Parsing query and getting list of tables used

Question

Hi All,
   I require to parse sql query (simple & complex as well) and to get the list of tables used in the query.
   And need to validate that, the list of tables against a whitelist to maintained in the file.
  I tried to write my own parser, since there are lots possible ways to write complex queries, i am unable to cover all the scenarios.
  I require a help, Is there any other ways to get the list of tables used in a query?
Thanks,
Mahendran..

Answer 1

user10279893 wrote:
Hi All,
   I require to parse sql query (simple & complex as well) and to get the list of tables used in the query.
   And need to validate that, the list of tables against a whitelist to maintained in the file.
  I tried to write my own parser, since there are lots possible ways to write complex queries, i am unable to cover all the scenarios.
  I require a help, Is there any other ways to get the list of tables used in a query?
Thanks,
Mahendran..
You'll never cover all scenarios by trying to parse a query.
For example, what about a query that queries views and those views query pipelined functions, and those pipelined functions query other views or tables or pipelined functions etc.
And what about a query that dynamically obtains the table names, such as...
SQL> select
  2    table_name,
  3    to_number(
  4      extractvalue(
  5        xmltype(
  6 dbms_xmlgen.getxml('select count(*) c from '||table_name))
  7        ,'/ROWSET/ROW/C')) count
  8  from user_tables
  9 where iot_type != 'IOT_OVERFLOW'
10 or    iot_type is null;
TABLE_NAME                      COUNT
DEPT                                4
EMP                                14
BONUS                               0
SALGRADE                            5
.. in this case the tables being accessed (to count their rows) are not even known within the code, they are dynamically generated queries at run-time, so the only way to know what tables are accessed is to actually run the query.
Of course, a well designed system, with proper documentation and version control would allow you to take any database object and know where and what is using it, and whether it's even being used.