17
17
"""A collection of function for handling URL operations."""
19
from __future__ import absolute_import
23
from bzrlib.lazy_import import lazy_import
24
lazy_import(globals(), """
25
from posixpath import split as _posix_split, normpath as _posix_normpath
28
from urllib import parse as urlparse
35
from .lazy_import import lazy_import
36
lazy_import(globals(), """
37
from posixpath import split as _posix_split
46
class InvalidURL(errors.PathError):
48
_fmt = 'Invalid url supplied to transport: "%(path)s"%(extra)s'
51
class InvalidURLJoin(errors.PathError):
53
_fmt = "Invalid URL join request: %(reason)s: %(base)r + %(join_args)r"
55
def __init__(self, reason, base, join_args):
58
self.join_args = join_args
59
errors.PathError.__init__(self, base, reason)
62
class InvalidRebaseURLs(errors.PathError):
64
_fmt = "URLs differ by more than path: %(from_)r and %(to)r"
66
def __init__(self, from_, to):
69
errors.PathError.__init__(self, from_, 'URLs differ by more than path.')
36
72
def basename(url, exclude_trailing_slash=True):
37
73
"""Return the last component of a URL.
60
96
return split(url, exclude_trailing_slash=exclude_trailing_slash)[0]
99
# Private copies of quote and unquote, copied from Python's
100
# urllib module because urllib unconditionally imports socket, which imports
103
always_safe = ('ABCDEFGHIJKLMNOPQRSTUVWXYZ'
104
'abcdefghijklmnopqrstuvwxyz'
107
for i, c in zip(range(256), ''.join(map(chr, range(256)))):
108
_safe_map[c] = c if (i < 128 and c in always_safe) else '%{0:02X}'.format(i)
112
def quote(s, safe='/'):
113
"""quote('abc def') -> 'abc%20def'
115
Each part of a URL, e.g. the path info, the query, etc., has a
116
different set of reserved characters that must be quoted.
118
RFC 2396 Uniform Resource Identifiers (URI): Generic Syntax lists
119
the following reserved characters.
121
reserved = ";" | "/" | "?" | ":" | "@" | "&" | "=" | "+" |
124
Each of these characters is reserved in some component of a URL,
125
but not necessarily in all of them.
127
By default, the quote function is intended for quoting the path
128
section of a URL. Thus, it will not encode '/'. This character
129
is reserved, but in typical usage the quote function is being
130
called on a path where the existing slash characters are used as
136
raise TypeError('None object cannot be quoted')
138
cachekey = (safe, always_safe)
140
(quoter, safe) = _safe_quoters[cachekey]
142
safe_map = _safe_map.copy()
143
safe_map.update([(c, c) for c in safe])
144
quoter = safe_map.__getitem__
145
safe = always_safe + safe
146
_safe_quoters[cachekey] = (quoter, safe)
147
if not s.rstrip(safe):
149
return ''.join(map(quoter, s))
152
unquote = urlparse.unquote
63
155
def escape(relpath):
64
156
"""Escape relpath to be a valid url."""
65
if isinstance(relpath, unicode):
157
if not isinstance(relpath, str) and sys.version_info[0] == 2:
66
158
relpath = relpath.encode('utf-8')
67
# After quoting and encoding, the path should be perfectly
68
# safe as a plain ASCII string, str() just enforces this
69
return str(urllib.quote(relpath, safe='/~'))
159
return quote(relpath, safe='/~')
72
162
def file_relpath(base, path):
121
match = _url_scheme_re.match(base)
124
scheme = match.group('scheme')
125
path = match.group('path').split('/')
126
if path[-1:] == ['']:
127
# Strip off a trailing slash
128
# This helps both when we are at the root, and when
129
# 'base' has an extra slash at the end
132
path = base.split('/')
134
if scheme is not None and len(path) >= 1:
136
# the path should be represented as an abs path.
137
# we know this must be absolute because of the presence of a URL scheme.
139
path = [''] + path[1:]
141
# create an empty host, but dont alter the path - this might be a
142
# relative url fragment.
216
scheme_end, path_start = _find_scheme_and_separator(base)
217
if scheme_end is None and path_start is None:
219
elif path_start is None:
220
path_start = len(base)
221
path = base[path_start:]
147
match = _url_scheme_re.match(arg)
150
scheme = match.group('scheme')
151
# this skips .. normalisation, making http://host/../../..
153
path = match.group('path').split('/')
154
# set the host and path according to new absolute URL, discarding
155
# any previous values.
156
# XXX: duplicates mess from earlier in this function. This URL
157
# manipulation code needs some cleaning up.
158
if scheme is not None and len(path) >= 1:
161
# url scheme implies absolute path.
164
# no url scheme we take the path as is.
223
arg_scheme_end, arg_path_start = _find_scheme_and_separator(arg)
224
if arg_scheme_end is None and arg_path_start is None:
226
elif arg_path_start is None:
227
arg_path_start = len(arg)
228
if arg_scheme_end is not None:
230
path = arg[arg_path_start:]
231
scheme_end = arg_scheme_end
232
path_start = arg_path_start
167
path = '/'.join(path)
168
234
path = joinpath(path, arg)
169
path = path.split('/')
170
if remove_root and path[0:1] == ['']:
173
# Remove the leading slash from the path, so long as it isn't also the
174
# trailing slash, which we want to keep if present.
175
if path and path[0] == '' and len(path) > 1:
180
return '/'.join(path)
181
return scheme + '://' + '/'.join(path)
235
return base[:path_start] + path
184
238
def joinpath(base, *args):
237
292
# importing directly from posixpath allows us to test this
238
293
# on non-posix platforms
239
return 'file://' + escape(_posix_normpath(
240
osutils._posix_abspath(path)))
294
return 'file://' + escape(osutils._posix_abspath(path))
243
297
def _win32_local_path_from_url(url):
244
298
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
245
299
if not url.startswith('file://'):
246
raise errors.InvalidURL(url, 'local urls must start with file:///, '
300
raise InvalidURL(url, 'local urls must start with file:///, '
247
301
'UNC path urls must start with file://')
302
url = split_segment_parameters_raw(url)[0]
248
303
# We strip off all 3 slashes
249
304
win32_url = url[len('file:'):]
250
305
# check for UNC path: //HOST/path
251
306
if not win32_url.startswith('///'):
252
307
if (win32_url[2] == '/'
253
308
or win32_url[3] in '|:'):
254
raise errors.InvalidURL(url, 'Win32 UNC path urls'
309
raise InvalidURL(url, 'Win32 UNC path urls'
255
310
' have form file://HOST/path')
256
311
return unescape(win32_url)
339
395
:param url: Either a hybrid URL or a local path
340
396
:return: A normalized URL which only includes 7-bit ASCII characters.
342
m = _url_scheme_re.match(url)
398
scheme_end, path_start = _find_scheme_and_separator(url)
399
if scheme_end is None:
344
400
return local_path_to_url(url)
345
scheme = m.group('scheme')
346
path = m.group('path')
347
if not isinstance(url, unicode):
401
prefix = url[:path_start]
402
path = url[path_start:]
403
if not isinstance(url, text_type):
349
405
if c not in _url_safe_characters:
350
raise errors.InvalidURL(url, 'URLs can only contain specific'
406
raise InvalidURL(url, 'URLs can only contain specific'
351
407
' safe characters (not %r)' % c)
352
408
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
353
return str(scheme + '://' + ''.join(path))
409
return str(prefix + ''.join(path))
355
411
# We have a unicode (hybrid) url
356
412
path_chars = list(path)
358
for i in xrange(len(path_chars)):
414
for i in range(len(path_chars)):
359
415
if path_chars[i] not in _url_safe_characters:
360
416
chars = path_chars[i].encode('utf-8')
361
417
path_chars[i] = ''.join(
362
418
['%%%02X' % ord(c) for c in path_chars[i].encode('utf-8')])
363
419
path = ''.join(path_chars)
364
420
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
365
return str(scheme + '://' + path)
421
return str(prefix + path)
368
424
def relative_url(base, other):
469
525
return url_base + head, tail
528
def split_segment_parameters_raw(url):
529
"""Split the subsegment of the last segment of a URL.
531
:param url: A relative or absolute URL
532
:return: (url, subsegments)
534
# GZ 2011-11-18: Dodgy removing the terminal slash like this, function
535
# operates on urls not url+segments, and Transport classes
536
# should not be blindly adding slashes in the first place.
537
lurl = strip_trailing_slash(url)
538
# Segments begin at first comma after last forward slash, if one exists
539
segment_start = lurl.find(",", lurl.rfind("/")+1)
540
if segment_start == -1:
542
return (lurl[:segment_start], [str(s) for s in lurl[segment_start+1:].split(",")])
545
def split_segment_parameters(url):
546
"""Split the segment parameters of the last segment of a URL.
548
:param url: A relative or absolute URL
549
:return: (url, segment_parameters)
551
(base_url, subsegments) = split_segment_parameters_raw(url)
553
for subsegment in subsegments:
554
(key, value) = subsegment.split("=", 1)
555
if not isinstance(key, str):
557
if not isinstance(value, str):
558
raise TypeError(value)
559
parameters[key] = value
560
return (base_url, parameters)
563
def join_segment_parameters_raw(base, *subsegments):
564
"""Create a new URL by adding subsegments to an existing one.
566
This adds the specified subsegments to the last path in the specified
567
base URL. The subsegments should be bytestrings.
569
:note: You probably want to use join_segment_parameters instead.
573
for subsegment in subsegments:
574
if not isinstance(subsegment, str):
575
raise TypeError("Subsegment %r is not a bytestring" % subsegment)
576
if "," in subsegment:
577
raise InvalidURLJoin(", exists in subsegments",
579
return ",".join((base,) + subsegments)
582
def join_segment_parameters(url, parameters):
583
"""Create a new URL by adding segment parameters to an existing one.
585
The parameters of the last segment in the URL will be updated; if a
586
parameter with the same key already exists it will be overwritten.
588
:param url: A URL, as string
589
:param parameters: Dictionary of parameters, keys and values as bytestrings
591
(base, existing_parameters) = split_segment_parameters(url)
593
new_parameters.update(existing_parameters)
594
for key, value in parameters.items():
595
if not isinstance(key, str):
596
raise TypeError("parameter key %r is not a str" % key)
597
if not isinstance(value, str):
598
raise TypeError("parameter value %r for %s is not a str" %
601
raise InvalidURLJoin("= exists in parameter key", url,
603
new_parameters[key] = value
604
return join_segment_parameters_raw(base,
605
*["%s=%s" % item for item in sorted(new_parameters.items())])
472
608
def _win32_strip_local_trailing_slash(url):
473
609
"""Strip slashes after the drive letter"""
474
610
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
524
660
This returns a Unicode path from a URL
526
662
# jam 20060427 URLs are supposed to be ASCII only strings
527
# If they are passed in as unicode, urllib.unquote
663
# If they are passed in as unicode, unquote
528
664
# will return a UNICODE string, which actually contains
529
665
# utf-8 bytes. So we have to ensure that they are
530
666
# plain ASCII strings, or the final .decode will
531
667
# try to encode the UNICODE => ASCII, and then decode
535
except UnicodeError, e:
536
raise errors.InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
538
unquoted = urllib.unquote(url)
669
if isinstance(url, text_type):
671
url = url.encode("ascii")
672
except UnicodeError as e:
673
raise InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
675
unquoted = urlparse.unquote_to_bytes(url)
677
unquoted = unquote(url)
540
679
unicode_path = unquoted.decode('utf-8')
541
except UnicodeError, e:
542
raise errors.InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
680
except UnicodeError as e:
681
raise InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
543
682
return unicode_path
611
750
except UnicodeDecodeError:
612
751
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
613
752
unescaped = ''.join(escaped_chunks)
615
decoded = unescaped.decode('utf-8')
616
except UnicodeDecodeError:
617
# If this path segment cannot be properly utf-8 decoded
618
# after doing unescaping we will just leave it alone
753
if sys.version_info[0] == 2:
622
decoded.encode(encoding)
623
except UnicodeEncodeError:
624
# If this chunk cannot be encoded in the local
625
# encoding, then we should leave it alone
755
decoded = unescaped.decode('utf-8')
756
except UnicodeDecodeError:
757
# If this path segment cannot be properly utf-8 decoded
758
# after doing unescaping we will just leave it alone
628
# Otherwise take the url decoded one
762
decoded.encode(encoding)
763
except UnicodeEncodeError:
764
# If this chunk cannot be encoded in the local
765
# encoding, then we should leave it alone
768
# Otherwise take the url decoded one
630
772
return u'/'.join(res)
693
835
return osutils.pathjoin(*segments)
841
def __init__(self, scheme, quoted_user, quoted_password, quoted_host,
844
self.quoted_host = quoted_host
845
self.host = unquote(self.quoted_host)
846
self.quoted_user = quoted_user
847
if self.quoted_user is not None:
848
self.user = unquote(self.quoted_user)
851
self.quoted_password = quoted_password
852
if self.quoted_password is not None:
853
self.password = unquote(self.quoted_password)
857
self.quoted_path = _url_hex_escapes_re.sub(_unescape_safe_chars, quoted_path)
858
self.path = unquote(self.quoted_path)
860
def __eq__(self, other):
861
return (isinstance(other, self.__class__) and
862
self.scheme == other.scheme and
863
self.host == other.host and
864
self.user == other.user and
865
self.password == other.password and
866
self.path == other.path)
869
return "<%s(%r, %r, %r, %r, %r, %r)>" % (
870
self.__class__.__name__,
871
self.scheme, self.quoted_user, self.quoted_password,
872
self.quoted_host, self.port, self.quoted_path)
875
def from_string(cls, url):
876
"""Create a URL object from a string.
878
:param url: URL as bytestring
880
# GZ 2017-06-09: Actually validate ascii-ness
881
# pad.lv/1696545: For the moment, accept both native strings and unicode.
882
if isinstance(url, str):
884
elif isinstance(url, text_type):
887
except UnicodeEncodeError:
888
raise InvalidURL(url)
890
raise InvalidURL(url)
891
(scheme, netloc, path, params,
892
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
893
user = password = host = port = None
895
user, host = netloc.rsplit('@', 1)
897
user, password = user.split(':', 1)
901
if ':' in host and not (host[0] == '[' and host[-1] == ']'):
903
host, port = host.rsplit(':', 1)
907
raise InvalidURL('invalid port number %s in url:\n%s' %
909
if host != "" and host[0] == '[' and host[-1] == ']': #IPv6
912
return cls(scheme, user, password, host, port, path)
915
netloc = self.quoted_host
917
netloc = "[%s]" % netloc
918
if self.quoted_user is not None:
919
# Note that we don't put the password back even if we
920
# have one so that it doesn't get accidentally
922
netloc = '%s@%s' % (self.quoted_user, netloc)
923
if self.port is not None:
924
netloc = '%s:%d' % (netloc, self.port)
925
return urlparse.urlunparse(
926
(self.scheme, netloc, self.quoted_path, None, None, None))
929
def _combine_paths(base_path, relpath):
930
"""Transform a Transport-relative path to a remote absolute path.
932
This does not handle substitution of ~ but does handle '..' and '.'
937
t._combine_paths('/home/sarah', 'project/foo')
938
=> '/home/sarah/project/foo'
939
t._combine_paths('/home/sarah', '../../etc')
941
t._combine_paths('/home/sarah', '/etc')
944
:param base_path: base path
945
:param relpath: relative url string for relative part of remote path.
946
:return: urlencoded string for final path.
948
# pad.lv/1696545: For the moment, accept both native strings and unicode.
949
if isinstance(relpath, str):
951
elif isinstance(relpath, text_type):
953
relpath = relpath.encode()
954
except UnicodeEncodeError:
955
raise InvalidURL(relpath)
957
raise InvalidURL(relpath)
958
relpath = _url_hex_escapes_re.sub(_unescape_safe_chars, relpath)
959
if relpath.startswith('/'):
962
base_parts = base_path.split('/')
963
if len(base_parts) > 0 and base_parts[-1] == '':
964
base_parts = base_parts[:-1]
965
for p in relpath.split('/'):
967
if len(base_parts) == 0:
968
# In most filesystems, a request for the parent
969
# of root, just returns root.
976
path = '/'.join(base_parts)
977
if not path.startswith('/'):
981
def clone(self, offset=None):
982
"""Return a new URL for a path relative to this URL.
984
:param offset: A relative path, already urlencoded
985
:return: `URL` instance
987
if offset is not None:
988
relative = unescape(offset)
989
if sys.version_info[0] == 2:
990
relative = relative.encode('utf-8')
991
path = self._combine_paths(self.path, relative)
992
path = quote(path, safe="/~")
994
path = self.quoted_path
995
return self.__class__(self.scheme, self.quoted_user,
996
self.quoted_password, self.quoted_host, self.port,
697
1000
def parse_url(url):
698
1001
"""Extract the server address, the credentials and the path from the url.
703
1006
:param url: an quoted url
705
1007
:return: (scheme, user, password, host, port, path) tuple, all fields
708
if isinstance(url, unicode):
709
raise errors.InvalidURL('should be ascii:\n%r' % url)
710
url = url.encode('utf-8')
711
(scheme, netloc, path, params,
712
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
713
user = password = host = port = None
715
user, host = netloc.rsplit('@', 1)
717
user, password = user.split(':', 1)
718
password = urllib.unquote(password)
719
user = urllib.unquote(user)
723
if ':' in host and not (host[0] == '[' and host[-1] == ']'): #there *is* port
724
host, port = host.rsplit(':',1)
728
raise errors.InvalidURL('invalid port number %s in url:\n%s' %
730
if host != "" and host[0] == '[' and host[-1] == ']': #IPv6
733
host = urllib.unquote(host)
734
path = urllib.unquote(path)
736
return (scheme, user, password, host, port, path)
1010
parsed_url = URL.from_string(url)
1011
return (parsed_url.scheme, parsed_url.user, parsed_url.password,
1012
parsed_url.host, parsed_url.port, parsed_url.path)