Previously, we noted one common approach to supporting multiple features. Some applications use multiple top-level main programs in separate .py script files. If we do this, then combining commands from separate files forces us to write a shell script. It doesn't seem optimal to introduce yet another tool and another language to do programming in-the-large (PITL).

A slightly more flexible alternative to creating separate script files is using a positional parameter to select a specific top-level Command object. For our example, we'd like to select either the simulation or the analysis command. To do this, we would add a parameter to the command-line argument parsing the following code:

parser.add_argument(
    "command", action="store", default='simulate', 
    choices=['simulate', 'analyze']) 
parser.add_argument("outputfile", action="store", metavar="output") 

This would change the command-line API to add the top-level verb to the command line. We can then map our argument values to class names that implement the desired command, as shown in the following code:

command_map = {
    'simulate': Simulate_Command, 
    'analyze': Analyze_Command
}
command = command_map[options.command] 
command.configure(options)
command.run()

This allows us to create even higher-level composite features. For example, we might want to combine simulation and analysis into a single, overall program. We also might like to do this without resorting to using the shell.