1. Write the connection string to a temp file. This would be your parameter file. The format is as below.
   userid=scott/tiger@
   As it is a parameter file, you can actually put all the other sqlldr parameters in there.
2. Call sqlldr with parfile parameter.
   Example : sqlldr parfile=paramfile.txt data=mydata.csv control=mytable.ctl log=mydata101.log
3. Delete the temp file

To be more clear, here’s the actual code I used.

umask 0077
PWDFILE=$(mktemp)
echo "userid=${SYS_JOB_BSS_ORACLE_UID}/${PASWD}@${SYS_JOB_BSS_ORACLE_SID}" > $PWDFILE
sqlldr parfile=$PWDFILE data=${data_file} control=${APP_CTL_FILE_NAME} log=${data_file}.log rows=100 errors=100
retCode=`echo $?`
rm $PWDFILE


I would like to explain it a bit but I realized that you won’t be here if you don’t understand what it does. If anything, the umask there just sets the permission of the temp file to be created as 077. The rest are pretty much self explanatory.

http://laurentschneider.com/wordpress/2009/05/hide-password-from-ps-output-sql-loader.html

SELECT * FROM ALL_DIRECTORIES;

The query above will display all the directories accessible to the current user. They are folders in the server where you can write your output files. For some reason, I forgot about this twice already and had to scour the web to find it. The problem is, I also forgot the keywords. So, I’m writing it now.

To register new directories, you can follow the example below.

Example
create or replace directory mydir as 'c:mydirectory';

To use, run the code below.

declare
f utl_file.file_type;
begin
f := utl_file.fopen('MYDIR', 'myfilename.txt', 'w');
utl_file.put_line(f, 'just another line text 1');
utl_file.put_line(f, 'just another line text 2');
utl_file.fclose(f);
end;

whereby
m is the base.
n is the exponent.

Example

power(36,5);

will return 1679616

Take note that this function returns NUMBER as the data type. That means the maximum value can only be 1.0 x 10^126. Later versions (10g onwards) can use binary_double as the base which makes the function output binary_double. I don’t know where you would need that large number, though. Also, if m is negative, then n must be an integer.

In case you are asking where I need it, there’s this code conversion whereby I needed to use the power function to get the value of a particular character. The exponent(n) is not fixed so I needed this function. To be specific, the function involves the pl/sql version of RowIdToRowIdNum of Siebel. I’ll post the code later.

References:

http://asktom.oracle.com/pls/asktom/f?p=100:11:0::::P11_QUESTION_ID:1594387900346571874


MERGE INTO tstg_idl_approach a
USING tmdb_idl_approach b
ON (a.approach_id = b.approach_id)
WHEN MATCHED THEN
UPDATE SET a.approach_code = b.approach_code ,
a.current_ind = b.current_ind
WHEN NOT MATCHED THEN
INSERT (approach_id ,
approach_code,
current_ind )
VALUES (b.approach_id ,
b.approach_code,
b.current_ind );

For further reading, I recommend visiting http://www.oracle-developer.net/display.php?id=203.

XMLAGG (Available since Oracle 9i)


select rtrim (xmlagg (xmlelement (e, email_add || ',')).extract ('//text()'), ',') email_add
from email_groups
where email_group = '1-A945-9';

In that example, I needed to display all the email addresses in an email group. The output should be separated by comma so that I can pass it onto the email client without any problem nor any additional codes.

There’s more at Burleson Consulting in case you want other techniques.

DATE ADD (SYSDATE +/-1)

SELECT DATE_ADD(SYSDATE(),INTERVAL 1 DAY);
SELECT DATE_ADD(SYSDATE(),INTERVAL -10 DAY);


ADD_MONTHS

SELECT DATE_ADD(SYSDATE(),INTERVAL 1 MONTH);


TRUNC()

SELECT DATE(SYSDATE());


TO_DATE()

SELECT STR_TO_DATE( '07-08-2010', '%d-%m-%Y' );

TIME DIFFERENCE

SELECT TIMESTAMPDIFF(
SECOND , '2010-01-07 23:50:59', SYSDATE( ) ) /3600 /24 day,
TIMESTAMPDIFF(
SECOND , '2010-01-07 23:50:59', SYSDATE( ) ) /3600 hours,
TIMESTAMPDIFF(
SECOND , '2010-01-07 23:50:59', SYSDATE( ) ) /60 mins,
TIMESTAMPDIFF(
SECOND , '2010-01-07 23:50:59', SYSDATE( ) ) secs,

In the preceding example, I could have just used HOUR, MINUTE, DAY, MONTH or YEAR interval but in real world, I mostly use SECOND for a more accurate results. That’s not to say timestampdiff is not accurate. Maybe because I’m used in how I do it in Oracle.

MONTHS_BETWEEN

SELECT TIMESTAMPDIFF(
MONTH , '2010-01-07', SYSDATE( ) ) no_months


Reference:

http://dev.mysql.com/doc/refman/5.1/en/date-and-time-functions.html

Syntax


IFNULL(expr1,expr2)


Usage

If expr1 is not NULL, IFNULL() returns expr1; otherwise it returns expr2. Take note that IFNULL() returns a numeric or string value, depending on the context in which it is used. I haven’t used it for other data types yet but I reckon it would have the same results. That is to return the same data type as that of expr1. Another important thing to remember is that expr1 and expr2 should have the same data type

Case in point is the alpha-numeric auto-generated code. In Oracle, it’s easy to generate AA123, AA124, AA125, etc., etc. All we have to do is concat the alpha part (AA) and with the nextval of sequence. For MySQL, the alpha part and the numeric part have to be saved in separate fields.

However, Maresa of Microshell has created an alternative solution for it. The article discusses a way to emulate the nextval function in mysql. It a nutshell, they created a table similar to the all_sequences of Oracle. This is where you insert your sequences. Then, create a function named nextval() function which increments and returns value of the sequence.
You can check the procedure here.

It’s pretty neat, actually. It’s an excellent alternative and you can even create similar functions like currval(), prevval(), among others. However, I have some reservations on his approach especially on the issue of concurrency. You see in Oracle, calling the nextval() will increment the value and that’s already persisted (saved) in the sequence table even without saving your query/dml statement. On this approach, other queries accessing the pseudo-nextval will have to wait until your query finishes. Not a good idea.

So for now, I’ll will still have to make do with whatever MySQL is providing. If there’s any consolation, MySQL provides LAST_INSERT_ID() function in case you want to get the latest id inserted. See details here here.

So I’m just hoping that MySQL will implement an equivalent feature in the near future. With the imminent takeover by Oracle, I’m actually hoping that development of MySQL will push through and take a steam. We all know that the development of this widely used open-source database engine has been hampered ever since Sun bought it.

Sequence is a little feature. But little as it seems, it solves a lot of problems and helps improve code development time.

WITH Clause Syntax

WITH
 alias_name         -- alias for the aggregate_query
AS
 (aggregate_query_here)
SELECT...            -- Beginning of the query main body

It should be noted that multiple aggregate queries can be defined in the WITH clause as below.

WITH
 alias_name1
AS
 (agg_query1) ,
 alias_name2
AS
 (agg_query2)
SELECT...

When using subquery factoring, think of the aggregate_query as an in-line view. Actually, it is a view but a temporary one. Instead of creating a permanent view accessible in the database, you are creating a temporary one exclusively used by your main query.

Usage

I primarily use this in creating adhoc queries, reports and extractions. Of course, you can just create a view to accomplish this task. However, there are times wherein you are not allowed to create additional objects in the database for one reason or another. In that case, this subquery optimization comes very handy.

Caveat

You can’t use this inside Oracle Forms. It also doesn’t work inside stored procedures or anything with PL/SQL in it. At least I’m talking about until 11g.

Syntax #1


CASE [ expression ]
WHEN condition_1 THEN result_1
WHEN condition_2 THEN result_2
...
WHEN condition_n THEN result_n
ELSE result
END


Example


SELECT last_name, commission_pct,
(CASE commission_pct
WHEN 0.1 THEN ‘Low’
WHEN 0.15 THEN ‘Average’
WHEN 0.2 THEN ‘High’
ELSE ‘N/A’
END ) Commission
FROM employees ORDER BY last_name;


Syntax #2


CASE
WHEN [ condition_1 ] THEN result_1
WHEN [ condition_2 ] THEN result_2
...
WHEN [ condition_n ] THEN result_n
ELSE result
END


Example


SELECT m.paper_type ,
m.paper_brand ,
m.paper_grammage,
m.paper_width ,
m.paper_length ,
m.paper_stock_no,
(
CASE
WHEN NVL(smc.qty_reserved_onhand,0) != NVL(st.sum_qty_r_oh,0) THEN 'QROH NOT TALLY'
WHEN NVL(smc.qty_reserved_incom,0) != NVL(st.sum_qty_r_incom,0) THEN 'QRINC NOT TALLY'
WHEN NVL(smc.ton_reserved_onhand,0) != NVL(st.sum_ton_r_oh,0) THEN 'TROH NOT TALLY'
WHEN NVL(smc.ton_reserved_incom,0) != NVL(st.sum_ton_r_incom,0) THEN 'TRINC NOT TALLY'
END
) ERROR_MESSAGE
from paper_stock_master m


Personally, I use Syntax #2 most of the time. I don’t know why but maybe it just happened that I find it more suitable to use in the situations that I needed the CASE statement.

Difference Between DECODE and CASE WHEN

The most fundamental difference between the two oracle statements is that in DECODE, you are comparing a field or value in a given set of values or fields. It’s actually a shorthand for IF..ELSIF..ELSE statement with the condition all set to equals (=). CASE WHEN is almost the same but you can use any conditional operator (i.e, =,!=, >, >= , <, <=, etc) thereby giving you more freedom and flexibility in writing your code.

In may instances CASE WHEN helped me reduce the number of lines of my codes as well as optimize the performance. On top of that, I no longer need a full pl/sql program with multiple queries and even cursors to be able to provide the result that a simple CASE statement can provide. However CASE, like any other sql commands, is only applicable for certain scenarios. It’s not a magic keyword that will solve/apply to everything. So you have to weigh your options if you need it or not.

Ciao!