On 07/20/2015 10:19 AM, Eric V. Smith wrote:
On 07/20/2015 10:08 AM, Ryan Gonzalez wrote:
I would prefer something more like:
def f(s): caller = inspect.stack()[1][0] return s.format(dict(caller.f_globals, **caller.f_locals))
You need to use format_map (or **dict(...)). And ChainMap might be a better choice, it would take some benchmarking to know.
Also, you don't get builtins using this approach. I'm using eval to exactly match what evaluating the variable in the parent context would give you. That might not matter depending on the actual requirements.
But I agree there are multiple ways to do this, and several of them could be made to work. Mine might have fatal flaws that more testing would show.
My quick testing comes up with this, largely based on the code by joejev: import sys import collections def f(str): frame = sys._getframe(1) return str.format_map(collections.ChainMap( frame.f_locals, frame.f_globals, frame.f_globals['__builtins__'].__dict__)) I'm not sure about the builtins, but this seems to work. Also, you might want to be able to pass in the frame depth to allow this to be callable more than 1 level deep. So, given that this is all basically possible to implement today (at the cost of using sys._getframe()), I'm -1 on adding any compiler tricks to support this via syntax. From what I know of PyPy, this should be supported there, albeit at a large performance cost. Eric.