qmk_firmware

QMK firmware for my keyboards (Corne, Sweep Ferris) and trackball (Ploopy Adept)
Log | Files | Refs | Submodules | LICENSE

license_check.py (4860B)

      1 # Copyright 2023 Nick Brassel (@tzarc)
      2 # SPDX-License-Identifier: GPL-2.0-or-later
      3 import re
      4 from milc import cli
      5 from qmk.constants import LICENSE_TEXTS
      6 from qmk.path import normpath
      7 
      8 L_PAREN = re.compile(r'\(\[\{\<')
      9 R_PAREN = re.compile(r'\)\]\}\>')
     10 PUNCTUATION = re.compile(r'[\.,;:]+')
     11 TRASH_PREFIX = re.compile(r'^(\s|/|\*|#)+')
     12 TRASH_SUFFIX = re.compile(r'(\s|/|\*|#|\\)+$')
     13 SPACE = re.compile(r'\s+')
     14 SUFFIXES = ['.c', '.h', '.cpp', '.cxx', '.hpp', '.hxx']
     15 
     16 
     17 def _simplify_text(input):
     18     lines = input.lower().split('\n')
     19     lines = [PUNCTUATION.sub('', line) for line in lines]
     20     lines = [TRASH_PREFIX.sub('', line) for line in lines]
     21     lines = [TRASH_SUFFIX.sub('', line) for line in lines]
     22     lines = [SPACE.sub(' ', line) for line in lines]
     23     lines = [L_PAREN.sub('(', line) for line in lines]
     24     lines = [R_PAREN.sub(')', line) for line in lines]
     25     lines = [line.strip() for line in lines]
     26     lines = [line for line in lines if line is not None and line != '']
     27     return ' '.join(lines)
     28 
     29 
     30 def _preformat_license_texts():
     31     # Pre-format all the licenses
     32     for _, long_licenses in LICENSE_TEXTS:
     33         for i in range(len(long_licenses)):
     34             long_licenses[i] = _simplify_text(long_licenses[i])
     35 
     36 
     37 def _determine_suffix_condition(extensions):
     38     def _default_suffix_condition(s):
     39         return s in SUFFIXES
     40 
     41     conditional = _default_suffix_condition
     42 
     43     if extensions is not None and len(extensions) > 0:
     44         suffixes = [f'.{s}' if not s.startswith('.') else s for s in extensions]
     45 
     46         def _specific_suffix_condition(s):
     47             return s in suffixes
     48 
     49         conditional = _specific_suffix_condition
     50 
     51     return conditional
     52 
     53 
     54 def _determine_file_list(inputs, conditional):
     55     check_list = set()
     56     for filename in inputs:
     57         if filename.is_dir():
     58             for file in sorted(filename.rglob('*')):
     59                 if file.is_file() and conditional(file.suffix):
     60                     check_list.add(file)
     61         elif filename.is_file():
     62             if conditional(filename.suffix):
     63                 check_list.add(filename)
     64 
     65     return list(sorted(check_list))
     66 
     67 
     68 def _detect_license_from_file_contents(filename, absolute=False, short=False):
     69     data = filename.read_text(encoding='utf-8', errors='ignore')
     70     filename_out = str(filename.absolute()) if absolute else str(filename)
     71 
     72     if 'SPDX-License-Identifier:' in data:
     73         res = data.split('SPDX-License-Identifier:')
     74         license = re.split(r'\s|//|\*', res[1].strip())[0].strip()
     75         found = False
     76         for short_license, _ in LICENSE_TEXTS:
     77             if license.lower() == short_license.lower():
     78                 license = short_license
     79                 found = True
     80                 break
     81 
     82         if not found:
     83             if short:
     84                 print(f'{filename_out} UNKNOWN')
     85             else:
     86                 cli.log.error(f'{{fg_cyan}}{filename_out}{{fg_reset}} -- unknown license, or no license detected!')
     87             return False
     88 
     89         if short:
     90             print(f'{filename_out} {license}')
     91         else:
     92             cli.log.info(f'{{fg_cyan}}{filename_out}{{fg_reset}} -- license detected: {license} (SPDX License Identifier)')
     93         return True
     94 
     95     else:
     96         simple_text = _simplify_text(data)
     97         for short_license, long_licenses in LICENSE_TEXTS:
     98             for long_license in long_licenses:
     99                 if long_license in simple_text:
    100                     if short:
    101                         print(f'{filename_out} {short_license}')
    102                     else:
    103                         cli.log.info(f'{{fg_cyan}}{filename_out}{{fg_reset}} -- license detected: {short_license} (Full text)')
    104                     return True
    105 
    106         if short:
    107             print(f'{filename_out} UNKNOWN')
    108         else:
    109             cli.log.error(f'{{fg_cyan}}{filename_out}{{fg_reset}} -- unknown license, or no license detected!')
    110 
    111     return False
    112 
    113 
    114 @cli.argument('inputs', nargs='*', arg_only=True, type=normpath, help='List of input files or directories.')
    115 @cli.argument('-s', '--short', action='store_true', help='Short output.')
    116 @cli.argument('-a', '--absolute', action='store_true', help='Print absolute paths.')
    117 @cli.argument('-e', '--extension', arg_only=True, action='append', default=[], help='Override list of extensions. Can be specified multiple times for multiple extensions.')
    118 @cli.subcommand('File license check.', hidden=False if cli.config.user.developer else True)
    119 def license_check(cli):
    120     _preformat_license_texts()
    121 
    122     conditional = _determine_suffix_condition(cli.args.extension)
    123     check_list = _determine_file_list(cli.args.inputs, conditional)
    124 
    125     failed = False
    126     for filename in sorted(check_list):
    127         if not _detect_license_from_file_contents(filename, absolute=cli.args.absolute, short=cli.args.short):
    128             failed = True
    129 
    130     if failed:
    131         return False