Path: csiph.com!usenet.pasdenom.info!gegeweb.org!de-l.enfer-du-nord.net!feeder2.enfer-du-nord.net!feeds.phibee-telecom.net!newsfeed.xs4all.nl!newsfeed6.news.xs4all.nl!xs4all!post.news.xs4all.nl!not-for-mail
To: python-list@python.org
From: Robert Kern <robert.kern@gmail.com>
Subject: Re: avoid import short-circuiting
Date: Fri, 16 Mar 2012 22:20:15 +0000
References: <4F636F0D.3060006@gmail.com> <jjvsnj$v9d$1@dough.gmane.org> <4F63B8D4.8010103@gmail.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.7; rv:10.0.2) Gecko/20120216 Thunderbird/10.0.2
In-Reply-To: <4F63B8D4.8010103@gmail.com>
Precedence: list
Newsgroups: comp.lang.python
Message-ID: <mailman.742.1331936429.3037.python-list@python.org>
Lines: 81
NNTP-Posting-Host: 2001:888:2000:d::a6
Xref: csiph.com comp.lang.python:21785

On 3/16/12 10:04 PM, Andrea Crotti wrote:
> On 03/16/2012 05:19 PM, Robert Kern wrote:
>> On 3/16/12 4:49 PM, Andrea Crotti wrote:
>>> I started the following small project:
>>>
>>> https://github.com/AndreaCrotti/import-tree
>>>
>>> because I would like to find out what exactly depends on what at run-time, using
>>> an import hook.
>>>
>>> It works quite well for small examples but the main problem is that once a
>>> module is imported
>>> it's added to sys.modules and then it doesn't go through the import hook
>>> anymore.
>>>
>>> I tried to mess around with sys.modules but it might not be a good idea, and it
>>> leads to easy
>>> infinite loops.
>>> Is there a good way to achieve this?
>>> I guess I could do the loop detection myself, but that should not be too hard..
>>
>> You want to monkeypatch __builtin__.__import__() instead. It always gets called.
>>
>
> Seems like a good idea :)
>
> My first attempt failes though
>
>
> def full(module):
> from __builtin__ import __import__
> ls = []
> orig = __import__
>
> def my_import(name):
> ls.append(name)
> orig(name)
>
> __import__ = my_import
> __import__(module)
> __import__ = orig
> return ls
>
>
> it imports only the first element and doesn't import the dependencies..
> Any hints?

You need to replace it in __builtin__. Don't forget to handle all of the arguments.


import __builtin__

orig_import = __builtin__.__import__

all_imports = []

def my_import(*args, **kwds):
     module = orig_import(*args, **kwds)
     # Get the fully-qualified module name from the module object itself
     # instead of trying to compute it yourself.
     all_imports.append(module.__name__)
     return module

__builtin__.__import__ = my_import


For extra points, make a context manager that hooks and then unhooks your
custom importer.

You may also want to take a look at an import profiler that I once made:

http://www.enthought.com/~rkern/cgi-bin/hgwebdir.cgi/import_profiler/file/tip/import_profiler.py#l23

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless enigma
  that is made terrible by our own mad attempt to interpret it as though it had
  an underlying truth."
   -- Umberto Eco