Skip to main content

Programming by contract library.

Project description

Deal – python library for design by contract (DbC) programming.

This library contain 3 main conception from DbC:

  • Precondition – condition that must always be true just prior to the execution of function.

  • Postcondition – condition that must always be true just after the execution of function.

  • Invariant – condition that can be relied upon to be true during execution of a program. In this library invariant condition check in 3 cases:

    1. Before class method execution.

    2. After class method execution.

    3. After some class attribute setting.

Features

  • Functional declaration for conditions.

  • Custom exceptions for best usage in try-except block.

  • Ability to set optional error message and (or) exception class.

  • Ability to return error message from contact.

  • Ability to use Django Form styled validators as contracts.

  • Attribute setting validation by invariant.

  • Validation by invariant dynamically assigned to object attributes and methods.

  • Readable source code (all decorators implemented by classes)

Installation

Stable:

pip install deal

Dev:

pip install -e git+https://github.com/orsinium/deal.git#egg=deal

TL;DR

  • @pre – validate function arguments (pre-validation).

  • @post – validate function result (post-validation).

  • @inv – validate class methods before and after some method calling and after attribute setting.

Exceptions structure:

  • ContractError (inherited from built-in AssertionError)

    • PreContractError

    • PostContractError

    • InvContractError

Library decorators doesn’t catch any exceptions raised from contracts.

Usage

Pre (pre, require):

In [1]: from deal import pre, post, inv, ContractError

In [2]: @pre(lambda *args: all(map(lambda x: x > 0, args)))
   ...: def my_sum(*args):
   ...:     return sum(args)
   ...:

In [3]: my_sum(2, 3, 4)
Out[3]: 9

In [4]: my_sum(2, -3, 4)
PreContractError:

Post (post, ensure):

In [5]: @post(lambda x: x > 0)
   ...: def my_sum(*args):
   ...:     return sum(args)
   ...:

In [6]: my_sum(2, -3, 4)
Out[6]: 3

In [7]: my_sum(2, -3, -4)
PostContractError:

Inv (inv, invariant):

In [8]: @inv(lambda obj: obj.x > 0)
   ...: class A:
   ...:     x = 4
   ...:

In [9]: a = A()

In [10]: a.x = 10

In [11]: a.x = -10
InvContractError:

In [12]: A
Out[12]: deal.core.AInvarianted

Custom message:

In [13]: @pre(lambda x: x > 0, "x must be > 0")
    ...: def f(x):
    ...:     return x * 2
    ...:

In [14]: f(-2)
PreContractError: x must be > 0

Custom exception:

In [15]: @pre(lambda x: x > 0, exception=AssertionError("x must be > 0"))
    ...: def f(x):
    ...:     return x * 2
    ...:

In [16]: f(-2)
AssertionError: x must be > 0

Validators (nearly Django Forms style, except initialization):

In [17]: class Validator:
    ...:     def __init__(self, x):
    ...:         self.x = x
    ...:
    ...:     def is_valid(self):
    ...:         if self.x <= 0:
    ...:             self.errors = ['x must be > 0']
    ...:             return False
    ...:         return True
    ...:

In [18]: @pre(Validator)
    ...: def f(x):
    ...:     return x * 2
    ...:

In [19]: f(5)
Out[19]: 10

In [20]: f(-5)
PreContractError: ['x must be > 0']

Return error message from contract:

In [21]: @pre(lambda x: x > 0 or "x must be > 0")
    ...: def f(x):
    ...:     return x * 2
    ...:

In [22]: f(-5)
PreContractError: x must be > 0

Contracts chaining:

In [23]: @pre(lambda x: x > 0)
   ...: @pre(lambda x: x < 10)
   ...: def f(x):
   ...:     return x * 2
   ...:

In [24]: f(5)
Out[24]: 10

In [25]: f(-1)
PreContractError:

In [26]: f(12)
PreContractError:

Contracts chaining order

  • @inv: from top to bottom.

  • @pre: from top to bottom.

  • @post: from bottom to top.

Perfomance

NOTICE: 1 µs == 1000 ns

@pre and @post:

In [27]: f = lambda x: x

In [28]: pre_f = pre(lambda x: True)(f)

In [29]: post_f = post(lambda x: True)(f)

In [30]: %timeit f(10)
92.3 ns ± 3.62 ns per loop (mean ± std. dev. of 7 runs, 10000000 loops each)

In [31]: %timeit pre_f(10)
2.07 µs ± 92.5 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)

In [32]: %timeit post_f(10)
2.03 µs ± 18.6 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)

+1 µs

@inv:

In [33]: class A:
    ...:     x = 4
    ...:

In [34]: InvA = inv(lambda obj: True)(A)

In [35]: a = A()

In [36]: inv_a = InvA()

In [37]: %timeit a.x = 10
76.4 ns ± 1.36 ns per loop (mean ± std. dev. of 7 runs, 10000000 loops each)

In [38]: %timeit inv_a.x = 10
6.89 µs ± 408 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)

+6 µs

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page