Search code examples
pythonpylintpep8pyflakespychecker

Pylint, PyChecker or PyFlakes?


I would like to get some feedback on these tools on:

  • features;
  • adaptability;
  • ease of use and learning curve.

Solution

  • Well, I am a bit curious, so I just tested the three myself right after asking the question ;-)

    Ok, this is not a very serious review, but here is what I can say:

    I tried the tools with the default settings (it's important because you can pretty much choose your check rules) on the following script:

    #!/usr/local/bin/python
    # by Daniel Rosengren modified by e-satis
    
    import sys, time
    stdout = sys.stdout
    
    BAILOUT = 16
    MAX_ITERATIONS = 1000
    
    class Iterator(object) :
    
        def __init__(self):
    
            print 'Rendering...'
            for y in xrange(-39, 39):
                stdout.write('\n')
                for x in xrange(-39, 39):
                    if self.mandelbrot(x/40.0, y/40.0) :
                        stdout.write(' ')
                    else:
                        stdout.write('*')
    
    
        def mandelbrot(self, x, y):
            cr = y - 0.5
            ci = x
            zi = 0.0
            zr = 0.0
    
            for i in xrange(MAX_ITERATIONS) :
                temp = zr * zi
                zr2 = zr * zr
                zi2 = zi * zi
                zr = zr2 - zi2 + cr
                zi = temp + temp + ci
    
                if zi2 + zr2 > BAILOUT:
                    return i
    
            return 0
    
    t = time.time()
    Iterator()
    print '\nPython Elapsed %.02f' % (time.time() - t)
    

    As a result:

    • PyChecker is troublesome because it compiles the module to analyze it. If you don't want your code to run (e.g, it performs a SQL query), that's bad.
    • PyFlakes is supposed to be light. Indeed, it decided that the code was perfect. I am looking for something quite severe so I don't think I'll go for it.
    • PyLint has been very talkative and rated the code 3/10 (OMG, I'm a dirty coder !).

    Strong points of PyLint:

    • Very descriptive and accurate report.
    • Detect some code smells. Here it told me to drop my class to write something with functions because the OO approach was useless in this specific case. Something I knew, but never expected a computer to tell me :-p
    • The fully corrected code run faster (no class, no reference binding...).
    • Made by a French team. OK, it's not a plus for everybody, but I like it ;-)

    Cons of Pylint:

    • Some rules are really strict. I know that you can change it and that the default is to match PEP8, but is it such a crime to write 'for x in seq'? Apparently yes because you can't write a variable name with less than 3 letters. I will change that.
    • Very very talkative. Be ready to use your eyes.

    Corrected script (with lazy doc strings and variable names):

    #!/usr/local/bin/python
    # by Daniel Rosengren, modified by e-satis
    """
    Module doctring
    """
    
    
    import time
    from sys import stdout
    
    BAILOUT = 16
    MAX_ITERATIONS = 1000
    
    def mandelbrot(dim_1, dim_2):
        """
        function doc string
        """
        cr1 = dim_1 - 0.5
        ci1 = dim_2
        zi1 = 0.0
        zr1 = 0.0
    
        for i in xrange(MAX_ITERATIONS) :
            temp = zr1 * zi1
            zr2 = zr1 * zr1
            zi2 = zi1 * zi1
            zr1 = zr2 - zi2 + cr1
            zi1 = temp + temp + ci1
    
            if zi2 + zr2 > BAILOUT:
                return i
    
        return 0
    
    def execute() :
        """
        func doc string
        """
        print 'Rendering...'
        for dim_1 in xrange(-39, 39):
            stdout.write('\n')
            for dim_2 in xrange(-39, 39):
                if mandelbrot(dim_1/40.0, dim_2/40.0) :
                    stdout.write(' ')
                else:
                    stdout.write('*')
    
    
    START_TIME = time.time()
    execute()
    print '\nPython Elapsed %.02f' % (time.time() - START_TIME)
    

    Thanks to Rudiger Wolf, I discovered pep8 that does exactly what its name suggests: matching PEP8. It has found several syntax no-nos that Pylint did not. But Pylint found stuff that was not specifically linked to PEP8 but interesting. Both tools are interesting and complementary.

    Eventually I will use both since there are really easy to install (via packages or setuptools) and the output text is so easy to chain.

    To give you a little idea of their output:

    pep8:

    ./python_mandelbrot.py:4:11: E401 multiple imports on one line
    ./python_mandelbrot.py:10:1: E302 expected 2 blank lines, found 1
    ./python_mandelbrot.py:10:23: E203 whitespace before ':'
    ./python_mandelbrot.py:15:80: E501 line too long (108 characters)
    ./python_mandelbrot.py:23:1: W291 trailing whitespace
    ./python_mandelbrot.py:41:5: E301 expected 1 blank line, found 3
    

    Pylint:

    ************* Module python_mandelbrot
    C: 15: Line too long (108/80)
    C: 61: Line too long (85/80)
    C:  1: Missing docstring
    C:  5: Invalid name "stdout" (should match (([A-Z_][A-Z0-9_]*)|(__.*__))$)
    C: 10:Iterator: Missing docstring
    C: 15:Iterator.__init__: Invalid name "y" (should match [a-z_][a-z0-9_]{2,30}$)
    C: 17:Iterator.__init__: Invalid name "x" (should match [a-z_][a-z0-9_]{2,30}$)
    
    [...] and a very long report with useful stats like :
    
    Duplication
    -----------
    
    +-------------------------+------+---------+-----------+
    |                         |now   |previous |difference |
    +=========================+======+=========+===========+
    |nb duplicated lines      |0     |0        |=          |
    +-------------------------+------+---------+-----------+
    |percent duplicated lines |0.000 |0.000    |=          |
    +-------------------------+------+---------+-----------+