wh

wh, a wormhole that helps you redirect time and space. Literally a decorator for recording the elapsed time and the number of calls for a certain Python function.

Notes: It brings slight overheads including additional function calls (e.g. incrementing time) and branches (e.g. for tweaking the context). Don't mess with tiny little stubs or highly precise benchmarks with this module.

Usage

Methods

• wh.trek(stream=sys.stdout): Decorator for recording the call info, which is logged in sys.stdout by default. Returns an instance of _WhContextManager
• _WhContextManager.reset(): Manually resets states for next contexts
• _WhContextManager.done(): Manually finishes the task and starts logging and self.reset()-ting

Attributes

• _WhContextManager.ncall: Number of calls
• _WhContextManager.elapsed: Elapsed time
• _WhContextManager.retval: Return value of the function. If it is not even called before, TypeError would be raised (e.g. foo().retval instead of foo.retval right after foo's declaration)

Example

import wh


@wh.trek()
def fib(n):
    """Fibonacci number"""
    return fib(n - 2) + fib(n - 1) if n > 2 else 1


# No benchmark output.
assert fib(10) == 55
fib.reset()  # reset for a next call


# Output triggered by `with`.
with fib(10) as ret:
    assert ret == 55
# Output:
#     [wh] fib: 109 calls, 0.157958984375(ms) elapsed
#


# File I/O.
f = open('foo.txt', 'w')


@wh.trek(f)
def fac(n):
    """Factorial"""
    return n * fac(n - 1) if n > 1 else 1


assert fac(4) == 24
fac.done() # manually writes to file
f.close()  # context should be managed manually

Install

$ pip install wh

License

MIT