Exist and Not Exists Query Misunderstanding?

问题

Consider the following schema of a product database:

Parts (pid, pname)
Suppliers (sid, sname)
Catalog (sid, pid)

sid in Catalog is a foreign key and references Suppliers and pid in Catalog is a foreign key and references the Parts table. (s1, p1) in Catalog table shows supplier s1 produces p1.

If I want find details of Parts that are not supplied by some suppliers, how we can do this? I read some parts as follows:

SELECT *
FROM parts P
WHERE ....... (SELECT S.sid 
               FROM suppliers 
                WHERE .....
                       (SELECT * 
                        FROM catalog C 
                        WHERE s.sid = C.sid AND P.pid = C.pid))

My problem is via - why at first? We use exist and in second ... we use not exist?

回答1:

If I want find details of Parts that are not supplied by some suppliers, how we can do this?

If you want "details for parts where there exists a supplier that doesn't supply that part" then your query, with EXISTS then NOT EXISTS, is correct.

My problem is via - why at first? We use exist and in second ... we use not exist?

I will explain why "details for parts where there exists a supplier that doesn't supply that part" can give your query. (With EXISTS then NOT EXISTS.)

A table holds that rows that make some predicate (statement template) into a true proposition (statement):

Parts "part P.pid is named P.pname"
Suppliers "supplier S.sid is named S.sname"
Catalog "supplier C.sid supplies part C.pid"

You need to express the predicate for your result. You want the P.pid-P.pname rows that make a true statement from this:

    "part P.pid is named P.pname"
AND EXISTS S.sid ["supplier S.sid is named something"
        AND NOT "supplier S.sid supplies part P.pid"]

But we must express this predicate in terms of the given predicates (plus conditions and logic operators) so that the DBMS can calculate the rows that satisfy it:

    "part P.pid is named P.pname"
AND EXISTS S.sid [EXISTS S.sname "supplier S.sid is named S.sname"
        AND NOT EXISTS C.sid, C.pid [
                "supplier C.sid supplies part P.pid" AND C.sid = S.sid AND C.pid = P.pid]]

Now convert to SQL:

initial predicate of T becomes FROMT
AND predicate of T becomes JOINT
AND condition becomes WHEREcondition or ONcondition
outermost EXISTS dropped columns becomes SELECTkept columns
other EXISTS all columns of T [predicate of T] becomes EXISTS (T)

SELECT *
FROM Parts P
WHERE EXISTS (SELECT S.sid FROM Suppliers S
        WHERE NOT EXISTS (SELECT 1 FROM Catalog C
               WHERE C.sid = S.sid AND C.pid = P.pid))

PS Your question's original "details of parts that are not supplied by some suppliers" is unclear/ambiguous. It could mean P.id-P.pname pairs where:

"the parts are not supplied by all the suppliers" (the interpretation agreeing with your SQL)
"the parts are not not supplied (by any of the suppliers)" (the interpretation agreeing with the two SQL versions in Shnugo's answer

"the parts don't have multiple suppliers"

SELECT * FROM Parts P
WHERE NOT EXISTS (SELECT *
    FROM Catalog C1 JOIN Catalog C2 ON C1.sid <> C2.sid
    AND C1.pid = P.pid AND C2.pid = P.pid)

If you have Parts 1-4 and Suppliers 1-3 then the above give three different answers after:

INSERT INTO Catalog VALUES (1,1),(2,1),(2,2),(3,1),(3,2),(3,3);

PPS Logic formulations for the above are:

-- 1. "the parts are not supplied by all the suppliers"  
    P(P.pid, P.pname)
AND EXISTS S.sid, S.sname [S(S.sid, S.sname) AND NOT C(S.sid, P.pid)]
-- 2 "the parts are not supplied (by some of the suppliers)"
P(P.pid, P.pname) AND NOT EXISTS C.sid C(C.sid, P.pid)
-- 3. "the parts don't have multiple suppliers"
    P(P.pid, P.pname)
AND NOT EXISTS C1.sid, C2.sid
        [C(C1.sid, P.pid) AND C(C2.sid, P.pid) AND C1.sid <> C2.sid]

回答2:

Wasn't it simple as that? If not please use my example tables and specify your expected output:

DECLARE @Parts TABLE(pid INT, pname VARCHAR(100));
DECLARE @Suppliers TABLE(sid INT, sname VARCHAR(100));
DECLARE @Catalog TABLE(pid INT, sid INT);

INSERT INTO @Parts VALUES(1,'Part 1'),(2,'Part 2'),(3,'Part 3'),(4,'Part 4');
INSERT INTO @Suppliers VALUES(1,'Suppl 1'),(2,'Suppl 2'),(3,'Suppl 3');
INSERT INTO @Catalog VALUES(1,1),(1,3) --part 1 from two suppliers
                          ,(2,1)       --part 2 from one supplier
                                       --part 3 no supplier
                                       --part 4 no supplier

--Get all parts *that not supplied by some suppliers*
SELECT * 
FROM @Parts AS p 
WHERE p.pid NOT IN(SELECT c.pid FROM @Catalog AS c)

The result

pid pname
3   Part 3
4   Part 4

EDIT: The same is reached with "your" query, but this would perform worse:

This query is looking for parts, where there is no supplier which does have an entry in "catalog". And I must admit, this smells a bit like homework :-)

SELECT * FROM @Parts P 
WHERE NOT EXISTS (SELECT S.sid 
                  FROM @Suppliers S 
                  WHERE EXISTS (SELECT 1 
                                FROM @Catalog C 
                                WHERE s.sid = C.sid AND P.pid = C.pid
                                )
                  )

来源：https://stackoverflow.com/questions/35398701/exist-and-not-exists-query-misunderstanding

标签

sql

database

database-design

relational-database

relational-algebra