Python function decorator calls

Question

I have a function that is called about 1742 times but when i do a decorator to calculate each call time of it i found that it prints only 647 times i don't know why that happens.

Update:

• My problem here is not with how to make timing its with the difference between the number of calls, I want a way to make a decorator like function that have accurate calls.

• My python version is 2.6 and here is the module i am working on http://pastebin.com/MXu1pLWM

• In the profiling output i found that the caller function to the decorated function calls it only 647 and it was the only caller function!!.

Figured out!

• The function that i decorated have a loop in it with length of 1742 but the actual calls to the function is only 647, although i still not understand why it says 1742 in calls section instead of 647 :)

---

number = 0
def timing(f):
    def wrap(*args):
        time1 = time.time()
        ret = f(*args)
        time2 = time.time()
        global number
        number+=1
        print '%s function took %0.6f ms No of calls: %s' % (f.func_name, ((time2-time1)), str(number))
        return ret
    return wrap

Answer 1

I saw your pasted code - on the pastebin. The problem is that you are wrapping the __iter__ method of a class. It is only called once for each loop that is started. And it shoud return imediatelly with an iterator object -- what would be profilable is the calls to the "next" method on the object returned by __iter__ (this is what is called in each for loop interacton).

Which means that without changing anything, instead of using your profiler as a decorator on a call to __iter__ you'd use that code in some other way around the yield statement inside taht __iter__

Other than that, you will be better of using timeit.timeit or other already existing profiles to avoid such traps, as are listed on the comments.