pl sql procedure for update - oracle

Question

this procedure for update rows of phone number with dashes and adding default area code if phone number does not have it.i do not want to use cursor
CREATE OR REPLACE procedure pro
AS
begin
update judge set phone# = substr(Phone#, 1, 3) || '-' || substr(Phone#, 4,3) || '-' ||
substr(Phone#, 7, 4) where length(trim(phone#))=10;
update judge set phone# = substr(Phone#, 0, 0) || '309-298' ||
substr(Phone#, 1, 5)
where length(trim(phone#))=5;
END;
/
i want to add dashes only if phone number length is 10 and add area code if length is 5.
this code is working but is there any more efficient way of doing it.

Answer 1

is there any more efficient way of doing it.
Yes, there may be a faster method, but it depends on how big the table is, and what percentage of records in the table will be changed.
If the entire table is small - let's say it has less than 100~500 records, then creating indexes will most likely not give you any profit, simple Full Table Scan will be fast enough. In this case use only ONE update command instead of two separate ones - in this way, the table will be read and updated only once instead of twice, and the execution time will be shorter by about half :
update judge set phone# =
CASE length(trim(phone#))
WHEN 10
THEN substr(Phone#,1,3) || '-' || substr(Phone#,4,3) || '-' || substr(Phone#,7,4)
WHEN 5
THEN '309-298' || substr(Phone#,1,5)
ELSE phone#
END
where length(trim(phone#)) in (5,10);
If the entire table is big (thousands or millions records), but a number of records with lengths of 5 and 10 is relatively small (let say that less than 10~15% of all records), then create a function based index:
CREATE INDEX some_name_ix ON judge( length(trim(phone#)) );
and then, after creating the index, refresh the statistics:
exec DBMS_STATS.gather_table_stats( user, 'judge' );
After the above steps check if oracle is willing to use this index by generating explain plans for the below three update commands
EXPLAIN PLAN FOR
UPDATE judge SET phone#= '123'
WHERE length( trim( phone# ) ) in ( 5, 10 )
SELECT * FROM table( dbms_xplan.display );
EXPLAIN PLAN FOR
UPDATE judge SET phone#= '123'
WHERE length( trim( phone# ) ) = 5
SELECT * FROM table( dbms_xplan.display );
EXPLAIN PLAN FOR
UPDATE judge SET phone#= '123'
WHERE length( trim( phone# ) ) = 10
SELECT * FROM table( dbms_xplan.display );
NOTE: SET phone#= '123' clause doesn't matter here, Oracle will not update the table, what's important to us - and what we are checking using explain plan command - is how Oracle will execute the query for different WHERE clauses
For each of the above commands you may see something like below - the keyword TABLE ACCESS FULL means, that Oracle is going to use the Full Table Scan method for this udate, and ignores our index:
--------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
--------------------------------------------------------------------------------
| 0 | UPDATE STATEMENT | | 53421 | 6468K| 36512 (1)| 00:00:02 |
|* 1 | TABLE ACCESS FULL| JUDGE | 53421 | 6468K| 36512 (1)| 00:00:02 |
--------------------------------------------------------------------------------
you may also see sometjing like below: TABLE ACCESS BY INDEX ROWID ... + INDEX RANGE SCAN ... index name - this means, that Oracle is willing to use the index for this update:
---------------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
---------------------------------------------------------------------------------------------------
| 0 | UPDATE STATEMENT | | 1 | 124 | 5 (0)| 00:00:01 |
| 1 | TABLE ACCESS BY INDEX ROWID BATCHED| JUDGE | 1 | 124 | 5 (0)| 00:00:01 |
|* 2 | INDEX RANGE SCAN | MY_INDEX_IX | 1 | | 3 (0)| 00:00:01 |
---------------------------------------------------------------------------------------------------
There is a Cost (%CPU) column in these tables, which say as what is relative cost of this concrete update (low cost = fast speed, high cost = slow speed).
Finally, depending of results you will get in these explain plans, you'll may decide to:
use only one single update in the procedure
or keep two separate updates
In a case when Oracle use the index in none of these 3 cases, then the index is useless and you can drop it using:
DROP INDEX indexname

Answer 2

For example you can try to combine updates:
update judge
set phone# = decode( length( trim( phone# ) ),
5, '309-298' || substr( Phone#, 1, 5 ),
10, substr( Phone#, 1, 3 ) || '-' || substr( Phone#, 4, 3 ) || '-' || substr( Phone#, 7, 4 ),
phone# )
where length( trim( phone# ) ) in ( 5, 10 );
and using regexp maybe more flexible than seaching by length and second of course insert already prepared and formated data.