A script to get the current list of CPython version's magic numbers from CPython repository

get_python_versions_magics.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

"""
A script to get the current list of CPython version's magic numbers from CPython repository.
It is parses "/Lib/importlib/_bootstrap_external.py" source file to get this list.

And this is a helper for my script: https://gist.github.com/delimitry/bad5496b52161449f6de,
which allows to get the version of Python by which the file was compiled
"""

from __future__ import print_function

import re
import requests
from collections import OrderedDict


START_TAG = '# Known values:'
END_TAG = '# MAGIC must change whenever the bytecode emitted'

PYTHON_MAGIC_PATTERN = re.compile(r'(Python .*?):*\s+(\d+)')
ONLY_MAGIC_PATTERN = re.compile(r'\s+(\d+)\s')

BOOTSTRAP_EXTERNAL_PY_URL = 'https://raw.githubusercontent.com/python/cpython/master/Lib/importlib/_bootstrap_external.py'


def main():
    """Get CPython version's magic numbers from official CPython repository"""
    response = requests.get(BOOTSTRAP_EXTERNAL_PY_URL)
    data = response.text

    magics = []
    cur_ver, magic = '', ''
    pos = data.find(START_TAG)
    if pos < 0:
        return
    pos_end = data.find(END_TAG)
    found_data = data[pos + len(START_TAG): pos_end]
    for line in found_data.split('\n'):
        uncomment_line = line[1:] if line.startswith('#') else line
        res = PYTHON_MAGIC_PATTERN.findall(uncomment_line)
        if res:
            cur_ver, magic = res[0]
            magics.append((magic, cur_ver,))
        else:
            res = ONLY_MAGIC_PATTERN.findall(uncomment_line)
            if res:
                magic = res[0]
                magics.append((magic, cur_ver,))

    # make an ordered dict to handle magic numbers' duplicates
    magic_version = OrderedDict()
    for item in magics:
        magic, version = item
        if magic not in magic_version:
            magic_version[magic] = [version]
        else:
            magic_version[magic].append(version)

    # write out magics
    out = ''
    out += 'MAGICS = {\n'
    for magic, version in magic_version.items():
        out += '    {}: \'{}\',\n'.format(magic, ', '.join(sorted(set(version))))
    out += '}\n'
    print(out)


if __name__ == '__main__':
    main()