| 1 | n/a | #!/usr/bin/env python3 |
|---|
| 2 | n/a | |
|---|
| 3 | n/a | """List all those Python files that require a coding directive |
|---|
| 4 | n/a | |
|---|
| 5 | n/a | Usage: findnocoding.py dir1 [dir2...] |
|---|
| 6 | n/a | """ |
|---|
| 7 | n/a | |
|---|
| 8 | n/a | __author__ = "Oleg Broytmann, Georg Brandl" |
|---|
| 9 | n/a | |
|---|
| 10 | n/a | import sys, os, re, getopt |
|---|
| 11 | n/a | |
|---|
| 12 | n/a | # our pysource module finds Python source files |
|---|
| 13 | n/a | try: |
|---|
| 14 | n/a | import pysource |
|---|
| 15 | n/a | except ImportError: |
|---|
| 16 | n/a | # emulate the module with a simple os.walk |
|---|
| 17 | n/a | class pysource: |
|---|
| 18 | n/a | has_python_ext = looks_like_python = can_be_compiled = None |
|---|
| 19 | n/a | def walk_python_files(self, paths, *args, **kwargs): |
|---|
| 20 | n/a | for path in paths: |
|---|
| 21 | n/a | if os.path.isfile(path): |
|---|
| 22 | n/a | yield path.endswith(".py") |
|---|
| 23 | n/a | elif os.path.isdir(path): |
|---|
| 24 | n/a | for root, dirs, files in os.walk(path): |
|---|
| 25 | n/a | for filename in files: |
|---|
| 26 | n/a | if filename.endswith(".py"): |
|---|
| 27 | n/a | yield os.path.join(root, filename) |
|---|
| 28 | n/a | pysource = pysource() |
|---|
| 29 | n/a | |
|---|
| 30 | n/a | |
|---|
| 31 | n/a | print("The pysource module is not available; " |
|---|
| 32 | n/a | "no sophisticated Python source file search will be done.", file=sys.stderr) |
|---|
| 33 | n/a | |
|---|
| 34 | n/a | |
|---|
| 35 | n/a | decl_re = re.compile(rb'^[ \t\f]*#.*?coding[:=][ \t]*([-\w.]+)') |
|---|
| 36 | n/a | blank_re = re.compile(rb'^[ \t\f]*(?:[#\r\n]|$)') |
|---|
| 37 | n/a | |
|---|
| 38 | n/a | def get_declaration(line): |
|---|
| 39 | n/a | match = decl_re.match(line) |
|---|
| 40 | n/a | if match: |
|---|
| 41 | n/a | return match.group(1) |
|---|
| 42 | n/a | return b'' |
|---|
| 43 | n/a | |
|---|
| 44 | n/a | def has_correct_encoding(text, codec): |
|---|
| 45 | n/a | try: |
|---|
| 46 | n/a | str(text, codec) |
|---|
| 47 | n/a | except UnicodeDecodeError: |
|---|
| 48 | n/a | return False |
|---|
| 49 | n/a | else: |
|---|
| 50 | n/a | return True |
|---|
| 51 | n/a | |
|---|
| 52 | n/a | def needs_declaration(fullpath): |
|---|
| 53 | n/a | try: |
|---|
| 54 | n/a | infile = open(fullpath, 'rb') |
|---|
| 55 | n/a | except IOError: # Oops, the file was removed - ignore it |
|---|
| 56 | n/a | return None |
|---|
| 57 | n/a | |
|---|
| 58 | n/a | with infile: |
|---|
| 59 | n/a | line1 = infile.readline() |
|---|
| 60 | n/a | line2 = infile.readline() |
|---|
| 61 | n/a | |
|---|
| 62 | n/a | if (get_declaration(line1) or |
|---|
| 63 | n/a | blank_re.match(line1) and get_declaration(line2)): |
|---|
| 64 | n/a | # the file does have an encoding declaration, so trust it |
|---|
| 65 | n/a | return False |
|---|
| 66 | n/a | |
|---|
| 67 | n/a | # check the whole file for non utf-8 characters |
|---|
| 68 | n/a | rest = infile.read() |
|---|
| 69 | n/a | |
|---|
| 70 | n/a | if has_correct_encoding(line1+line2+rest, "utf-8"): |
|---|
| 71 | n/a | return False |
|---|
| 72 | n/a | |
|---|
| 73 | n/a | return True |
|---|
| 74 | n/a | |
|---|
| 75 | n/a | |
|---|
| 76 | n/a | usage = """Usage: %s [-cd] paths... |
|---|
| 77 | n/a | -c: recognize Python source files trying to compile them |
|---|
| 78 | n/a | -d: debug output""" % sys.argv[0] |
|---|
| 79 | n/a | |
|---|
| 80 | n/a | if __name__ == '__main__': |
|---|
| 81 | n/a | |
|---|
| 82 | n/a | try: |
|---|
| 83 | n/a | opts, args = getopt.getopt(sys.argv[1:], 'cd') |
|---|
| 84 | n/a | except getopt.error as msg: |
|---|
| 85 | n/a | print(msg, file=sys.stderr) |
|---|
| 86 | n/a | print(usage, file=sys.stderr) |
|---|
| 87 | n/a | sys.exit(1) |
|---|
| 88 | n/a | |
|---|
| 89 | n/a | is_python = pysource.looks_like_python |
|---|
| 90 | n/a | debug = False |
|---|
| 91 | n/a | |
|---|
| 92 | n/a | for o, a in opts: |
|---|
| 93 | n/a | if o == '-c': |
|---|
| 94 | n/a | is_python = pysource.can_be_compiled |
|---|
| 95 | n/a | elif o == '-d': |
|---|
| 96 | n/a | debug = True |
|---|
| 97 | n/a | |
|---|
| 98 | n/a | if not args: |
|---|
| 99 | n/a | print(usage, file=sys.stderr) |
|---|
| 100 | n/a | sys.exit(1) |
|---|
| 101 | n/a | |
|---|
| 102 | n/a | for fullpath in pysource.walk_python_files(args, is_python): |
|---|
| 103 | n/a | if debug: |
|---|
| 104 | n/a | print("Testing for coding: %s" % fullpath) |
|---|
| 105 | n/a | result = needs_declaration(fullpath) |
|---|
| 106 | n/a | if result: |
|---|
| 107 | n/a | print(fullpath) |
|---|