it looks like custodian_extract_header is a cursor.
If this is the case, then the simplest way to make sure that your ce_line variable is of the appropriate type is to define it as a record of the cursor record. Declare the variable after you’ve declared the cursor like this :

ce_line custodian_extract_header%ROWTYPE;

You can then fetch the cursor into this record. However, you will have to convert the separate cursor elements into a VARCHAR2 before writing to the file.
So, if for example your cursor is returning a number, a date, and a varchar like this :

CURSOR custodian_extract_header IS
   SELECT my_num, my_date, my_char
   FROM some_table;

… you would need to build the string to write to the file as follows ( assuming that l_buffer is declared as a varchar2(4000) and that you want to output comma separated values)…

...
l_buffer := ce_line.my_num||','||TO_CHAR(ce_line.my_date)||','||ce_line.my_char;
UTL_FILE(ce_file, ce_line, FALSE);

HTH

Mike

If i select only VARCHAR columns for my V_CHR_SQL, the below code runs perfectly fine.

I need to know what data type has to be defined for ce_line so that it accepts any datatype.

Decleare:
ce_line varchar2(20000);
V_DB_DIR_NAME varchar2 (34) := ‘ ‘;
ce_file UTL_FILE.file_type;

Body:

open custodian_extract_header for V_CHR_SQL;

IF UTL_FILE.is_open (ce_file)
THEN
LOOP
FETCH custodian_extract_header
INTO ce_line;

EXIT WHEN custodian_extract_header%NOTFOUND;

UTL_FILE.put_line (ce_file, ce_line, FALSE);
END LOOP;
END IF;

CLOSE custodian_extract_header;
UTL_FILE.fclose (ce_file);

From what you’ve said, I’m assuming that :

1) The structure of the ref cursor you are querying (i.e. the fields it contains) is known at runtime
2) You are running on a unix based os ( hence CRON)
3) You want this job to run at regular intervals ( presumably once per day, overnight).

There are also a couple of questions :

1) Do you have the option of using DBMS_SCHEDULER to run the job entirely from within the database ? ( You can see examples of this here.
2) Are you restricted as to where you can create the output file ( does it have to be in a certain directory ?)

If you want to go down the DBMS_SCHEDULER route, you can use the UTL_FILE package to write the file out to a directory that you’ve defined in the database.
There’s an example of how to use UTL_FILE here.
Alternatively, you could create an anonymous PL/SQL block saved in a file ( and use DBMS_OUTPUT), call this from a shell script and use the SQL*Plus spool command to create your file.
There’s an example of invoking SQL*Plus from a shell script here.

HTH

Mike

Expample
v_chr_sql := ‘SELECT ‘ || v_chr_sql || ‘ FROM powerseller.loan L, powerseller.gmacm_loan_search B’
|| ‘ WHERE L.servicing_number = B.servicing_number’
|| ‘ AND B.pool_name IN (SELECT column_value pool_name FROM THE’
|| ‘ (SELECT CAST(in_list_fnc(UPPER(”’ || pi_pool_id_list || ”’)) AS ListTableType) FROM DUAL))’
|| ‘ ORDER BY L.pool_name, L.servicing_number’;
dbms_output.put_line(v_chr_sql);
OPEN pio_cstdn_extract_cur FOR v_chr_sql;

<?xml version="1.0"?>
<ROWSET>
<ROW>
<NAME>HART</NAME>
<POSITION>GOALKEEPER</POSITION>
<CAPS>6</CAPS>
<GOALS>0</GOALS>
</ROW>
<ROW>
<NAME>CARSON</NAME>
<POSITION>GOALKEEPER</POSITION>
<CAPS>3</CAPS>
<GOALS>0</GOALS>
</ROW>
<ROW>
<NAME>FOSTER</NAME>
<POSITION>GOALKEEPER</POSITION>
<CAPS>4</CAPS>
<GOALS>0</GOALS>
</ROW>
<ROW>
<NAME>CAHILL</NAME>
<POSITION>DEFENDER</POSITION>
<CAPS>1</CAPS>
<GOALS>0</GOALS>
</ROW>
<ROW>
<NAME>COLE A</NAME>
<POSITION>DEFENDER</POSITION>
<CAPS>85</CAPS>
<GOALS>0</GOALS>
</ROW>
<ROW>
<NAME>DAWSON</NAME>
<POSITION>DEFENDER</POSITION>
<CAPS>2</CAPS>
<GOALS>0</GOALS>
</ROW>
<ROW>
<NAME>GIBBS</NAME>
<POSITION>DEFENDER</POSITION>
<CAPS>1</CAPS>
<GOALS>0</GOALS>
</ROW>
<ROW>
<NAME>JAGIELKA</NAME>
<POSITION>DEFENDER</POSITION>
<CAPS>6</CAPS>
<GOALS>0</GOALS>
</ROW>
<ROW>
<NAME>JOHNSON G</NAME>
<POSITION>DEFENDER</POSITION>
<CAPS>29</CAPS>
<GOALS>1</GOALS>
</ROW>
<ROW>
<NAME>LESCOTT</NAME>
<POSITION>DEFENDER</POSITION>
<CAPS>10</CAPS>
<GOALS>0</GOALS>
</ROW>
<ROW>
<NAME>UPSON</NAME>
<POSITION>DEFENDER</POSITION>
<CAPS>21</CAPS>
<GOALS>2</GOALS>
</ROW>
<ROW>
<NAME>BARRY</NAME>
<POSITION>MIDFIELDER</POSITION>
<CAPS>42</CAPS>
<GOALS>2</GOALS>
</ROW>
<ROW>
<NAME>CARRICK</NAME>
<POSITION>MIDFIELDER</POSITION>
<CAPS>22</CAPS>
<GOALS>0</GOALS>
</ROW>
<ROW>
<NAME>GERRARD</NAME>
<POSITION>MIDFIELDER</POSITION>
<CAPS>87</CAPS>
<GOALS>19</GOALS>
</ROW>
<ROW>
<NAME>JOHNSON A</NAME>
<POSITION>MIDFIELDER</POSITION>
<CAPS>4</CAPS>
<GOALS>2</GOALS>
</ROW>
<ROW>
<NAME>MILNER</NAME>
<POSITION>MIDFIELDER</POSITION>
<CAPS>14</CAPS>
<GOALS>0</GOALS>
</ROW>
<ROW>
<NAME>WALCOTT</NAME>
<POSITION>MIDFIELDER</POSITION>
<CAPS>14</CAPS>
<GOALS>3</GOALS>
</ROW>
<ROW>
<NAME>WRIGHT-PHILLIPS</NAME>
<POSITION>MIDFIELDER</POSITION>
<CAPS>35</CAPS>
<GOALS>6</GOALS>
</ROW>
<ROW>
<NAME>YOUNG</NAME>
<POSITION>MIDFIELDER</POSITION>
<CAPS>9</CAPS>
<GOALS>0</GOALS>
</ROW>
<ROW>
<NAME>BENT</NAME>
<POSITION>STRIKER</POSITION>
<CAPS>7</CAPS>
<GOALS>1</GOALS>
</ROW>
<ROW>
<NAME>COLE C</NAME>
<POSITION>STRIKER</POSITION>
<CAPS>7</CAPS>
<GOALS>0</GOALS>
</ROW>
<ROW>
<NAME>CROUCH</NAME>
<POSITION>STRIKER</POSITION>
<CAPS>40</CAPS>
<GOALS>21</GOALS>
</ROW>
<ROW>
<NAME>DEFOE</NAME>
<POSITION>STRIKER</POSITION>
<CAPS>45</CAPS>
<GOALS>15</GOALS>
</ROW>
<ROW>
<NAME>ROONEY</NAME>
<POSITION>STRIKER</POSITION>
<CAPS>67</CAPS>
<GOALS>26</GOALS>
</ROW>
</ROWSET>

PL/SQL procedure successfully completed.

SQL> 

This is slightly different to the starting point for this post, which was getting output in a CSV format. However, I’m sure there are people reading this who will find that it’s just what they’re looking for.
Thanks for sharing,

Mike

declare
cur sys_refcursor;

procedure p(p_str in varchar2) is
l_str long := p_str || chr(10);
l_piece long;
n number;
begin
loop
exit when l_str is null;
n := instr(l_str, chr(10));
l_piece := substr(l_str, 1, n – 1);
l_str := substr(l_str, n + 1);
loop
exit when l_piece is null;
dbms_output.put_line(substr(l_piece, 1, 250));
l_piece := substr(l_piece, 251);
end loop;
end loop;
end;

function print_refcursor(cur sys_refcursor) return varchar2 as
begin
return xmltype(cur).getstringval();
end print_refcursor;

begin
open cur for
select * from tab;

p(print_refcursor(cur));
end;
/