PostgreSQL组合没有重复

set search_path='tmp';

WITH ztab AS (
SELECT idx as idx,substring ( 'WTF!' FROM idx FOR 1) as str
FROM generate_series(1,char_length( 'WTF!' )) idx
)
SELECT t1.str,t2.str
FROM ztab t1
JOIN ztab t2 ON t2.idx > t1.idx
        ;

结果：

str | str 
-----+-----
 W   | T
 W   | F
 W   | !
 T   | F
 T   | !
 F   | !
(6 rows)

不幸的是我找不到避免双弦常量的方法. (但整个事情可以打包成一个函数)如果没有重复的字符(或者你想要它们),你可以在str而不是idx上进行反连接.

UPDATE(来自ypercube的提示)似乎OP希望连接字符串.就这样吧：：

WITH ztab AS (
SELECT idx as idx,char_length( 'WTF!' )) idx
)
SELECT t1.str || t2.str AS results
FROM ztab t1
JOIN ztab t2 ON t2.idx > t1.idx
        ;

结果：

results 
---------
 WT
 WF
 W!
 TF
 T!
 F!
(6 rows)

UPDATE2 :(这是递归的东西…)

WITH RECURSIVE xtab AS (
        WITH no_cte AS (
        SELECT
        1::int AS len,idx as idx,substring ( 'WTF!' FROM idx FOR 1) as str
        FROM generate_series(1,char_length( 'WTF!' )) idx
        )
        SELECT t0.len as len,t0.idx,t0.str
        FROM no_cte t0
        UNION SELECT 1+t1.len,tc.idx,t1.str || tc.str AS str
        FROM xtab t1
        JOIN no_cte tc ON tc.idx > t1.idx
        )
SELECT * FROM xtab
ORDER BY len,str
-- WHERE len=2
        ;

结果3：

len | idx | str  
-----+-----+------
   1 |   4 | !
   1 |   3 | F
   1 |   2 | T
   1 |   1 | W
   2 |   4 | F!
   2 |   4 | T!
   2 |   3 | TF
   2 |   4 | W!
   2 |   3 | WF
   2 |   2 | WT
   3 |   4 | TF!
   3 |   4 | WF!
   3 |   4 | WT!
   3 |   3 | WTF
   4 |   4 | WTF!
(15 rows)