Description of JobSchedulerPLSQLJob - Execute PL/SQL procedure

What is the purpose of this job?

The JITL JobSchedulerPLSQLJob job provide an standardized and parameterized  interface to execute Oracle PL/SQLs.  The JobScheduler offers out of the box capability to execute PL/SQLs, pass parameters to the PL/SQL or retrive and pass on the results of a PL/SQL execution to next job step as JobScheduler Order parameter.  The JobSchedulerPLSQLJob can be used to execute existing PL/SQL files just by referring them in the command parameter. 

Simple Example

Following is simple example to demonstrate capabilities of the JITL JobSchedulerPLSQLJob.

<?xml version="1.0" encoding="ISO-8859-1"?>
<job  title="Execute PL/SQL procedure" order="yes">
    <description >
        <include  file="jobs/JobSchedulerPLSQLJob.xml"/>
    </description>
    <params >
        
        <param  name="db_url"      value="jdbc:oracle:thin:@:1521:DORCL01"/>
        <param  name="db_user"     value="sos_scheduler"/>
        <param  name="db_password" value="sos"/>			
		<param  name="command"     value="
		DECLARE   
		   v_date DATE := SYSDATE; 
		BEGIN      		    
			SELECT SYSDATE    
			INTO v_date   
			FROM DUAL;     
			DBMS_OUTPUT.PUT_LINE(' +++              +++');   
			DBMS_OUTPUT.PUT_LINE('SET today_date IS '|| v_date);  
			DBMS_OUTPUT.PUT_LINE(' +++              +++');   
		END;
		"/>
        <param  name="variable_parser_reg_expr" value="^SET\s+([^\s]+)\s*IS\s+(.*)$"/>
    </params>
    <script  language="java" java_class="sos.scheduler.db.JobSchedulerPLSQLJobJSAdapterClass"/>
    <run_time />
</job>

How is the PL/SQL script defined?

• The PL/SQL code be 

  • saved to a separate file i.e. get_order.sql , and subsequently sql file can be referred as the value of the job parameter "command", this is a recommended  approach to achieve "separation of concern" in application architecture. 

     <param  name="command" value="config/live/commn/sqls/get_order.sql"/>

  • PL/SQL code can also be specified as the value of the parameter command, entire PL/CODE can be written as part of the Job.XML, this approach is preferred 

     <param  name="command" value="
    		DECLARE   
    		   v_order_date DATE := SYSDATE; 
    		BEGIN      		    
    			SELECT SYSDATE    
    			INTO v_order_date   
    			FROM DUAL;     
    			DBMS_OUTPUT.PUT_LINE(' +++              +++');   
    			DBMS_OUTPUT.PUT_LINE('SET order_date IS '|| v_order_date);  
    			DBMS_OUTPUT.PUT_LINE(' +++              +++');   
    		END;
    	"/>

• Before the script is executed, the script will be analyzed by the job. 
  • Should JobScheduler parameter names be containted in that script then they are substituted by their current value.
  • The following notations are supported: %parameter_name%, ${SCHEDULER_PARAM_parameter_name}

How can results be used in subsequent jobs?

• The job makes use of the appropriate parameters in the order. 
• The first of these are static parameters, such as those described in the section "return parameters" of the template documentation. 
• At the same time these are dynamic parameters. The job anlyzes the output of the PL/SQL statements and extracts values by use of a regular expression that catches both names and values of these parameters. The regular expression is defined by the parameter variable_parser_reg_exp.

These parameters are readable by all jobs in all job steps. The value of the parameters can be modified by subsequent jobs.

    declare
      howmany NUMBER;
      p_id varchar2(20) := null;
    begin
      dbms_output.put_line('set variable1=value1');
      p_id := '12345';
      --
      -- If the job or order contains the parameter
      --
      -- table_name=scheduler_variables
      --
      -- then all following selects are substituted to 'select count(*) into howmany from scheduler_variables;'
      --
      select count(*) into howmany from $\{SCHEDULER_PARAM_table_name\};
      select count(*) into howmany from %table_name%;
      select count(*) into howmany from %TABLE_NAME%;
      select count(*) into howmany from $\{SCHEDULER_PARAM_TABLE_NAME\};
      select count(*) into howmany from $\{scheduler_param_table_name\};
      select count(*) into howmany from $\{sChEdUlEr_pArAm_tAbLe_nAmE\};
      --
      -- now put the results to the buffer
      -- JS will get the results from the buffer
      --
      dbms_output.put_line('The table %table_name% has ' || howmany || ' rows.');
      dbms_output.put_line('set howmany is ' || howmany);
      dbms_output.put_line('set variable1 is ' || p_id);
      dbms_output.put_line('set variable2 is value2');
    end;

Example: PL/SQL code

• If the job or order contains the parameter table_name with the value scheduler_variables then all following SQL "select" statements are substituted to: select count(*) into howmany from scheduler_variables;
• If the job is started in a job chain by an order then the following parameters are added to the order 
  • see the parameter variable_parser_reg_expr 
  • For this purpose, the output of the PL/SQL statement is parsed with the regular expression: ^SET\\s+([^\\s]+)\\s*IS\\s+(.*)$

From the above example the following parameters are returned:

• howmany=8
• variable1=12345
• variable2=value2

If the same parameter is set multiple times then the last value is used as the order parameter.

An example for a job xml file:

  <job>
     <params>
       <param name="command" value="select sysdate from dual" />
       <param name="db_url" value="jdbc:oracle:thin:@localhost:1521:XE" />
       <param name="db_user" value="test" />
       <param name="db_password" value="test" />
       <param name="variable_parser_reg_expr" value="^SET\\s+([^\\s]+)\\s*IS\\s+(.*)$" />
     </params>
     <script language="java" java_class="sos.scheduler.db.JobSchedulerPLSQLJobJSAdapterClass" />
  </job>

Parameters used by JobSchedulerPLSQLJob

Parameter: command

• It is possible to define more than one PL/SQL statement in the command parameter. Such statements are then carried out in the order in which they are specified and have to be separated by a semicolon and a subsequent new line.
• Parameters are substituted in database statements. This is effected by using parameter names in the form §{param} at any given position in a statement. The parameter is then substituted by the value of the job parameter with the same name. Substitution takes place before database statements are executed.

Parameter: variable_parser_reg_expr

• The output of the PL/SQL statement will be parsed by this regular expression. The matches are available as order parameters.
• The regular expression has to specify to two groups: the first group for the parameter name and the second group for the value.

The prompt statement in SQL*Plus writes some output like this:

WHENEVER SQLERROR EXIT SQL.SQLCODE
WHENEVER OSERROR EXIT FAILURE
prompt ready;
prompt foo bar;
prompt SET hello IS world;
exit;

The dbms_output.put_line() function in PL/SQL writes some output like this:

begin
  dbms_output.put_line('ready');
  dbms_output.put_line('Set hello IS world');
  dbms_output.put_line('fooh1. bar');
end;

Return parameters created by JobSchedulerPLSQLJob

The order parameters described below are automatically created and returned by the job:

Parameter: sql_error

• If during the execution of the PL/SQL statements one or multiple errors occur, then the error message will be stored in this order parameter. If no error occurs then the contents of the parameter will be empty.

Parameter: std_out_output

• The output of the PL/SQL statements to stdout is reported as the value of this parameter.