Wed, 13 Jul 2022 15:34:50 +0200
Revisions <no_multi_processing, Variables Viewer, with_python2> closed.
5714
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
1 | #!/usr/bin/env python |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
2 | """ |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
3 | Script which takes one or more file paths and reports on their detected |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
4 | encodings |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
5 | |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
6 | Example:: |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
7 | |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
8 | % chardetect somefile someotherfile |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
9 | somefile: windows-1252 with confidence 0.5 |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
10 | someotherfile: ascii with confidence 1.0 |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
11 | |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
12 | If no paths are provided, it takes its input from stdin. |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
13 | |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
14 | """ |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
15 | |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
16 | from __future__ import absolute_import, print_function, unicode_literals |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
17 | |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
18 | import argparse |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
19 | import sys |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
20 | from io import open |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
21 | |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
22 | from chardet import __version__ |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
23 | from chardet.universaldetector import UniversalDetector |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
24 | |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
25 | |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
26 | def description_of(lines, name='stdin'): |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
27 | """ |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
28 | Return a string describing the probable encoding of a file or |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
29 | list of strings. |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
30 | |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
31 | :param lines: The lines to get the encoding of. |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
32 | :type lines: Iterable of bytes |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
33 | :param name: Name of file or collection of lines |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
34 | :type name: str |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
35 | """ |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
36 | u = UniversalDetector() |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
37 | for line in lines: |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
38 | u.feed(line) |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
39 | u.close() |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
40 | result = u.result |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
41 | if result['encoding']: |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
42 | return '{0}: {1} with confidence {2}'.format(name, result['encoding'], |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
43 | result['confidence']) |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
44 | else: |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
45 | return '{0}: no result'.format(name) |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
46 | |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
47 | |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
48 | def main(argv=None): |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
49 | ''' |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
50 | Handles command line arguments and gets things started. |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
51 | |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
52 | :param argv: List of arguments, as if specified on the command-line. |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
53 | If None, ``sys.argv[1:]`` is used instead. |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
54 | :type argv: list of str |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
55 | ''' |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
56 | # Get command line arguments |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
57 | parser = argparse.ArgumentParser( |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
58 | description="Takes one or more file paths and reports their detected \ |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
59 | encodings", |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
60 | formatter_class=argparse.ArgumentDefaultsHelpFormatter, |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
61 | conflict_handler='resolve') |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
62 | parser.add_argument('input', |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
63 | help='File whose encoding we would like to determine.', |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
64 | type=argparse.FileType('rb'), nargs='*', |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
65 | default=[sys.stdin]) |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
66 | parser.add_argument('--version', action='version', |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
67 | version='%(prog)s {0}'.format(__version__)) |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
68 | args = parser.parse_args(argv) |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
69 | |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
70 | for f in args.input: |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
71 | if f.isatty(): |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
72 | print("You are running chardetect interactively. Press " + |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
73 | "CTRL-D twice at the start of a blank line to signal the " + |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
74 | "end of your input. If you want help, run chardetect " + |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
75 | "--help\n", file=sys.stderr) |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
76 | print(description_of(f, f.name)) |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
77 | |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
78 | |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
79 | if __name__ == '__main__': |
90c57b50600f
Updated chardet to 3.0.2.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
80 | main() |