左外连接结果在 hive 上变大

Question

所有，我想从这个查询中得到可靠的结果：

SELECT ..

FROM
(
SELECT 
 CO_CODE,
REP.cua cua,
PRD.PRODUCT_DESC, 
REGEXP_EXTRACT(B.rfbbn,'^(?:[^*]*\\*){2}([^*]*)',1) cllt,
NVL(CCY_bbce,0) bbce, 
B.TYPE, 
A.conn_keyy

FROM 
(
SELECT conn_keyy , ext_date FROM 
(tablee.aa) A
)aaxyz 
WHERE flag = 'Y'
)A

LEFT OUTER JOIN 
tablee.B
ON A.conn_keyy = B.conn_keyy

LEFT OUTER JOIN (SELECT DISTINCT * FROM tablee.cc) CPLCUR
ON CPLCUR.conn_keyy = A.conn_keyy
AND CPLCUR.cllt = REGEXP_EXTRACT(B.rfbbn,'^(?:[^*]*\\*){2}([^*]*)',1)
AND CPLCUR.dtdt = '1999' 

LEFT OUTER JOIN (SELECT DISTINCT * FROM tablee.dd) CPLBAL
ON CPLBAL.conn_keyy = A.conn_keyy
AND CPLBAL.SEQUENCE = CPLCUR.SEQUENCE
AND CPLBAL.dtdt = '1999' 

LEFT OUTER JOIN (SELECT DISTINCT * FROM tablee.ee) CPLCCY
ON CPLCCY.conn_keyy = A.conn_keyy
AND CPLCCY.SEQUENCE = CPLCUR.SEQUENCE
AND CPLCCY.dtdt = '1999' 

LEFT OUTER JOIN (SELECT DISTINCT * FROM tablee.ff) CPLMOV
ON CPLMOV.conn_keyy = A.conn_keyy
AND CPLMOV.SEQUENCE = CPLCUR.SEQUENCE
AND CPLMOV.dtdt = '1999' 

LEFT OUTER JOIN
 (tablee.REP)REP 
ON REP.relino = B.lnido

LEFT OUTER JOIN tablee.P PRD
ON PRD.PRODUCT_CODE = REGEXP_EXTRACT(A.conn_keyy,'[.]([^.]+)',1)
AND PRD.dtdt = '1999'

WHERE B.lnido LIKE 'PLCONS1%'
) rrvv;

fyi，A 的选择计数 (*) 约为 60,000

我只是想知道为什么我的查询结果变成了 15 亿.. 我错过了什么？我操作左外连接时出了什么问题？

Answer 1

Join 可以复制行如果连接键在第二个表中不是唯一的，并且如果连接键在两个表中都不是唯一的，则会产生更多的重复。

例如：

with 

A as (
select 1 key, 'one' name
union all
select 1 key, 'two' name
),

B as (
select 1 key, 'one' name
union all
select 1 key, 'two' name
)

select *
  from A left join B on A.key=B.key

结果有四行，每个表只包含两个：

a.key   a.name  b.key   b.name
1       one     1       one
1       one     1       two
1       two     1       one
1       two     1       two

如何找到重复的键：

select B.conn_keyy, count(*) cnt 
  from tablee.B
group by B.conn_keyy
 having count(*)>1
order by cnt desc limit 100;

检查您要加入的每个表并决定您可以执行的操作：应用过滤、区分或添加更多连接键以进行连接（一对一或零）或（多对一或零）