121
111
join('http://foo', 'bar') => 'http://foo/bar'
122
112
join('http://foo', 'bar', '../baz') => 'http://foo/baz'
126
scheme_end, path_start = _find_scheme_and_separator(base)
127
if scheme_end is None and path_start is None:
129
elif path_start is None:
130
path_start = len(base)
131
path = base[path_start:]
114
m = _url_scheme_re.match(base)
117
scheme = m.group('scheme')
118
path = m.group('path').split('/')
120
path = base.split('/')
133
arg_scheme_end, arg_path_start = _find_scheme_and_separator(arg)
134
if arg_scheme_end is None and arg_path_start is None:
136
elif arg_path_start is None:
137
arg_path_start = len(arg)
138
if arg_scheme_end is not None:
140
path = arg[arg_path_start:]
141
scheme_end = arg_scheme_end
142
path_start = arg_path_start
123
m = _url_scheme_re.match(arg)
126
scheme = m.group('scheme')
127
path = m.group('path').split('/')
144
path = joinpath(path, arg)
145
return base[:path_start] + path
148
def joinpath(base, *args):
149
"""Join URL path segments to a URL path segment.
151
This is somewhat like osutils.joinpath, but intended for URLs.
153
XXX: this duplicates some normalisation logic, and also duplicates a lot of
154
path handling logic that already exists in some Transport implementations.
155
We really should try to have exactly one place in the code base responsible
156
for combining paths of URLs.
158
path = base.split('/')
159
if len(path) > 1 and path[-1] == '':
160
#If the path ends in a trailing /, remove it.
163
if arg.startswith('/'):
165
for chunk in arg.split('/'):
170
raise errors.InvalidURLJoin('Cannot go above root',
129
for chunk in arg.split('/'):
134
# Don't pop off the host portion
137
raise errors.InvalidURLJoin('Cannot go above root',
178
143
return '/'.join(path)
144
return scheme + '://' + '/'.join(path)
181
147
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
182
148
def _posix_local_path_from_url(url):
183
149
"""Convert a url like file:///path/to/foo into /path/to/foo"""
184
url = split_segment_parameters_raw(url)[0]
185
file_localhost_prefix = 'file://localhost/'
186
if url.startswith(file_localhost_prefix):
187
path = url[len(file_localhost_prefix) - 1:]
188
elif not url.startswith('file:///'):
189
raise errors.InvalidURL(
190
url, 'local urls must start with file:/// or file://localhost/')
192
path = url[len('file://'):]
150
if not url.startswith('file:///'):
151
raise errors.InvalidURL(url, 'local urls must start with file:///')
193
152
# We only strip off 2 slashes
194
return unescape(path)
153
return unescape(url[len('file://'):])
197
156
def _posix_local_path_to_url(path):
200
159
This also handles transforming escaping unicode characters, etc.
202
# importing directly from posixpath allows us to test this
161
# importing directly from posixpath allows us to test this
203
162
# on non-posix platforms
204
163
return 'file://' + escape(_posix_normpath(
205
osutils._posix_abspath(path)))
164
bzrlib.osutils._posix_abspath(path)))
208
167
def _win32_local_path_from_url(url):
209
168
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
210
if not url.startswith('file://'):
211
raise errors.InvalidURL(url, 'local urls must start with file:///, '
212
'UNC path urls must start with file://')
213
url = split_segment_parameters_raw(url)[0]
169
if not url.startswith('file:///'):
170
raise errors.InvalidURL(url, 'local urls must start with file:///')
214
171
# We strip off all 3 slashes
215
win32_url = url[len('file:'):]
216
# check for UNC path: //HOST/path
217
if not win32_url.startswith('///'):
218
if (win32_url[2] == '/'
219
or win32_url[3] in '|:'):
220
raise errors.InvalidURL(url, 'Win32 UNC path urls'
221
' have form file://HOST/path')
222
return unescape(win32_url)
224
# allow empty paths so we can serve all roots
225
if win32_url == '///':
228
# usual local path with drive letter
229
if (len(win32_url) < 6
230
or win32_url[3] not in ('abcdefghijklmnopqrstuvwxyz'
231
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
232
or win32_url[4] not in '|:'
233
or win32_url[5] != '/'):
172
win32_url = url[len('file:///'):]
173
if (win32_url[0] not in ('abcdefghijklmnopqrstuvwxyz'
174
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
175
or win32_url[1] not in '|:'
176
or win32_url[2] != '/'):
234
177
raise errors.InvalidURL(url, 'Win32 file urls start with'
235
' file:///x:/, where x is a valid drive letter')
236
return win32_url[3].upper() + u':' + unescape(win32_url[5:])
178
' file:///X:/, where X is a valid drive letter')
179
return win32_url[0].upper() + u':' + unescape(win32_url[2:])
239
182
def _win32_local_path_to_url(path):
242
185
This also handles transforming escaping unicode characters, etc.
244
# importing directly from ntpath allows us to test this
187
# importing directly from ntpath allows us to test this
245
188
# on non-win32 platform
246
189
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
247
190
# which actually strips trailing space characters.
248
# The worst part is that on linux ntpath.abspath has different
191
# The worst part is that under linux ntpath.abspath has different
249
192
# semantics, since 'nt' is not an available module.
253
win32_path = osutils._win32_abspath(path)
254
# check for UNC path \\HOST\path
255
if win32_path.startswith('//'):
256
return 'file:' + escape(win32_path)
257
return ('file:///' + str(win32_path[0].upper()) + ':' +
258
escape(win32_path[2:]))
193
win32_path = bzrlib.osutils._nt_normpath(
194
bzrlib.osutils._win32_abspath(path)).replace('\\', '/')
195
return 'file:///' + win32_path[0].upper() + ':' + escape(win32_path[2:])
261
198
local_path_to_url = _posix_local_path_to_url
262
199
local_path_from_url = _posix_local_path_from_url
263
200
MIN_ABS_FILEURL_LENGTH = len('file:///')
264
WIN32_MIN_ABS_FILEURL_LENGTH = len('file:///C:/')
266
202
if sys.platform == 'win32':
267
203
local_path_to_url = _win32_local_path_to_url
268
204
local_path_from_url = _win32_local_path_from_url
270
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
273
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,}):(//)?(?P<path>.*)$')
274
_url_hex_escapes_re = re.compile(r'(%[0-9a-fA-F]{2})')
277
def _unescape_safe_chars(matchobj):
278
"""re.sub callback to convert hex-escapes to plain characters (if safe).
280
e.g. '%7E' will be converted to '~'.
282
hex_digits = matchobj.group(0)[1:]
283
char = chr(int(hex_digits, 16))
284
if char in _url_dont_escape_characters:
287
return matchobj.group(0).upper()
206
MIN_ABS_FILEURL_LENGTH = len('file:///C:/')
209
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
290
212
def normalize_url(url):
291
213
"""Make sure that a path string is in fully normalized URL form.
293
This handles URLs which have unicode characters, spaces,
215
This handles URLs which have unicode characters, spaces,
294
216
special characters, etc.
296
218
It has two basic modes of operation, depending on whether the
306
228
:param url: Either a hybrid URL or a local path
307
229
:return: A normalized URL which only includes 7-bit ASCII characters.
309
scheme_end, path_start = _find_scheme_and_separator(url)
310
if scheme_end is None:
231
m = _url_scheme_re.match(url)
311
233
return local_path_to_url(url)
312
prefix = url[:path_start]
313
path = url[path_start:]
314
234
if not isinstance(url, unicode):
316
236
if c not in _url_safe_characters:
317
237
raise errors.InvalidURL(url, 'URLs can only contain specific'
318
238
' safe characters (not %r)' % c)
319
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
320
return str(prefix + ''.join(path))
322
240
# We have a unicode (hybrid) url
323
path_chars = list(path)
241
scheme = m.group('scheme')
242
path = list(m.group('path'))
325
for i in xrange(len(path_chars)):
326
if path_chars[i] not in _url_safe_characters:
327
chars = path_chars[i].encode('utf-8')
328
path_chars[i] = ''.join(
329
['%%%02X' % ord(c) for c in path_chars[i].encode('utf-8')])
330
path = ''.join(path_chars)
331
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
332
return str(prefix + path)
244
for i in xrange(len(path)):
245
if path[i] not in _url_safe_characters:
246
chars = path[i].encode('utf-8')
247
path[i] = ''.join(['%%%02X' % ord(c) for c in path[i].encode('utf-8')])
248
return scheme + '://' + ''.join(path)
335
251
def relative_url(base, other):
436
337
return url_base + head, tail
439
def split_segment_parameters_raw(url):
440
"""Split the subsegment of the last segment of a URL.
442
:param url: A relative or absolute URL
443
:return: (url, subsegments)
445
(parent_url, child_dir) = split(url)
446
subsegments = child_dir.split(",")
447
if len(subsegments) == 1:
449
return (join(parent_url, subsegments[0]), subsegments[1:])
452
def split_segment_parameters(url):
453
"""Split the segment parameters of the last segment of a URL.
455
:param url: A relative or absolute URL
456
:return: (url, segment_parameters)
458
(base_url, subsegments) = split_segment_parameters_raw(url)
460
for subsegment in subsegments:
461
(key, value) = subsegment.split("=", 1)
462
parameters[key] = value
463
return (base_url, parameters)
466
def join_segment_parameters_raw(base, *subsegments):
467
"""Create a new URL by adding subsegments to an existing one.
469
This adds the specified subsegments to the last path in the specified
470
base URL. The subsegments should be bytestrings.
472
:note: You probably want to use join_segment_parameters instead.
476
for subsegment in subsegments:
477
if type(subsegment) is not str:
478
raise TypeError("Subsegment %r is not a bytestring" % subsegment)
479
if "," in subsegment:
480
raise errors.InvalidURLJoin(", exists in subsegments",
482
return ",".join((base,) + subsegments)
485
def join_segment_parameters(url, parameters):
486
"""Create a new URL by adding segment parameters to an existing one.
488
The parameters of the last segment in the URL will be updated; if a
489
parameter with the same key already exists it will be overwritten.
491
:param url: A URL, as string
492
:param parameters: Dictionary of parameters, keys and values as bytestrings
494
(base, existing_parameters) = split_segment_parameters(url)
496
new_parameters.update(existing_parameters)
497
for key, value in parameters.iteritems():
498
if type(key) is not str:
499
raise TypeError("parameter key %r is not a bytestring" % key)
500
if type(value) is not str:
501
raise TypeError("parameter value %r for %s is not a bytestring" %
504
raise errors.InvalidURLJoin("= exists in parameter key", url,
506
new_parameters[key] = value
507
return join_segment_parameters_raw(base,
508
*["%s=%s" % item for item in sorted(new_parameters.items())])
511
def _win32_strip_local_trailing_slash(url):
512
"""Strip slashes after the drive letter"""
513
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
519
340
def strip_trailing_slash(url):
520
341
"""Strip trailing slash, except for root paths.
585
411
# These are characters that if escaped, should stay that way
586
412
_no_decode_chars = ';/?:@&=+$,#'
587
413
_no_decode_ords = [ord(c) for c in _no_decode_chars]
588
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
414
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
589
415
+ ['%02X' % o for o in _no_decode_ords])
590
416
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
591
417
+ [('%02X' % o, chr(o)) for o in range(256)]))
592
418
#These entries get mapped to themselves
593
419
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
595
# These characters shouldn't be percent-encoded, and it's always safe to
596
# unencode them if they are.
597
_url_dont_escape_characters = set(
598
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
599
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
600
"0123456789" # Numbers
601
"-._~" # Unreserved characters
604
421
# These characters should not be escaped
605
_url_safe_characters = set(
606
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
607
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
608
"0123456789" # Numbers
609
"_.-!~*'()" # Unreserved characters
610
"/;?:@&=+$," # Reserved characters
611
"%#" # Extra reserved characters
422
_url_safe_characters = set('abcdefghijklmnopqrstuvwxyz'
423
'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
614
428
def unescape_for_display(url, encoding):
615
429
"""Decode what you can for a URL, so that we get a nice looking path.
667
480
# Otherwise take the url decoded one
669
482
return u'/'.join(res)
672
def derive_to_location(from_location):
673
"""Derive a TO_LOCATION given a FROM_LOCATION.
675
The normal case is a FROM_LOCATION of http://foo/bar => bar.
676
The Right Thing for some logical destinations may differ though
677
because no / may be present at all. In that case, the result is
678
the full name without the scheme indicator, e.g. lp:foo-bar => foo-bar.
679
This latter case also applies when a Windows drive
680
is used without a path, e.g. c:foo-bar => foo-bar.
681
If no /, path separator or : is found, the from_location is returned.
683
if from_location.find("/") >= 0 or from_location.find(os.sep) >= 0:
684
return os.path.basename(from_location.rstrip("/\\"))
686
sep = from_location.find(":")
688
return from_location[sep+1:]
693
def _is_absolute(url):
694
return (osutils.pathjoin('/foo', url) == url)
697
def rebase_url(url, old_base, new_base):
698
"""Convert a relative path from an old base URL to a new base URL.
700
The result will be a relative path.
701
Absolute paths and full URLs are returned unaltered.
703
scheme, separator = _find_scheme_and_separator(url)
704
if scheme is not None:
706
if _is_absolute(url):
708
old_parsed = urlparse.urlparse(old_base)
709
new_parsed = urlparse.urlparse(new_base)
710
if (old_parsed[:2]) != (new_parsed[:2]):
711
raise errors.InvalidRebaseURLs(old_base, new_base)
712
return determine_relative_path(new_parsed[2],
713
join(old_parsed[2], url))
716
def determine_relative_path(from_path, to_path):
717
"""Determine a relative path from from_path to to_path."""
718
from_segments = osutils.splitpath(from_path)
719
to_segments = osutils.splitpath(to_path)
721
for count, (from_element, to_element) in enumerate(zip(from_segments,
723
if from_element != to_element:
727
unique_from = from_segments[count:]
728
unique_to = to_segments[count:]
729
segments = (['..'] * len(unique_from) + unique_to)
730
if len(segments) == 0:
732
return osutils.pathjoin(*segments)
738
def __init__(self, scheme, quoted_user, quoted_password, quoted_host,
741
self.quoted_host = quoted_host
742
self.host = urllib.unquote(self.quoted_host)
743
self.quoted_user = quoted_user
744
if self.quoted_user is not None:
745
self.user = urllib.unquote(self.quoted_user)
748
self.quoted_password = quoted_password
749
if self.quoted_password is not None:
750
self.password = urllib.unquote(self.quoted_password)
754
self.quoted_path = quoted_path
755
self.path = urllib.unquote(self.quoted_path)
757
def __eq__(self, other):
758
return (isinstance(other, self.__class__) and
759
self.scheme == other.scheme and
760
self.host == other.host and
761
self.user == other.user and
762
self.password == other.password and
763
self.path == other.path)
766
return "<%s(%r, %r, %r, %r, %r, %r)>" % (
767
self.__class__.__name__,
768
self.scheme, self.quoted_user, self.quoted_password,
769
self.quoted_host, self.port, self.quoted_path)
772
def from_string(cls, url):
773
"""Create a URL object from a string.
775
:param url: URL as bytestring
777
if isinstance(url, unicode):
778
raise errors.InvalidURL('should be ascii:\n%r' % url)
779
url = url.encode('utf-8')
780
(scheme, netloc, path, params,
781
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
782
user = password = host = port = None
784
user, host = netloc.rsplit('@', 1)
786
user, password = user.split(':', 1)
790
if ':' in host and not (host[0] == '[' and host[-1] == ']'):
792
host, port = host.rsplit(':',1)
796
raise errors.InvalidURL('invalid port number %s in url:\n%s' %
798
if host != "" and host[0] == '[' and host[-1] == ']': #IPv6
801
return cls(scheme, user, password, host, port, path)
804
netloc = self.quoted_host
806
netloc = "[%s]" % netloc
807
if self.quoted_user is not None:
808
# Note that we don't put the password back even if we
809
# have one so that it doesn't get accidentally
811
netloc = '%s@%s' % (self.quoted_user, netloc)
812
if self.port is not None:
813
netloc = '%s:%d' % (netloc, self.port)
814
return urlparse.urlunparse(
815
(self.scheme, netloc, self.quoted_path, None, None, None))
818
def _combine_paths(base_path, relpath):
819
"""Transform a Transport-relative path to a remote absolute path.
821
This does not handle substitution of ~ but does handle '..' and '.'
826
t._combine_paths('/home/sarah', 'project/foo')
827
=> '/home/sarah/project/foo'
828
t._combine_paths('/home/sarah', '../../etc')
830
t._combine_paths('/home/sarah', '/etc')
833
:param base_path: base path
834
:param relpath: relative url string for relative part of remote path.
835
:return: urlencoded string for final path.
837
if not isinstance(relpath, str):
838
raise errors.InvalidURL(relpath)
839
if relpath.startswith('/'):
842
base_parts = base_path.split('/')
843
if len(base_parts) > 0 and base_parts[-1] == '':
844
base_parts = base_parts[:-1]
845
for p in relpath.split('/'):
847
if len(base_parts) == 0:
848
# In most filesystems, a request for the parent
849
# of root, just returns root.
856
path = '/'.join(base_parts)
857
if not path.startswith('/'):
861
def clone(self, offset=None):
862
"""Return a new URL for a path relative to this URL.
864
:param offset: A relative path, already urlencoded
865
:return: `URL` instance
867
if offset is not None:
868
relative = unescape(offset).encode('utf-8')
869
path = self._combine_paths(self.path, relative)
870
path = urllib.quote(path)
872
path = self.quoted_path
873
return self.__class__(self.scheme, self.quoted_user,
874
self.quoted_password, self.quoted_host, self.port,
879
"""Extract the server address, the credentials and the path from the url.
881
user, password, host and path should be quoted if they contain reserved
884
:param url: an quoted url
885
:return: (scheme, user, password, host, port, path) tuple, all fields
888
parsed_url = URL.from_string(url)
889
return (parsed_url.scheme, parsed_url.user, parsed_url.password,
890
parsed_url.host, parsed_url.port, parsed_url.path)