Create record variable that refers dynamic query assigned to a ref cursor?

Question

Hi ,
Just explaining what I am trying to achieve:
1) i have a hr.departments table that was loaded in hr schema on 1st oct 2012 with 4 columns(department_id, department_name, manager_id, location_id)
2) now I have a new schema by my name 'rahul' and I have loaded departments table but now an additional column has come into picture,ie created_date, this table got loaded on 1st-Nov-2012
3) Now going forward my columns could be dropped from the departments table (it can be a case), for example might be my departments table in my schema 'rahul' one day could comprise of only 3 columns(department_id,department_name,manager_id)
4) Now in the next step, I have managed to extract common column names(in a single line where columns are delimited using a comma) from both the tables(hr.departments and rahul.departments) which are (department_id, department_name, manager_id, location_id) using all_tab_cols table and I have written a function for it which i will be pasting below.
5) now going forward, using the above column names line with column names delimited using comma, I have used a ref cursor and assigned a query to it using the line of columns that I have extracted from the above point
6) Now I want to create a record variable which refers to my ref cursor, something like we do when we create a record variable by reffering to an explicit cursor defination that we give in the declaration block.
PS:
1) I have been out of touch with plsql for a long time so I have lost a lot of mmeory regarding plsql.
2) basically I need to compare data in hr.departments table with rahul.departments table for only columns that are common to both the tables, rest new or discarded columns information will go in one of the log tables that I have created(this is done already)
Please help me, I did try searching on google for the same but it really confused me very badly, any kind of help is appreciated, please find my code below:
Regards
Rahul
Code :
===================================================================================================
create or replace procedure p_compare_data(fp_old_table_name in varchar2, fp_new_table_name in varchar2)
is
type ref_cursor_old_table is ref cursor;
v_ref_cursor_old_table ref_cursor_old_table;
--record_v_ref_cursor_old_table v_ref_cursor_old_table;
--record_ref_cursor_old_table v_ref_cursor_old_table%ROWTYPE;
Lv_all_column_names varchar2(2000);
function f_fetch_common_column_names(fp_old_table_name in varchar2, fp_new_table_name in varchar2)
return varchar2
is
          Lv_all_column_names varchar2(2000);
begin
          execute immediate 'select ltrim(all_column_names,'','') all_column_names_in_line from (select * from (select sys_connect_by_path(hr_e_column_name,'','') all_column_names from (select hr_e_column_name, rownum curr, rownum - 1 prev from (select hr_e.* , rahul_e.* , case when (hr_e_column_name = rahul_e_column_name) then 1 when (rahul_e_column_name is NULL) then 2 when (hr_e_column_name is NULL) then 3 end decision from (select column_name hr_e_column_name from all_tab_cols where owner ='''||substr(fp_old_table_name,1,instr(fp_old_table_name,'.',1,1)-1)||''' and table_name = '''||substr(fp_old_table_name,instr(fp_old_table_name,'.',1,1)+1)||''') hr_e full outer join (select column_name rahul_e_column_name from all_tab_cols where owner = '''||substr(fp_new_table_name,1,instr(fp_new_table_name,'.',1,1)-1)||''' and table_name = '''||substr(fp_new_table_name,instr(fp_new_table_name,'.',1,1)+1)||''') rahul_e on hr_e.hr_e_column_name = rahul_e.rahul_e_column_name) decision_table where decision = 1) a start with a.curr = 1 connect by prior curr = prev) b order by length(b.all_column_names) desc) all_column_names_in_line where rownum = 1' into Lv_all_column_names;
          return (Lv_all_column_names);
end;
begin
     Lv_all_column_names := f_fetch_common_column_names(fp_old_table_name, fp_new_table_name);
     --dbms_output.put_line(Lv_all_column_names);
     open v_ref_cursor_old_table for ('select '||Lv_all_column_names||' from '||fp_old_table_name);
end;
=====================================================================================================

Answer 1

>
6) Now I want to create a record variable which refers to my ref cursor, something like we do when we create a record variable by reffering to an explicit cursor defination that we give in the declaration block.
>
This thread is basically nothing more than a continuation of your original thread except now you are finally explaining what you are really trying to do.
Re: passing table name to a procedure and then need to open a cursor ..
In that original thread you said you found the solution
>
Well Mate thanks for your suggestion but I got it working through ref cusror, thanks for your time.
>
So I ask you to post your 'solution' and when you finally did it was clear that you hadn't solved anything at all. Your solution used a ref cursor all right but the code relied on a record ('record_employees') that was based on a table name ('employees') that was declared within the procedure. There isn't much point in passing in a table name if you have to hard-code the table name in the procedure.
create or replace procedure p_ref_cursor(fp_old_table in varchar2)
is
type ref_cursor is REF CURSOR;
v_ref_cursor ref_cursor;
record_employees hr.employees%ROWTYPE;
begin
open v_ref_cursor for 'select * from '||fp_old_table;
loop
fetch v_ref_cursor into record_employees;
exit when v_ref_cursor%NOTFOUND;
dbms_output.put_line(record_employees.employee_id);
end loop;
end;Then sb92075 ask you the question that illustrates what I just said
>
what happens when you pass in "HR.DEPARTMENTS" ; besides throwing errors?
>
And you blew him off with this
>
Mate, departments never came in my context, in my prior message I explained what I was trying to achieve ... so I dont know what problem you are understanding reading my posts.
>
And now here you are asking how to get this to work for the departments table.
It is very difficult to help someone that won't tell us what it is they are really trying to do so we can try to suggest some better ways of doing it. Hopefully, in the future, you wil start by explaining your problems instead of focusing on the solution you think you should use.
Back to the issue -
The first thing you should do is finish defining the requirements. Assuming the above actually works to identify columns that have different data what are you going to do with that information?
1. Do you need to save that different data from TABLE1 somewhere?
2. If you don't save it how will anyone look at it to decide which table has the correct data?
3. If you do save it how will you save it 'generically' since other tables will have different columns and datatypes?
4. What about the data from the same record in TABLE2; do you need to save that data somewhere?
5. Will these two tables have primary keys? Are they on the same columns in each table? If not what if TABLE1 has one record but there are TWO records in TABLE2 that are identical. Is that a match? Or is that a problem because TABLE2 has an extra record even though the record is identical?
In short detecting the differences is just one small part of the entire problem. You also need to save those differences somewhere so someone can examine the data and decide what action to take. That is the more difficult part of trying to implement a 'generic' solution.
But now that we know what you are really trying to do take a read through this thread from 6 years ago. It has three different ways to pass a query to a procedure and get different output. You may want to save a copy of the thread since it has some very advanced techniques in it.
How to pipeline a function with a dynamic number of columns?
See ascheffer's reply Posted: May 9, 2006 4:53 AM for using data cartridge functionality with a pipelined function.
See Kamal's reply Posted: May 10, 2006 4:49 AM - it shows how to get XML output.
See BluShadow's reply Posted: Mar 27, 2009 1:50 AM - for using dynamic sql