SQL & PL/SQL :: Joining But Avoiding Duplicate Rows
Oct 16, 2010
I have two tables A and B
CREATE TABLE A(EMP_ID NUMBER, EMP_NAME VARCHAR2(100))
CREATE TABLE B(EMP_ID NUMBER, EMP_ATT1 VARCHAR2(10), EMP_ATT2 VARCHAR2(10))
INSERT INTO A VALUES(1, 'ONE');
INSERT INTO A VALUES(2, 'TWO');
INSERT INTO A VALUES(3, 'THREE');
[Code]....
This query returns all the matching row of A and B
SELECT A.EMP_ID, A.EMP_NAME, B.EMP_ATT1, B.EMP_ATT2
FROM A
INNER JOIN B ON A.EMP_ID=B.EMP_ID
The output for this shows:
EMP_ID EMP_NAME EMP_ATT1 EMP_ATT2
1 ONE 1ATT1 1ATT2
2 TWO 2ATT1 2ATT2
2 TWO 2ATT1.1 2ATT2.1
3 THREE 3ATT1 3ATT2
The requirement is to avoid duplicate rows even if matched:
EMP_ID EMP_NAME EMP_ATT1 EMP_ATT2
1 ONE 1ATT1 1ATT2
2 TWO 2ATT1 2ATT2
3 THREE 3ATT1 3ATT2
I create a second "DATE_PAIRS" table, and join a few subqueries. It works with conditions (ID=1), but scanning the whole table (2+ million records) is abysmal.
I have two tables. The first contains all Segment information within Oracle i.e;
Table 1 Segment Description Flex_Value 1 North 1234 1 South 1235 1 East 1236 2 Car C001 2 Boat B001
and the second table contains financial data, but only the segment code;
Table 2 Date Segment1 Segment2 Value 01/01/11 1234 C001 10,000 02/01/11 1235 C001 10,000
what I want to return is some of the columns within table two as well as additional columns for the segment descriptions.
Below is an extract of what I have put together, and it does return what I need, but I'm sure there is a much more efficient way of creating the query;
SSD@ermd> desc person_pos_history Name Null? Type ------------------------------------------------------------------------ -------- ------------------------
PERSON_POSITION_HISTORY_ID NOT NULL NUMBER(10) POSITION_TYPE_ID NOT NULL NUMBER(10) PERSON_ID NOT NULL NUMBER(10) EVENT_ID NOT NULL NUMBER(10) USER_INFO_ID NUMBER(10) TIMESTAMP NOT NULL DATE
We found out that few person_id's are repeating for a particular event (3):
select PERSON_ID, count(*) from person_pos_history group by PERSON_ID, EVENT_ID having event_id=3 and count(*) > 1 order by 2
If we look at the 1st person id "217045", we can see that it is repeating 356 times for event id 3.
SSD@ermd> select POSITION_ASSIGNMENT_HISTORY_ID, POSITION_TYPE_ID, PERSON_ID,EVENT_ID, to_char(timestamp, 'YYYY-MM-DD HH24:MI:SS') 2 from person_pos_history 3 where EVENT_ID=3 4 and person_id=217045 5 order by timestamp;
356 rows selected.It is safe to assume that the person id/event id with the earliest timestamp is the one that was loaded 1st, hence, the one we want to keep and the rest should be deleted.
I am trying to find sum for one record for each partition but while taking that timestamp giving me bit trouble, i have tried to reproduce the table and some little data
CREATE TABLE TEST_COUNT (END_TIME DATE ,SUCCESSFUL_ROWS NUMBER ,FAILED_ROWS NUMBER ,TBL_NAME VARCHAR (4) ,PARTITION_NAME VARCHAR (240) )
column sid format 'a5' column serial# format 'a10' column mins_running format 'a15' column sql_text format 'a100' set linesize 200 set pagesize 30
[Code]..
I am running this code, and the output shows multiple lines.
TRIM(S.SID) TRIM(S.SERIAL#) MINS_RUNNING SUBSTR(Q.SQL_TEXT,1,70) ---------------------------------------- ---------------------------------------- --------------- ---------------------------------------------------------------- 700 46592 242.08 Select count(*) as count, case when count(*)>0 then 'FAIL' else 700 46592 242.08 'PASS' end as result from (SELECT cv.code_value FROM code_valu
[Code]...
Is there a way to wrap up the column for SQL_TEXT VARCHAR2(64) so that I can 1 row for the output?
I can get it through this query : select PARTY_ID from XXWFS_CUSTOMER_EXT group by PARTY_ID having count (PARTY_ID) > 1;
Now for the records which i got for each duplicate row i want to update the second row with a specific value.. so that duplicate rows does not exist anymore
Ex: I got party id's 12, 14, 16, 18 two times each
Now as 12 is two times.. i want to update the second row of 12 with some x value same is the case for other values like 14,16, etc
I have a view and in that view i need to remove duplicate rows from output. For that i need to run select query in where clause of view if select query return true then we need to execute second condition.
my requirement in view like
And.......... And ((select count(*) from table A where conditions)=1 )then name is null AND
in that code first we need to check first select query condition then we need to apply name is null condition. but i tried to run it but select query not run properly. because tables is used in View.
Name _____ Smith Street Smith Street John Street Ed Street Ed Street Ed Street
and need to assign sequence numbers only when the record (Name) changes, e.g. :
Name Seq _____ ____ Smith Street 1 Smith Street 1 John Street 2 Ed Street 3 Ed Street 3 Ed Street 3
I have experimented with row_number partition but then i just get the sequence returning to 1 when the name value changes.
If I grouped the records by Name I would like to have unique, sequential numbers: 1, 2, 3 but where there is the same name I would like the sequence to stop and the number to replicate?
I have one table in which I want to restrict some records from being inserted. I don't want to put any checked constraints. e.g. consider following table
transaction( id number primary key, txn_date timestamp(7), payee varchar2(40), amount number, memo varchar2(40), ref_num number )
I want to write SQL which should not inset duplicate record.
e.g.
I have written one as bellow:
insert into transaction select 1, to_date('2009-12-12','YYYY-MM-DD'), 'Payee1', 12, 'Test', 212 from dual where (select count(*) from transaction where txn_date=to_date('2009-12-12','YYYY-MM-DD') and payee='Payee1' and amount=12)=0;
Can I use exists/not exists, which query will be more appropriate. (Please consider that fields which I am using to filter out the duplicate transactions does not contain primary key.)
Can I write such SQL. Or do i check for duplicate rows one by one and then filter the duplicate records.
My requirement if id, join_date, join_time, result of table1 is matched with table2 at least one time then if repeating rows associated with the id should not come.Here is the test case.
create table table1 ( id number , join_date varchar2(8), join_time varchar2(6), status varchar2(10)); create table table2 ( id number , join_date varchar2(8), join_time varchar2(6), status varchar2(10));
insert into table1 values (01, '20010101', '0500', 'PASS'); insert into table1 values (01, '20010102', '0501', 'FAIL'); insert into table1 values (02, '20010103', '0502', 'PASS'); insert into table1 values (03, '20010104', '0503', 'FAIL'); insert into table1 values (04, '20010105', '0504', 'PASS'); insert into table1 values (05, '20010106', '0505', 'FAIL'); [code]...
I have tried the below mentioned query, whether any better query is there than this because in real-time data have 2 millions of record in table 1 and 60 thousand in table2.
select distinct a.id, a.join_date, a.join_time, a.status from table1 a, table2 b where a.id = b.id and (a.id, a.join_date, a.join_time, a.status) not in (select b.id, b.join_date, b.join_time, b.status from table2 b) and a.id = ( select distinct a.id [code]....
What would cause Oracle to insert duplicate rows into a table? Could a join of two tables in the initial query assigned to an application page cause ORacle to insert an extra row into a table when an update to data value occurs? I have no insert triggers and no foreign keys assigned to the table. I am not sure what would cause Oracle to assume that an insert of a row must occur. I want to prevent that insert.
I've got situation where one of the procedure causing insert into table rows but some of them are not needed Temporary I need to disable this and exclude some type of data being inserted by changing procedure.
I wouldn't want to change the procedure is there a way to write a trigger that would not allow to insert a row into a table if condition is met? Something of a logic:
CREATE OR REPLACE TRIGGER avoid_insert BEFORE (AFTER?) INSERT if entity_id = 123 DO NOT DO THE INSERT (DO NOT DO ANYTHING)
CODE AMOUNT DESCRIPTION AAA 5 five dollars for pizza AAA 2 two dollars for tips AAA 1 one dollar for dogsitting BBB 6 six dollars for babysitting BBB 1 one dollar for tips
My goal is to list all records, "grouping" by code, with sum(amount), but the final display has to show all descriptions, one for row, avoiding to repeat the "CODE" column and "sum(AMOUNT)" column.The result should be like this:
CODE SUM(AMOUNT) DESCRIPTION AAA 8 five dollars for pizza two dollars for tips one dollar for dogsitting BBB 7 six dollars for babysitting one dollar for tips
That is, the "CODE" is displayed only the first row, with its sum of "amount".I think I have to use the analytics functions, but I was a little stuck.
Given two tables in a 1:N relationship, I want to retrieve, for each parent row, any child row which either satisfies some condition cond1 or, if no child rows matching cond1 exist, which satisfies another condition cond2.
I can easily write such a query using NOT EXISTS, but I am forced to repeat the condition cond1. This is inconvenient, as the repeated condition is quite long (involving other tables as well), and I would like to write it only once.
For example (using the HR schema included in Oracle 10g XE): I want to retrieve, for each department, all the "new" employees (hired after a certain date) or, if no "new" employees exist, the department manager.
We are on oracle 10.2.0.4 on Solaris 10 and have a perf. issue with a bind variable using query. The query is in java application. I want to test its performance when the query doesn't use bind variable and instead uses the passed value as literal. How can it be done?
EMPNO ENAME JOB MGR HIREDATE SAL COMM DEPTNO ---------- ---------- --------- ---------- --------- ---------- ---------- ---------- 7839 KING PRESIDENT 17-NOV-81 5000 10
1 row selected.
But the statements will be taken as similar statements by oracle (due to :vn). Now I want oracle to take it as literal and the change for this has to be done in java code in my actual scenario which has a different query (but conceptually it uses bind variable and I want it to use passed value as literal). How can it be done?
can we link a table that contains a foreign key with a sequence table? i have a table that has a sequence bookid and i want to link this table to another table that contains a foreign key of bookid.can i link this two table and how?
I have a table (promo_custom) with following colimns (id, id_promo, button_promo, button_submit, css, user, date) and other table (promo_buttons) with following columns (id, id_design, name, url, username, date) .
Columns promo_custom.button_promo and promo_custom.button_submit are foreign keys (references promo_buttons.id).
I want to create a view showing table promo_custom but instead of columns button_promo and button_submit I would like to show the column url of them.
SELECT c1.c_no, c1.c_name, c1.city FROM customers c1 LEFT OUTER JOIN customers c2 ON (c1.city = c2.city AND c1.c_name<>c2.c_name); SELECT c1.c_no, c1.c_name, c1.city FROM customers c1 RIGHT OUTER JOIN customers c2 ON (c1.city = c2.city AND c1.c_name<>c2.c_name);
We have a table for reports. If user A submits a report ...and say the sequence # is 242. When this report goes to Admin ...he submits this request then in same table we add another row with say sequence # 245. THEN we update column called 'Asctd ID' for 242 and add 245 in there. and then update Asctd ID for 245 and add 242 in there.
(This table has many fields, one of which is report Name field)
Now i am running a query like this...
SELECT b.JOB_ID, a.DESC, TO_CHAR(a.CREATE_DATE,'MM/DD/YYYY'), DECODE(a.DLVRY_TYPE,'','PDF',a.DLVRY_TYPE), DECODE(a.FLG,'','N/A', a.FLG), ((TO_DATE(a.CREATE_DATE,'DD-MON-YY')) + 21) - TO_DATE(SYSDATE, 'DD-MON-YY'), c.STATUS_DESC, a.SIZE_NUM, b.PRVDR_ID, a.asctd_id, d.NM FROM REQUEST_DIM a, PROVIDER_DIM b, STATUS_DIM c, DIM d WHERE a.FLAG = 'P' and RTRIM(a.RPT_RQST_USER_ID) = 'TEST02' AND a.RPT_RQST_TYPE = 'D' AND a.RPT_RQST_ACTIVE_IND = 'A' and a.asctd_id is not null and a.RPT_RQST_ID = b.RPT_RQST_ID and b.JOB_ID = c.JOB_ID and b.PRVDR_ID = d.PRVDR_ID AND c.CREATE_DT = (SELECT MAX(d.CREATE_DT) FROM STATUS_DIM d WHERE d.RPT_RQST_ID = b.RPT_RQST_ID and d.JOB_ID = b.JOB_ID ) ORDER BY a.RPT_RQST_ID, a.CREATE_DATE
Now this query is run by the admin (job 245) ...it returns a bunch of stuff and also the name of the report that the admin gave this. But when admin sees this we want to be able to displace the report name that user A gave it (asctd ID 242). so the row 245 HAS a asctd ID 242. and there is a row 242 from which we can get the name easily. But i dont know if this is possible in ONE QUERY??