Rough code to scrape pycodestyle for errors and output .rst-ish

pycodestyle_errors.py

"""Rough script to scrape pycodestyle.py and intro.rst for error codes and descriptions, then generate something .rst-ish"""

import collections
import re

import pycodestyle


ErrorCode = collections.namedtuple("ErrorCode", field_names=("code", "short_msg", "long_msg"))

### scrape intro.rst
def generate_error_codes(file_name="intro.rst"):
    """ Read the docs/intro.rst file and extract error and warning codes.
    Relatively easy, since they're in a table and start with "| W" or "| E"
    """
    intro_codes = {}
    for line in open(file_name).read().split('\n'):
        if not (line.startswith("| E") or
                line.startswith("| W")):
            continue
        
        code = line[2:6]
        short_msg = line.split("|")[2].strip()
        
        #print((code, short_msg))
        intro_codes[code] = short_msg
    return intro_codes
    
intro_codes = generate_error_codes()
#print (intro_codes)


warn_or_error = re.compile('([EW]\d\d\d):')

### now scrape pycodestyle.py for warning/error functions
err_funcs = {}

# TODO: Check: multiple codes per function, but one function per code??
for f in dir(pycodestyle):
    func = getattr(pycodestyle, f)
    if type(func).__name__ != 'function' or not func.__doc__:
        continue
    
    codes = warn_or_error.findall(func.__doc__)
    if codes:
        #print (f"Match! {codes}")
        for match in set(codes):
            err_funcs[match] = func

#print ("-" * 42)

### Build 'master list' of errors
error_codes = {}
for code in sorted(intro_codes.keys()):
    if code in err_funcs:
        doc = err_funcs[code].__doc__
    else:
        doc = "No doc found for " + code
    ec = ErrorCode(
            code=code,
            short_msg=intro_codes[code],
            long_msg=doc)
    error_codes[code] = ec
    
    #print (f"{ec.code}: {ec.short_msg}")
    #print (ec.long_msg)
    #print ("-----\n")

print (""".. currentmodule:: pycodestyle

Error codes
============

These are the `PEP 8`_ errors that pycodestyle checks for.

.. contents::
   :local:
""")

for code, ec in error_codes.items():
    print (f"""
{ec.code}: {ec.short_msg}
{"-"*( len(ec.code) + len(ec.short_msg) + 2)}

{ec.long_msg}""")
# for some weird reason long message comes out with multiple trailing lines -- might be a Windows thing?

print ("""
.. _PEP 8: http://www.python.org/dev/peps/pep-0008/
""")