17
17
"""A collection of function for handling URL operations."""
19
from __future__ import absolute_import
28
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
35
from .lazy_import import lazy_import
36
lazy_import(globals(), """
37
from posixpath import split as _posix_split
48
class InvalidURL(errors.PathError):
50
_fmt = 'Invalid url supplied to transport: "%(path)s"%(extra)s'
53
class InvalidURLJoin(errors.PathError):
55
_fmt = "Invalid URL join request: %(reason)s: %(base)r + %(join_args)r"
57
def __init__(self, reason, base, join_args):
60
self.join_args = join_args
61
errors.PathError.__init__(self, base, reason)
64
class InvalidRebaseURLs(errors.PathError):
66
_fmt = "URLs differ by more than path: %(from_)r and %(to)r"
68
def __init__(self, from_, to):
71
errors.PathError.__init__(
72
self, from_, 'URLs differ by more than path.')
75
36
def basename(url, exclude_trailing_slash=True):
76
37
"""Return the last component of a URL.
99
60
return split(url, exclude_trailing_slash=exclude_trailing_slash)[0]
103
quote_from_bytes = urlparse.quote_from_bytes
104
quote = urlparse.quote
105
unquote_to_bytes = urlparse.unquote_to_bytes
107
# Private copies of quote and unquote, copied from Python's urllib module
108
# because urllib unconditionally imports socket, which imports ssl.
110
always_safe = ('ABCDEFGHIJKLMNOPQRSTUVWXYZ'
111
'abcdefghijklmnopqrstuvwxyz'
114
for i, c in zip(range(256), ''.join(map(chr, range(256)))):
115
_safe_map[c] = c if (
116
i < 128 and c in always_safe) else '%{0:02X}'.format(i)
119
def quote_from_bytes(s, safe='/'):
120
"""quote('abc def') -> 'abc%20def'
122
Each part of a URL, e.g. the path info, the query, etc., has a
123
different set of reserved characters that must be quoted.
125
RFC 2396 Uniform Resource Identifiers (URI): Generic Syntax lists
126
the following reserved characters.
128
reserved = ";" | "/" | "?" | ":" | "@" | "&" | "=" | "+" |
131
Each of these characters is reserved in some component of a URL,
132
but not necessarily in all of them.
134
By default, the quote function is intended for quoting the path
135
section of a URL. Thus, it will not encode '/'. This character
136
is reserved, but in typical usage the quote function is being
137
called on a path where the existing slash characters are used as
143
raise TypeError('None object cannot be quoted')
145
cachekey = (safe, always_safe)
147
(quoter, safe) = _safe_quoters[cachekey]
149
safe_map = _safe_map.copy()
150
safe_map.update([(c, c) for c in safe])
151
quoter = safe_map.__getitem__
152
safe = always_safe + safe
153
_safe_quoters[cachekey] = (quoter, safe)
154
if not s.rstrip(safe):
156
return ''.join(map(quoter, s))
158
quote = quote_from_bytes
159
unquote_to_bytes = urlparse.unquote
162
unquote = urlparse.unquote
165
63
def escape(relpath):
166
64
"""Escape relpath to be a valid url."""
167
if not isinstance(relpath, str) and sys.version_info[0] == 2:
65
if isinstance(relpath, unicode):
168
66
relpath = relpath.encode('utf-8')
169
return quote(relpath, safe='/~')
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='/~'))
172
72
def file_relpath(base, path):
226
scheme_end, path_start = _find_scheme_and_separator(base)
227
if scheme_end is None and path_start is None:
229
elif path_start is None:
230
path_start = len(base)
231
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.
233
arg_scheme_end, arg_path_start = _find_scheme_and_separator(arg)
234
if arg_scheme_end is None and arg_path_start is None:
236
elif arg_path_start is None:
237
arg_path_start = len(arg)
238
if arg_scheme_end is not None:
240
path = arg[arg_path_start:]
241
scheme_end = arg_scheme_end
242
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)
244
168
path = joinpath(path, arg)
245
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)
248
184
def joinpath(base, *args):
302
237
# importing directly from posixpath allows us to test this
303
238
# on non-posix platforms
304
return 'file://' + escape(osutils._posix_abspath(path))
239
return 'file://' + escape(_posix_normpath(
240
osutils._posix_abspath(path)))
307
243
def _win32_local_path_from_url(url):
308
244
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
309
245
if not url.startswith('file://'):
310
raise InvalidURL(url, 'local urls must start with file:///, '
311
'UNC path urls must start with file://')
312
url = split_segment_parameters_raw(url)[0]
246
raise errors.InvalidURL(url, 'local urls must start with file:///, '
247
'UNC path urls must start with file://')
313
248
# We strip off all 3 slashes
314
249
win32_url = url[len('file:'):]
315
250
# check for UNC path: //HOST/path
316
251
if not win32_url.startswith('///'):
317
252
if (win32_url[2] == '/'
318
or win32_url[3] in '|:'):
319
raise InvalidURL(url, 'Win32 UNC path urls'
320
' have form file://HOST/path')
253
or win32_url[3] in '|:'):
254
raise errors.InvalidURL(url, 'Win32 UNC path urls'
255
' have form file://HOST/path')
321
256
return unescape(win32_url)
323
258
# allow empty paths so we can serve all roots
405
339
:param url: Either a hybrid URL or a local path
406
340
:return: A normalized URL which only includes 7-bit ASCII characters.
408
scheme_end, path_start = _find_scheme_and_separator(url)
409
if scheme_end is None:
342
m = _url_scheme_re.match(url)
410
344
return local_path_to_url(url)
411
prefix = url[:path_start]
412
path = url[path_start:]
413
if not isinstance(url, text_type):
345
scheme = m.group('scheme')
346
path = m.group('path')
347
if not isinstance(url, unicode):
415
349
if c not in _url_safe_characters:
416
raise InvalidURL(url, 'URLs can only contain specific'
417
' safe characters (not %r)' % c)
350
raise errors.InvalidURL(url, 'URLs can only contain specific'
351
' safe characters (not %r)' % c)
418
352
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
419
return str(prefix + ''.join(path))
353
return str(scheme + '://' + ''.join(path))
421
355
# We have a unicode (hybrid) url
422
356
path_chars = list(path)
424
for i in range(len(path_chars)):
358
for i in xrange(len(path_chars)):
425
359
if path_chars[i] not in _url_safe_characters:
360
chars = path_chars[i].encode('utf-8')
426
361
path_chars[i] = ''.join(
427
['%%%02X' % c for c in bytearray(path_chars[i].encode('utf-8'))])
362
['%%%02X' % ord(c) for c in path_chars[i].encode('utf-8')])
428
363
path = ''.join(path_chars)
429
364
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
430
return str(prefix + path)
365
return str(scheme + '://' + path)
433
368
def relative_url(base, other):
534
469
return url_base + head, tail
537
def split_segment_parameters_raw(url):
538
"""Split the subsegment of the last segment of a URL.
540
:param url: A relative or absolute URL
541
:return: (url, subsegments)
543
# GZ 2011-11-18: Dodgy removing the terminal slash like this, function
544
# operates on urls not url+segments, and Transport classes
545
# should not be blindly adding slashes in the first place.
546
lurl = strip_trailing_slash(url)
547
# Segments begin at first comma after last forward slash, if one exists
548
segment_start = lurl.find(",", lurl.rfind("/") + 1)
549
if segment_start == -1:
551
return (lurl[:segment_start],
552
[str(s) for s in lurl[segment_start + 1:].split(",")])
555
def split_segment_parameters(url):
556
"""Split the segment parameters of the last segment of a URL.
558
:param url: A relative or absolute URL
559
:return: (url, segment_parameters)
561
(base_url, subsegments) = split_segment_parameters_raw(url)
563
for subsegment in subsegments:
564
(key, value) = subsegment.split("=", 1)
565
if not isinstance(key, str):
567
if not isinstance(value, str):
568
raise TypeError(value)
569
parameters[key] = value
570
return (base_url, parameters)
573
def join_segment_parameters_raw(base, *subsegments):
574
"""Create a new URL by adding subsegments to an existing one.
576
This adds the specified subsegments to the last path in the specified
577
base URL. The subsegments should be bytestrings.
579
:note: You probably want to use join_segment_parameters instead.
583
for subsegment in subsegments:
584
if not isinstance(subsegment, str):
585
raise TypeError("Subsegment %r is not a bytestring" % subsegment)
586
if "," in subsegment:
587
raise InvalidURLJoin(", exists in subsegments",
589
return ",".join((base,) + subsegments)
592
def join_segment_parameters(url, parameters):
593
"""Create a new URL by adding segment parameters to an existing one.
595
The parameters of the last segment in the URL will be updated; if a
596
parameter with the same key already exists it will be overwritten.
598
:param url: A URL, as string
599
:param parameters: Dictionary of parameters, keys and values as bytestrings
601
(base, existing_parameters) = split_segment_parameters(url)
603
new_parameters.update(existing_parameters)
604
for key, value in parameters.items():
605
if not isinstance(key, str):
606
raise TypeError("parameter key %r is not a str" % key)
607
if not isinstance(value, str):
608
raise TypeError("parameter value %r for %r is not a str" %
611
raise InvalidURLJoin("= exists in parameter key", url,
613
new_parameters[key] = value
614
return join_segment_parameters_raw(
615
base, *["%s=%s" % item for item in sorted(new_parameters.items())])
618
472
def _win32_strip_local_trailing_slash(url):
619
473
"""Strip slashes after the drive letter"""
620
474
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
670
524
This returns a Unicode path from a URL
672
526
# jam 20060427 URLs are supposed to be ASCII only strings
673
# If they are passed in as unicode, unquote
527
# If they are passed in as unicode, urllib.unquote
674
528
# will return a UNICODE string, which actually contains
675
529
# utf-8 bytes. So we have to ensure that they are
676
530
# plain ASCII strings, or the final .decode will
677
531
# 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,))
681
if isinstance(url, text_type):
684
except UnicodeError as e:
686
url, 'URL was not a plain ASCII url: %s' % (e,))
687
return urlparse.unquote(url)
689
if isinstance(url, text_type):
691
url = url.encode("ascii")
692
except UnicodeError as e:
694
url, 'URL was not a plain ASCII url: %s' % (e,))
695
unquoted = unquote(url)
697
unicode_path = unquoted.decode('utf-8')
698
except UnicodeError as e:
700
url, 'Unable to encode the URL as utf-8: %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,))
704
546
# These are characters that if escaped, should stay that way
705
547
_no_decode_chars = ';/?:@&=+$,#'
706
548
_no_decode_ords = [ord(c) for c in _no_decode_chars]
707
549
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
708
+ ['%02X' % o for o in _no_decode_ords])
709
_hex_display_map = dict(([('%02x' % o, int2byte(o)) for o in range(256)]
710
+ [('%02X' % o, int2byte(o)) for o in range(256)]))
711
# These entries get mapped to themselves
712
_hex_display_map.update((hex, b'%' + hex.encode('ascii'))
713
for hex in _no_decode_hex)
550
+ ['%02X' % o for o in _no_decode_ords])
551
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
552
+ [('%02X' % o, chr(o)) for o in range(256)]))
553
#These entries get mapped to themselves
554
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
715
556
# These characters shouldn't be percent-encoded, and it's always safe to
716
557
# unencode them if they are.
717
558
_url_dont_escape_characters = set(
718
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
719
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
720
"0123456789" # Numbers
721
"-._~" # Unreserved characters
559
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
560
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
561
"0123456789" # Numbers
562
"-._~" # Unreserved characters
724
565
# These characters should not be escaped
725
566
_url_safe_characters = set(
726
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
727
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
728
"0123456789" # Numbers
729
"_.-!~*'()" # Unreserved characters
730
"/;?:@&=+$," # Reserved characters
731
"%#" # Extra reserved characters
567
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
568
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
569
"0123456789" # Numbers
570
"_.-!~*'()" # Unreserved characters
571
"/;?:@&=+$," # Reserved characters
572
"%#" # Extra reserved characters
735
def _unescape_segment_for_display(segment, encoding):
736
"""Unescape a segment for display.
738
Helper for unescape_for_display
740
:param url: A 7-bit ASCII URL
741
:param encoding: The final output encoding
743
:return: A unicode string which can be safely encoded into the
746
escaped_chunks = segment.split('%')
747
escaped_chunks[0] = escaped_chunks[0].encode('utf-8')
748
for j in range(1, len(escaped_chunks)):
749
item = escaped_chunks[j]
751
escaped_chunks[j] = _hex_display_map[item[:2]]
753
# Put back the percent symbol
754
escaped_chunks[j] = b'%' + \
755
(item[:2].encode('utf-8') if PY3 else item[:2])
756
except UnicodeDecodeError:
757
escaped_chunks[j] = unichr(int(item[:2], 16)).encode('utf-8')
758
escaped_chunks[j] += (item[2:].encode('utf-8') if PY3 else item[2:])
759
unescaped = b''.join(escaped_chunks)
761
decoded = unescaped.decode('utf-8')
762
except UnicodeDecodeError:
763
# If this path segment cannot be properly utf-8 decoded
764
# after doing unescaping we will just leave it alone
768
decoded.encode(encoding)
769
except UnicodeEncodeError:
770
# If this chunk cannot be encoded in the local
771
# encoding, then we should leave it alone
774
# Otherwise take the url decoded one
778
575
def unescape_for_display(url, encoding):
779
576
"""Decode what you can for a URL, so that we get a nice looking path.
803
600
# Split into sections to try to decode utf-8
804
601
res = url.split('/')
805
for i in range(1, len(res)):
806
res[i] = _unescape_segment_for_display(res[i], encoding)
602
for i in xrange(1, len(res)):
603
escaped_chunks = res[i].split('%')
604
for j in xrange(1, len(escaped_chunks)):
605
item = escaped_chunks[j]
607
escaped_chunks[j] = _hex_display_map[item[:2]] + item[2:]
609
# Put back the percent symbol
610
escaped_chunks[j] = '%' + item
611
except UnicodeDecodeError:
612
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
613
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
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
628
# Otherwise take the url decoded one
807
630
return u'/'.join(res)
871
693
return osutils.pathjoin(*segments)
877
def __init__(self, scheme, quoted_user, quoted_password, quoted_host,
880
self.quoted_host = quoted_host
881
self.host = unquote(self.quoted_host)
882
self.quoted_user = quoted_user
883
if self.quoted_user is not None:
884
self.user = unquote(self.quoted_user)
887
self.quoted_password = quoted_password
888
if self.quoted_password is not None:
889
self.password = unquote(self.quoted_password)
893
self.quoted_path = _url_hex_escapes_re.sub(
894
_unescape_safe_chars, quoted_path)
895
self.path = unquote(self.quoted_path)
897
def __eq__(self, other):
898
return (isinstance(other, self.__class__) and
899
self.scheme == other.scheme and
900
self.host == other.host and
901
self.user == other.user and
902
self.password == other.password and
903
self.path == other.path)
906
return "<%s(%r, %r, %r, %r, %r, %r)>" % (
907
self.__class__.__name__,
908
self.scheme, self.quoted_user, self.quoted_password,
909
self.quoted_host, self.port, self.quoted_path)
912
def from_string(cls, url):
913
"""Create a URL object from a string.
915
:param url: URL as bytestring
917
# GZ 2017-06-09: Actually validate ascii-ness
918
# pad.lv/1696545: For the moment, accept both native strings and
920
if isinstance(url, str):
922
elif isinstance(url, text_type):
925
except UnicodeEncodeError:
926
raise InvalidURL(url)
928
raise InvalidURL(url)
929
(scheme, netloc, path, params,
930
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
931
user = password = host = port = None
933
user, host = netloc.rsplit('@', 1)
935
user, password = user.split(':', 1)
939
if ':' in host and not (host[0] == '[' and host[-1] == ']'):
941
host, port = host.rsplit(':', 1)
946
raise InvalidURL('invalid port number %s in url:\n%s' %
950
if host != "" and host[0] == '[' and host[-1] == ']': # IPv6
953
return cls(scheme, user, password, host, port, path)
956
netloc = self.quoted_host
958
netloc = "[%s]" % netloc
959
if self.quoted_user is not None:
960
# Note that we don't put the password back even if we
961
# have one so that it doesn't get accidentally
963
netloc = '%s@%s' % (self.quoted_user, netloc)
964
if self.port is not None:
965
netloc = '%s:%d' % (netloc, self.port)
966
return urlparse.urlunparse(
967
(self.scheme, netloc, self.quoted_path, None, None, None))
970
def _combine_paths(base_path, relpath):
971
"""Transform a Transport-relative path to a remote absolute path.
973
This does not handle substitution of ~ but does handle '..' and '.'
978
t._combine_paths('/home/sarah', 'project/foo')
979
=> '/home/sarah/project/foo'
980
t._combine_paths('/home/sarah', '../../etc')
982
t._combine_paths('/home/sarah', '/etc')
985
:param base_path: base path
986
:param relpath: relative url string for relative part of remote path.
987
:return: urlencoded string for final path.
989
# pad.lv/1696545: For the moment, accept both native strings and
991
if isinstance(relpath, str):
993
elif isinstance(relpath, text_type):
995
relpath = relpath.encode()
996
except UnicodeEncodeError:
997
raise InvalidURL(relpath)
999
raise InvalidURL(relpath)
1000
relpath = _url_hex_escapes_re.sub(_unescape_safe_chars, relpath)
1001
if relpath.startswith('/'):
1004
base_parts = base_path.split('/')
1005
if len(base_parts) > 0 and base_parts[-1] == '':
1006
base_parts = base_parts[:-1]
1007
for p in relpath.split('/'):
1009
if len(base_parts) == 0:
1010
# In most filesystems, a request for the parent
1011
# of root, just returns root.
1017
base_parts.append(p)
1018
path = '/'.join(base_parts)
1019
if not path.startswith('/'):
1023
def clone(self, offset=None):
1024
"""Return a new URL for a path relative to this URL.
1026
:param offset: A relative path, already urlencoded
1027
:return: `URL` instance
1029
if offset is not None:
1030
relative = unescape(offset)
1031
if sys.version_info[0] == 2:
1032
relative = relative.encode('utf-8')
1033
path = self._combine_paths(self.path, relative)
1034
path = quote(path, safe="/~")
1036
path = self.quoted_path
1037
return self.__class__(self.scheme, self.quoted_user,
1038
self.quoted_password, self.quoted_host, self.port,
1042
697
def parse_url(url):
1043
698
"""Extract the server address, the credentials and the path from the url.
1048
703
:param url: an quoted url
1049
705
:return: (scheme, user, password, host, port, path) tuple, all fields
1052
parsed_url = URL.from_string(url)
1053
return (parsed_url.scheme, parsed_url.user, parsed_url.password,
1054
parsed_url.host, parsed_url.port, parsed_url.path)
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)
added
removed
bzrlib/urlutils.py
