623.py

import re
import sys
from pathlib import Path

REGEXES = [(re.compile(p, f), r) for p, r, f in [
    (r'^#(?: -\*-) coding: ?utf-8(?: -\*-)\n', '', 0),
    (r"\.(encode|decode)\('utf-?8'\)", r'.\1()', 0),
    ('(class \w+)\(object\):\n', r'\1:\n', 0),
    ('super\(\w+, \w+\)\.', 'super().', 0),
    ('u"""(.*?)"""', r'"""\1"""', re.S),
    ("u?'(.*?)'", r"'\1'", 0),
    ('u?"(.*?)"', r'"\1"', 0),
    ('^from __future__.*?\n', '\n', re.M),
    ('^import six\n', '\n', re.M),
    ('^from six\.moves import (range|map|zip|filter|input)\n', '\n', re.M),
    ('^from six import StringIO\n', '\nfrom io import StringIO\n', re.M),
    ('^from six.moves.urllib.parse import urlparse', '\nfrom urllib.parse import urlparse', re.M),
    ('six.(text_type|string_types)', 'str', re.M),
    ('six.binary_type', 'bytes', 0),
    ('six.iteritems\((.+?)\)', r'\1.items()', 0),
    ('def __unicode__\(self\):', 'def __str__(self):', 0),
    ('\n +def __str__\(self\): *\n +return self.__unicode__\(\)\n', '', 0),
    ('^\n+', '', 0),
]]


def traverse(p: Path) -> int:
    changed = 0
    for pp in p.iterdir():
        if pp.is_dir():
            changed += traverse(pp)
        elif pp.is_file() and pp.suffix == '.py':
            text = pp.read_text()
            b4 = hash(text)
            for pattern, repl in REGEXES:
                text = pattern.sub(repl, text)
            if b4 != hash(text):
                pp.write_text(text)
                changed += 1
    return changed


changed_files = traverse(Path(sys.argv[1]))
print('{} files changed'.format(changed_files))