SQL选择具有可互换值的两列的唯一组合

Question

I have a flights table that has columns like "DEP" (departure) and "ARR" (arrival).

It is obvious that Departure and arrival columns can have interchangeable city names. For example, 1 row can have "SEA" as departure and "NYC" as arrival while another row can have "NYC" as departure and "SEA" as arrival.

I want to fetch only the unique combinations of departure and arrival cities. So, in this case, I want to see either of the 2 rows as output but not both the rows.

Here is how my table looks more realistically with data:

CREATE TABLE flights(DEP CHAR(3), ARR CHAR(3), COST INT)

INSERT INTO flights VALUES ('SEA', 'CHG', 100)
INSERT INTO flights VALUES ('CHG', 'SEA', 100)
INSERT INTO flights VALUES ('SEA', 'SFO', 100)
INSERT INTO flights VALUES ('SEA', 'LSA', 100)
INSERT INTO flights VALUES ('SEA', 'SJO', 100)
INSERT INTO flights VALUES ('SFO', 'CHG', 100)
INSERT INTO flights VALUES ('SFO', 'SEA', 100)
INSERT INTO flights VALUES ('BOS', 'SEA', 100)
INSERT INTO flights VALUES ('NYC', 'CHG', 100)
INSERT INTO flights VALUES ('NYC', 'SEA', 100)
INSERT INTO flights VALUES ('SEA', 'NYC', 100)

SELECT * FROM flights --(11 rows)

DEP  ARR  COST
---- ---- ----
BOS  SEA  100
CHG  SEA  100
NYC  CHG  100
NYC  SEA  100
SEA  CHG  100
SEA  LSA  100
SEA  NYC  100
SEA  SFO  100
SEA  SJO  100
SFO  CHG  100
SFO  SEA  100

For above table my output should be like (8 rows):

COL1 COL2 
---- ----
SFO  SEA
SFO  CHG
SEA  SJO
SEA  NYC
SEA  LSA
SEA  CHG
NYC  CHG
BOS  SEA

I was able to write a TSQL code using temp tables, variables, and loop etc but I strongly feel that this can be achieved in more simpler SQL probably by using UNION/INTERSECT/EXCEPT/EXISTS etc.

Here is my solution:

DECLARE @i INT = 1  --loop counter
DECLARE @exist BIT = 0  --flag to check if the combination  already exists
DECLARE @dep CHAR(3), @arr CHAR(3)

SELECT @i = COUNT(*) FROM dbo.flights   --get the row count

CREATE TABLE #tResult(dep CHAR(3), arr CHAR(3)) --output table
CREATE TABLE #tTemp (id TINYINT identity, dep CHAR(3), arr CHAR(3))

INSERT INTO #tTemp (dep, arr)
SELECT DISTINCT dep, arr FROM flights


WHILE (@i > 0)
BEGIN

    --get 
    SELECT @dep = dep, @arr = arr
    FROM #tTemp WHERE id = @i

    SET @exist = (SELECT count(*) FROM #tResult WHERE arr = @dep AND dep = @arr)    --check if this combination has been inserted in output

    --if not exists, insert this combination
    IF (@exist = 0)
    BEGIN
        INSERT INTO #tResult (dep, arr)
        VALUES (@dep, @arr)
    END

    SET @i = @i - 1 --decreament loop counter
END

SELECT * FROM #tResult

Can anyone please share a better or simpler solution to this problem. Thanks!

Answer 1

One way is using UNION ALL :

select dep,
    arr
from flights
where dep < arr

union all

select dep,
    arr
from flights f
where dep > arr
    and not exists (
        select 1 from flights f2 
        where f.dep = f2.arr and f.arr = f2.dep
        );

Demo

Answer 2

Something like this usually works:

Select distinct case when dep < arr then dep else arr end as col1,
                case when dep < arr then arr else dep end as col2
From flights