On 11/10/2011 05:02 PM, Eric Snow wrote:
Yeah, I'm working on a reference for imports in Python. They're just
a little too mysterious relative to the rest of the language. But
it's not too helpful yet. In the meantime...
Yes it's quite mysterious, and it's actually not as hard as it looks..
Anyway I'm glad to say that I reached what I basically wanted.
This script actually compiles the code and runs it, reporting the
imports done in the end.
Any suggestion is still welcome :)
"""
This script is used to analyse the imports which are actually being done
"""
import argparse
import os
import sys
class CollectImports(object):
"""
Loader object
"""
def __init__(self):
self.loaded = set()
def __str__(self):
return str(self.loaded)
def find_module(self, module_name, package=None):
print("requesting %s" % module_name)
self.loaded.add(module_name)
if __name__ == '__main__':
parser = argparse.ArgumentParser(description='analyse the imports
made')
parser.add_argument('script')
parser.add_argument('arguments', nargs='*')
ns = parser.parse_args()
progname = ns.script
sys.path.insert(0, os.path.dirname(progname))
cl = CollectImports()
# TODO: maybe we can create a with clause also for this thing
sys.meta_path.append(cl)
# TODO: catch the exit signal and present the output
code = compile(open(progname).read(), progname, 'exec')
exec(code)
print("imports done: %s" % str(cl))
That _is_ pretty strange.
After what I recommended above, are you still getting the wierdness?
It could just be a side-effect of your use of the imp functions. in
load_module(), so getting rid of it would help.
What version of Python are you using? If not 2.7 or 3.2, do you get
the same problem when you run the code under one of those latest
versions?
Even when the number of imports is different, are they always in the
same order? Highly unlikely, but if you say no then this is extra
fishy.
It was python 2.7 on Arch-linux, but thanks to your suggestions
everything was actually fixed..
--
http://mail.python.org/mailman/listinfo/python-list
