Can't compare a row with other rows within a table.. (many to many compare)

Question

Hi all..
I am very new to PL/SQL..
I need to travers through a table for comparing it's rows with in the table with other rows. For this I am trying to use bellow Pl/sql.
create or replace compare()
Declare
VAR_HIT CHAR(1);
SEARCH_RECORD_DATA UDB.table1%ROWTYPE;
CANDIDATE_RECORD_DATA UDB.table1%ROWTYPE;
CURSOR SEARCH_RECORDS_CURSOR IS SELECT * FROM UDB.table1 order by registration_id;
CURSOR CANDIDATE_RECORDS_CURSOR IS SELECT * FROM UDB.table1 order by registration_id;
BEGIN
FOR SEARCH_RECORD_DATA IN SEARCH_RECORDS_CURSOR LOOP
FOR CANDIDATE_RECORD_DATA IN CANDIDATE_RECORDS_CURSOR LOOP
IF (CANDIDATE_RECORD_DATA.DECISION='P') THEN
VAR_HIT:='y';
IF(CANDIDATE_RECORD_DATA.FIRST_NAME!='unknown') AND (CANDIDATE_RECORD_DATA.FIRST_NAME!=SEARCH_RECORD_DATA.FIRST_NAME) THEN
VAR_HIT:='n';
ELSIF(CANDIDATE_RECORD_DATA.LAST_NAME!='unknown') AND (CANDIDATE_RECORD_DATA.LAST_NAME!=SEARCH_RECORD_DATA.LAST_NAME) THEN
VAR_HIT:='n';
ELSIF(CANDIDATE_RECORD_DATA.BIRTH_DATE!='unknown') AND (CANDIDATE_RECORD_DATA.BIRTH_DATE!=SEARCH_RECORD_DATA.BIRTH_DATE) THEN
VAR_HIT:='n';
ELSIF(CANDIDATE_RECORD_DATA.GENDER!='U') AND (CANDIDATE_RECORD_DATA.GENDER!=SEARCH_RECORD_DATA.GENDER) THEN
VAR_HIT:='n';
ELSIF(CANDIDATE_RECORD_DATA.FATHER_NAME!='unknown') AND (CANDIDATE_RECORD_DATA.FATHER_NAME!=SEARCH_RECORD_DATA.FATHER_NAME) THEN
VAR_HIT:='n';
ELSIF (CANDIDATE_RECORD_DATA.MOTHER_NAME!='unknown') AND (CANDIDATE_RECORD_DATA.MOTHER_NAME!=SEARCH_RECORD_DATA.MOTHER_NAME) THEN
VAR_HIT:='n';
END IF;
IF(VAR_HIT='y') THEN
INSERT INTO UDB.BIO_DI_HIT_RESULT(REGISTRATION_ID,SEARCH_ID,HIT_CANDIDATE_ID,SEARCH_DETAILS,CANDIDATE_DETAILS) VALUES (SEARCH_RECORD_DATA.REGISTRATION_ID,SEARCH_RECORD_DATA.EGM_NO,CANDIDATE_RECORD_DATA.EGM_NO,VAR_SEARCH_DETAILS,VAR_CANDIDATE_DETAILS);
UPDATE UDB.BIO_RECORDS_DEMOGRAPHICS SET DECISION='D';
END IF;
END IF;
END LOOP;
commit;
END LOOP;
END;
Outer loop is working fine (it is raversing throughout the table) (say for 8000 records 8000 times)
But Enner loop is not working fine e.i. it's running just for 8000 times for 8000 records. while it should run more then 8000 time..
Can any one help me..
Is the way of using two cursor on one table for comparing each row of record is correct????????? :(

Answer 1

Finally I come up with this:
INSERT INTO UDB.BIO_DI_HIT_RESULT(REGISTRATION_ID,SEARCH_ID,HIT_CANDIDATE_ID,SEARCH_DETAILS,CANDIDATE_DETAILS)
SELECT SEARCH.REGISTRATION_ID, SEARCH.Key_NO,CANDIDATE.Key_NO,
'First name='||SEARCH.FIRST_NAME||',Last name='||SEARCH.LAST_NAME||',Birth date='||SEARCH.BIRTH_DATE||',Gender='||SEARCH.GENDER||',Fathers name='||SEARCH.FATHER_NAME||',Mother name='||SEARCH.MOTHER_NAME,
'First name='||CANDIDATE.FIRST_NAME||',Last name='||CANDIDATE.LAST_NAME||',Birth date='||CANDIDATE.BIRTH_DATE||',Gender='||CANDIDATE.GENDER||',Fathers name='||CANDIDATE.FATHER_NAME||',Mother name='||CANDIDATE.MOTHER_NAME
FROM UDB.BIO_RECORDS_DEMOGRAPHICS SEARCH,UDB.BIO_RECORDS_DEMOGRAPHICS CANDIDATE
WHERE (SEARCH.FIRST_NAME!='unknown' OR SEARCH.LAST_NAME!='unknown' OR SEARCH.BIRTH_DATE!='unknown' OR SEARCH.GENDER!='unknown' OR SEARCH.FATHER_NAME!='unknown' OR SEARCH.MOTHER_NAME!='unknown')
AND SEARCH.REGISTRATION_ID != CANDIDATE.REGISTRATION_ID
AND SEARCH.REGISTRATION_ID < CANDIDATE.REGISTRATION_ID
AND (SEARCH.FIRST_NAME='unknown' OR CANDIDATE.FIRST_NAME IN (SEARCH.FIRST_NAME,'unknown'))
AND (SEARCH.LAST_NAME='unknown' OR CANDIDATE.LAST_NAME IN (SEARCH.LAST_NAME,'unknown'))
AND (SEARCH.BIRTH_DATE='unknown' OR CANDIDATE.BIRTH_DATE IN (SEARCH.BIRTH_DATE,'unknown'))
AND (SEARCH.GENDER='U' OR CANDIDATE.GENDER IN (SEARCH.GENDER,'U'))
AND (SEARCH.BIRTH_DATE='unknown' OR CANDIDATE.BIRTH_DATE IN (SEARCH.BIRTH_DATE,'unknown'))
AND (SEARCH.FATHER_NAME='unknown'OR CANDIDATE.FATHER_NAME IN (SEARCH.FATHER_NAME,'unknown'))
AND (SEARCH.MOTHER_NAME='unknown' OR CANDIDATE.MOTHER_NAME IN (CANDIDATE.MOTHER_NAME,'unknown'))
ORDER BY SEARCH.REGISTRATION_ID;
Now it realy meet all my final table requirement.. I am happy.. :)
But again last requirement is like..
In the final BIO_DI_HIT_RESULT table I need to add one more column named as match_count
This column will have values of total no. of exact match field (not unknown) for each records in the BIO_DI_HIT_RESULT.....
One way of duing this is--
Step1: I will use a cursor for traversing through each of the rows in BIO_DI_HIT_RESULT.
Step 2: And based on count of matches in search_details(substring) and candiate_details(substring) I update each and every rows of the table.
It will be very slow.. And for records like 2 million records.. I can't think of this idea. :(
Sooooo,, Is there ANY BETTER WAY OF DOING THIS AS WELL.. BETTER MEANS FAST WAY.. :(
Thanks in advance....
Edited by: user13367608 on Jan 8, 2011 2:13 AM