|
1
|
|
|
# Stolen from brentp: |
|
2
|
|
|
# <https://github.com/brentp/toolshed/blob/master/toolshed/files.py> |
|
3
|
|
|
|
|
4
|
|
|
import bz2 |
|
5
|
|
|
import gzip |
|
6
|
|
|
import sys |
|
7
|
|
|
import urllib |
|
8
|
|
|
|
|
9
|
|
|
|
|
10
|
|
|
if sys.version_info[0] < 3: |
|
11
|
|
|
int_types = (int, long) |
|
12
|
|
|
urlopen = urllib.urlopen |
|
13
|
|
|
else: |
|
14
|
|
|
int_types = (int,) |
|
15
|
|
|
basestring = str |
|
16
|
|
|
from urllib.request import urlopen |
|
17
|
|
|
|
|
18
|
|
|
|
|
19
|
|
|
def nopen(f, mode="r"): |
|
20
|
|
|
r""" |
|
21
|
|
|
open a file that's gzipped or return stdin for '-' |
|
22
|
|
|
if f is a number, the result of nopen(sys.argv[f]) is returned. |
|
23
|
|
|
>>> nopen('-') == sys.stdin, nopen('-', 'w') == sys.stdout |
|
24
|
|
|
(True, True) |
|
25
|
|
|
>>> nopen(sys.argv[0]) |
|
26
|
|
|
<...file...> |
|
27
|
|
|
# expands user and vars ($HOME) |
|
28
|
|
|
>>> nopen("~/.bashrc").name == nopen("$HOME/.bashrc").name |
|
29
|
|
|
True |
|
30
|
|
|
# an already open file. |
|
31
|
|
|
>>> nopen(open(sys.argv[0])) |
|
32
|
|
|
<...file...> |
|
33
|
|
|
>>> nopen(0) |
|
34
|
|
|
<...file...> |
|
35
|
|
|
Or provide nicer access to Popen.stdout |
|
36
|
|
|
>>> files = list(nopen("|ls")) |
|
37
|
|
|
>>> assert 'setup.py\n' in files or b'setup.py\n' in files, files |
|
38
|
|
|
""" |
|
39
|
|
|
if isinstance(f, int_types): |
|
40
|
|
|
return nopen(sys.argv[f], mode) |
|
41
|
|
|
|
|
42
|
|
|
if not isinstance(f, basestring): |
|
43
|
|
|
return f |
|
44
|
|
|
if f.startswith("|"): |
|
45
|
|
|
# using shell explicitly makes things like process substitution work: |
|
46
|
|
|
# http://stackoverflow.com/questions/7407667/python-subprocess-subshells-and-redirection |
|
47
|
|
|
# use sys.stderr so we dont have to worry about checking it... |
|
48
|
|
|
p = Popen(f[1:], stdout=PIPE, stdin=PIPE, |
|
49
|
|
|
stderr=sys.stderr if mode == "r" else PIPE, |
|
50
|
|
|
shell=True, bufsize=-1, # use system default for buffering |
|
51
|
|
|
preexec_fn=prefunc, |
|
52
|
|
|
close_fds=False, executable=os.environ.get('SHELL')) |
|
53
|
|
|
if sys.version_info[0] > 2: |
|
54
|
|
|
import io |
|
55
|
|
|
p.stdout = io.TextIOWrapper(p.stdout) |
|
56
|
|
|
p.stdin = io.TextIOWrapper(p.stdin) |
|
57
|
|
|
if mode != "r": |
|
58
|
|
|
p.stderr = io.TextIOWrapper(p.stderr) |
|
59
|
|
|
|
|
60
|
|
|
if mode and mode[0] == "r": |
|
61
|
|
|
return process_iter(p, f[1:]) |
|
62
|
|
|
return p |
|
63
|
|
|
|
|
64
|
|
|
if f.startswith(("http://", "https://", "ftp://")): |
|
65
|
|
|
fh = urlopen(f) |
|
66
|
|
|
if f.endswith(".gz"): |
|
67
|
|
|
return ungzipper(fh) |
|
68
|
|
|
if sys.version_info[0] < 3: |
|
69
|
|
|
return fh |
|
70
|
|
|
import io |
|
71
|
|
|
return io.TextIOWrapper(fh) |
|
72
|
|
|
f = op.expanduser(op.expandvars(f)) |
|
73
|
|
|
if f.endswith((".gz", ".Z", ".z")): |
|
74
|
|
|
fh = gzip.open(f, mode) |
|
75
|
|
|
if sys.version_info[0] < 3: |
|
76
|
|
|
return fh |
|
77
|
|
|
import io |
|
78
|
|
|
return io.TextIOWrapper(fh) |
|
79
|
|
|
elif f.endswith((".bz", ".bz2", ".bzip2")): |
|
80
|
|
|
fh = bz2.BZ2File(f, mode) |
|
81
|
|
|
if sys.version_info[0] < 3: |
|
82
|
|
|
return fh |
|
83
|
|
|
import io |
|
84
|
|
|
return io.TextIOWrapper(fh) |
|
85
|
|
|
|
|
86
|
|
|
return {"r": sys.stdin, "w": sys.stdout}[mode[0]] if f == "-" \ |
|
87
|
|
|
else open(f, mode) |
|
88
|
|
|
|
|
89
|
|
|
|
|
90
|
|
|
def ungzipper(fh, blocksize=16384): |
|
91
|
|
|
""" |
|
92
|
|
|
work-around to get streaming download of http://.../some.gz |
|
93
|
|
|
""" |
|
94
|
|
|
import zlib |
|
95
|
|
|
uzip = zlib.decompressobj(16 + zlib.MAX_WBITS) |
|
96
|
|
|
data = uzip.decompress(fh.read(blocksize)).split("\n") |
|
97
|
|
|
|
|
98
|
|
|
while len(data[0]): |
|
99
|
|
|
# last chunk might not be a full line. |
|
100
|
|
|
save = data.pop() |
|
101
|
|
|
for line in data: |
|
102
|
|
|
yield line |
|
103
|
|
|
data = uzip.decompress(fh.read(blocksize)).split("\n") |
|
104
|
|
|
# first line is prepended with saved chunk from end of last set. |
|
105
|
|
|
data[0] = save + data[0] |
|
106
|
|
|
|
tanghaibao /
goatools
