连接来自多列的不同值

Question

I have this in a table

party1 | party2
A      | E 
B      | D
C      | H
D      | B
E      | A
F      | H 
G      | I

Result I am trying to get is distinct concatenation of party1 and party2 , ie

AE (will same as EA so only one instance is required)
BD (will same as DB so only one instance is required)
CH
FH
GI

Party1 and Party2 are formed using CTE. Parties AI flows from single table Party and are identified as party1 and party2 using business logic of code 4 and 1.

Below is the query i could form

WITH CTE AS
(
      SELECT 
    DISTINCT CPR1.PARTY_IDENTIFIER AS PARTY1,
             CPR2.PARTY_IDENTIFIER AS PARTY2,
             CCR.CLIENT_RELATIONSHIP_TYPE_CODE,
             CPR1.CLIENT_TYPE_CODE
        FROM GOLD.CLIENT_TO_PARTY_RELATIONSHIP CPR1
        JOIN GOLD.CLIENT_TO_CLIENT_RELATIONSHIP CCR 
          ON     CPR1.CLIENT_IDENTIFIER = CCR.CLIENT_IDENTIFIER 
             AND CCR.CLIENT_RELATIONSHIP_TYPE_CODE = '04'
             AND CPR1.CLIENT_TYPE_CODE = '1'
        JOIN GOLD.CLIENT_TO_PARTY_RELATIONSHIP CPR2 
          ON CPR2.CLIENT_IDENTIFIER = CCR.CLIENT_IDENTIFIER
       WHERE (CPR1.PARTY_IDENTIFIER_INACTIVE_DATE IS NULL OR CPR1.PARTY_IDENTIFIER_INACTIVE_DATE ='')
         AND (CPR2.PARTY_IDENTIFIER_INACTIVE_DATE IS NULL OR CPR2.PARTY_IDENTIFIER_INACTIVE_DATE ='')
         AND CPR1.PARTY_IDENTIFIER<>CPR2.PARTY_IDENTIFIER
         AND CPR1.DATA_AS_OF_DATE LIKE '201912%' 
         AND CPR2.DATA_AS_OF_DATE LIKE '201912%'
         AND CCR.DATA_AS_OF_DATE LIKE '201912%'
)

  SELECT 
DISTINCT CONCAT("PG","_",TRIM(CAST(PARTY1 AS STRING)) ,"_",TRIM(CAST(PARTY1 AS STRING))) AS PG_KEY,
         party1,
         party2
    FROM CTE

Answer 1

Your description says that you are actually not interested in the concatenation but in the (2-element) sets that are composed from the paired data. Representing them (as you do) as a 2-letter string, go along the following template.

WITH CTE AS
(
      SELECT CPR1.PARTY_IDENTIFIER AS PARTY1,
             CPR2.PARTY_IDENTIFIER AS PARTY2,
             CCR.CLIENT_RELATIONSHIP_TYPE_CODE,
             CPR1.CLIENT_TYPE_CODE
        FROM GOLD.CLIENT_TO_PARTY_RELATIONSHIP CPR1
        JOIN GOLD.CLIENT_TO_CLIENT_RELATIONSHIP CCR 
          ON     CPR1.CLIENT_IDENTIFIER = CCR.CLIENT_IDENTIFIER 
             AND CCR.CLIENT_RELATIONSHIP_TYPE_CODE = '04'
             AND CPR1.CLIENT_TYPE_CODE = '1'
        JOIN GOLD.CLIENT_TO_PARTY_RELATIONSHIP CPR2 
          ON CPR2.CLIENT_IDENTIFIER = CCR.CLIENT_IDENTIFIER
       WHERE (CPR1.PARTY_IDENTIFIER_INACTIVE_DATE IS NULL OR CPR1.PARTY_IDENTIFIER_INACTIVE_DATE ='')
         AND (CPR2.PARTY_IDENTIFIER_INACTIVE_DATE IS NULL OR CPR2.PARTY_IDENTIFIER_INACTIVE_DATE ='')
         AND CPR1.PARTY_IDENTIFIER<>CPR2.PARTY_IDENTIFIER
         AND CPR1.DATA_AS_OF_DATE LIKE '201912%' 
         AND CPR2.DATA_AS_OF_DATE LIKE '201912%'
         AND CCR.DATA_AS_OF_DATE LIKE '201912%'
)
  select
distinct CASE WHEN party1 < party2 
              THEN party1 || party2 
              ELSE party2 || party1
         END pair
         FROM CTE
       ;

Caveats

Pull the CTE definition out of the subquery. Concat syntax may vary depending on the sql dialect / database product you are using.