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):
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], 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
parameters[key] = value
556
return (base_url, parameters)
559
def join_segment_parameters_raw(base, *subsegments):
560
"""Create a new URL by adding subsegments to an existing one.
562
This adds the specified subsegments to the last path in the specified
563
base URL. The subsegments should be bytestrings.
565
:note: You probably want to use join_segment_parameters instead.
569
for subsegment in subsegments:
570
if not isinstance(subsegment, str):
571
raise TypeError("Subsegment %r is not a bytestring" % subsegment)
572
if "," in subsegment:
573
raise InvalidURLJoin(", exists in subsegments",
575
return ",".join((base,) + subsegments)
578
def join_segment_parameters(url, parameters):
579
"""Create a new URL by adding segment parameters to an existing one.
581
The parameters of the last segment in the URL will be updated; if a
582
parameter with the same key already exists it will be overwritten.
584
:param url: A URL, as string
585
:param parameters: Dictionary of parameters, keys and values as bytestrings
587
(base, existing_parameters) = split_segment_parameters(url)
589
new_parameters.update(existing_parameters)
590
for key, value in parameters.items():
591
if not isinstance(key, str):
592
raise TypeError("parameter key %r is not a bytestring" % key)
593
if not isinstance(value, str):
594
raise TypeError("parameter value %r for %s is not a bytestring" %
597
raise InvalidURLJoin("= exists in parameter key", url,
599
new_parameters[key] = value
600
return join_segment_parameters_raw(base,
601
*["%s=%s" % item for item in sorted(new_parameters.items())])
472
604
def _win32_strip_local_trailing_slash(url):
473
605
"""Strip slashes after the drive letter"""
474
606
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
524
656
This returns a Unicode path from a URL
526
658
# jam 20060427 URLs are supposed to be ASCII only strings
527
# If they are passed in as unicode, urllib.unquote
659
# If they are passed in as unicode, unquote
528
660
# will return a UNICODE string, which actually contains
529
661
# utf-8 bytes. So we have to ensure that they are
530
662
# plain ASCII strings, or the final .decode will
531
663
# 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)
665
if isinstance(url, text_type):
667
url = url.encode("ascii")
668
except UnicodeError as e:
669
raise InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
671
unquoted = urlparse.unquote_to_bytes(url)
673
unquoted = unquote(url)
540
675
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,))
676
except UnicodeError as e:
677
raise InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
543
678
return unicode_path
693
828
return osutils.pathjoin(*segments)
834
def __init__(self, scheme, quoted_user, quoted_password, quoted_host,
837
self.quoted_host = quoted_host
838
self.host = unquote(self.quoted_host)
839
self.quoted_user = quoted_user
840
if self.quoted_user is not None:
841
self.user = unquote(self.quoted_user)
844
self.quoted_password = quoted_password
845
if self.quoted_password is not None:
846
self.password = unquote(self.quoted_password)
850
self.quoted_path = _url_hex_escapes_re.sub(_unescape_safe_chars, quoted_path)
851
self.path = unquote(self.quoted_path)
853
def __eq__(self, other):
854
return (isinstance(other, self.__class__) and
855
self.scheme == other.scheme and
856
self.host == other.host and
857
self.user == other.user and
858
self.password == other.password and
859
self.path == other.path)
862
return "<%s(%r, %r, %r, %r, %r, %r)>" % (
863
self.__class__.__name__,
864
self.scheme, self.quoted_user, self.quoted_password,
865
self.quoted_host, self.port, self.quoted_path)
868
def from_string(cls, url):
869
"""Create a URL object from a string.
871
:param url: URL as bytestring
873
# GZ 2017-06-09: Actually validate ascii-ness
874
if not isinstance(url, str):
875
raise InvalidURL('should be ascii:\n%r' % url)
876
(scheme, netloc, path, params,
877
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
878
user = password = host = port = None
880
user, host = netloc.rsplit('@', 1)
882
user, password = user.split(':', 1)
886
if ':' in host and not (host[0] == '[' and host[-1] == ']'):
888
host, port = host.rsplit(':', 1)
892
raise InvalidURL('invalid port number %s in url:\n%s' %
894
if host != "" and host[0] == '[' and host[-1] == ']': #IPv6
897
return cls(scheme, user, password, host, port, path)
900
netloc = self.quoted_host
902
netloc = "[%s]" % netloc
903
if self.quoted_user is not None:
904
# Note that we don't put the password back even if we
905
# have one so that it doesn't get accidentally
907
netloc = '%s@%s' % (self.quoted_user, netloc)
908
if self.port is not None:
909
netloc = '%s:%d' % (netloc, self.port)
910
return urlparse.urlunparse(
911
(self.scheme, netloc, self.quoted_path, None, None, None))
914
def _combine_paths(base_path, relpath):
915
"""Transform a Transport-relative path to a remote absolute path.
917
This does not handle substitution of ~ but does handle '..' and '.'
922
t._combine_paths('/home/sarah', 'project/foo')
923
=> '/home/sarah/project/foo'
924
t._combine_paths('/home/sarah', '../../etc')
926
t._combine_paths('/home/sarah', '/etc')
929
:param base_path: base path
930
:param relpath: relative url string for relative part of remote path.
931
:return: urlencoded string for final path.
933
if not isinstance(relpath, str):
934
raise InvalidURL(relpath)
935
relpath = _url_hex_escapes_re.sub(_unescape_safe_chars, relpath)
936
if relpath.startswith('/'):
939
base_parts = base_path.split('/')
940
if len(base_parts) > 0 and base_parts[-1] == '':
941
base_parts = base_parts[:-1]
942
for p in relpath.split('/'):
944
if len(base_parts) == 0:
945
# In most filesystems, a request for the parent
946
# of root, just returns root.
953
path = '/'.join(base_parts)
954
if not path.startswith('/'):
958
def clone(self, offset=None):
959
"""Return a new URL for a path relative to this URL.
961
:param offset: A relative path, already urlencoded
962
:return: `URL` instance
964
if offset is not None:
965
relative = unescape(offset).encode('utf-8')
966
path = self._combine_paths(self.path, relative)
967
path = quote(path, safe="/~")
969
path = self.quoted_path
970
return self.__class__(self.scheme, self.quoted_user,
971
self.quoted_password, self.quoted_host, self.port,
697
975
def parse_url(url):
698
976
"""Extract the server address, the credentials and the path from the url.
703
981
:param url: an quoted url
705
982
: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)
985
parsed_url = URL.from_string(url)
986
return (parsed_url.scheme, parsed_url.user, parsed_url.password,
987
parsed_url.host, parsed_url.port, parsed_url.path)