13
15
# You should have received a copy of the GNU General Public License
14
16
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
19
"""A collection of function for handling URL operations."""
19
from __future__ import absolute_import
28
from urllib import parse as urlparse
25
from bzrlib.lazy_import import lazy_import
26
lazy_import(globals(), """
27
from posixpath import split as _posix_split, normpath as _posix_normpath
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.')
72
37
def basename(url, exclude_trailing_slash=True):
73
38
"""Return the last component of a URL.
96
61
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
155
64
def escape(relpath):
156
65
"""Escape relpath to be a valid url."""
157
if not isinstance(relpath, str) and sys.version_info[0] == 2:
66
if isinstance(relpath, unicode):
158
67
relpath = relpath.encode('utf-8')
159
return quote(relpath, safe='/~')
68
# After quoting and encoding, the path should be perfectly
69
# safe as a plain ASCII string, str() just enforces this
70
return str(urllib.quote(relpath))
162
73
def file_relpath(base, path):
163
74
"""Compute just the relative sub-portion of a url
165
76
This assumes that both paths are already fully specified file:// URLs.
167
if len(base) < MIN_ABS_FILEURL_LENGTH:
168
raise ValueError('Length of base (%r) must equal or'
169
' exceed the platform minimum url length (which is %d)' %
170
(base, MIN_ABS_FILEURL_LENGTH))
171
base = osutils.normpath(local_path_from_url(base))
172
path = osutils.normpath(local_path_from_url(path))
78
assert len(base) >= MIN_ABS_FILEURL_LENGTH, ('Length of base must be equal or'
79
' exceed the platform minimum url length (which is %d)' %
80
MIN_ABS_FILEURL_LENGTH)
82
base = local_path_from_url(base)
83
path = local_path_from_url(path)
173
84
return escape(osutils.relpath(base, path))
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:]
122
match = _url_scheme_re.match(base)
125
scheme = match.group('scheme')
126
path = match.group('path').split('/')
127
if path[-1:] == ['']:
128
# Strip off a trailing slash
129
# This helps both when we are at the root, and when
130
# 'base' has an extra slash at the end
133
path = base.split('/')
135
if scheme is not None and len(path) >= 1:
137
# the path should be represented as an abs path.
138
# we know this must be absolute because of the presence of a URL scheme.
140
path = [''] + path[1:]
142
# create an empty host, but dont alter the path - this might be a
143
# relative url fragment.
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
148
match = _url_scheme_re.match(arg)
151
scheme = match.group('scheme')
152
# this skips .. normalisation, making http://host/../../..
154
path = match.group('path').split('/')
155
# set the host and path according to new absolute URL, discarding
156
# any previous values.
157
# XXX: duplicates mess from earlier in this function. This URL
158
# manipulation code needs some cleaning up.
159
if scheme is not None and len(path) >= 1:
162
# url scheme implies absolute path.
165
# no url scheme we take the path as is.
168
path = '/'.join(path)
234
169
path = joinpath(path, arg)
235
return base[:path_start] + path
170
path = path.split('/')
171
if remove_root and path[0:1] == ['']:
174
# Remove the leading slash from the path, so long as it isn't also the
175
# trailing slash, which we want to keep if present.
176
if path and path[0] == '' and len(path) > 1:
181
return '/'.join(path)
182
return scheme + '://' + '/'.join(path)
238
185
def joinpath(base, *args):
239
186
"""Join URL path segments to a URL path segment.
241
188
This is somewhat like osutils.joinpath, but intended for URLs.
243
190
XXX: this duplicates some normalisation logic, and also duplicates a lot of
290
230
This also handles transforming escaping unicode characters, etc.
292
# importing directly from posixpath allows us to test this
232
# importing directly from posixpath allows us to test this
293
233
# on non-posix platforms
294
return 'file://' + escape(osutils._posix_abspath(path))
234
return 'file://' + escape(_posix_normpath(
235
osutils._posix_abspath(path)))
297
238
def _win32_local_path_from_url(url):
298
239
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
299
240
if not url.startswith('file://'):
300
raise InvalidURL(url, 'local urls must start with file:///, '
241
raise errors.InvalidURL(url, 'local urls must start with file:///, '
301
242
'UNC path urls must start with file://')
302
url = split_segment_parameters_raw(url)[0]
303
243
# We strip off all 3 slashes
304
244
win32_url = url[len('file:'):]
305
245
# check for UNC path: //HOST/path
306
246
if not win32_url.startswith('///'):
307
247
if (win32_url[2] == '/'
308
248
or win32_url[3] in '|:'):
309
raise InvalidURL(url, 'Win32 UNC path urls'
249
raise errors.InvalidURL(url, 'Win32 UNC path urls'
310
250
' have form file://HOST/path')
311
251
return unescape(win32_url)
313
# allow empty paths so we can serve all roots
314
if win32_url == '///':
317
252
# usual local path with drive letter
318
if (len(win32_url) < 6
319
or win32_url[3] not in ('abcdefghijklmnopqrstuvwxyz'
320
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
253
if (win32_url[3] not in ('abcdefghijklmnopqrstuvwxyz'
254
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
321
255
or win32_url[4] not in '|:'
322
256
or win32_url[5] != '/'):
323
raise InvalidURL(url, 'Win32 file urls start with'
257
raise errors.InvalidURL(url, 'Win32 file urls start with'
324
258
' file:///x:/, where x is a valid drive letter')
325
259
return win32_url[3].upper() + u':' + unescape(win32_url[5:])
525
430
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())])
608
433
def _win32_strip_local_trailing_slash(url):
609
434
"""Strip slashes after the drive letter"""
610
435
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
660
485
This returns a Unicode path from a URL
662
487
# jam 20060427 URLs are supposed to be ASCII only strings
663
# If they are passed in as unicode, unquote
488
# If they are passed in as unicode, urllib.unquote
664
489
# will return a UNICODE string, which actually contains
665
490
# utf-8 bytes. So we have to ensure that they are
666
491
# plain ASCII strings, or the final .decode will
667
492
# try to encode the UNICODE => ASCII, and then decode
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)
496
except UnicodeError, e:
497
raise errors.InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
499
unquoted = urllib.unquote(url)
679
501
unicode_path = unquoted.decode('utf-8')
680
except UnicodeError as e:
681
raise InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
502
except UnicodeError, e:
503
raise errors.InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
682
504
return unicode_path
685
507
# These are characters that if escaped, should stay that way
686
508
_no_decode_chars = ';/?:@&=+$,#'
687
509
_no_decode_ords = [ord(c) for c in _no_decode_chars]
688
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
510
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
689
511
+ ['%02X' % o for o in _no_decode_ords])
690
512
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
691
513
+ [('%02X' % o, chr(o)) for o in range(256)]))
692
514
#These entries get mapped to themselves
693
_hex_display_map.update((hex, '%'+hex) for hex in _no_decode_hex)
695
# These characters shouldn't be percent-encoded, and it's always safe to
696
# unencode them if they are.
697
_url_dont_escape_characters = set(
698
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
699
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
700
"0123456789" # Numbers
701
"-._~" # Unreserved characters
515
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
704
517
# These characters should not be escaped
705
518
_url_safe_characters = set(
750
562
except UnicodeDecodeError:
751
563
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
752
564
unescaped = ''.join(escaped_chunks)
753
if sys.version_info[0] == 2:
566
decoded = unescaped.decode('utf-8')
567
except UnicodeDecodeError:
568
# If this path segment cannot be properly utf-8 decoded
569
# after doing unescaping we will just 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
573
decoded.encode(encoding)
574
except UnicodeEncodeError:
575
# If this chunk cannot be encoded in the local
576
# encoding, then we should leave it alone
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
579
# Otherwise take the url decoded one
772
581
return u'/'.join(res)
775
def derive_to_location(from_location):
776
"""Derive a TO_LOCATION given a FROM_LOCATION.
778
The normal case is a FROM_LOCATION of http://foo/bar => bar.
779
The Right Thing for some logical destinations may differ though
780
because no / may be present at all. In that case, the result is
781
the full name without the scheme indicator, e.g. lp:foo-bar => foo-bar.
782
This latter case also applies when a Windows drive
783
is used without a path, e.g. c:foo-bar => foo-bar.
784
If no /, path separator or : is found, the from_location is returned.
786
if from_location.find("/") >= 0 or from_location.find(os.sep) >= 0:
787
return os.path.basename(from_location.rstrip("/\\"))
789
sep = from_location.find(":")
791
return from_location[sep+1:]
796
def _is_absolute(url):
797
return (osutils.pathjoin('/foo', url) == url)
800
def rebase_url(url, old_base, new_base):
801
"""Convert a relative path from an old base URL to a new base URL.
803
The result will be a relative path.
804
Absolute paths and full URLs are returned unaltered.
806
scheme, separator = _find_scheme_and_separator(url)
807
if scheme is not None:
809
if _is_absolute(url):
811
old_parsed = urlparse.urlparse(old_base)
812
new_parsed = urlparse.urlparse(new_base)
813
if (old_parsed[:2]) != (new_parsed[:2]):
814
raise InvalidRebaseURLs(old_base, new_base)
815
return determine_relative_path(new_parsed[2],
816
join(old_parsed[2], url))
819
def determine_relative_path(from_path, to_path):
820
"""Determine a relative path from from_path to to_path."""
821
from_segments = osutils.splitpath(from_path)
822
to_segments = osutils.splitpath(to_path)
824
for count, (from_element, to_element) in enumerate(zip(from_segments,
826
if from_element != to_element:
830
unique_from = from_segments[count:]
831
unique_to = to_segments[count:]
832
segments = (['..'] * len(unique_from) + unique_to)
833
if len(segments) == 0:
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,
1001
"""Extract the server address, the credentials and the path from the url.
1003
user, password, host and path should be quoted if they contain reserved
1006
:param url: an quoted url
1007
:return: (scheme, user, password, host, port, path) tuple, all fields
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)