
and  user_id not in 
(select user_id 
from surveys 
where user_id = u.user_id 
and  class_id = cr.class_id) 
having count(*) > 1