加入2个多对多表，得到3个相邻的boolean列

Question

Let's say there are three tables A , B and C where

Table A is:

╔═══════╗
║ col_1 ║
╠═══════╣
║ a     ║
║ b     ║
║ c     ║
╚═══════╝

Table B is:

╔═══════╗
║ col_2 ║
╠═══════╣
║ o     ║
║ p     ║
║ q     ║
╚═══════╝

Table C is:

╔═══════╗
║ c_col ║
╠═══════╣
║ x     ║
║ y     ║
║ z     ║
╚═══════╝

Let's say there is a many-to-many relationship between table A and table C via a mapping table A_to_C shown below:

╔═══════╦═══════╗
║ col_1 ║ c_col ║
╠═══════╬═══════╣
║ a     ║ x     ║
║ a     ║ y     ║
║ b     ║ y     ║
║ b     ║ z     ║
║ c     ║ x     ║
╚═══════╩═══════╝

Also there is a many-to-many relationship between table B and table C via a mapping table B_to_C shown below:

╔═══════╦═══════╗
║ col_2 ║ c_col ║
╠═══════╬═══════╣
║ o     ║ y     ║
║ o     ║ z     ║
║ p     ║ x     ║
║ q     ║ x     ║
║ q     ║ y     ║
╚═══════╩═══════╝

For value 'b' in col_1 and value 'q' in col_2 I need to get the result as:

╔═══════╦═══════╦════════╗
║ c_col ║ col_1 ║ col_2  ║
╠═══════╬═══════╬════════╣
║ x     ║ NULL  ║ q      ║
║ y     ║ b     ║ q      ║
║ z     ║ b     ║ NULL   ║
╚═══════╩═══════╩════════╝

which can later be transformed into:

╔═══════╦═══════╦═══════╗
║ c_col ║ col_1 ║ col_2 ║
╠═══════╬═══════╬═══════╣
║ x     ║ False ║ True  ║
║ y     ║ True  ║ True  ║
║ z     ║ True  ║ False ║
╚═══════╩═══════╩═══════╝

I asked a similar question here but in this case due to many to many relationships with duplicate column values I am not able to achieve the above result without duplicate columns using the solution in the above link.

Answer 1

You can use the following query:

select c.c_col, ac.col_1 is not null, bc.col_2 is not null
from C c
left join (select * from A_to_C where col_1 = 'b') ac on c.c_col = ac.c_col
left join (select * from B_to_C where col_2 = 'b') bc on c.c_col = bc.c_col
where (ac.col_1, bc.col2) = ('b', 'q')

Answer 2

You don't need the two LEFT JOINS and the IS NOT NULL; EXISTS() already yields a boolean:

---

SELECT C.c_col
        , EXISTS( SELECT *
                FROM A_to_C ac
                WHERE ac.c_col = C.c_col AND ac.col_1 = 'b'
                ) AS col_1
        , EXISTS( SELECT *
                FROM B_to_C bc
                WHERE bc.c_col = C.c_col AND bc.col_2 = 'q'
                ) AS col_2
FROM C
        ;

Answer 3

You can generate this result with a series of LEFT JOIN s, placing your col_1 and col_2 filters in the JOIN conditions:

SELECT C.c_col, ac.col_1, bc.col_2
FROM C
LEFT JOIN A_to_C ac ON ac.c_col = C.c_col AND ac.col_1 = 'b'
LEFT JOIN B_to_C bc ON bc.c_col = C.c_col AND bc.col_2 = 'q'

Output:

c_col   col_1   col_2
x       (null)  q
y       b       q
z       b       (null)

Demo on SQLFiddle

Those values can then be converted into booleans by testing they are not NULL :

SELECT C.c_col, ac.col_1 IS NOT NULL AS col_1, bc.col_2 IS NOT NULL AS col_2
FROM C
LEFT JOIN A_to_C ac ON ac.c_col = C.c_col AND ac.col_1 = 'b'
LEFT JOIN B_to_C bc ON bc.c_col = C.c_col AND bc.col_2 = 'q'

Output:

c_col   col_1   col_2
x       false   true
y       true    true
z       true    false

Demo on SQLFiddle