On Wed, Sep 2, 2020 at 11:29 AM Andras Tantos <python-l...@andras.tantosonline.com> wrote: > OK, and sorry about that. May third time a charm?
Thank you! Whatever you did this time, that worked :) > Indeed, I'm executing a code object and not a string. Is there any way > to get to the string version of the function and force a re-compile of > it during an exec? Would it help? Eeeuuuuhhhhhhhh.... normally you'd do that by, well, *not* compiling it. But maybe the inspect module can help here: https://docs.python.org/3/library/inspect.html#inspect.getsource There's no guarantee that it will work though. > I guess if I was doing that, I would need to execute the 'body' of the > function within the exec and by passing in a local parameter I can > pretend it has its own call-stack entry. But is that truly equivalent to > calling the original function? I think it is getting an actual call stack entry anyway. Now, if your goal is to do something with an expression evaluator, then there are LOTS of options. Let's say you want to allow people to write code that implicitly calls on some data source, so typing "x ** 2 + y ** 2" would look up x and y in some way and then do arithmetic with them. If you get the original source as a string, you can most certainly do things with the globals as you compile it. Or if you want more reliability, compile it to AST, walk the tree, and replace nodes as required, before compiling it the rest of the way for execution. Or run the code at global scope, with the globals doing your magic and the builtins doing everything else. Lots of ways to do things, as long as you start with the source code. ChrisA -- https://mail.python.org/mailman/listinfo/python-list
