Skip to main content

Use context managers with a function instead of a statement.

Project description

Use context managers with a function instead of a statement.

Provides a minimal, clean and portable interface for using context managers with all the advantages of functions over syntax.


Because context managers are awesome, but currently can’t be used in as many places as I would like, and this is the first step towards making that possible with less boilerplate and more portability.


This library’s version numbers follow the SemVer 2.0.0 specification.

The current version number is available in the variable __version__ as is normal for Python modules.


pip install with-as-a-function


The with_ function implements the raw basic logic of executing a context manager as described in PEP-343:

from with_ import with_

With it we can do things like this:

data = with_(open('my_file.txt'), lambda my_file:

Which is equivalent to:

with open('my_file.txt') as my_file:
    data =

You can think of it as being functionally equivalent to:

def with_(manager, action):
    with manager as value:
        return action(value)
    return None

except that it’s also portable to Python implementations and versions that don’t support the with statement.


Portable to all releases of both Python 3 and Python 2.

Even those without the with statement.

(The oldest tested is 2.5, but it will likely work on all Python 2 versions and probably on even earlier versions.)

For Python implementations that do not support sys.exc_info, a “no traceback” variant can be installed manually, by grabbing the file and saving it as (Saving it as has the advantage that your code can just do from with_ import with_ and it’ll just work consistently, without version-detecting boilerplate.)

You are of course welcome to just copy-paste the tiny with_ function definition into your code.

Design Decisions

  • Not exposing anything special to support the equivalent of wrapping yield from and yield within with blocks, because that is a more complex problem, and I want to have more experience seeing those usecases before I start committing to an interface.

  • Not using finally as the reference code in PEP-343 does because the only purpose of the finally was to catch escaping statements like return, break, and continue. Since we use a single function instead of a “suite” of code, those cannot happen.

  • Not using an else clause for the clean exit variant, because the code is still equally clear without it, and arguably is even clearer because it’s the same number of lines but code paths terminate faster.

  • Also not using either else or finally because of “portability conservatism”: the principle that when portability matters, it is safer to bet on the most conservative feature-set possible. You’d think every pragmatically usable Python implementation would get try block else and finally clauses right, but it turns out no, some don’t.

    If it made a big difference in code clarity or correness, I would be reticent to consider this a sufficient reason by itself, but it is a factor worth considering.

  • To aid portability of code to Python implementations that do not support getting a traceback object for the exception being handled, passing None as traceback to __exit__ helps writing code that portably follows “progressive enhancement” or “graceful degradation” practices: tracebacks are properly used where possible, but ignored where not.

    This matches the behavior of MicroPython and Transcrypt, two implementations of Python which support with but don’t support traceback objects, so this suggests that it is a reasonable choice.

    This is not always the wisest choice: some features and behavior are relied on for security, correctness, or debugging, and in those cases the inability to fulfill the contract of an interface must not be silently hidden.

    Because of this, the “no traceback” variant is “opt-in”: if you’re using it, you deliberately included it into your project, or a dependency of yours did.

  • Using from sys import exc_info as _exc_info instead of import sys as _sys and then calling _sys.exc_info because:

    1. Python’s dynamic semantics mean that any other piece of code is allowed to change sys.exc_info globally at any time.
    2. Therefore first and foremost, we are communicating our intent that we would rather be immune from that as much as possible.
    3. Someone could still modify sys.exc_info before with_ is first imported, but at that point odds are higher that this was done with due care of how it will effect all code using it.
    4. Someone could modify our with_._exc_info name directly, but at that point they definitely deliberately intended to do so.
    5. The important thing is that we opt-out of being effected by casual, careless, or experimental reassignments of sys.exc_info.

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for with-as-a-function, version 1.0.2.post2
Filename, size File type Python version Upload date Hashes
Filename, size with_as_a_function-1.0.2.post2-py2.py3-none-any.whl (5.6 kB) File type Wheel Python version py2.py3 Upload date Hashes View hashes
Filename, size with-as-a-function-1.0.2.post2.tar.gz (5.2 kB) File type Source Python version None Upload date Hashes View hashes

Supported by

Elastic Elastic Search Pingdom Pingdom Monitoring Google Google BigQuery Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page