Olliesに基づいてdbms_outputを使用しますが、カーソルを動的に通過するために
テスト用に設定
/*create table temp (aa varchar2(50) , bb number , cc date ) ;
insert into temp (aa,bb,cc)
select chr(level+100) , level, sysdate+level
from dual
connect by level < 15 ;
/
*/
テストを表示するためのブロック（これは11gを想定しています）
set serveroutput on
declare
l_cur SYS_REFCURSOR ;
PROCEDURE CursorOutput(
p_refcursor IN OUT SYS_REFCURSOR
)
AS
l_desc DBMS_SQL.DESC_TAB ;
l_cols BINARY_INTEGER ;
l_cursor BINARY_INTEGER ;
v_varchar2 VARCHAR2( 4000 ) ;
v_number NUMBER ;
v_date DATE ;
l_data varchar2( 32767 ) ;
l_columnValue VARCHAR2( 32767 ) ;
l_processedRows Number := 0;
BEGIN
/* Convert refcursor "parameter" to DBMS_SQL cursor... */
l_cursor := DBMS_SQL.TO_CURSOR_NUMBER( p_refcursor );
/* Describe the cursor... */
DBMS_SQL.DESCRIBE_COLUMNS( l_cursor, l_cols, l_desc );
/* Define columns to be fetched. We're only using V2, NUM, DATE for example...
for a complete list of the col_types this link is accessible.
http://download.oracle.com/docs/cd/B10501_01/server.920/a96540/sql_elements2a.htm#45504
http://forums.oracle.com/forums/thread.jspa?threadID=912475
if not a usable type, will throw new exception
*/
FOR i IN 1 .. l_cols LOOP
IF l_desc(i).col_type = 2 THEN
DBMS_SQL.DEFINE_COLUMN(l_cursor, i, v_number);
ELSIF l_desc(i).col_type = 12 THEN
DBMS_SQL.DEFINE_COLUMN(l_cursor, i, v_date);
ELSif l_desc(i).col_type = 01 or l_desc(i).col_type = 96 then
DBMS_SQL.DEFINE_COLUMN(l_cursor, i, v_varchar2, 4000);
else
--raise an exception if the user's query contains a datatype not (yet) supported by this procedure
RAISE_APPLICATION_ERROR(-20000, 'Invalid Data Type for conversion to delimited file. {' || l_desc(i).col_name || '}');
END IF;
END LOOP;
/* -- print out the column names if desired
FOR i IN 1 .. l_cols LOOP
dbms_output.put_line('** ' || l_desc(i).col_name) ;
END LOOP;
*/
/* Fetch all data... */
WHILE DBMS_SQL.FETCH_ROWS(l_cursor) > 0 LOOP
dbms_output.put_line('LINE: ' || l_processedRows || '');
FOR i IN 1 .. l_cols LOOP
if l_desc(i).col_type = 12 THEN --we are in a date
DBMS_SQL.COLUMN_VALUE(l_cursor, i, v_date);
v_varchar2 := to_char(v_date , 'dd-MON-yyyy' ) ;
elsif l_desc(i).col_type = 2 THEN --we are in a number
DBMS_SQL.COLUMN_VALUE(l_cursor, i, v_number);
v_varchar2 := to_char(v_number) ;
else --treat it as a string (should be varchar2,char,etc)
DBMS_SQL.COLUMN_VALUE(l_cursor, i, v_varchar2);
IF v_varchar2 IS NOT NULL THEN
v_varchar2 := '"' || v_varchar2 || '"' ;
ELSE
v_varchar2 := '';
END IF ;
end if ;
dbms_output.put_line(l_desc(i).col_name || '=>' || v_varchar2) ;
END LOOP;
l_processedRows := l_processedRows + 1 ;
END LOOP;
dbms_sql.close_cursor(l_cursor);
dbms_output.put_line('I found and processed ' || l_processedRows || ' rows .');
END;
begin
open l_cur for select * from temp;
CursorOutput(p_refcursor => l_cur) ;
end ;
/
この結果が得られます
LINE: 0
AA=>"e"
BB=>1
CC=>04-JAN-2012
LINE: 1
AA=>"f"
BB=>2
CC=>05-JAN-2012
LINE: 2
AA=>"g"
BB=>3
CC=>06-JAN-2012
LINE: 3
AA=>"h"
BB=>4
CC=>07-JAN-2012
LINE: 4
AA=>"i"
BB=>5
CC=>08-JAN-2012
LINE: 5
AA=>"j"
BB=>6
CC=>09-JAN-2012
LINE: 6
AA=>"k"
BB=>7
CC=>10-JAN-2012
LINE: 7
AA=>"l"
BB=>8
CC=>11-JAN-2012
LINE: 8
AA=>"m"
BB=>9
CC=>12-JAN-2012
LINE: 9
AA=>"n"
BB=>10
CC=>13-JAN-2012
LINE: 10
AA=>"o"
BB=>11
CC=>14-JAN-2012
LINE: 11
AA=>"p"
BB=>12
CC=>15-JAN-2012
LINE: 12
AA=>"q"
BB=>13
CC=>16-JAN-2012
LINE: 13
AA=>"r"
BB=>14
CC=>17-JAN-2012
I found and processed 14 rows .
これらの2つのリンクをソースとして使用してcsvファイルを動的に構築するためにこれと同様のことを行いました
ただし、目的によっては、SQL Developer（またはToad）で実行して、結果をエクスポートすることもできます。