annotate lib/python3.8/site-packages/pip/_vendor/requests/utils.py @ 0:9e54283cc701 draft

"planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
author guerler
date Mon, 27 Jul 2020 03:47:31 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1 # -*- coding: utf-8 -*-
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
3 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
4 requests.utils
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
5 ~~~~~~~~~~~~~~
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
6
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
7 This module provides utility functions that are used within Requests
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
8 that are also useful for external consumption.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
9 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
10
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
11 import codecs
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
12 import contextlib
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
13 import io
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
14 import os
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
15 import re
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
16 import socket
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
17 import struct
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
18 import sys
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
19 import tempfile
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
20 import warnings
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
21 import zipfile
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
22
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
23 from .__version__ import __version__
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
24 from . import certs
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
25 # to_native_string is unused here, but imported here for backwards compatibility
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
26 from ._internal_utils import to_native_string
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
27 from .compat import parse_http_list as _parse_list_header
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
28 from .compat import (
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
29 quote, urlparse, bytes, str, OrderedDict, unquote, getproxies,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
30 proxy_bypass, urlunparse, basestring, integer_types, is_py3,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
31 proxy_bypass_environment, getproxies_environment, Mapping)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
32 from .cookies import cookiejar_from_dict
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
33 from .structures import CaseInsensitiveDict
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
34 from .exceptions import (
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
35 InvalidURL, InvalidHeader, FileModeWarning, UnrewindableBodyError)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
36
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
37 NETRC_FILES = ('.netrc', '_netrc')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
38
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
39 DEFAULT_CA_BUNDLE_PATH = certs.where()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
40
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
41 DEFAULT_PORTS = {'http': 80, 'https': 443}
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
42
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
43
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
44 if sys.platform == 'win32':
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
45 # provide a proxy_bypass version on Windows without DNS lookups
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
46
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
47 def proxy_bypass_registry(host):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
48 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
49 if is_py3:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
50 import winreg
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
51 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
52 import _winreg as winreg
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
53 except ImportError:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
54 return False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
55
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
56 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
57 internetSettings = winreg.OpenKey(winreg.HKEY_CURRENT_USER,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
58 r'Software\Microsoft\Windows\CurrentVersion\Internet Settings')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
59 # ProxyEnable could be REG_SZ or REG_DWORD, normalizing it
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
60 proxyEnable = int(winreg.QueryValueEx(internetSettings,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
61 'ProxyEnable')[0])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
62 # ProxyOverride is almost always a string
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
63 proxyOverride = winreg.QueryValueEx(internetSettings,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
64 'ProxyOverride')[0]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
65 except OSError:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
66 return False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
67 if not proxyEnable or not proxyOverride:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
68 return False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
69
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
70 # make a check value list from the registry entry: replace the
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
71 # '<local>' string by the localhost entry and the corresponding
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
72 # canonical entry.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
73 proxyOverride = proxyOverride.split(';')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
74 # now check if we match one of the registry values.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
75 for test in proxyOverride:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
76 if test == '<local>':
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
77 if '.' not in host:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
78 return True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
79 test = test.replace(".", r"\.") # mask dots
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
80 test = test.replace("*", r".*") # change glob sequence
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
81 test = test.replace("?", r".") # change glob char
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
82 if re.match(test, host, re.I):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
83 return True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
84 return False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
85
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
86 def proxy_bypass(host): # noqa
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
87 """Return True, if the host should be bypassed.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
88
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
89 Checks proxy settings gathered from the environment, if specified,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
90 or the registry.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
91 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
92 if getproxies_environment():
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
93 return proxy_bypass_environment(host)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
94 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
95 return proxy_bypass_registry(host)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
96
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
97
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
98 def dict_to_sequence(d):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
99 """Returns an internal sequence dictionary update."""
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
100
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
101 if hasattr(d, 'items'):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
102 d = d.items()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
103
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
104 return d
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
105
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
106
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
107 def super_len(o):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
108 total_length = None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
109 current_position = 0
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
110
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
111 if hasattr(o, '__len__'):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
112 total_length = len(o)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
113
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
114 elif hasattr(o, 'len'):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
115 total_length = o.len
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
116
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
117 elif hasattr(o, 'fileno'):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
118 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
119 fileno = o.fileno()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
120 except io.UnsupportedOperation:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
121 pass
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
122 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
123 total_length = os.fstat(fileno).st_size
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
124
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
125 # Having used fstat to determine the file length, we need to
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
126 # confirm that this file was opened up in binary mode.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
127 if 'b' not in o.mode:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
128 warnings.warn((
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
129 "Requests has determined the content-length for this "
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
130 "request using the binary size of the file: however, the "
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
131 "file has been opened in text mode (i.e. without the 'b' "
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
132 "flag in the mode). This may lead to an incorrect "
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
133 "content-length. In Requests 3.0, support will be removed "
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
134 "for files in text mode."),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
135 FileModeWarning
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
136 )
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
137
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
138 if hasattr(o, 'tell'):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
139 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
140 current_position = o.tell()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
141 except (OSError, IOError):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
142 # This can happen in some weird situations, such as when the file
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
143 # is actually a special file descriptor like stdin. In this
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
144 # instance, we don't know what the length is, so set it to zero and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
145 # let requests chunk it instead.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
146 if total_length is not None:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
147 current_position = total_length
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
148 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
149 if hasattr(o, 'seek') and total_length is None:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
150 # StringIO and BytesIO have seek but no useable fileno
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
151 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
152 # seek to end of file
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
153 o.seek(0, 2)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
154 total_length = o.tell()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
155
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
156 # seek back to current position to support
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
157 # partially read file-like objects
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
158 o.seek(current_position or 0)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
159 except (OSError, IOError):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
160 total_length = 0
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
161
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
162 if total_length is None:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
163 total_length = 0
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
164
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
165 return max(0, total_length - current_position)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
166
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
167
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
168 def get_netrc_auth(url, raise_errors=False):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
169 """Returns the Requests tuple auth for a given url from netrc."""
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
170
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
171 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
172 from netrc import netrc, NetrcParseError
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
173
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
174 netrc_path = None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
175
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
176 for f in NETRC_FILES:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
177 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
178 loc = os.path.expanduser('~/{}'.format(f))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
179 except KeyError:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
180 # os.path.expanduser can fail when $HOME is undefined and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
181 # getpwuid fails. See https://bugs.python.org/issue20164 &
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
182 # https://github.com/requests/requests/issues/1846
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
183 return
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
184
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
185 if os.path.exists(loc):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
186 netrc_path = loc
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
187 break
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
188
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
189 # Abort early if there isn't one.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
190 if netrc_path is None:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
191 return
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
192
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
193 ri = urlparse(url)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
194
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
195 # Strip port numbers from netloc. This weird `if...encode`` dance is
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
196 # used for Python 3.2, which doesn't support unicode literals.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
197 splitstr = b':'
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
198 if isinstance(url, str):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
199 splitstr = splitstr.decode('ascii')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
200 host = ri.netloc.split(splitstr)[0]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
201
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
202 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
203 _netrc = netrc(netrc_path).authenticators(host)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
204 if _netrc:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
205 # Return with login / password
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
206 login_i = (0 if _netrc[0] else 1)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
207 return (_netrc[login_i], _netrc[2])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
208 except (NetrcParseError, IOError):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
209 # If there was a parsing error or a permissions issue reading the file,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
210 # we'll just skip netrc auth unless explicitly asked to raise errors.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
211 if raise_errors:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
212 raise
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
213
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
214 # AppEngine hackiness.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
215 except (ImportError, AttributeError):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
216 pass
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
217
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
218
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
219 def guess_filename(obj):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
220 """Tries to guess the filename of the given object."""
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
221 name = getattr(obj, 'name', None)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
222 if (name and isinstance(name, basestring) and name[0] != '<' and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
223 name[-1] != '>'):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
224 return os.path.basename(name)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
225
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
226
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
227 def extract_zipped_paths(path):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
228 """Replace nonexistent paths that look like they refer to a member of a zip
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
229 archive with the location of an extracted copy of the target, or else
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
230 just return the provided path unchanged.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
231 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
232 if os.path.exists(path):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
233 # this is already a valid path, no need to do anything further
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
234 return path
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
235
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
236 # find the first valid part of the provided path and treat that as a zip archive
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
237 # assume the rest of the path is the name of a member in the archive
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
238 archive, member = os.path.split(path)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
239 while archive and not os.path.exists(archive):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
240 archive, prefix = os.path.split(archive)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
241 member = '/'.join([prefix, member])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
242
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
243 if not zipfile.is_zipfile(archive):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
244 return path
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
245
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
246 zip_file = zipfile.ZipFile(archive)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
247 if member not in zip_file.namelist():
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
248 return path
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
249
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
250 # we have a valid zip archive and a valid member of that archive
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
251 tmp = tempfile.gettempdir()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
252 extracted_path = os.path.join(tmp, *member.split('/'))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
253 if not os.path.exists(extracted_path):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
254 extracted_path = zip_file.extract(member, path=tmp)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
255
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
256 return extracted_path
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
257
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
258
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
259 def from_key_val_list(value):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
260 """Take an object and test to see if it can be represented as a
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
261 dictionary. Unless it can not be represented as such, return an
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
262 OrderedDict, e.g.,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
263
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
264 ::
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
265
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
266 >>> from_key_val_list([('key', 'val')])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
267 OrderedDict([('key', 'val')])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
268 >>> from_key_val_list('string')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
269 ValueError: cannot encode objects that are not 2-tuples
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
270 >>> from_key_val_list({'key': 'val'})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
271 OrderedDict([('key', 'val')])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
272
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
273 :rtype: OrderedDict
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
274 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
275 if value is None:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
276 return None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
277
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
278 if isinstance(value, (str, bytes, bool, int)):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
279 raise ValueError('cannot encode objects that are not 2-tuples')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
280
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
281 return OrderedDict(value)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
282
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
283
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
284 def to_key_val_list(value):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
285 """Take an object and test to see if it can be represented as a
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
286 dictionary. If it can be, return a list of tuples, e.g.,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
287
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
288 ::
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
289
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
290 >>> to_key_val_list([('key', 'val')])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
291 [('key', 'val')]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
292 >>> to_key_val_list({'key': 'val'})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
293 [('key', 'val')]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
294 >>> to_key_val_list('string')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
295 ValueError: cannot encode objects that are not 2-tuples.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
296
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
297 :rtype: list
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
298 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
299 if value is None:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
300 return None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
301
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
302 if isinstance(value, (str, bytes, bool, int)):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
303 raise ValueError('cannot encode objects that are not 2-tuples')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
304
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
305 if isinstance(value, Mapping):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
306 value = value.items()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
307
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
308 return list(value)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
309
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
310
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
311 # From mitsuhiko/werkzeug (used with permission).
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
312 def parse_list_header(value):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
313 """Parse lists as described by RFC 2068 Section 2.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
314
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
315 In particular, parse comma-separated lists where the elements of
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
316 the list may include quoted-strings. A quoted-string could
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
317 contain a comma. A non-quoted string could have quotes in the
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
318 middle. Quotes are removed automatically after parsing.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
319
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
320 It basically works like :func:`parse_set_header` just that items
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
321 may appear multiple times and case sensitivity is preserved.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
322
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
323 The return value is a standard :class:`list`:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
324
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
325 >>> parse_list_header('token, "quoted value"')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
326 ['token', 'quoted value']
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
327
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
328 To create a header from the :class:`list` again, use the
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
329 :func:`dump_header` function.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
330
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
331 :param value: a string with a list header.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
332 :return: :class:`list`
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
333 :rtype: list
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
334 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
335 result = []
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
336 for item in _parse_list_header(value):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
337 if item[:1] == item[-1:] == '"':
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
338 item = unquote_header_value(item[1:-1])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
339 result.append(item)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
340 return result
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
341
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
342
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
343 # From mitsuhiko/werkzeug (used with permission).
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
344 def parse_dict_header(value):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
345 """Parse lists of key, value pairs as described by RFC 2068 Section 2 and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
346 convert them into a python dict:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
347
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
348 >>> d = parse_dict_header('foo="is a fish", bar="as well"')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
349 >>> type(d) is dict
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
350 True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
351 >>> sorted(d.items())
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
352 [('bar', 'as well'), ('foo', 'is a fish')]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
353
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
354 If there is no value for a key it will be `None`:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
355
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
356 >>> parse_dict_header('key_without_value')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
357 {'key_without_value': None}
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
358
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
359 To create a header from the :class:`dict` again, use the
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
360 :func:`dump_header` function.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
361
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
362 :param value: a string with a dict header.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
363 :return: :class:`dict`
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
364 :rtype: dict
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
365 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
366 result = {}
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
367 for item in _parse_list_header(value):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
368 if '=' not in item:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
369 result[item] = None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
370 continue
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
371 name, value = item.split('=', 1)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
372 if value[:1] == value[-1:] == '"':
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
373 value = unquote_header_value(value[1:-1])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
374 result[name] = value
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
375 return result
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
376
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
377
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
378 # From mitsuhiko/werkzeug (used with permission).
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
379 def unquote_header_value(value, is_filename=False):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
380 r"""Unquotes a header value. (Reversal of :func:`quote_header_value`).
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
381 This does not use the real unquoting but what browsers are actually
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
382 using for quoting.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
383
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
384 :param value: the header value to unquote.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
385 :rtype: str
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
386 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
387 if value and value[0] == value[-1] == '"':
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
388 # this is not the real unquoting, but fixing this so that the
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
389 # RFC is met will result in bugs with internet explorer and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
390 # probably some other browsers as well. IE for example is
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
391 # uploading files with "C:\foo\bar.txt" as filename
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
392 value = value[1:-1]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
393
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
394 # if this is a filename and the starting characters look like
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
395 # a UNC path, then just return the value without quotes. Using the
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
396 # replace sequence below on a UNC path has the effect of turning
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
397 # the leading double slash into a single slash and then
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
398 # _fix_ie_filename() doesn't work correctly. See #458.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
399 if not is_filename or value[:2] != '\\\\':
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
400 return value.replace('\\\\', '\\').replace('\\"', '"')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
401 return value
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
402
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
403
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
404 def dict_from_cookiejar(cj):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
405 """Returns a key/value dictionary from a CookieJar.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
406
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
407 :param cj: CookieJar object to extract cookies from.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
408 :rtype: dict
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
409 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
410
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
411 cookie_dict = {}
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
412
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
413 for cookie in cj:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
414 cookie_dict[cookie.name] = cookie.value
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
415
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
416 return cookie_dict
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
417
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
418
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
419 def add_dict_to_cookiejar(cj, cookie_dict):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
420 """Returns a CookieJar from a key/value dictionary.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
421
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
422 :param cj: CookieJar to insert cookies into.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
423 :param cookie_dict: Dict of key/values to insert into CookieJar.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
424 :rtype: CookieJar
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
425 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
426
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
427 return cookiejar_from_dict(cookie_dict, cj)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
428
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
429
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
430 def get_encodings_from_content(content):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
431 """Returns encodings from given content string.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
432
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
433 :param content: bytestring to extract encodings from.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
434 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
435 warnings.warn((
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
436 'In requests 3.0, get_encodings_from_content will be removed. For '
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
437 'more information, please see the discussion on issue #2266. (This'
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
438 ' warning should only appear once.)'),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
439 DeprecationWarning)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
440
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
441 charset_re = re.compile(r'<meta.*?charset=["\']*(.+?)["\'>]', flags=re.I)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
442 pragma_re = re.compile(r'<meta.*?content=["\']*;?charset=(.+?)["\'>]', flags=re.I)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
443 xml_re = re.compile(r'^<\?xml.*?encoding=["\']*(.+?)["\'>]')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
444
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
445 return (charset_re.findall(content) +
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
446 pragma_re.findall(content) +
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
447 xml_re.findall(content))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
448
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
449
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
450 def _parse_content_type_header(header):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
451 """Returns content type and parameters from given header
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
452
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
453 :param header: string
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
454 :return: tuple containing content type and dictionary of
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
455 parameters
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
456 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
457
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
458 tokens = header.split(';')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
459 content_type, params = tokens[0].strip(), tokens[1:]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
460 params_dict = {}
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
461 items_to_strip = "\"' "
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
462
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
463 for param in params:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
464 param = param.strip()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
465 if param:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
466 key, value = param, True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
467 index_of_equals = param.find("=")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
468 if index_of_equals != -1:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
469 key = param[:index_of_equals].strip(items_to_strip)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
470 value = param[index_of_equals + 1:].strip(items_to_strip)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
471 params_dict[key.lower()] = value
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
472 return content_type, params_dict
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
473
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
474
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
475 def get_encoding_from_headers(headers):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
476 """Returns encodings from given HTTP Header Dict.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
477
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
478 :param headers: dictionary to extract encoding from.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
479 :rtype: str
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
480 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
481
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
482 content_type = headers.get('content-type')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
483
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
484 if not content_type:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
485 return None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
486
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
487 content_type, params = _parse_content_type_header(content_type)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
488
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
489 if 'charset' in params:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
490 return params['charset'].strip("'\"")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
491
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
492 if 'text' in content_type:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
493 return 'ISO-8859-1'
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
494
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
495
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
496 def stream_decode_response_unicode(iterator, r):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
497 """Stream decodes a iterator."""
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
498
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
499 if r.encoding is None:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
500 for item in iterator:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
501 yield item
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
502 return
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
503
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
504 decoder = codecs.getincrementaldecoder(r.encoding)(errors='replace')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
505 for chunk in iterator:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
506 rv = decoder.decode(chunk)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
507 if rv:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
508 yield rv
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
509 rv = decoder.decode(b'', final=True)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
510 if rv:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
511 yield rv
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
512
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
513
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
514 def iter_slices(string, slice_length):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
515 """Iterate over slices of a string."""
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
516 pos = 0
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
517 if slice_length is None or slice_length <= 0:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
518 slice_length = len(string)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
519 while pos < len(string):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
520 yield string[pos:pos + slice_length]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
521 pos += slice_length
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
522
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
523
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
524 def get_unicode_from_response(r):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
525 """Returns the requested content back in unicode.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
526
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
527 :param r: Response object to get unicode content from.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
528
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
529 Tried:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
530
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
531 1. charset from content-type
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
532 2. fall back and replace all unicode characters
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
533
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
534 :rtype: str
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
535 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
536 warnings.warn((
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
537 'In requests 3.0, get_unicode_from_response will be removed. For '
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
538 'more information, please see the discussion on issue #2266. (This'
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
539 ' warning should only appear once.)'),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
540 DeprecationWarning)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
541
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
542 tried_encodings = []
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
543
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
544 # Try charset from content-type
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
545 encoding = get_encoding_from_headers(r.headers)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
546
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
547 if encoding:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
548 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
549 return str(r.content, encoding)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
550 except UnicodeError:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
551 tried_encodings.append(encoding)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
552
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
553 # Fall back:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
554 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
555 return str(r.content, encoding, errors='replace')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
556 except TypeError:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
557 return r.content
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
558
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
559
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
560 # The unreserved URI characters (RFC 3986)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
561 UNRESERVED_SET = frozenset(
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
562 "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz" + "0123456789-._~")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
563
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
564
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
565 def unquote_unreserved(uri):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
566 """Un-escape any percent-escape sequences in a URI that are unreserved
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
567 characters. This leaves all reserved, illegal and non-ASCII bytes encoded.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
568
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
569 :rtype: str
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
570 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
571 parts = uri.split('%')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
572 for i in range(1, len(parts)):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
573 h = parts[i][0:2]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
574 if len(h) == 2 and h.isalnum():
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
575 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
576 c = chr(int(h, 16))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
577 except ValueError:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
578 raise InvalidURL("Invalid percent-escape sequence: '%s'" % h)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
579
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
580 if c in UNRESERVED_SET:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
581 parts[i] = c + parts[i][2:]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
582 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
583 parts[i] = '%' + parts[i]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
584 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
585 parts[i] = '%' + parts[i]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
586 return ''.join(parts)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
587
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
588
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
589 def requote_uri(uri):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
590 """Re-quote the given URI.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
591
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
592 This function passes the given URI through an unquote/quote cycle to
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
593 ensure that it is fully and consistently quoted.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
594
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
595 :rtype: str
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
596 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
597 safe_with_percent = "!#$%&'()*+,/:;=?@[]~"
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
598 safe_without_percent = "!#$&'()*+,/:;=?@[]~"
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
599 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
600 # Unquote only the unreserved characters
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
601 # Then quote only illegal characters (do not quote reserved,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
602 # unreserved, or '%')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
603 return quote(unquote_unreserved(uri), safe=safe_with_percent)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
604 except InvalidURL:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
605 # We couldn't unquote the given URI, so let's try quoting it, but
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
606 # there may be unquoted '%'s in the URI. We need to make sure they're
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
607 # properly quoted so they do not cause issues elsewhere.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
608 return quote(uri, safe=safe_without_percent)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
609
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
610
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
611 def address_in_network(ip, net):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
612 """This function allows you to check if an IP belongs to a network subnet
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
613
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
614 Example: returns True if ip = 192.168.1.1 and net = 192.168.1.0/24
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
615 returns False if ip = 192.168.1.1 and net = 192.168.100.0/24
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
616
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
617 :rtype: bool
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
618 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
619 ipaddr = struct.unpack('=L', socket.inet_aton(ip))[0]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
620 netaddr, bits = net.split('/')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
621 netmask = struct.unpack('=L', socket.inet_aton(dotted_netmask(int(bits))))[0]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
622 network = struct.unpack('=L', socket.inet_aton(netaddr))[0] & netmask
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
623 return (ipaddr & netmask) == (network & netmask)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
624
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
625
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
626 def dotted_netmask(mask):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
627 """Converts mask from /xx format to xxx.xxx.xxx.xxx
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
628
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
629 Example: if mask is 24 function returns 255.255.255.0
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
630
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
631 :rtype: str
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
632 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
633 bits = 0xffffffff ^ (1 << 32 - mask) - 1
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
634 return socket.inet_ntoa(struct.pack('>I', bits))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
635
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
636
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
637 def is_ipv4_address(string_ip):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
638 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
639 :rtype: bool
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
640 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
641 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
642 socket.inet_aton(string_ip)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
643 except socket.error:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
644 return False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
645 return True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
646
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
647
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
648 def is_valid_cidr(string_network):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
649 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
650 Very simple check of the cidr format in no_proxy variable.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
651
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
652 :rtype: bool
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
653 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
654 if string_network.count('/') == 1:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
655 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
656 mask = int(string_network.split('/')[1])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
657 except ValueError:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
658 return False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
659
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
660 if mask < 1 or mask > 32:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
661 return False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
662
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
663 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
664 socket.inet_aton(string_network.split('/')[0])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
665 except socket.error:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
666 return False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
667 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
668 return False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
669 return True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
670
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
671
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
672 @contextlib.contextmanager
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
673 def set_environ(env_name, value):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
674 """Set the environment variable 'env_name' to 'value'
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
675
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
676 Save previous value, yield, and then restore the previous value stored in
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
677 the environment variable 'env_name'.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
678
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
679 If 'value' is None, do nothing"""
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
680 value_changed = value is not None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
681 if value_changed:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
682 old_value = os.environ.get(env_name)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
683 os.environ[env_name] = value
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
684 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
685 yield
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
686 finally:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
687 if value_changed:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
688 if old_value is None:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
689 del os.environ[env_name]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
690 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
691 os.environ[env_name] = old_value
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
692
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
693
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
694 def should_bypass_proxies(url, no_proxy):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
695 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
696 Returns whether we should bypass proxies or not.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
697
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
698 :rtype: bool
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
699 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
700 # Prioritize lowercase environment variables over uppercase
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
701 # to keep a consistent behaviour with other http projects (curl, wget).
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
702 get_proxy = lambda k: os.environ.get(k) or os.environ.get(k.upper())
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
703
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
704 # First check whether no_proxy is defined. If it is, check that the URL
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
705 # we're getting isn't in the no_proxy list.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
706 no_proxy_arg = no_proxy
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
707 if no_proxy is None:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
708 no_proxy = get_proxy('no_proxy')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
709 parsed = urlparse(url)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
710
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
711 if parsed.hostname is None:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
712 # URLs don't always have hostnames, e.g. file:/// urls.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
713 return True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
714
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
715 if no_proxy:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
716 # We need to check whether we match here. We need to see if we match
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
717 # the end of the hostname, both with and without the port.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
718 no_proxy = (
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
719 host for host in no_proxy.replace(' ', '').split(',') if host
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
720 )
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
721
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
722 if is_ipv4_address(parsed.hostname):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
723 for proxy_ip in no_proxy:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
724 if is_valid_cidr(proxy_ip):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
725 if address_in_network(parsed.hostname, proxy_ip):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
726 return True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
727 elif parsed.hostname == proxy_ip:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
728 # If no_proxy ip was defined in plain IP notation instead of cidr notation &
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
729 # matches the IP of the index
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
730 return True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
731 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
732 host_with_port = parsed.hostname
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
733 if parsed.port:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
734 host_with_port += ':{}'.format(parsed.port)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
735
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
736 for host in no_proxy:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
737 if parsed.hostname.endswith(host) or host_with_port.endswith(host):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
738 # The URL does match something in no_proxy, so we don't want
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
739 # to apply the proxies on this URL.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
740 return True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
741
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
742 with set_environ('no_proxy', no_proxy_arg):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
743 # parsed.hostname can be `None` in cases such as a file URI.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
744 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
745 bypass = proxy_bypass(parsed.hostname)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
746 except (TypeError, socket.gaierror):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
747 bypass = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
748
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
749 if bypass:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
750 return True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
751
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
752 return False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
753
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
754
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
755 def get_environ_proxies(url, no_proxy=None):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
756 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
757 Return a dict of environment proxies.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
758
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
759 :rtype: dict
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
760 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
761 if should_bypass_proxies(url, no_proxy=no_proxy):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
762 return {}
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
763 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
764 return getproxies()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
765
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
766
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
767 def select_proxy(url, proxies):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
768 """Select a proxy for the url, if applicable.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
769
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
770 :param url: The url being for the request
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
771 :param proxies: A dictionary of schemes or schemes and hosts to proxy URLs
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
772 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
773 proxies = proxies or {}
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
774 urlparts = urlparse(url)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
775 if urlparts.hostname is None:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
776 return proxies.get(urlparts.scheme, proxies.get('all'))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
777
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
778 proxy_keys = [
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
779 urlparts.scheme + '://' + urlparts.hostname,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
780 urlparts.scheme,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
781 'all://' + urlparts.hostname,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
782 'all',
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
783 ]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
784 proxy = None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
785 for proxy_key in proxy_keys:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
786 if proxy_key in proxies:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
787 proxy = proxies[proxy_key]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
788 break
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
789
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
790 return proxy
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
791
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
792
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
793 def default_user_agent(name="python-requests"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
794 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
795 Return a string representing the default user agent.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
796
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
797 :rtype: str
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
798 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
799 return '%s/%s' % (name, __version__)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
800
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
801
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
802 def default_headers():
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
803 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
804 :rtype: requests.structures.CaseInsensitiveDict
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
805 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
806 return CaseInsensitiveDict({
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
807 'User-Agent': default_user_agent(),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
808 'Accept-Encoding': ', '.join(('gzip', 'deflate')),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
809 'Accept': '*/*',
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
810 'Connection': 'keep-alive',
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
811 })
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
812
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
813
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
814 def parse_header_links(value):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
815 """Return a list of parsed link headers proxies.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
816
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
817 i.e. Link: <http:/.../front.jpeg>; rel=front; type="image/jpeg",<http://.../back.jpeg>; rel=back;type="image/jpeg"
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
818
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
819 :rtype: list
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
820 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
821
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
822 links = []
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
823
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
824 replace_chars = ' \'"'
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
825
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
826 value = value.strip(replace_chars)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
827 if not value:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
828 return links
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
829
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
830 for val in re.split(', *<', value):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
831 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
832 url, params = val.split(';', 1)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
833 except ValueError:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
834 url, params = val, ''
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
835
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
836 link = {'url': url.strip('<> \'"')}
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
837
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
838 for param in params.split(';'):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
839 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
840 key, value = param.split('=')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
841 except ValueError:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
842 break
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
843
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
844 link[key.strip(replace_chars)] = value.strip(replace_chars)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
845
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
846 links.append(link)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
847
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
848 return links
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
849
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
850
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
851 # Null bytes; no need to recreate these on each call to guess_json_utf
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
852 _null = '\x00'.encode('ascii') # encoding to ASCII for Python 3
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
853 _null2 = _null * 2
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
854 _null3 = _null * 3
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
855
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
856
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
857 def guess_json_utf(data):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
858 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
859 :rtype: str
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
860 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
861 # JSON always starts with two ASCII characters, so detection is as
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
862 # easy as counting the nulls and from their location and count
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
863 # determine the encoding. Also detect a BOM, if present.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
864 sample = data[:4]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
865 if sample in (codecs.BOM_UTF32_LE, codecs.BOM_UTF32_BE):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
866 return 'utf-32' # BOM included
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
867 if sample[:3] == codecs.BOM_UTF8:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
868 return 'utf-8-sig' # BOM included, MS style (discouraged)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
869 if sample[:2] in (codecs.BOM_UTF16_LE, codecs.BOM_UTF16_BE):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
870 return 'utf-16' # BOM included
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
871 nullcount = sample.count(_null)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
872 if nullcount == 0:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
873 return 'utf-8'
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
874 if nullcount == 2:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
875 if sample[::2] == _null2: # 1st and 3rd are null
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
876 return 'utf-16-be'
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
877 if sample[1::2] == _null2: # 2nd and 4th are null
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
878 return 'utf-16-le'
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
879 # Did not detect 2 valid UTF-16 ascii-range characters
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
880 if nullcount == 3:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
881 if sample[:3] == _null3:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
882 return 'utf-32-be'
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
883 if sample[1:] == _null3:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
884 return 'utf-32-le'
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
885 # Did not detect a valid UTF-32 ascii-range character
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
886 return None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
887
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
888
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
889 def prepend_scheme_if_needed(url, new_scheme):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
890 """Given a URL that may or may not have a scheme, prepend the given scheme.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
891 Does not replace a present scheme with the one provided as an argument.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
892
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
893 :rtype: str
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
894 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
895 scheme, netloc, path, params, query, fragment = urlparse(url, new_scheme)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
896
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
897 # urlparse is a finicky beast, and sometimes decides that there isn't a
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
898 # netloc present. Assume that it's being over-cautious, and switch netloc
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
899 # and path if urlparse decided there was no netloc.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
900 if not netloc:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
901 netloc, path = path, netloc
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
902
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
903 return urlunparse((scheme, netloc, path, params, query, fragment))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
904
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
905
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
906 def get_auth_from_url(url):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
907 """Given a url with authentication components, extract them into a tuple of
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
908 username,password.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
909
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
910 :rtype: (str,str)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
911 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
912 parsed = urlparse(url)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
913
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
914 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
915 auth = (unquote(parsed.username), unquote(parsed.password))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
916 except (AttributeError, TypeError):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
917 auth = ('', '')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
918
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
919 return auth
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
920
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
921
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
922 # Moved outside of function to avoid recompile every call
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
923 _CLEAN_HEADER_REGEX_BYTE = re.compile(b'^\\S[^\\r\\n]*$|^$')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
924 _CLEAN_HEADER_REGEX_STR = re.compile(r'^\S[^\r\n]*$|^$')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
925
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
926
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
927 def check_header_validity(header):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
928 """Verifies that header value is a string which doesn't contain
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
929 leading whitespace or return characters. This prevents unintended
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
930 header injection.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
931
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
932 :param header: tuple, in the format (name, value).
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
933 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
934 name, value = header
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
935
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
936 if isinstance(value, bytes):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
937 pat = _CLEAN_HEADER_REGEX_BYTE
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
938 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
939 pat = _CLEAN_HEADER_REGEX_STR
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
940 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
941 if not pat.match(value):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
942 raise InvalidHeader("Invalid return character or leading space in header: %s" % name)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
943 except TypeError:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
944 raise InvalidHeader("Value for header {%s: %s} must be of type str or "
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
945 "bytes, not %s" % (name, value, type(value)))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
946
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
947
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
948 def urldefragauth(url):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
949 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
950 Given a url remove the fragment and the authentication part.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
951
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
952 :rtype: str
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
953 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
954 scheme, netloc, path, params, query, fragment = urlparse(url)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
955
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
956 # see func:`prepend_scheme_if_needed`
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
957 if not netloc:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
958 netloc, path = path, netloc
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
959
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
960 netloc = netloc.rsplit('@', 1)[-1]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
961
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
962 return urlunparse((scheme, netloc, path, params, query, ''))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
963
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
964
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
965 def rewind_body(prepared_request):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
966 """Move file pointer back to its recorded starting position
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
967 so it can be read again on redirect.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
968 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
969 body_seek = getattr(prepared_request.body, 'seek', None)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
970 if body_seek is not None and isinstance(prepared_request._body_position, integer_types):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
971 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
972 body_seek(prepared_request._body_position)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
973 except (IOError, OSError):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
974 raise UnrewindableBodyError("An error occurred when rewinding request "
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
975 "body for redirect.")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
976 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
977 raise UnrewindableBodyError("Unable to rewind request body for redirect.")