"Parse" SQL in bash or ksh

[stagebi]

Hi!

I've a text file programme1.out :

CONNECT TO ORACLE (&CONXORA);

CREATE TABLE TABLE1 AS
    SELECT  ID_PERS

        FROM CONNECTION TO ORACLE
        (
        SELECT  DISTINCT PERS.ID_PERS

        FROM TAB_GRP GRP , TAB_PERS PERS

        WHERE DATE_DEB_GRP <= &DATE_QUOTE
              AND (DATE_FIN_GRP IS NULL OR DATE_FIN_GRP > &DATE_QUOTE)
              AND PERS.ID_PERS = GRP.ID_PERS
              AND (PERS.DATE_SUPP_SI IS NULL OR PERS.DATE_SUPP_SI > &DATE_QUOTE)
ORDER BY ID_PERS
);
DISCONNECT FROM ORACLE;
CONNECT TO ORACLE (&CONXORA);

CREATE TABLE TABLE2 AS
  SELECT    ID_PERS,
            ID_PRODUIT,
            CODE_PRODUIT,
            TOP_1
            TOP_2
                
    FROM CONNECTION TO ORACLE
    (
        SELECT ID_PERS,
               ID_PRODUIT,
                CODE_PRODUIT,
                MAX (CASE WHEN (CODE ='01' ) THEN 1 ELSE 0 END) AS TOP_1
                MAX (CASE WHEN (CODE='02' ) THEN 1 ELSE 0 END) AS TOP_2
        FROM TABLE_FROM1
        WHERE  CODE IN ('01','02')
        AND    ID_MOIS <= &ID_MOIS
        GROUP BY ID_PERS,ID_PRODUIT
    );
DISCONNECT FROM ORACLE;

For each query (between connect to oracle and disconnect from oracle), I have to display the name of the create table, the name of tables used (real name of the table) and the name of the columns in the SELECT.

At the end, I would like an output like this : table_create ; table_from ; column_name

Example :

table1;tab_grp;
table1;tab_pers;ID_PERS
table2;table_from1;ID_PRODUIT
table2;table_from1;CODE_PRODUIT
table2;table_from1;TOP_1
table2;table_from1;TOP_2

Is it possible to do this easily with a script in ksh or bash (UNIX AIX IBM)? (with sed, awk..)

Regards

[tpl]

welcome to the forum

can you show the output from that Oracle macro?
then perhaps we can help you convert that output into
your desired final form