That Pesky Error in PyLint’s Results#
How to avoid PyLint’s Module ‘hashlib’ has no ‘md5’ member error when using hashlib’s md5() function in your code, and the consequences of doing so.
I do not use md5 much at all these days, as it is known to be fairly unsafe; having said that, md5 does still have its uses when it comes to comparing file likeness in non-critical applications. It just so happens that PyLint does not seem to like invoking hashlib with specific constructors and will therefore mark down code ratings as a result. See the example below:
>>> import hashlib
>>> HASH = hashlib.md5()
>>> HASH.update('Test my hash')
>>> HASH.hexdigest()
'f94b71b31c1d09d352db8b59d4f98892'Checking the source code for a particular file using this method of invocating md5, PyLint invariably gives me an error message similar to the following:
raz@foobox:~/Python/soap$ pylint soap_client.py
No config file found, using default configuration
************* Module somemodule
E:231:SomeClass._some_method: Module 'hashlib' has no 'md5' member
[...]
Messages
--------
+-----------+------------+
|message id |occurrences |
+===========+============+
|E1101 |1 |
+-----------+------------+
Global evaluation
-----------------
Your code has been rated at 9.68/10Hmm, that is not good at all. I want a better PyLint score, something like 10/10 would do just fine, thanks.
What to do Slightly Differently#
Seeing as I would like my code to score as highly as possible in PyLint tests, what to do? The answer is to use md5 in a subtly different way to get rid of that error…
Instead of invocating the named constructor, such as hashlib.md5(), use the generic new(), constructor `hashlib.new(‘md5’)`:
>>> import hashlib
>>> HASH = hashlib.new('md5')
>>> HASH.update('Test my hash')
>>> HASH.hexdigest()
'f94b71b31c1d09d352db8b59d4fq98892'The interesting bit of test output now shows:
Global evaluation
-----------------
Your code has been rated at 10.00/10 (previous run: 9.68/10)Hooray, full marks at last! But, read on…!
A Minor Cautionary Note#
This does come with a minor health warning: as briefly mentioned in the hashlib documentation, the new() constructor runs slower than the named constructors. Let me put this to a very basic test, using Python’s timeit module.
Firstly with the generic constructor:
>>> import timeit
>>> s = """\
... import hashlib
... HASH = hashlib.new('md5')
... HASH.update('Test my hash')
... HASH.hexdigest()
... """
>>> t = timeit.Timer(stmt=s)
>>> print "%.2f usec/pass" % (1000000 * t.timeit(number=100000)/100000)
3.02 usec/passNow, with the named constructor:
>>> import timeit
>>> s = """\
... import hashlib
... HASH = hashlib.md5()
... HASH.update('Test my hash')
... HASH.hexdigest()
... """
>>> t = timeit.Timer(stmt=s)
>>> print "%.2f usec/pass" % (1000000 * t.timeit(number=100000)/100000)
2.41 usec/passSo we have knocked off approximately 20% of execution time in this code by using the named constructor at the expense of a drop in the overall PyLint score.
Finally#
As with most things, this is not a black and white issue. If performance is an quality of important consideration in your code, you may just have to put up with lower PyLint code ratings!
