When you run an interpreter like python (or other languages like Mathematica or a linux prompt, or a "DOS" prompt), you don't need source files: you just type commands and they happen immediately.
Should you want to make a source file (called a "script"), you can, with whatever editor you like, and then invoke python to run the script. I confess that I've never used python in this way in an IDE, so I'm not sure if I can help if you're specifically keeping that type of solution. My uses of python have been of the form:
Run
python and execute commands directly.
In this mode, I do sometimes edit commands in a text editor, then copy/paste those commands into python. Or, I might sometimes use the import command to load and execute a "module"
Run a python GUI, such as IDLE (it's a free download from somewhere).
I use this mode pretty much the same as the previous one. But the GUI makes it easier to deal with mistakes in typing. (e.g. so I don't have to use a text editor and use copy/paste). I have never tried to create/use scripts in this mode
Create python scripts, and invoke them with
python <script name>.
In this mode, I am typically executing python from a command line (e.g. a linux prompt, or a DOS prompt), and editing my python scripts in my favorite text editor, such as xemacs