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]
100
quote_from_bytes = urlparse.quote_from_bytes
101
quote = urlparse.quote
102
unquote_to_bytes = urlparse.unquote_to_bytes
104
# Private copies of quote and unquote, copied from Python's
105
# urllib module because urllib unconditionally imports socket, which imports
108
always_safe = ('ABCDEFGHIJKLMNOPQRSTUVWXYZ'
109
'abcdefghijklmnopqrstuvwxyz'
112
for i, c in zip(range(256), ''.join(map(chr, range(256)))):
113
_safe_map[c] = c if (i < 128 and c in always_safe) else '%{0:02X}'.format(i)
116
def quote_from_bytes(s, safe='/'):
117
"""quote('abc def') -> 'abc%20def'
119
Each part of a URL, e.g. the path info, the query, etc., has a
120
different set of reserved characters that must be quoted.
122
RFC 2396 Uniform Resource Identifiers (URI): Generic Syntax lists
123
the following reserved characters.
125
reserved = ";" | "/" | "?" | ":" | "@" | "&" | "=" | "+" |
128
Each of these characters is reserved in some component of a URL,
129
but not necessarily in all of them.
131
By default, the quote function is intended for quoting the path
132
section of a URL. Thus, it will not encode '/'. This character
133
is reserved, but in typical usage the quote function is being
134
called on a path where the existing slash characters are used as
140
raise TypeError('None object cannot be quoted')
142
cachekey = (safe, always_safe)
144
(quoter, safe) = _safe_quoters[cachekey]
146
safe_map = _safe_map.copy()
147
safe_map.update([(c, c) for c in safe])
148
quoter = safe_map.__getitem__
149
safe = always_safe + safe
150
_safe_quoters[cachekey] = (quoter, safe)
151
if not s.rstrip(safe):
153
return ''.join(map(quoter, s))
155
quote = quote_from_bytes
156
unquote_to_bytes = urlparse.unquote
159
unquote = urlparse.unquote
63
162
def escape(relpath):
64
163
"""Escape relpath to be a valid url."""
65
if isinstance(relpath, unicode):
164
if not isinstance(relpath, str) and sys.version_info[0] == 2:
66
165
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='/~'))
166
return quote(relpath, safe='/~')
72
169
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.
223
scheme_end, path_start = _find_scheme_and_separator(base)
224
if scheme_end is None and path_start is None:
226
elif path_start is None:
227
path_start = len(base)
228
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.
230
arg_scheme_end, arg_path_start = _find_scheme_and_separator(arg)
231
if arg_scheme_end is None and arg_path_start is None:
233
elif arg_path_start is None:
234
arg_path_start = len(arg)
235
if arg_scheme_end is not None:
237
path = arg[arg_path_start:]
238
scheme_end = arg_scheme_end
239
path_start = arg_path_start
167
path = '/'.join(path)
168
241
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)
242
return base[:path_start] + path
184
245
def joinpath(base, *args):
237
299
# importing directly from posixpath allows us to test this
238
300
# on non-posix platforms
239
return 'file://' + escape(_posix_normpath(
240
osutils._posix_abspath(path)))
301
return 'file://' + escape(osutils._posix_abspath(path))
243
304
def _win32_local_path_from_url(url):
244
305
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
245
306
if not url.startswith('file://'):
246
raise errors.InvalidURL(url, 'local urls must start with file:///, '
307
raise InvalidURL(url, 'local urls must start with file:///, '
247
308
'UNC path urls must start with file://')
309
url = split_segment_parameters_raw(url)[0]
248
310
# We strip off all 3 slashes
249
311
win32_url = url[len('file:'):]
250
312
# check for UNC path: //HOST/path
251
313
if not win32_url.startswith('///'):
252
314
if (win32_url[2] == '/'
253
315
or win32_url[3] in '|:'):
254
raise errors.InvalidURL(url, 'Win32 UNC path urls'
316
raise InvalidURL(url, 'Win32 UNC path urls'
255
317
' have form file://HOST/path')
256
318
return unescape(win32_url)
339
402
:param url: Either a hybrid URL or a local path
340
403
:return: A normalized URL which only includes 7-bit ASCII characters.
342
m = _url_scheme_re.match(url)
405
scheme_end, path_start = _find_scheme_and_separator(url)
406
if scheme_end is None:
344
407
return local_path_to_url(url)
345
scheme = m.group('scheme')
346
path = m.group('path')
347
if not isinstance(url, unicode):
408
prefix = url[:path_start]
409
path = url[path_start:]
410
if not isinstance(url, text_type):
349
412
if c not in _url_safe_characters:
350
raise errors.InvalidURL(url, 'URLs can only contain specific'
413
raise InvalidURL(url, 'URLs can only contain specific'
351
414
' safe characters (not %r)' % c)
352
415
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
353
return str(scheme + '://' + ''.join(path))
416
return str(prefix + ''.join(path))
355
418
# We have a unicode (hybrid) url
356
419
path_chars = list(path)
358
for i in xrange(len(path_chars)):
421
for i in range(len(path_chars)):
359
422
if path_chars[i] not in _url_safe_characters:
360
423
chars = path_chars[i].encode('utf-8')
361
424
path_chars[i] = ''.join(
362
['%%%02X' % ord(c) for c in path_chars[i].encode('utf-8')])
425
['%%%02X' % c for c in bytearray(path_chars[i].encode('utf-8'))])
363
426
path = ''.join(path_chars)
364
427
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
365
return str(scheme + '://' + path)
428
return str(prefix + path)
368
431
def relative_url(base, other):
469
532
return url_base + head, tail
535
def split_segment_parameters_raw(url):
536
"""Split the subsegment of the last segment of a URL.
538
:param url: A relative or absolute URL
539
:return: (url, subsegments)
541
# GZ 2011-11-18: Dodgy removing the terminal slash like this, function
542
# operates on urls not url+segments, and Transport classes
543
# should not be blindly adding slashes in the first place.
544
lurl = strip_trailing_slash(url)
545
# Segments begin at first comma after last forward slash, if one exists
546
segment_start = lurl.find(",", lurl.rfind("/")+1)
547
if segment_start == -1:
549
return (lurl[:segment_start], [str(s) for s in lurl[segment_start+1:].split(",")])
552
def split_segment_parameters(url):
553
"""Split the segment parameters of the last segment of a URL.
555
:param url: A relative or absolute URL
556
:return: (url, segment_parameters)
558
(base_url, subsegments) = split_segment_parameters_raw(url)
560
for subsegment in subsegments:
561
(key, value) = subsegment.split("=", 1)
562
if not isinstance(key, str):
564
if not isinstance(value, str):
565
raise TypeError(value)
566
parameters[key] = value
567
return (base_url, parameters)
570
def join_segment_parameters_raw(base, *subsegments):
571
"""Create a new URL by adding subsegments to an existing one.
573
This adds the specified subsegments to the last path in the specified
574
base URL. The subsegments should be bytestrings.
576
:note: You probably want to use join_segment_parameters instead.
580
for subsegment in subsegments:
581
if not isinstance(subsegment, str):
582
raise TypeError("Subsegment %r is not a bytestring" % subsegment)
583
if "," in subsegment:
584
raise InvalidURLJoin(", exists in subsegments",
586
return ",".join((base,) + subsegments)
589
def join_segment_parameters(url, parameters):
590
"""Create a new URL by adding segment parameters to an existing one.
592
The parameters of the last segment in the URL will be updated; if a
593
parameter with the same key already exists it will be overwritten.
595
:param url: A URL, as string
596
:param parameters: Dictionary of parameters, keys and values as bytestrings
598
(base, existing_parameters) = split_segment_parameters(url)
600
new_parameters.update(existing_parameters)
601
for key, value in parameters.items():
602
if not isinstance(key, str):
603
raise TypeError("parameter key %r is not a str" % key)
604
if not isinstance(value, str):
605
raise TypeError("parameter value %r for %r is not a str" %
608
raise InvalidURLJoin("= exists in parameter key", url,
610
new_parameters[key] = value
611
return join_segment_parameters_raw(base,
612
*["%s=%s" % item for item in sorted(new_parameters.items())])
472
615
def _win32_strip_local_trailing_slash(url):
473
616
"""Strip slashes after the drive letter"""
474
617
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
524
667
This returns a Unicode path from a URL
526
669
# jam 20060427 URLs are supposed to be ASCII only strings
527
# If they are passed in as unicode, urllib.unquote
670
# If they are passed in as unicode, unquote
528
671
# will return a UNICODE string, which actually contains
529
672
# utf-8 bytes. So we have to ensure that they are
530
673
# plain ASCII strings, or the final .decode will
531
674
# 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)
540
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,))
678
if isinstance(url, text_type):
681
except UnicodeError as e:
682
raise InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
683
return urlparse.unquote(url)
685
if isinstance(url, text_type):
687
url = url.encode("ascii")
688
except UnicodeError as e:
689
raise InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
690
unquoted = unquote(url)
692
unicode_path = unquoted.decode('utf-8')
693
except UnicodeError as e:
694
raise InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
546
698
# These are characters that if escaped, should stay that way
611
763
except UnicodeDecodeError:
612
764
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
613
765
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
766
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
768
decoded = unescaped.decode('utf-8')
769
except UnicodeDecodeError:
770
# If this path segment cannot be properly utf-8 decoded
771
# after doing unescaping we will just leave it alone
628
# Otherwise take the url decoded one
775
decoded.encode(encoding)
776
except UnicodeEncodeError:
777
# If this chunk cannot be encoded in the local
778
# encoding, then we should leave it alone
781
# Otherwise take the url decoded one
630
785
return u'/'.join(res)
693
848
return osutils.pathjoin(*segments)
854
def __init__(self, scheme, quoted_user, quoted_password, quoted_host,
857
self.quoted_host = quoted_host
858
self.host = unquote(self.quoted_host)
859
self.quoted_user = quoted_user
860
if self.quoted_user is not None:
861
self.user = unquote(self.quoted_user)
864
self.quoted_password = quoted_password
865
if self.quoted_password is not None:
866
self.password = unquote(self.quoted_password)
870
self.quoted_path = _url_hex_escapes_re.sub(_unescape_safe_chars, quoted_path)
871
self.path = unquote(self.quoted_path)
873
def __eq__(self, other):
874
return (isinstance(other, self.__class__) and
875
self.scheme == other.scheme and
876
self.host == other.host and
877
self.user == other.user and
878
self.password == other.password and
879
self.path == other.path)
882
return "<%s(%r, %r, %r, %r, %r, %r)>" % (
883
self.__class__.__name__,
884
self.scheme, self.quoted_user, self.quoted_password,
885
self.quoted_host, self.port, self.quoted_path)
888
def from_string(cls, url):
889
"""Create a URL object from a string.
891
:param url: URL as bytestring
893
# GZ 2017-06-09: Actually validate ascii-ness
894
# pad.lv/1696545: For the moment, accept both native strings and unicode.
895
if isinstance(url, str):
897
elif isinstance(url, text_type):
900
except UnicodeEncodeError:
901
raise InvalidURL(url)
903
raise InvalidURL(url)
904
(scheme, netloc, path, params,
905
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
906
user = password = host = port = None
908
user, host = netloc.rsplit('@', 1)
910
user, password = user.split(':', 1)
914
if ':' in host and not (host[0] == '[' and host[-1] == ']'):
916
host, port = host.rsplit(':', 1)
920
raise InvalidURL('invalid port number %s in url:\n%s' %
922
if host != "" and host[0] == '[' and host[-1] == ']': #IPv6
925
return cls(scheme, user, password, host, port, path)
928
netloc = self.quoted_host
930
netloc = "[%s]" % netloc
931
if self.quoted_user is not None:
932
# Note that we don't put the password back even if we
933
# have one so that it doesn't get accidentally
935
netloc = '%s@%s' % (self.quoted_user, netloc)
936
if self.port is not None:
937
netloc = '%s:%d' % (netloc, self.port)
938
return urlparse.urlunparse(
939
(self.scheme, netloc, self.quoted_path, None, None, None))
942
def _combine_paths(base_path, relpath):
943
"""Transform a Transport-relative path to a remote absolute path.
945
This does not handle substitution of ~ but does handle '..' and '.'
950
t._combine_paths('/home/sarah', 'project/foo')
951
=> '/home/sarah/project/foo'
952
t._combine_paths('/home/sarah', '../../etc')
954
t._combine_paths('/home/sarah', '/etc')
957
:param base_path: base path
958
:param relpath: relative url string for relative part of remote path.
959
:return: urlencoded string for final path.
961
# pad.lv/1696545: For the moment, accept both native strings and unicode.
962
if isinstance(relpath, str):
964
elif isinstance(relpath, text_type):
966
relpath = relpath.encode()
967
except UnicodeEncodeError:
968
raise InvalidURL(relpath)
970
raise InvalidURL(relpath)
971
relpath = _url_hex_escapes_re.sub(_unescape_safe_chars, relpath)
972
if relpath.startswith('/'):
975
base_parts = base_path.split('/')
976
if len(base_parts) > 0 and base_parts[-1] == '':
977
base_parts = base_parts[:-1]
978
for p in relpath.split('/'):
980
if len(base_parts) == 0:
981
# In most filesystems, a request for the parent
982
# of root, just returns root.
989
path = '/'.join(base_parts)
990
if not path.startswith('/'):
994
def clone(self, offset=None):
995
"""Return a new URL for a path relative to this URL.
997
:param offset: A relative path, already urlencoded
998
:return: `URL` instance
1000
if offset is not None:
1001
relative = unescape(offset)
1002
if sys.version_info[0] == 2:
1003
relative = relative.encode('utf-8')
1004
path = self._combine_paths(self.path, relative)
1005
path = quote(path, safe="/~")
1007
path = self.quoted_path
1008
return self.__class__(self.scheme, self.quoted_user,
1009
self.quoted_password, self.quoted_host, self.port,
697
1013
def parse_url(url):
698
1014
"""Extract the server address, the credentials and the path from the url.
703
1019
:param url: an quoted url
705
1020
: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)
1023
parsed_url = URL.from_string(url)
1024
return (parsed_url.scheme, parsed_url.user, parsed_url.password,
1025
parsed_url.host, parsed_url.port, parsed_url.path)