• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1import os
2import re
3import sys
4
5def check_encoding(encoding, scan_dir, regex_pattern):
6  fname = None
7  try:
8    assert encoding in ['ascii', 'utf-8'], "unexpected encoding"
9    cmp = re.compile(regex_pattern)
10    for root, dirs, files in os.walk(scan_dir):
11      fname = root
12      cmp_list = [f for f in files if cmp.search(f) is not None]
13      for f in cmp_list:
14        fname = os.path.join(root, f)
15        with open(fname, mode='rb') as test_file:
16          btext = test_file.read()
17        # check encoding
18        btext.decode(encoding=encoding, errors="strict")
19        if encoding == "utf-8" and btext.startswith(b'\xEF\xBB\xBF'):
20          raise ValueError("unexpected BOM in file")
21        # check LF line endings
22        LF = btext.count(b'\n')
23        CR = btext.count(b'\r')
24        if CR!=0:
25          raise ValueError("invalid line endings: LF({})/CR({})".format(LF, CR))
26  except Exception as err:
27    print("ERROR with [{}]: {}".format(fname, err))
28    return -1
29  else:
30    return 0
31
32if __name__ == "__main__":
33  # python check-sources.sh.py 'ascii' '.' '.*\.(cpp|h)$'
34  res = check_encoding(sys.argv[1], sys.argv[2], sys.argv[3])
35  sys.exit(0 if res == 0 else -1)
36