23
from urllib import parse as urlparse
23
from bzrlib.lazy_import import lazy_import
24
lazy_import(globals(), """
25
from posixpath import split as _posix_split, normpath as _posix_normpath
30
from .lazy_import import lazy_import
31
lazy_import(globals(), """
32
from posixpath import split as _posix_split
37
class InvalidURL(errors.PathError):
39
_fmt = 'Invalid url supplied to transport: "%(path)s"%(extra)s'
42
class InvalidURLJoin(errors.PathError):
44
_fmt = "Invalid URL join request: %(reason)s: %(base)r + %(join_args)r"
46
def __init__(self, reason, base, join_args):
49
self.join_args = join_args
50
errors.PathError.__init__(self, base, reason)
53
class InvalidRebaseURLs(errors.PathError):
55
_fmt = "URLs differ by more than path: %(from_)r and %(to)r"
57
def __init__(self, from_, to):
60
errors.PathError.__init__(
61
self, from_, 'URLs differ by more than path.')
64
36
def basename(url, exclude_trailing_slash=True):
65
37
"""Return the last component of a URL.
156
scheme_end, path_start = _find_scheme_and_separator(base)
157
if scheme_end is None and path_start is None:
159
elif path_start is None:
160
path_start = len(base)
161
path = base[path_start:]
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.
163
arg_scheme_end, arg_path_start = _find_scheme_and_separator(arg)
164
if arg_scheme_end is None and arg_path_start is None:
166
elif arg_path_start is None:
167
arg_path_start = len(arg)
168
if arg_scheme_end is not None:
170
path = arg[arg_path_start:]
171
scheme_end = arg_scheme_end
172
path_start = arg_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.
167
path = '/'.join(path)
174
168
path = joinpath(path, arg)
175
return base[:path_start] + path
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)
178
184
def joinpath(base, *args):
232
231
# importing directly from posixpath allows us to test this
233
232
# on non-posix platforms
234
return 'file://' + escape(osutils._posix_abspath(path))
233
return 'file://' + escape(_posix_normpath(
234
osutils._posix_abspath(path)))
237
237
def _win32_local_path_from_url(url):
238
238
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
239
239
if not url.startswith('file://'):
240
raise InvalidURL(url, 'local urls must start with file:///, '
241
'UNC path urls must start with file://')
242
url = strip_segment_parameters(url)
240
raise errors.InvalidURL(url, 'local urls must start with file:///, '
241
'UNC path urls must start with file://')
243
242
# We strip off all 3 slashes
244
243
win32_url = url[len('file:'):]
245
244
# check for UNC path: //HOST/path
246
245
if not win32_url.startswith('///'):
247
246
if (win32_url[2] == '/'
248
or win32_url[3] in '|:'):
249
raise InvalidURL(url, 'Win32 UNC path urls'
250
' have form file://HOST/path')
247
or win32_url[3] in '|:'):
248
raise errors.InvalidURL(url, 'Win32 UNC path urls'
249
' have form file://HOST/path')
251
250
return unescape(win32_url)
253
252
# allow empty paths so we can serve all roots
335
333
:param url: Either a hybrid URL or a local path
336
334
:return: A normalized URL which only includes 7-bit ASCII characters.
338
scheme_end, path_start = _find_scheme_and_separator(url)
339
if scheme_end is None:
336
m = _url_scheme_re.match(url)
340
338
return local_path_to_url(url)
341
prefix = url[:path_start]
342
path = url[path_start:]
343
if not isinstance(url, str):
339
scheme = m.group('scheme')
340
path = m.group('path')
341
if not isinstance(url, unicode):
345
343
if c not in _url_safe_characters:
346
raise InvalidURL(url, 'URLs can only contain specific'
347
' safe characters (not %r)' % c)
344
raise errors.InvalidURL(url, 'URLs can only contain specific'
345
' safe characters (not %r)' % c)
348
346
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
349
return str(prefix + ''.join(path))
347
return str(scheme + '://' + ''.join(path))
351
349
# We have a unicode (hybrid) url
352
350
path_chars = list(path)
354
for i in range(len(path_chars)):
352
for i in xrange(len(path_chars)):
355
353
if path_chars[i] not in _url_safe_characters:
354
chars = path_chars[i].encode('utf-8')
356
355
path_chars[i] = ''.join(
357
['%%%02X' % c for c in bytearray(path_chars[i].encode('utf-8'))])
356
['%%%02X' % ord(c) for c in path_chars[i].encode('utf-8')])
358
357
path = ''.join(path_chars)
359
358
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
360
return str(prefix + path)
359
return str(scheme + '://' + path)
363
362
def relative_url(base, other):
464
463
return url_base + head, tail
467
def split_segment_parameters_raw(url):
468
"""Split the subsegment of the last segment of a URL.
470
:param url: A relative or absolute URL
471
:return: (url, subsegments)
473
# GZ 2011-11-18: Dodgy removing the terminal slash like this, function
474
# operates on urls not url+segments, and Transport classes
475
# should not be blindly adding slashes in the first place.
476
lurl = strip_trailing_slash(url)
477
# Segments begin at first comma after last forward slash, if one exists
478
segment_start = lurl.find(",", lurl.rfind("/") + 1)
479
if segment_start == -1:
481
return (lurl[:segment_start],
482
[str(s) for s in lurl[segment_start + 1:].split(",")])
485
def split_segment_parameters(url):
486
"""Split the segment parameters of the last segment of a URL.
488
:param url: A relative or absolute URL
489
:return: (url, segment_parameters)
491
(base_url, subsegments) = split_segment_parameters_raw(url)
493
for subsegment in subsegments:
495
(key, value) = subsegment.split("=", 1)
497
raise InvalidURL(url, "missing = in subsegment")
498
if not isinstance(key, str):
500
if not isinstance(value, str):
501
raise TypeError(value)
502
parameters[key] = value
503
return (base_url, parameters)
506
def strip_segment_parameters(url):
507
"""Strip the segment parameters from a URL.
509
:param url: A relative or absolute URL
512
base_url, subsegments = split_segment_parameters_raw(url)
516
def join_segment_parameters_raw(base, *subsegments):
517
"""Create a new URL by adding subsegments to an existing one.
519
This adds the specified subsegments to the last path in the specified
520
base URL. The subsegments should be bytestrings.
522
:note: You probably want to use join_segment_parameters instead.
526
for subsegment in subsegments:
527
if not isinstance(subsegment, str):
528
raise TypeError("Subsegment %r is not a bytestring" % subsegment)
529
if "," in subsegment:
530
raise InvalidURLJoin(", exists in subsegments",
532
return ",".join((base,) + subsegments)
535
def join_segment_parameters(url, parameters):
536
"""Create a new URL by adding segment parameters to an existing one.
538
The parameters of the last segment in the URL will be updated; if a
539
parameter with the same key already exists it will be overwritten.
541
:param url: A URL, as string
542
:param parameters: Dictionary of parameters, keys and values as bytestrings
544
(base, existing_parameters) = split_segment_parameters(url)
546
new_parameters.update(existing_parameters)
547
for key, value in parameters.items():
548
if not isinstance(key, str):
549
raise TypeError("parameter key %r is not a str" % key)
550
if not isinstance(value, str):
551
raise TypeError("parameter value %r for %r is not a str" %
554
raise InvalidURLJoin("= exists in parameter key", url,
556
new_parameters[key] = value
557
return join_segment_parameters_raw(
558
base, *["%s=%s" % item for item in sorted(new_parameters.items())])
561
466
def _win32_strip_local_trailing_slash(url):
562
467
"""Strip slashes after the drive letter"""
563
468
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
613
518
This returns a Unicode path from a URL
615
520
# jam 20060427 URLs are supposed to be ASCII only strings
616
# If they are passed in as unicode, unquote
521
# If they are passed in as unicode, urllib.unquote
617
522
# will return a UNICODE string, which actually contains
618
523
# utf-8 bytes. So we have to ensure that they are
619
524
# plain ASCII strings, or the final .decode will
620
525
# try to encode the UNICODE => ASCII, and then decode
529
except UnicodeError, e:
530
raise errors.InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
623
if isinstance(url, str):
626
except UnicodeError as e:
628
url, 'URL was not a plain ASCII url: %s' % (e,))
629
return urlparse.unquote(url)
532
unquoted = urllib.unquote(url)
534
unicode_path = unquoted.decode('utf-8')
535
except UnicodeError, e:
536
raise errors.InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
632
540
# These are characters that if escaped, should stay that way
633
541
_no_decode_chars = ';/?:@&=+$,#'
634
542
_no_decode_ords = [ord(c) for c in _no_decode_chars]
635
543
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
636
+ ['%02X' % o for o in _no_decode_ords])
637
_hex_display_map = dict(([('%02x' % o, bytes([o])) for o in range(256)]
638
+ [('%02X' % o, bytes([o])) for o in range(256)]))
639
# These entries get mapped to themselves
640
_hex_display_map.update((hex, b'%' + hex.encode('ascii'))
641
for hex in _no_decode_hex)
544
+ ['%02X' % o for o in _no_decode_ords])
545
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
546
+ [('%02X' % o, chr(o)) for o in range(256)]))
547
#These entries get mapped to themselves
548
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
643
550
# These characters shouldn't be percent-encoded, and it's always safe to
644
551
# unencode them if they are.
645
552
_url_dont_escape_characters = set(
646
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
647
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
648
"0123456789" # Numbers
649
"-._~" # Unreserved characters
553
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
554
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
555
"0123456789" # Numbers
556
"-._~" # Unreserved characters
652
559
# These characters should not be escaped
653
560
_url_safe_characters = set(
654
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
655
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
656
"0123456789" # Numbers
657
"_.-!~*'()" # Unreserved characters
658
"/;?:@&=+$," # Reserved characters
659
"%#" # Extra reserved characters
561
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
562
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
563
"0123456789" # Numbers
564
"_.-!~*'()" # Unreserved characters
565
"/;?:@&=+$," # Reserved characters
566
"%#" # Extra reserved characters
663
def _unescape_segment_for_display(segment, encoding):
664
"""Unescape a segment for display.
666
Helper for unescape_for_display
668
:param url: A 7-bit ASCII URL
669
:param encoding: The final output encoding
671
:return: A unicode string which can be safely encoded into the
674
escaped_chunks = segment.split('%')
675
escaped_chunks[0] = escaped_chunks[0].encode('utf-8')
676
for j in range(1, len(escaped_chunks)):
677
item = escaped_chunks[j]
679
escaped_chunks[j] = _hex_display_map[item[:2]]
681
# Put back the percent symbol
682
escaped_chunks[j] = b'%' + (item[:2].encode('utf-8'))
683
except UnicodeDecodeError:
684
escaped_chunks[j] = chr(int(item[:2], 16)).encode('utf-8')
685
escaped_chunks[j] += (item[2:].encode('utf-8'))
686
unescaped = b''.join(escaped_chunks)
688
decoded = unescaped.decode('utf-8')
689
except UnicodeDecodeError:
690
# If this path segment cannot be properly utf-8 decoded
691
# after doing unescaping we will just leave it alone
695
decoded.encode(encoding)
696
except UnicodeEncodeError:
697
# If this chunk cannot be encoded in the local
698
# encoding, then we should leave it alone
701
# Otherwise take the url decoded one
705
569
def unescape_for_display(url, encoding):
706
570
"""Decode what you can for a URL, so that we get a nice looking path.
730
594
# Split into sections to try to decode utf-8
731
595
res = url.split('/')
732
for i in range(1, len(res)):
733
res[i] = _unescape_segment_for_display(res[i], encoding)
596
for i in xrange(1, len(res)):
597
escaped_chunks = res[i].split('%')
598
for j in xrange(1, len(escaped_chunks)):
599
item = escaped_chunks[j]
601
escaped_chunks[j] = _hex_display_map[item[:2]] + item[2:]
603
# Put back the percent symbol
604
escaped_chunks[j] = '%' + item
605
except UnicodeDecodeError:
606
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
607
unescaped = ''.join(escaped_chunks)
609
decoded = unescaped.decode('utf-8')
610
except UnicodeDecodeError:
611
# If this path segment cannot be properly utf-8 decoded
612
# after doing unescaping we will just leave it alone
616
decoded.encode(encoding)
617
except UnicodeEncodeError:
618
# If this chunk cannot be encoded in the local
619
# encoding, then we should leave it alone
622
# Otherwise take the url decoded one
734
624
return u'/'.join(res)
798
687
return osutils.pathjoin(*segments)
804
def __init__(self, scheme, quoted_user, quoted_password, quoted_host,
807
self.quoted_host = quoted_host
808
self.host = unquote(self.quoted_host)
809
self.quoted_user = quoted_user
810
if self.quoted_user is not None:
811
self.user = unquote(self.quoted_user)
814
self.quoted_password = quoted_password
815
if self.quoted_password is not None:
816
self.password = unquote(self.quoted_password)
820
self.quoted_path = _url_hex_escapes_re.sub(
821
_unescape_safe_chars, quoted_path)
822
self.path = unquote(self.quoted_path)
824
def __eq__(self, other):
825
return (isinstance(other, self.__class__) and
826
self.scheme == other.scheme and
827
self.host == other.host and
828
self.user == other.user and
829
self.password == other.password and
830
self.path == other.path)
833
return "<%s(%r, %r, %r, %r, %r, %r)>" % (
834
self.__class__.__name__,
835
self.scheme, self.quoted_user, self.quoted_password,
836
self.quoted_host, self.port, self.quoted_path)
839
def from_string(cls, url):
840
"""Create a URL object from a string.
842
:param url: URL as bytestring
844
# GZ 2017-06-09: Actually validate ascii-ness
845
# pad.lv/1696545: For the moment, accept both native strings and
847
if isinstance(url, str):
849
elif isinstance(url, str):
852
except UnicodeEncodeError:
853
raise InvalidURL(url)
855
raise InvalidURL(url)
856
(scheme, netloc, path, params,
857
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
858
user = password = host = port = None
860
user, host = netloc.rsplit('@', 1)
862
user, password = user.split(':', 1)
866
if ':' in host and not (host[0] == '[' and host[-1] == ']'):
868
host, port = host.rsplit(':', 1)
873
raise InvalidURL('invalid port number %s in url:\n%s' %
877
if host != "" and host[0] == '[' and host[-1] == ']': # IPv6
880
return cls(scheme, user, password, host, port, path)
883
netloc = self.quoted_host
885
netloc = "[%s]" % netloc
886
if self.quoted_user is not None:
887
# Note that we don't put the password back even if we
888
# have one so that it doesn't get accidentally
890
netloc = '%s@%s' % (self.quoted_user, netloc)
891
if self.port is not None:
892
netloc = '%s:%d' % (netloc, self.port)
893
return urlparse.urlunparse(
894
(self.scheme, netloc, self.quoted_path, None, None, None))
897
def _combine_paths(base_path, relpath):
898
"""Transform a Transport-relative path to a remote absolute path.
900
This does not handle substitution of ~ but does handle '..' and '.'
905
t._combine_paths('/home/sarah', 'project/foo')
906
=> '/home/sarah/project/foo'
907
t._combine_paths('/home/sarah', '../../etc')
909
t._combine_paths('/home/sarah', '/etc')
912
:param base_path: base path
913
:param relpath: relative url string for relative part of remote path.
914
:return: urlencoded string for final path.
916
# pad.lv/1696545: For the moment, accept both native strings and
918
if isinstance(relpath, str):
920
elif isinstance(relpath, str):
922
relpath = relpath.encode()
923
except UnicodeEncodeError:
924
raise InvalidURL(relpath)
926
raise InvalidURL(relpath)
927
relpath = _url_hex_escapes_re.sub(_unescape_safe_chars, relpath)
928
if relpath.startswith('/'):
931
base_parts = base_path.split('/')
932
if len(base_parts) > 0 and base_parts[-1] == '':
933
base_parts = base_parts[:-1]
934
for p in relpath.split('/'):
936
if len(base_parts) == 0:
937
# In most filesystems, a request for the parent
938
# of root, just returns root.
945
path = '/'.join(base_parts)
946
if not path.startswith('/'):
950
def clone(self, offset=None):
951
"""Return a new URL for a path relative to this URL.
953
:param offset: A relative path, already urlencoded
954
:return: `URL` instance
956
if offset is not None:
957
relative = unescape(offset)
958
path = self._combine_paths(self.path, relative)
959
path = quote(path, safe="/~")
961
path = self.quoted_path
962
return self.__class__(self.scheme, self.quoted_user,
963
self.quoted_password, self.quoted_host, self.port,
967
691
def parse_url(url):
968
692
"""Extract the server address, the credentials and the path from the url.
973
697
:param url: an quoted url
974
699
:return: (scheme, user, password, host, port, path) tuple, all fields
977
parsed_url = URL.from_string(url)
978
return (parsed_url.scheme, parsed_url.user, parsed_url.password,
979
parsed_url.host, parsed_url.port, parsed_url.path)
702
if isinstance(url, unicode):
703
raise errors.InvalidURL('should be ascii:\n%r' % url)
704
url = url.encode('utf-8')
705
(scheme, netloc, path, params,
706
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
707
user = password = host = port = None
709
user, host = netloc.rsplit('@', 1)
711
user, password = user.split(':', 1)
712
password = urllib.unquote(password)
713
user = urllib.unquote(user)
717
if ':' in host and not (host[0] == '[' and host[-1] == ']'): #there *is* port
718
host, port = host.rsplit(':',1)
722
raise errors.InvalidURL('invalid port number %s in url:\n%s' %
724
if host != "" and host[0] == '[' and host[-1] == ']': #IPv6
727
host = urllib.unquote(host)
728
path = urllib.unquote(path)
730
return (scheme, user, password, host, port, path)