1. 程式人生 > >Oracle使用row_number() over (partition order by)和DISTINCT去除重複記錄

Oracle使用row_number() over (partition order by)和DISTINCT去除重複記錄

            最近做的一個模組涉及到8張表的聯合查詢,由於這8張表中有很多主從表的關聯,結果在使用模糊查詢的時候查詢結果集出現了重複記錄。如下:                                                                                                   

      

      執行的SQL語句如下:

select 
cf.id id, 
o.OBJECTCODE,
o.OBJECTNAME objNAME1,
cf.name name,
o1.OBJECTNAME objNAME2,
o2.OBJECTNAME objNAME3,
cf.no no,
bc.cert_no,
bc.FIELD3_VALUE,
to_char(bc.CREATE_DATE,'yyyy-MM-dd'),
trim(to_char(nvl(cf.FORMAMOUNT,0),'99,999,999,999,999,990.99')),
o3.OBJECTNAME objNAME4,
ap.PERSONNELNAME apName,
t.actorname,
to_char(cf.CREATETIME,'yyyy-MM-dd'),
cf.TEMPLETEID,
cf.orgentityid 
FROM 
t_cc_ct_bc bc,
T_CC_BillDetailData detail,
T_CC_BILLMAINDATA main,
cc_form cf,
t_sys_user ap,
T_CC_OBJECT o,
T_CC_OBJECT o1,
T_CC_OBJECT o2,
T_CC_OBJECT o3,
t_sys_flow_task t where 1=1 
and cf.no=main.BILLNUMBER  
and main.item109=o.OBJECTID  
and main.REQUISITIONUSER=ap.USERID 
and detail.BILLMAINDATAID=main.BILLMAINDATAID  
and main.billMainDataID=bc.BILLMAINDATAID  
and o1.OBJECTID=detail.DIMACCOUNT  
and o2.OBJECTID=detail.COMPUTATIONACCOUNT  
and o3.OBJECTID=main.RequisitionUserDepartment   
and main.billnumber=t.bono   
and t.tasktype='finishTask'   
and t.activityname='共享稽核會計'  
and (t.ACTORNAME LIKE '%黃%')    
order by cf.createtime desc; 

一、使用DISTINCT去重

select 
DISTINCT(cf.ID), 
o.OBJECTCODE,
o.OBJECTNAME objNAME1,
cf.name name,
o1.OBJECTNAME objNAME2,
o2.OBJECTNAME objNAME3,
cf.no no,
bc.cert_no,
bc.FIELD3_VALUE,
to_char(bc.CREATE_DATE,'yyyy-MM-dd'),
trim(to_char(nvl(cf.FORMAMOUNT,0),'99,999,999,999,999,990.99')),
o3.OBJECTNAME objNAME4,
ap.PERSONNELNAME apName,
t.actorname,
to_char(cf.CREATETIME,'yyyy-MM-dd') as cfcreatetime,
cf.TEMPLETEID,
cf.orgentityid 
FROM 
t_cc_ct_bc bc,
T_CC_BillDetailData detail,
T_CC_BILLMAINDATA main,
cc_form cf,
t_sys_user ap,
T_CC_OBJECT o,
T_CC_OBJECT o1,
T_CC_OBJECT o2,
T_CC_OBJECT o3,
t_sys_flow_task t  
where 1=1 and cf.no=main.BILLNUMBER  
and main.item109=o.OBJECTID  
and main.REQUISITIONUSER=ap.USERID 
and detail.BILLMAINDATAID=main.BILLMAINDATAID  
and main.billMainDataID=bc.BILLMAINDATAID  
and o1.OBJECTID=detail.DIMACCOUNT  
and o2.OBJECTID=detail.COMPUTATIONACCOUNT  
and o3.OBJECTID=main.RequisitionUserDepartment   
and main.billnumber=t.bono   
and t.tasktype='finishTask'   
and t.activityname='共享稽核會計'  
and (t.ACTORNAME LIKE '%劉%') 
order by cfcreatetime desc

     這裡就不過多的解釋了,在這裡我們使用ID去重,只需要將ID欄位使用DISTINCT()處理一下就可以了。

     注意,在這之前遇到一個問題:

     

     這是由於DISTINCT函式和ORDER BY衝突導致的,從上面的SQL中可以看出,最後是要根據某一表的日期欄位進行排序的,之前我用的是cf.createtime,引發了上面的sql錯誤,後來將cf.createtime這一列添加了一個別名,就這樣,問題完美的解決了,然後再ORDER BY的時候使用別名,這樣就完美的解決了。

      to_char(cf.CREATETIME,'yyyy-MM-dd') as cfcreatetime

二、使用row_number() over (partition order by)去重

with ect as(
select 
cf.id id, 
o.OBJECTCODE,
o.OBJECTNAME objNAME1,
cf.name name,
o1.OBJECTNAME objNAME2,
o2.OBJECTNAME objNAME3,
cf.no no,
bc.cert_no,
bc.FIELD3_VALUE,
to_char(bc.CREATE_DATE,'yyyy-MM-dd'),
trim(to_char(nvl(cf.FORMAMOUNT,0),'99,999,999,999,999,990.99')),
o3.OBJECTNAME objNAME4,
ap.PERSONNELNAME apName,
t.actorname,
to_char(cf.CREATETIME,'yyyy-MM-dd') as create1,
cf.TEMPLETEID,
cf.orgentityid 
FROM 
t_cc_ct_bc bc,
T_CC_BillDetailData detail,
T_CC_BILLMAINDATA main,
cc_form cf,
t_sys_user ap,
T_CC_OBJECT o,
T_CC_OBJECT o1,
T_CC_OBJECT o2,
T_CC_OBJECT o3,
t_sys_flow_task t  where 1=1 
and cf.no=main.BILLNUMBER  
and main.item109=o.OBJECTID  
and main.REQUISITIONUSER=ap.USERID 
and detail.BILLMAINDATAID=main.BILLMAINDATAID  
and main.billMainDataID=bc.BILLMAINDATAID  
and o1.OBJECTID=detail.DIMACCOUNT  
and o2.OBJECTID=detail.COMPUTATIONACCOUNT  
and o3.OBJECTID=main.RequisitionUserDepartment   
and main.billnumber=t.bono   
and t.tasktype='finishTask'   
and t.activityname='共享稽核會計'  
and (t.ACTORNAME LIKE '%黃%'))
select * from(
	select ss.*,row_number() over (partition by id order by create1) rid from ect ss 
) a where rid=1

    說明:

    1.with ect as() 

    括號內的內容,就是查詢的sql結果,其中包含重複資料,於是用with as() 建立一個臨時表ect為臨時表的名字。

    2.select ss.*,row_number() over (partition by id order by create1) rid from ect ss

    這句sql的意思是,查詢臨時表ect 別名為ss,row_number() over(partition by 需要檢索重複的列 order by 排序的列名)  別名為 rid  form ect ss,這時候就會查詢獲得一個rid列,如果id列存在多條相同值就以1開始遞增。

    3.select * from (↑) a where rid = 1

    這句sql是篩選rid為1,也就是id只出現1次的資料,這時候就去重複了。

    最後執行結果如下;

    

    從ID上看,我們是去重成功的。