|
1
|
|
|
"""Low-level utilities for reading a variety of source formats.""" |
|
2
|
|
|
import codecs |
|
3
|
|
|
import gzip |
|
4
|
|
|
import io |
|
5
|
|
|
import tarfile |
|
6
|
|
|
import six |
|
7
|
|
|
|
|
8
|
|
|
|
|
9
|
|
|
def open_(filename, mode='r', encoding=None): |
|
10
|
|
|
"""Open a text file with encoding and optional gzip compression. |
|
11
|
|
|
|
|
12
|
|
|
Note that on legacy Python any encoding other than ``None`` or opening |
|
13
|
|
|
GZipped files will return an unpicklable file-like object. |
|
14
|
|
|
|
|
15
|
|
|
Parameters |
|
16
|
|
|
---------- |
|
17
|
|
|
filename : str |
|
18
|
|
|
The filename to read. |
|
19
|
|
|
mode : str, optional |
|
20
|
|
|
The mode with which to open the file. Defaults to `r`. |
|
21
|
|
|
encoding : str, optional |
|
22
|
|
|
The encoding to use (see the codecs documentation_ for supported |
|
23
|
|
|
values). Defaults to ``None``. |
|
24
|
|
|
|
|
25
|
|
|
.. _documentation: |
|
26
|
|
|
https://docs.python.org/3/library/codecs.html#standard-encodings |
|
27
|
|
|
|
|
28
|
|
|
""" |
|
29
|
|
|
if filename.endswith('.gz'): |
|
30
|
|
|
if six.PY2: |
|
31
|
|
|
zf = io.BufferedReader(gzip.open(filename, mode)) |
|
32
|
|
|
if encoding: |
|
33
|
|
|
return codecs.getreader(encoding)(zf) |
|
34
|
|
|
else: |
|
35
|
|
|
return zf |
|
36
|
|
|
else: |
|
37
|
|
|
return io.BufferedReader(gzip.open(filename, mode, |
|
38
|
|
|
encoding=encoding)) |
|
39
|
|
|
if six.PY2: |
|
40
|
|
|
if encoding: |
|
41
|
|
|
return codecs.open(filename, mode, encoding=encoding) |
|
42
|
|
|
else: |
|
43
|
|
|
return open(filename, mode) |
|
44
|
|
|
else: |
|
45
|
|
|
return open(filename, mode, encoding=encoding) |
|
46
|
|
|
|
|
47
|
|
|
|
|
48
|
|
|
def tar_open(f): |
|
49
|
|
|
"""Open either a filename or a file-like object as a TarFile. |
|
50
|
|
|
|
|
51
|
|
|
Parameters |
|
52
|
|
|
---------- |
|
53
|
|
|
f : str or file-like object |
|
54
|
|
|
The filename or file-like object from which to read. |
|
55
|
|
|
|
|
56
|
|
|
Returns |
|
57
|
|
|
------- |
|
58
|
|
|
TarFile |
|
59
|
|
|
A `TarFile` instance. |
|
60
|
|
|
|
|
61
|
|
|
""" |
|
62
|
|
|
if isinstance(f, six.string_types): |
|
63
|
|
|
return tarfile.open(name=f) |
|
64
|
|
|
else: |
|
65
|
|
|
return tarfile.open(fileobj=f) |
|
66
|
|
|
|
mila-udem /
fuel
