About PL/SQL Functions

Hello All,
Say, I have define a PL/SQL function. Is it posible to view its codes using SQLPlus and not SQL Developer? If it is posible how do I do that.
Thanks a lot.
Regards,
Talifhani

You can query USER_SOURCE
SELECT line, text
FROM user_source
WHERE NAME = 'YOUR_FUNCTION_NAME'
AND TYPE = 'FUNCTION'
order by line;
The text column will contain your source code.
Dilipkumar Patel.

Similar Messages

Question about PL/SQL Function in report

Hi all. I have a (basic) question about using a PL/SQL function in a report. In my report page I have an item (PL/SQL function) that has the following syntax:
return username from t1
where username like 'Le%'
I reference this item in an email process. Ideally this function would return all usernames that begin with 'Le' in the drafted email; however, its not looping so I'm only getting a return on the first name.
I have a few books on PL/SQL but I'm not exactly sure where to start - I imagine there is a loop process I need to integrate into the function but, like I said, I'm not sure.
I would have posted this to the PL/SQL forum but also wanted some feedback on whether or not using this method is the best way forward. Essentially I have a table with usernames and just want to send email messages based on some simple logic.
Thanks!
Len

Hi Sergio,
First, it's an honor to have you reply to my post. Thanks!
To be more specific,
T1 is a table containing the results from a user security scan. Username is a column in this table containing user names. What I have done is create a page in HTML DB with the following elements:
1. An 'Item' containing the PL/SQL function that is hidden. The purpose of this Item is to generate information from T1 that meets the simple logic parameters - in this case, the user names that begin with "Le". I have called this Item P145_TT
2. A 'Process' containing references to HTMLDB_MAIL. The 'Process' looks similar to this:
HTMLDB_MAIL.SEND(
P_TO => '[email protected]',
P_FROM => 'Anon Email Account',
P_BODY => 'Note: This message was automatically generated. Do not attempt to reply to this email.
These are the users that begin with Le: '||:P145_TT||' ',
P_SUBJ => 'Summary');
3. A 'Button' to submit the process and (hopefully) send the message containing all users that adhere to the logic in the function.
4. If I use 'return username from t1...' I will get a username generated, but like I said previously, it only gives me the first and does not loop.

SQL function extract

Hello XML Guru,
Could you tell me where i can find informations about how SQL function extract works for extracting XML data stored as relational-object?
Is VIEW always created for extracting XML data when executing this SQL function extract?
Thanks for your help.

You'll find documentation on this function in the [Oracle XML DB Developer's Guide| http://download.oracle.com/docs/cd/B19306_01/appdev.102/b14259/xdb03usg.htm#sthref336] .
HTH
Chris

Multiple Select List looping thru PL/SQL function body returning SQL query

Hi,
I have a Multiple Select List. I want to loop through the values from the Select List and process them in a PL/SQL function body returning a SQL query. Currently, my code only returns the SQL SELECT results of one item in the select list. How do I change my code to make it return the results of all of the items in the select list? (I tested it and it is definitely picking up all the values in the select list).
<b>
DECLARE
selected_items HTMLDB_APPLICATION_GLOBAL.VC_ARR2;
s VARCHAR2(20);
q varchar2(32767);
BEGIN
selected_items := HTMLDB_UTIL.STRING_TO_TABLE(:P50_SELECTED_INSTRUMENTS);
-- htp.p('COUNT: '||selected_items.count);
FOR i in 1..selected_items.count LOOP
s := TO_CHAR(selected_items(i));
-- htp.p('First: '||s);
-- htp.p('Second: '||:s);
-- htp.p('Third: '||TO_CHAR(selected_items(i)));
q:= 'SELECT '||
'SUBSTR(orig_geo_loc_sys,1,INSTR(orig_geo_loc_sys,''-'')-1) AS INSTRUMENT, '||
'SUBSTR(orig_geo_loc_sys,INSTR(orig_geo_loc_sys,''-'')+1, LENGTH'||
' (orig_geo_loc_sys)) AS ORIG_LINENUM, '||
'sum(orig_intrl) orig_intrl, '||
'sum(orig_extrl) orig_extrl, '||
'sum(recv_intrl) recv_intrl, '||
'sum(recv_extrl) recv_extrl '||
'FROM line_usage_sum_view '||
'WHERE TO_CHAR(orig_geo_loc_sys) LIKE ''' || s ||'%'' '||
--'WHERE TO_CHAR(orig_geo_loc_sys) LIKE ''2213003%'' '||
'AND switch_id = ''' || :P1_SWITCH_ID || ''' ' ||
'AND call_start_date > TO_DATE(''30-NOV-1999'') ' ||
'AND call_clear_time > TO_DATE(''30-NOV-1999'') '||
'AND '||
:SORTFIELD||' BETWEEN '||
'TO_DATE(:STARTDATE,''dd-MON-YYYY HH24:MI'') AND '||
'TO_DATE(:STOPDATE, ''dd-MON-YYYY HH24:MI'') '||
'GROUP BY GROUPING SETS (orig_geo_loc_sys)';
-- htp.p('SQL query: '||q);
RETURN q;
END LOOP;
END;</b>
Thank you,
Laura

Laura,
First, I would be careful of introducing SQL Injection possibilities. Any time I see
'Select ... ' || :P123_FOO || ' ... '
I worry about sql injection. In your case you are converting :P50_SELECTED_INSTRUMENTS into selected_items and then selected_items into s. So when I see
'WHERE TO_CHAR(orig_geo_loc_sys) LIKE ''' || s ||'%'' '||
I think, "I could use sql Injection and hack this."
So, I would do some validation on :P50_SELECTED_INSTRUMENTS or some other method to avoid this.
I'm not certain I understand your query. Do you really intend to allow the user to select the beginning of a string and then find all rows that start with that string? Or, do you just want to let them find when it matches the string. This is one way if you want to do matching:
DECLARE
selected_items HTMLDB_APPLICATION_GLOBAL.VC_ARR2;
s VARCHAR2(32767);
q varchar2(32767);
BEGIN
-- Change the : separate string to be comma separated with quoted strings
s := '''' || replace(:P50_SELECTED_INSTRUMENTS, ',', ''',''')|| '''' ;
-- htp.p('COUNT: '||selected_items.count);
q:= 'SELECT '||
'SUBSTR(orig_geo_loc_sys,1,INSTR(orig_geo_loc_sys,''-'')-1) AS INSTRUMENT, '||
'SUBSTR(orig_geo_loc_sys,INSTR(orig_geo_loc_sys,''-'')+1, LENGTH'||
' (orig_geo_loc_sys)) AS ORIG_LINENUM, '||
'sum(orig_intrl) orig_intrl, '||
'sum(orig_extrl) orig_extrl, '||
'sum(recv_intrl) recv_intrl, '||
'sum(recv_extrl) recv_extrl '||
'FROM line_usage_sum_view '||
'WHERE TO_CHAR(orig_geo_loc_sys) in (' || s ||' ) '||
--'WHERE TO_CHAR(orig_geo_loc_sys) LIKE ''2213003%'' '||
'AND switch_id = ''' || :P1_SWITCH_ID || ''' ' ||
'AND call_start_date > TO_DATE(''30-NOV-1999'') ' ||
'AND call_clear_time > TO_DATE(''30-NOV-1999'') '||
'AND '||
:SORTFIELD||' BETWEEN '||
'TO_DATE(:STARTDATE,''dd-MON-YYYY HH24:MI'') AND '||
'TO_DATE(:STOPDATE, ''dd-MON-YYYY HH24:MI'') '||
'GROUP BY GROUPING SETS (orig_geo_loc_sys)';
-- htp.p('SQL query: '||q);
RETURN q;
END;
If you want to do something more like you originally stated, try this:
DECLARE
selected_items HTMLDB_APPLICATION_GLOBAL.VC_ARR2;
s VARCHAR2(20);
q varchar2(32767);
BEGIN
selected_items := HTMLDB_UTIL.STRING_TO_TABLE(:P50_SELECTED_INSTRUMENTS);
-- htp.p('COUNT: '||selected_items.count);
q:= 'SELECT '||
'SUBSTR(orig_geo_loc_sys,1,INSTR(orig_geo_loc_sys,''-'')-1) AS INSTRUMENT, '||
'SUBSTR(orig_geo_loc_sys,INSTR(orig_geo_loc_sys,''-'')+1, LENGTH'||
' (orig_geo_loc_sys)) AS ORIG_LINENUM, '||
'sum(orig_intrl) orig_intrl, '||
'sum(orig_extrl) orig_extrl, '||
'sum(recv_intrl) recv_intrl, '||
'sum(recv_extrl) recv_extrl '||
'FROM line_usage_sum_view '||
'WHERE 1=1 ';
FOR i in 1..selected_items.count LOOP
s := TO_CHAR(selected_items(i));
q := q || ' and TO_CHAR(orig_geo_loc_sys) LIKE '''|| s ||'%'' ' ;
END LOOP;
q := q || ||'%'' '||
--'WHERE TO_CHAR(orig_geo_loc_sys) LIKE ''2213003%'' '||
'AND switch_id = ''' || :P1_SWITCH_ID || ''' ' ||
'AND call_start_date > TO_DATE(''30-NOV-1999'') ' ||
'AND call_clear_time > TO_DATE(''30-NOV-1999'') '||
'AND '||
:SORTFIELD||' BETWEEN '||
'TO_DATE(:STARTDATE,''dd-MON-YYYY HH24:MI'') AND '||
'TO_DATE(:STOPDATE, ''dd-MON-YYYY HH24:MI'') '||
'GROUP BY GROUPING SETS (orig_geo_loc_sys)';
-- htp.p('SQL query: '||q);
RETURN q;
END;
Hope this helps...
Anton

How to user HTMLDB_COLLECTIONS in a PL/SQL function?

Hi!
I am trying to validate/check that an item from an updateable report is not in a collection before I add the item. I wrote a PL/SQL function to do this. It appears that a "normal" PL/SQL function cannot correctly see the HTMLDB_COLLLECTIONS view. The idea for this was posted to the forum about using a decode statement and a function to detemine whether to display the link value or not for each row. If an item has been added to the collection I want to not display the link. If the item is removed from the collection I need to redisplay the link.
I have tried validations but they depend on the page being resumbitted. And I tried to force that using a branch where the request was = 'ADD'. However everytime I set it up as a redirect to a URL it kept resetting the branch to a 'branch to a page within this application'.
Any suggestions on how to check the HTMLDB_COLLECTIONS before it is updated?
This is V2.0.
Thanks! Dave Venus

Hi!
Yes things work much better when run from within the "application". I had been trying to test the code from both SQL*Plus and from the SQL WorkShop.
The problem I was having even within the application was that I was getting back NULL return values. So I put some exception handling into the code and trapped the
data_not_found and others exceptions. I also had SQL%NOTFOUND code in my function. It appears the SQL%NOTFOUND was not being processed and the exception code was. I can paste the code I was using here if you want.
I went another way around the problem and rewrote the query (esp. the where clause) for the report and it seems to be doing what I need now. I would love to get the function working and understand how it should have worked so that I can use that functionallity in other applications.
Thanks!
Dave Venus

For each row of a table call a pl/sql function

Hi,
i have a search form in adf like this:
parameter1:___
parameter2:____
buttonSearch
Table with results
field1 field2 field3
row1 ------ --------- -------
row2 ------ --------- -------
row3 ------ --------- -------
The user inputs the parameters 1 and 2 then press buttonSearch and the query execute and returns rows 1 to 3.
What i need is for each row call pl/sql function and passed the parameter 1 and 2 and field 1 to 3 (plsql function recives 5 parameters (parameter1, parameter2, field1 , field2 and field3) )
my buttonSearch call a java class that execute ExecuteWithParamters method.
I create the call to my plsql function on Application module class and then export as a java interface.
So i have the function to use in the viewcontroller layer, but i don't know where to use it, and how to pass the paramters: the parameter 1 and 2 that user inputs and the row fields.....
any ideas....
thanks!!

Hi,
for this you need to call the PLSQL function upon table rendering, which means that you need a field in the table referencing a managed bean. In the managed bean you can use #{row} and resolve it using a ValueExpression. #{row} gives you access to the current rendered row (this is why you need to do it when the table renders) and thus allows you to call getAttribute(name) to get the values of field 1 - 3. The search field value you should get through the bindings reference (assuming the search form uses ADF). Then you create an operation binding for the executeWithParameters and call operationBindingName.getParamsMap().put(argname, argvalue); on it.
Frank
Ps.: I am concerned about the performance you get and wonder if it isn't possible to create a transient attribute that executes the function and displays the results. As I understand, the search parameters are only to filter the result set, which you still can do

Data passed from Java to a PL/SQL function

OK, I am at the beginner-to-intermediate level when it comes to PL/SQL...and a virtual novice when it comes to Java, so please forgive me if my question seems strange or naive.
We are developing a Java-based web app. Some of the screens are reminiscent of a spreadsheet-type layout--potentially LOTS of rows and LOTS of columns (actually, an open-ended # of rows/columns--the rows represent a particular 'thing' and the columns represent the dates on which that 'thing' occurred), where the user can check off any and all the cells that apply. We are also 'auditing' all database activity (who did what to the data, and when). The approach we have chosen is as follows: When the user clicks save, the first thing we do is 'inactivate' all of the existing active rows in the db for that screen, whether or not they have changed(i.e., there is an 'active' flag that we set to false & we also save the user id of the person making the change and the date time of the change). Then the Java code calls the database once per row/contiguous column unit. For instance, on a single row, if the user checks off columns 1-4 and 7-15 and column 21 the Java code will send three calls to the database for that row with info about the start/stop column for each unit.
OK--here is my concern...the majority of the time there will be a reasonably small #of 'units'. Occasionally there will be a moderate, and once in awhile a LARGE # of 'units'. So, let's take an extreme case and say that currently the db has 200 such row/contiguous column units for a given screen. The user goes into that screen and adds the 201st unit, then clicks Save. Java calls the db to first inactivate the 200 rows. Then one by one Java calls the db to add the rows back in, only on row #40, the internet connection is lost. The only way the user can tell what happened is to look at the entire screen and reverify ALL the data in order to tell what got saved/resaved and what didn't. Obviously this is a bad situation that we want to avoid.
I should add that the reason this approach was chosen is that the Java developers thought it would be way too complex in this situation to track the changes the user made on the screen, sort out what has been added/modified/deleted, and only call the db for changes.
So given this background, can anyone offer any suggestions/ideas on how we can prevent unintended data loss given the auditing constraints and concern about minimizing complexity on the Java side (redesigning the GUI screen is NOT an option, as the users are very attached to this design)?
Is there a way for Java to pass a multidimensional (& variable-sized) array to a PL/SQL function? Can Oracle's complex (row, table) data types be used in a function's parameter string, and if so...how?...Or else is there some way we can send a single call to the db on a screen like this one? We thought of calling the db once and sending one very large string packed with all the data (could get very big, not sure of Oracle's limatations on passed data--is it 32K for varchar2? )
Advice, ideas, even random thoughts would be greatly appreciated!

Tracy,
<quote>â¦ only on row #40, the internet connection is lost. The only way the user can tell what happened is to look at the entire screen and reverify ALL the data in order to tell what got saved/resaved and what didn't</quote>
That would be "the only way" if the Java programmers had decided to totally bury the error â¦ what should happen here is the end-user should get a proper error message â¦ no need to re-verify at this point since all bets are off (in fact, nothing shouldâve been changed in the database).
Nonetheless, your concerns regarding the chosen approach are valid â¦ making multiple calls to the database after a Save makes controlling a business transaction rather complex (in an n-tier system) â¦ they should make one call only to the database passing in all the data to be saved â¦ one database transaction â¦ either all changes get applied or none.
The fact that lots of data may need to be passed in for one Save is of no relevance â¦ if you have to carry 500Kb of data for that screen then, well, you have to do it â¦ 1 bucket of 500Kb or 50 buckets of 10Kb? â¦ it doesnât really matter as far as the actual transport part is concerned.
As already answered, one can pass complex types in pl/sql â¦ so one database call per Save would be my first random thought. There are lots of suspect things in here â¦ some things you didnât mentioned at all â¦ like how is concurrency addressed? â¦ have you/they implemented some sort of optimistic locking mechanism? Of course the architecture of your system wonât be solved in a forum â¦ so, finding a good system/data/Oracle architect on site for the project would be my second random thought.
PS. One last random thought â¦ fight off that idea of packing and sending large strings (XML or not) â¦ if the data is structured then contemplate XML for no more than few seconds.

T-SQL functions in ORACLE

Use T-SQL functions in ORACLE too, so you can write one script version that works in SYBASE, SQLSERVER and ORACLE when need to use the following functions:
ceiling, charindex, dateadd, datediff, datename, datepart, day, db_id, db_name, getdate, host_id, host_name, left, len, month, replicate, right, space, str, str_replace, stuff, substring, suser_id, suser_name, user_id, user_name and year.
The file [comp_tsql_ORA_schema.txt|http://forums.databasejournal.com/attachment.php?attachmentid=564&d=1258547015] creates the tablespace and schema to put the objects, and the file [comp_tsql_ORA.txt|http://forums.databasejournal.com/attachment.php?attachmentid=569&d=1259256898] creates the functions into the new schema. They will be avaliable for any schema of the oracle instance.
Hope this help!
Any suggestion please contact.
aklein2003
Edited by: user1958693 on 26/11/2009 10:16

jgarry wrote:
J2EE beta released around 1999. Religious programming wars far preceded that. Here's but [one example|http://groups.google.com/group/comp.lang.lisp/browse_thread/thread/1c50bc13e9302f00/34bb8f3ac77e2388?q=programming+language+religion#34bb8f3ac77e2388] from a quick google. I'm no longer convinced that elegance is a desirable trait for a language. So call me a heretic.
Ever tried XLISP? :-)
It's different. Back then it was simply the language itself. And it was not really a religious kind of battle, but more a mine-is-bigger-and-better-than-yours one.
With Java, it is all about The Acrhitecture and the blessed language of Java. ;-)
I wouldn't so much blame J2EE believers for that. I think there is a fundamental complexity to web type paradigms that hasn't been described in a precise and encompassing enough manner to enable correct project planning.The basic problem is that the database is treated as a persistent storage layer only. A bit bucket. And that is exactly what a database is not.
But as it is used that, loads of database features now need to be duplicated in the application layer. Which is done. Very poorly.
Heck, I've even been told that J2EE's 3-tier architecture is not client server. Such ignorance is really unbelievable. Or is it pure stupidity?Until you just made me look it up, I would have thought that too, I'm sure I saw client/server defined as two-tier in the last century. But I have much more stupid misdefinitions to deal with on a daily basis, simply from the multiplicity of paradigms. I tend to retreat into my shell. Hehehe. Know that feeling... These days I rather run away, or set shields to full power, than try to get in yet another frustrating, fruitless and futile discussion with a Java head about the wonders of the J2EE architecture. Or trying to explain that this "new" architecture is predated with what we used in the 80's on mainframe systems with transaction monitors, block devices (the 80's web browsers) and databases - which software layer for software layer was almost identical to "+The Great Architecture+" of today. (and we did it without buzzwords too) ;-)
Client-server is a software architecture and in essence describes 3 basic software components. The User Interface (UI), the Application (APP) and the Database (DB). And there are various ways to put these components together in a client-server architecture. The mistake that the Java fanbois make is thinking that client-server means having the client as the UI and APP as a single component (so-called fat client) and then the DB component as the server.
There are numerous versions of these - including the APP component being a server component (as we now these days more commonly used in over the web).
My beef with the J2EE "+religion+" always has been that client-server is client-server. The fundamentals stay the same.

SQL, PL/SQL functions, and ORA-04091 table is mutating

Dears,
Recently a question came up in an Oracle French forum about an insert/select that is throwing ORA-04091: table xxxx is mutating, trigger/function may not see it error in 11g while the same insert/select was working very well in 10g. The original poster gave a scenario that is easily reproducible. I am wondering what database release is correct the one throwing the error(11g) or the other one accepting the insert/select(10g)?
*10g*
mhouri.world > select * from v$version;
BANNER
Oracle Database 10g Enterprise Edition Release 10.2.0.4.0 - 64bi
PL/SQL Release 10.2.0.4.0 - Production
CORE     10.2.0.4.0     Production
TNS for Solaris: Version 10.2.0.4.0 - Production
NLSRTL Version 10.2.0.4.0 - Production
mhouri.world > create table t_read_consistency (id number, vc varchar2(15), primary key (id));
Table created.
mhouri.world > insert into
2        t_read_consistency
3      select
4        rownum id,
5        rpad('a',15,'a')
6      from
7        dual
8      connect by
9        level<=1000;
1000 rows created.
mhouri.world > commit;
Commit complete.
mhouri.world > create or replace function f_read_consistency return varchar2
2       as
3        lv_vc t_read_consistency.vc%type;
4       begin
5         select trc.vc
6         into lv_vc
7          from t_read_consistency trc
8         where trc.id = 70 ;
9        return lv_vc;
10     end f_read_consistency;
11    /
Function created.
mhouri.world >insert into
2        t_read_consistency (id, vc)
3      select
4         1001
5        ,f_read_consistency
6      from dual;
      ,f_read_consistency
ERROR at line 5:
ORA-04091: table MHOURI.T_READ_CONSISTENCY is mutating, trigger/function may not see it
ORA-06512: at "MHOURI.F_READ_CONSISTENCY", line 5
_11g_
mohamed@mhouri> select * from v$version;
BANNER
Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - Production
PL/SQL Release 11.2.0.1.0 - Production
CORE     11.2.0.1.0     Production
TNS for 32-bit Windows: Version 11.2.0.1.0 - Production
NLSRTL Version 11.2.0.1.0 - Production
mohamed@mhouri> create table t_read_consistency (id number, vc varchar2(15), primary key (id));
Table created.
mohamed@mhouri> insert into
2    t_read_consistency
3 select
4    rownum id,
5    rpad('a',15,'a')
6 from
7    dual
8 connect by
9    level<=1000;
1000 rows created.
mohamed@mhouri> create or replace function f_read_consistency return varchar2
2   as
3    lv_vc t_read_consistency.vc%type;
4   begin
5     select trc.vc
6     into lv_vc
7      from t_read_consistency trc
8     where trc.id = 70 ;
9    return lv_vc;
10 end f_read_consistency;
11 /
Function created.
mohamed@mhouri> insert into
2    t_read_consistency (id, vc)
3 select
4     1001
5    ,f_read_consistency
6 from dual;
,f_read_consistency
ERROR at line 5:
ORA-04091: table MOHAMED.T_READ_CONSISTENCY is mutating, trigger/function may
not see it
ORA-06512: at "MOHAMED.F_READ_CONSISTENCY", line 5 So far so good. Same behaviour for both releases. But let's bring a small change to the PL/SQL function to be as close as the example given in the French Forum
_10g where the select/insert was working without error_:
mhouri.world > select * from v$version;
BANNER
Oracle Database 10g Enterprise Edition Release 10.2.0.4.0 - 64bi
PL/SQL Release 10.2.0.4.0 - Production
CORE     10.2.0.4.0     Production
TNS for Solaris: Version 10.2.0.4.0 - Production
NLSRTL Version 10.2.0.4.0 - Production
mhouri.world > create type t_read_cs as object (id number,vc varchar2(15));
2 /
Type created.
mhouri.world > create type t_read_cs_tab as table of t_read_cs;
2 /
Type created.
mhouri.world > create or replace function f_read_consistency_tab
2 return t_read_cs_tab
3    as
4      lc_t_read_cs_tab t_read_cs_tab := t_read_cs_tab();
5      j binary_integer := 0;
6    begin
7        for x in (select
8                      id,
9                      vc
10                   from t_read_consistency trs
11                  where trs.id <= 10
12       ) loop
13
14          j := j +1;
15          lc_t_read_cs_tab.extend;
16          lc_t_read_cs_tab(j) := t_read_cs(x.id, x.vc);
17     end loop;
18     RETURN lc_t_read_cs_tab;
19   end f_read_consistency_tab;
20 /
Function created.
mhouri.world > select count(1) from t_read_consistency;
COUNT(1)
      1000
mhouri.world > select count(1)
2 from (select * from table(f_read_consistency_tab));
COUNT(1)
        10
mhouri.world > insert into t_read_consistency
2         (id,vc)
3      select id,vc
4   from table(f_read_consistency_tab)
5 ;
10 rows created.
mhouri.world > select count(1) from t_read_consistency;
COUNT(1)
      1010
_11g where the same insert/select is throwing an error:_
mohamed@mhouri> select * from v$version;
BANNER
Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - Production
PL/SQL Release 11.2.0.1.0 - Production
CORE    11.2.0.1.0      Production
TNS for 32-bit Windows: Version 11.2.0.1.0 - Production
NLSRTL Version 11.2.0.1.0 - Production
mohamed@mhouri> create type t_read_cs as object (id number,vc varchar2(15));
2 /
Type created.
mohamed@mhouri> create type t_read_cs_tab as table of t_read_cs;
2 /
Type created.
mohamed@mhouri> create or replace function f_read_consistency_tab
2      return t_read_cs_tab
3        as
4          lc_t_read_cs_tab t_read_cs_tab := t_read_cs_tab();
5          j binary_integer := 0;
6        begin
7            for x in (select
8                          id,
9                          vc
10                      from t_read_consistency trs
11                     where trs.id <= 10
12          ) loop
13
14            j := j +1;
15             lc_t_read_cs_tab.extend;
16             lc_t_read_cs_tab(j) := t_read_cs(x.id, x.vc);
17        end loop;
18        RETURN lc_t_read_cs_tab;
19      end f_read_consistency_tab;
20    /
Function created.
mohamed@mhouri> select count(1) from t_read_consistency;
COUNT(1)
      1000
mohamed@mhouri> select count(1) from (select * from table(f_read_consistency_tab));
COUNT(1)
        10
mohamed@mhouri> insert into t_read_consistency
2            (id,vc)
3          select id,vc
4      from table(f_read_consistency_tab)
5     ;
    from table(f_read_consistency_tab)
ERROR at line 4:
ORA-04091: table MOHAMED.T_READ_CONSISTENCY is mutating, trigger/function may
not see it
ORA-06512: at "MOHAMED.F_READ_CONSISTENCY_TAB", line 7 In addition, one of the posters spotted out very judiciously that if we slightly change the definition of the table t_read_consistency in 11g, strangely the insert/select will work correctly in this data base as shown below:
ohamed@mhouri> drop table tr_read_consistency;
Table dropped.
mohamed@mhouri> create table tr_read_consistency
2      as select rownum rn,
3                trs.*
4      from
5         t_read_consistency trs;
Table created.
mohamed@mhouri> insert into tr_read_consistency
2                 (rn, id,vc)
3              select rownum, id,vc
4           from table(f_read_consistency_tab);
10 rows created.So is this a regression? or a corrected bug during upgrade?
Thanks in advance
Mohamed Houri

I just followed the doc links provided by Tubby, which have 100% Correct answer. See below :
SQL> select * from v$version;
BANNER
Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - Production
PL/SQL Release 11.2.0.1.0 - Production
CORE    11.2.0.1.0      Production
TNS for 32-bit Windows: Version 11.2.0.1.0 - Production
NLSRTL Version 11.2.0.1.0 - Production
SQL> show user;
USER is "SCOTT"
SQL> create type t_read_cs as object (id number,vc varchar2(15));
2 /
Type created.
SQL> create type t_read_cs_tab as table of t_read_cs;
2 /
Type created.
SQL> create table t_read_consistency (id number, vc varchar2(15), primary key (id));
Table created.
SQL> create or replace function f_read_consistency_tab
2 return t_read_cs_tab
3 as
4 lc_t_read_cs_tab t_read_cs_tab := t_read_cs_tab();
5 j binary_integer := 0;
6 begin
7 for x in (select
8 id,
9 vc
10 from t_read_consistency trs
11 where trs.id <= 10
12 ) loop
13 j := j +1;
14 lc_t_read_cs_tab.extend;
15 lc_t_read_cs_tab(j) := t_read_cs(x.id, x.vc);
16 end loop;
17 RETURN lc_t_read_cs_tab;
18 end f_read_consistency_tab;
19 /
Function created.
SQL> insert into
2 t_read_consistency
3 select
4 rownum id,
5 rpad('a',15,'a')
6 from
7 dual
8 connect by
9 level<=1000;
1000 rows created.
SQL> select count(1) from t_read_consistency;
COUNT(1)
      1000
SQL> select count(1) from (select * from table(f_read_consistency_tab));
COUNT(1)
        10
SQL> insert into t_read_consistency
2 (id,vc)
3 select id,vc
4 from table(f_read_consistency_tab);
from table(f_read_consistency_tab)
ERROR at line 4:
ORA-04091: table SCOTT.T_READ_CONSISTENCY is mutating, trigger/function may not see it
ORA-06512: at "SCOTT.F_READ_CONSISTENCY_TAB", line 7
SQL> ed
Wrote file afiedt.buf
1 create or replace function f_read_consistency_tab
2 return t_read_cs_tab
3 as
4 lc_t_read_cs_tab t_read_cs_tab := t_read_cs_tab();
5 j binary_integer := 0;
6 PRAGMA AUTONOMOUS_TRANSACTION; <--- This works as documented in 11.2.0.1
7 begin
8 for x in (select
9 id,
10 vc
11 from t_read_consistency trs
12 where trs.id <= 10
13 ) loop
14 j := j +1;
15 lc_t_read_cs_tab.extend;
16 lc_t_read_cs_tab(j) := t_read_cs(x.id, x.vc);
17 end loop;
18 RETURN lc_t_read_cs_tab;
19* end f_read_consistency_tab;
SQL> /
Function created.
SQL> insert into t_read_consistency
2 (id,vc)
3 select id,vc
4 from table(f_read_consistency_tab);
insert into t_read_consistency
ERROR at line 1:
ORA-00001: unique constraint (SCOTT.SYS_C0011307) violated
SQL> drop table t_read_consistency purge;
Table dropped.
SQL> create table t_read_consistency (id number, vc varchar2(15));
Table created.
SQL> insert into
2 t_read_consistency
3 select
4 rownum id,
5 rpad('a',15,'a')
6 from
7 dual
8 connect by
9 level<=1000;
1000 rows created.
SQL> commit;
Commit complete.
SQL> insert into t_read_consistency
2 (id,vc)
3 select id,vc
4 from table(f_read_consistency_tab);
10 rows created.
SQL> commit;
Commit complete.
SQL>So, you have to add only PRAGMA AUTONOMOUS_TRANSACTION; before begin in your function code to avoid ORA-04091 in 11.2.0.1
But, All thanks to Tubby who pointed us to the correct documentation link.
Regards
Girish Sharma

How to find out PL/SQL function usage in Discoverer workbooks?

We have to make changes to one PL/SQL function that has been registered in Administrator. Is it possible to find out in which reports the function is used? The function is called in calculations and it returns certain time information. We have to add one argument to the function and therefore have to change calculations where the function is called. We have too many workbooks to manually check them all.
I am aware how to find out which folders and items are used in workbooks but can't find the same information about functions. We are using Discoverer 10.1.2.2
Thanks in advance!

Hi,
If you calculation is in the EUL you can look at the item dependencies, but if your calculation is in the workbook there is no way other than using the workbook dump (d51wkdmp.exe) utility or opening the workbook and manually checking.
You might want to consider overlaying the PL/SQL function definition so that there are 2 variants of the function. You can then add the extra parameter as an optional parameter in Discoverer. The correct variant will be called depending on how many parameters are used.
Hope that helps,
Rod West

EBS ISG using custom PL/SQL functions that return XMLType

Hi,
We have a custom PL/SQL package that we use for interfacing systems and some of the functions in this package ruturn an XMLType. We want to deploy the package functions as web services through the ISG, but it is not working as expected. When deployed through the ISG, the functions with XMLType return type produce a null response from the ISG (they work fine when called in SQL or PL/SQL; functions with non-XMLTypes work fine).
If we change the return type to CLOB (and use getClobVal() on the XMLType) then we get a response from the ISG, but it changes all the angle-brackets in the CLOB (which is still arbitrary XML text) to < > ...
What is the proper way to get the complex XMLType output through the ISG? Anyone have any more experience?
Thanks,
--Walt

Alex,
For the predicate groups that are indexed/stored, the exact operator types (as in equality, inequality, like etc) that are indexed are specified while assigning the default index parameters. In the following example, exf$indexoper is used to specify the list of indexed operators.
BEGIN
DBMS_EXPFIL.DEFAULT_INDEX_PARAMETERS('Car4Sale',
    exf$attribute_list (
       exf$attribute (attr_name => 'HorsePower(Model, Year)',
                      attr_oper => exf$indexoper('=','<','>','>=','<='),
                      attr_indexed => 'FALSE')    --- stored predicate group
END;
/You can find more information about exf$indexoper at
http://download.oracle.com/docs/cd/B28359_01/appdev.111/b28419/t_expfilobj.htm#ARPLS153
Could you confirm that you chose to index 'is null' and 'is not null' while assigning the default index parameters ? This information is available in OPERATOR_LIST column of the USER_EXPFIL_DEF_INDEX_PARAMS view.
Hope this helps,
-Aravind.

How to use Pl/Sql function in Discoverer report

Hi Everybody,
Would you be able to provide a workflow/guidance how to use sql function in Discoverer right from creating the function using PLSQL Developer.
(I am new to Oracle environment... prev. worked as analyst with Microsoft Access... took PL/Sql class in school a few years ago)
We are using Discoverer Desktop.
a) Just emailed dba for permission to create pl/sql function.
b) Recently got Discoverer Administrator installed on my pc.
can you pls add to the list:
c)....
d)....
etc.?
tx for your help, sandra

Hi Kranthi,
Thx for your warm words....
The links you provided are great! Thx soooo much!
Thanks to you and the rest of forum members, am able to use what you've all taught me
with other Discoverer reports.... so i was able to do more on my own....
About a month ago, I got access to Discoverer Admin, and was able to paste SQL to create a custom folder.
Then i joined the custom folder to an existing baseline folder.
I have found that sometimes when I use calc within a calc when the calcs use analytic functions, that Discoverer
just displays null in the column. I thought that maybe instead I could create my own function as a work-around.
I wrote IT for permissions to create views, functions etc. Hopefully, they'll approve and can move toward learning
pl/sql functions for Discoverer.
tx again, sandra

How to call pl/sql function from element values

EBS 11.5.10.2
XMLP 5.6.3
Hello,
I noticed that the output of the rdf-to-data template conversion process makes use of an undocumented feature of data templates, and I would like to get input from experts as to which situations this feature is usable.
The closest thing I can find in the documentation is a sample in the user guide. There is a "General Ledger Journals Data Template Example" that has a <dataStructure> section that contains <element> nodes which are NOT children of a <group> node. I can't find any explanation of this in the user guide.
I've noticed from converted templates that in these un-grouped elements you can make calls to PL/SQL functions in the "value" attribute, like this:
<dataStructure>
<group name="G_LINES" source="Q_MAIN">
    <element name="Line_Num"           value="Line_Num"/>
</group>
<element name="C_CALCULATED_VALUE" dataType="number" value="XX_CUSTOMPROCS.SOME_FUNCTION"/>
</dataStructure>Has anyone had any success being able to call PL/SQL functions from grouped elements? Whenever I try, it doesn't seem to work.
When I try something like this:
<dataStructure>
<group name="G_LINES" source="Q_MAIN">
    <element name="Line_Num"           value="Line_Num"/>
    <element name="some_calculation"   value="XX_CUSTOMPROCS.SOME_FUNCTION"/>
    <element name="some_calculation_b" value="XX_CUSTOMPROCS.SOME_FUNCTION_B(:Line_Num)"/>
</group>
<element name="C_CALCULATED_VALUE" dataType="number" value="XX_CUSTOMPROCS.SOME_FUNCTION"/>
</dataStructure>The <SOME_CALCULATION/> and <SOME_CALCULATION_B/> nodes come out empty in the output data xml file, but <C_CALCULATED_VALUE> would have a value as desired.

ah - perfect. That makes sense. Thank you for the response!
But what about when we need to pass parameters to those functions whos values are the results of aggregate element values?
This happens a lot in the converted data templates, where pl/sql package functions are meant to replace formula columns from the original Oracle Report. Take this example from the conversion of ARXAGMW.rdf (Aging Report, 7 Buckets):
(note the function call in the value of "Set_Percent_Inv_Inv" is using aggregate results from subgroups)
<group name="G_INV_INV" dataType="varchar2" source="Q_Invoice">
    <element name="Total_Inv_Inv_Amt" function="sum" dataType="number" value="G_Invoice.C_Amt_Due_Rem_Inv"/>
    <element name="Total_Inv_Inv_B0" function="sum" dataType="number" value="G_Invoice.C_Inv_B0"/>
    <element name="Total_Inv_Inv_B1" function="sum" dataType="number" value="G_Invoice.C_Inv_B1"/>
    <element name="Total_Inv_Inv_B2" function="sum" dataType="number" value="G_Invoice.C_Inv_B2"/>
    <element name="Total_Inv_Inv_B3" function="sum" dataType="number" value="G_Invoice.C_Inv_B3"/>
    <element name="Total_Inv_Inv_B4" function="sum" dataType="number" value="G_Invoice.C_Inv_B4"/>
    <element name="Total_Inv_Inv_B5" function="sum" dataType="number" value="G_Invoice.C_Inv_B5"/>
    <element name="Total_Inv_Inv_B6" function="sum" dataType="number" value="G_Invoice.C_Inv_B6"/>
    <element name="Set_Percent_Inv_Inv" dataType="number" value="XX_CUSTOMPROCS.XXC_ARXAGMW.set_percent_inv_invformula(:Total_Inv_Inv_Amt, :Total_Inv_Inv_B0, :Total_Inv_Inv_B1, :Total_Inv_Inv_B2, :Total_Inv_Inv_B3, :Total_Inv_Inv_B4, :Total_Inv_Inv_B5, :Total_Inv_Inv_B6)"/>
    <element name="Sum_Percent_B0_Inv" dataType="number" value="XX_CUSTOMPROCS.XXC_ARXAGMW.Sum_Percent_B0_Inv_p"/>
    <element name="Sum_Percent_B1_Inv" dataType="number" value="XX_CUSTOMPROCS.XXC_ARXAGMW.Sum_Percent_B1_Inv_p"/>
    <element name="Sum_Percent_B2_Inv" dataType="number" value="XX_CUSTOMPROCS.XXC_ARXAGMW.Sum_Percent_B2_Inv_p"/>
    <element name="Sum_Percent_B3_Inv" dataType="number" value="XX_CUSTOMPROCS.XXC_ARXAGMW.Sum_Percent_B3_Inv_p"/>
    <element name="Sum_Percent_B4_Inv" dataType="number" value="XX_CUSTOMPROCS.XXC_ARXAGMW.Sum_Percent_B4_Inv_p"/>
    <element name="Sum_Percent_B5_Inv" dataType="number" value="XX_CUSTOMPROCS.XXC_ARXAGMW.Sum_Percent_B5_Inv_p"/>
    <element name="Sum_Percent_B6_Inv" dataType="number" value="XX_CUSTOMPROCS.XXC_ARXAGMW.Sum_Percent_B6_Inv_p"/>
    <group name="G_Cust_Inv" dataType="varchar2" source="Q_Invoice">
      <group name="G_Site_Inv" dataType="varchar2" source="Q_Invoice">
        <group name="G_1" dataType="varchar2" source="Q_Invoice">
          <group name="G_Invoice" dataType="varchar2" source="Q_Invoice">
            <element name="C_Amt_Due_Rem_Inv" dataType="number" value="C_Amt_Due_Rem_Inv"/>
            <element name="C_Inv_B0" dataType="number" value="C_Inv_B0"/>
            <element name="C_Inv_B1" dataType="number" value="C_Inv_B1"/>
            <element name="C_Inv_B2" dataType="number" value="C_Inv_B2"/>
            <element name="C_Inv_B3" dataType="number" value="C_Inv_B3"/>
            <element name="C_Inv_B4" dataType="number" value="C_Inv_B4"/>
            <element name="C_Inv_B5" dataType="number" value="C_Inv_B5"/>
            <element name="C_Inv_B6" dataType="number" value="C_Inv_B6"/>
          </group>
        </group>
      </group>
    </group>
</group>
...All of these groups and sub-groups are based on one single query, so I am not sure how I would move the function call into the query without changing the results of the function.
In the example above, elements Sum_Percent_B0_Inv through Sum_Percent_B6_Inv grab the results of the calculation done in set_percent_inv_invformula. Here is the essence of that function:
sum_percent_b0_inv := ROUND ((total_inv_inv_b0 / total_inv_inv_amt) * 100, 2);
sum_percent_b1_inv := ROUND ((total_inv_inv_b1 / total_inv_inv_amt) * 100, 2);
sum_percent_b2_inv := ROUND ((total_inv_inv_b2 / total_inv_inv_amt) * 100, 2);
sum_percent_b3_inv := ROUND ((total_inv_inv_b3 / total_inv_inv_amt) * 100, 2);
sum_percent_b4_inv := ROUND ((total_inv_inv_b4 / total_inv_inv_amt) * 100, 2);
sum_percent_b5_inv := ROUND ((total_inv_inv_b5 / total_inv_inv_amt) * 100, 2);
sum_percent_b6_inv := ROUND ((total_inv_inv_b6 / total_inv_inv_amt) * 100, 2);The only solution I can think of is to have separate queries, one for each subgroup, that do the "sum" in sql; but that seems terribly inefficient.

Using Package to produce pl/sql function body returning sql query Report

I have existing code that we want to use in building reports in APEX. We are needing to modify it slightly to handle some new requirements, but would like to use them in reports based upon SQL query (pl/sql function body returning sql query) functionality.
Any suggestions as how to call these in an APEX report region?
Thank you,
Tony Miller
UTMB/EHN

Hi Tony-
I am also v new to Apex and you may have answered a question I was in search of. I, however, now have a couple more. First a bit of background-- Like in your situation, my college has a lot of existing code (400 +) that we need to get into Apex. The majority of this canned code contains one/both: 1) many lines, sometimes containing multiple select statements 2) imbedded create table & view statements (temp user owned). Up until reading your post, I was unable to figure out an easy way of getting this existing code into the product. Thanks. Now the questions, do you know how I would go about dealing with the create tables/views. I've read some posts on this forum which suggest temporary tables being unstable in this environment. Also, do you know if there's a size limitation when passing sql code via a function?
As you may/may not be able to tell, I'm a bit lost right now... so any info you can provide would be appreciated. Thanks.
Don

How to return less than 4000 characters from pl/sql function in SQL call?

Hello,
Is there a way to limit length for varchar when calling pl/sql function from SQL? No matter how I write it it always returns 4000 bytes.
If there is none, then does it make sense ever to specify lenght of the return variable?
My goal is to encapsulate business rules within pl/sql functions. But if all varchar columns are returned as 4000 it is not feasible solution. Not only this is a performance issue in a data warehousing environment, bet when using those rules within SQL views user experiance would suffer as well. Are we left with the rule hardcoding solution? Also, I think that using SUBSTRING or TRUNC functions on top of business rules function defeats the purpose.
Please see my attempt below. Your thoughts are appreciated.
Thank you.
/* Formatted on 06/11/2009 2:26:41 PM (QP5 v5.126.903.23003) */
CREATE OR REPLACE FUNCTION mytest (myvar_in VARCHAR2)
RETURN VARCHAR2
AS
l_return VARCHAR2 (15);
BEGIN
l_return := 'TEST_' || myvar_in;
RETURN l_return;
END mytest;
CREATE TABLE TEST_ME
AS
SELECT mytest ('ME') AS VERYLONG FROM DUAL;
describe TEST_ME;
RUN ABOVE CODE:
Function created.
Table created.
TABLE TEST_ME
Name Null? Type
VERYLONG VARCHAR2(4000)
Edited by: Ilmars2 on Jun 11, 2009 2:46 PM

Pointless,
Thanks for jumping in on this and I am glad you asked :).
I do not doubt that it is an architectural challenge. Otherwise it would have been done already! I am struggling with the fact that SQL knows what data type the function will return, but does not know the length of it, precision or scale. I will leave it at that.
I will go with some high level requirements to allow for alternative thoughts:
1)     Business defined rules. There are multiple types of business rules. Simple lookups, bucketing, complex calculations, data retrieved from other tables etc. We have about 500 different rules. Some of them are even overloaded – different inputs will produce the same output.
Some simple examples are:
Rule1 - Fruit
     when ‘A’ then ‘Apple’
     When ‘O’ then ‘Orange’
     Else ‘N/A’ end
Rule2 – Bonus
     when STATUS =’Active’ and LEVEL=’CEO’ then bonus=salary*1.0
     when STATUS =’Active’ and LEVEL=’nobody’ then bonus=salary*0.01
     else bonus=0
Rule 3 – Income Bracket
     When more than 0 and less or equal to 30000 then INC_B=’LOW’
     When more than 30000 and less of equal to 60000 then INC_B=’MIDDLE’
Etc.
2)     Challenge: All data users in an organization need to use the same rules (let’s assume data source is Oracle database). How to expose all the rules to different types of users in manageable way? Types of users – analysts, application/web developers, data warehouse teams etc.
3)     Current system: Not only each user has coded their own rules (luckily based on the common specification), but hard-coding is taking place for each query within the confines of one project. The project I just looked at had about 12 modules with 30 hardcoded queries. Oops! Few rules just changed.
My take: I was leaning toward encapsulating business logic within UDF’s. UDF’s provide all the flexibility we need + overloading. All the functions could be consumed by data warehouse team (building summary tables, cubes etc.) and application developers. For power users we could build views by applying the same functions on top of the source data. Thus avoid data duplication. It seemed win-win until this 4000 issue :).
Your thoughts on alternative approaches are appreciated.
Thank you.

About PL/SQL Functions

Similar Messages

Maybe you are looking for