61
61
return split(url, exclude_trailing_slash=exclude_trailing_slash)[0]
64
# Private copies of quote and unquote, copied from Python's
65
# urllib module because urllib unconditionally imports socket, which imports
68
always_safe = ('ABCDEFGHIJKLMNOPQRSTUVWXYZ'
69
'abcdefghijklmnopqrstuvwxyz'
72
for i, c in zip(xrange(256), str(bytearray(xrange(256)))):
73
_safe_map[c] = c if (i < 128 and c in always_safe) else '%{0:02X}'.format(i)
77
def quote(s, safe='/'):
78
"""quote('abc def') -> 'abc%20def'
80
Each part of a URL, e.g. the path info, the query, etc., has a
81
different set of reserved characters that must be quoted.
83
RFC 2396 Uniform Resource Identifiers (URI): Generic Syntax lists
84
the following reserved characters.
86
reserved = ";" | "/" | "?" | ":" | "@" | "&" | "=" | "+" |
89
Each of these characters is reserved in some component of a URL,
90
but not necessarily in all of them.
92
By default, the quote function is intended for quoting the path
93
section of a URL. Thus, it will not encode '/'. This character
94
is reserved, but in typical usage the quote function is being
95
called on a path where the existing slash characters are used as
101
raise TypeError('None object cannot be quoted')
103
cachekey = (safe, always_safe)
105
(quoter, safe) = _safe_quoters[cachekey]
107
safe_map = _safe_map.copy()
108
safe_map.update([(c, c) for c in safe])
109
quoter = safe_map.__getitem__
110
safe = always_safe + safe
111
_safe_quoters[cachekey] = (quoter, safe)
112
if not s.rstrip(safe):
114
return ''.join(map(quoter, s))
117
_hexdig = '0123456789ABCDEFabcdef'
118
_hextochr = dict((a + b, chr(int(a + b, 16)))
119
for a in _hexdig for b in _hexdig)
122
"""unquote('abc%20def') -> 'abc def'."""
130
s += _hextochr[item[:2]] + item[2:]
133
except UnicodeDecodeError:
134
s += unichr(int(item[:2], 16)) + item[2:]
138
64
def escape(relpath):
139
65
"""Escape relpath to be a valid url."""
140
66
if isinstance(relpath, unicode):
141
67
relpath = relpath.encode('utf-8')
142
68
# After quoting and encoding, the path should be perfectly
143
69
# safe as a plain ASCII string, str() just enforces this
144
return str(quote(relpath, safe='/~'))
70
return str(urllib.quote(relpath))
147
73
def file_relpath(base, path):
148
74
"""Compute just the relative sub-portion of a url
150
76
This assumes that both paths are already fully specified file:// URLs.
152
if len(base) < MIN_ABS_FILEURL_LENGTH:
153
raise ValueError('Length of base (%r) must equal or'
154
' exceed the platform minimum url length (which is %d)' %
155
(base, MIN_ABS_FILEURL_LENGTH))
156
base = osutils.normpath(local_path_from_url(base))
157
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)
158
84
return escape(osutils.relpath(base, path))
196
117
join('http://foo', 'bar') => 'http://foo/bar'
197
118
join('http://foo', 'bar', '../baz') => 'http://foo/baz'
201
scheme_end, path_start = _find_scheme_and_separator(base)
202
if scheme_end is None and path_start is None:
204
elif path_start is None:
205
path_start = len(base)
206
path = base[path_start:]
120
m = _url_scheme_re.match(base)
123
scheme = m.group('scheme')
124
path = m.group('path').split('/')
125
if path[-1:] == ['']:
126
# Strip off a trailing slash
127
# This helps both when we are at the root, and when
128
# 'base' has an extra slash at the end
131
path = base.split('/')
208
arg_scheme_end, arg_path_start = _find_scheme_and_separator(arg)
209
if arg_scheme_end is None and arg_path_start is None:
211
elif arg_path_start is None:
212
arg_path_start = len(arg)
213
if arg_scheme_end is not None:
215
path = arg[arg_path_start:]
216
scheme_end = arg_scheme_end
217
path_start = arg_path_start
134
m = _url_scheme_re.match(arg)
137
scheme = m.group('scheme')
138
# this skips .. normalisation, making http://host/../../..
140
path = m.group('path').split('/')
219
path = joinpath(path, arg)
220
return base[:path_start] + path
223
def joinpath(base, *args):
224
"""Join URL path segments to a URL path segment.
226
This is somewhat like osutils.joinpath, but intended for URLs.
228
XXX: this duplicates some normalisation logic, and also duplicates a lot of
229
path handling logic that already exists in some Transport implementations.
230
We really should try to have exactly one place in the code base responsible
231
for combining paths of URLs.
233
path = base.split('/')
234
if len(path) > 1 and path[-1] == '':
235
#If the path ends in a trailing /, remove it.
238
if arg.startswith('/'):
240
for chunk in arg.split('/'):
245
raise errors.InvalidURLJoin('Cannot go above root',
142
for chunk in arg.split('/'):
147
# Don't pop off the host portion
150
raise errors.InvalidURLJoin('Cannot go above root',
253
156
return '/'.join(path)
157
return scheme + '://' + '/'.join(path)
256
160
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
257
161
def _posix_local_path_from_url(url):
258
162
"""Convert a url like file:///path/to/foo into /path/to/foo"""
259
url = split_segment_parameters_raw(url)[0]
260
file_localhost_prefix = 'file://localhost/'
261
if url.startswith(file_localhost_prefix):
262
path = url[len(file_localhost_prefix) - 1:]
263
elif not url.startswith('file:///'):
264
raise errors.InvalidURL(
265
url, 'local urls must start with file:/// or file://localhost/')
267
path = url[len('file://'):]
163
if not url.startswith('file:///'):
164
raise errors.InvalidURL(url, 'local urls must start with file:///')
268
165
# We only strip off 2 slashes
269
return unescape(path)
166
return unescape(url[len('file://'):])
272
169
def _posix_local_path_to_url(path):
275
172
This also handles transforming escaping unicode characters, etc.
277
# importing directly from posixpath allows us to test this
174
# importing directly from posixpath allows us to test this
278
175
# on non-posix platforms
279
return 'file://' + escape(osutils._posix_abspath(path))
176
return 'file://' + escape(_posix_normpath(
177
osutils._posix_abspath(path)))
282
180
def _win32_local_path_from_url(url):
283
181
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
284
if not url.startswith('file://'):
285
raise errors.InvalidURL(url, 'local urls must start with file:///, '
286
'UNC path urls must start with file://')
287
url = split_segment_parameters_raw(url)[0]
182
if not url.startswith('file:///'):
183
raise errors.InvalidURL(url, 'local urls must start with file:///')
288
184
# We strip off all 3 slashes
289
win32_url = url[len('file:'):]
290
# check for UNC path: //HOST/path
291
if not win32_url.startswith('///'):
292
if (win32_url[2] == '/'
293
or win32_url[3] in '|:'):
294
raise errors.InvalidURL(url, 'Win32 UNC path urls'
295
' have form file://HOST/path')
296
return unescape(win32_url)
298
# allow empty paths so we can serve all roots
299
if win32_url == '///':
302
# usual local path with drive letter
303
if (len(win32_url) < 6
304
or win32_url[3] not in ('abcdefghijklmnopqrstuvwxyz'
305
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
306
or win32_url[4] not in '|:'
307
or win32_url[5] != '/'):
185
win32_url = url[len('file:///'):]
186
if (win32_url[0] not in ('abcdefghijklmnopqrstuvwxyz'
187
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
188
or win32_url[1] not in '|:'
189
or win32_url[2] != '/'):
308
190
raise errors.InvalidURL(url, 'Win32 file urls start with'
309
191
' file:///x:/, where x is a valid drive letter')
310
return win32_url[3].upper() + u':' + unescape(win32_url[5:])
192
return win32_url[0].upper() + u':' + unescape(win32_url[2:])
313
195
def _win32_local_path_to_url(path):
380
242
:param url: Either a hybrid URL or a local path
381
243
:return: A normalized URL which only includes 7-bit ASCII characters.
383
scheme_end, path_start = _find_scheme_and_separator(url)
384
if scheme_end is None:
245
m = _url_scheme_re.match(url)
385
247
return local_path_to_url(url)
386
prefix = url[:path_start]
387
path = url[path_start:]
388
248
if not isinstance(url, unicode):
390
250
if c not in _url_safe_characters:
391
251
raise errors.InvalidURL(url, 'URLs can only contain specific'
392
252
' safe characters (not %r)' % c)
393
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
394
return str(prefix + ''.join(path))
396
254
# We have a unicode (hybrid) url
397
path_chars = list(path)
255
scheme = m.group('scheme')
256
path = list(m.group('path'))
399
for i in xrange(len(path_chars)):
400
if path_chars[i] not in _url_safe_characters:
401
chars = path_chars[i].encode('utf-8')
402
path_chars[i] = ''.join(
403
['%%%02X' % ord(c) for c in path_chars[i].encode('utf-8')])
404
path = ''.join(path_chars)
405
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
406
return str(prefix + path)
258
for i in xrange(len(path)):
259
if path[i] not in _url_safe_characters:
260
chars = path[i].encode('utf-8')
261
path[i] = ''.join(['%%%02X' % ord(c) for c in path[i].encode('utf-8')])
262
return scheme + '://' + ''.join(path)
409
265
def relative_url(base, other):
510
361
return url_base + head, tail
513
def split_segment_parameters_raw(url):
514
"""Split the subsegment of the last segment of a URL.
516
:param url: A relative or absolute URL
517
:return: (url, subsegments)
519
# GZ 2011-11-18: Dodgy removing the terminal slash like this, function
520
# operates on urls not url+segments, and Transport classes
521
# should not be blindly adding slashes in the first place.
522
lurl = strip_trailing_slash(url)
523
# Segments begin at first comma after last forward slash, if one exists
524
segment_start = lurl.find(",", lurl.rfind("/")+1)
525
if segment_start == -1:
527
return (lurl[:segment_start], lurl[segment_start+1:].split(","))
530
def split_segment_parameters(url):
531
"""Split the segment parameters of the last segment of a URL.
533
:param url: A relative or absolute URL
534
:return: (url, segment_parameters)
536
(base_url, subsegments) = split_segment_parameters_raw(url)
538
for subsegment in subsegments:
539
(key, value) = subsegment.split("=", 1)
540
parameters[key] = value
541
return (base_url, parameters)
544
def join_segment_parameters_raw(base, *subsegments):
545
"""Create a new URL by adding subsegments to an existing one.
547
This adds the specified subsegments to the last path in the specified
548
base URL. The subsegments should be bytestrings.
550
:note: You probably want to use join_segment_parameters instead.
554
for subsegment in subsegments:
555
if type(subsegment) is not str:
556
raise TypeError("Subsegment %r is not a bytestring" % subsegment)
557
if "," in subsegment:
558
raise errors.InvalidURLJoin(", exists in subsegments",
560
return ",".join((base,) + subsegments)
563
def join_segment_parameters(url, parameters):
564
"""Create a new URL by adding segment parameters to an existing one.
566
The parameters of the last segment in the URL will be updated; if a
567
parameter with the same key already exists it will be overwritten.
569
:param url: A URL, as string
570
:param parameters: Dictionary of parameters, keys and values as bytestrings
572
(base, existing_parameters) = split_segment_parameters(url)
574
new_parameters.update(existing_parameters)
575
for key, value in parameters.iteritems():
576
if type(key) is not str:
577
raise TypeError("parameter key %r is not a bytestring" % key)
578
if type(value) is not str:
579
raise TypeError("parameter value %r for %s is not a bytestring" %
582
raise errors.InvalidURLJoin("= exists in parameter key", url,
584
new_parameters[key] = value
585
return join_segment_parameters_raw(base,
586
*["%s=%s" % item for item in sorted(new_parameters.items())])
589
364
def _win32_strip_local_trailing_slash(url):
590
365
"""Strip slashes after the drive letter"""
591
366
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
663
438
# These are characters that if escaped, should stay that way
664
439
_no_decode_chars = ';/?:@&=+$,#'
665
440
_no_decode_ords = [ord(c) for c in _no_decode_chars]
666
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
441
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
667
442
+ ['%02X' % o for o in _no_decode_ords])
668
443
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
669
444
+ [('%02X' % o, chr(o)) for o in range(256)]))
670
445
#These entries get mapped to themselves
671
446
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
673
# These characters shouldn't be percent-encoded, and it's always safe to
674
# unencode them if they are.
675
_url_dont_escape_characters = set(
676
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
677
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
678
"0123456789" # Numbers
679
"-._~" # Unreserved characters
682
448
# These characters should not be escaped
683
_url_safe_characters = set(
684
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
685
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
686
"0123456789" # Numbers
687
"_.-!~*'()" # Unreserved characters
688
"/;?:@&=+$," # Reserved characters
689
"%#" # Extra reserved characters
449
_url_safe_characters = set('abcdefghijklmnopqrstuvwxyz'
450
'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
692
455
def unescape_for_display(url, encoding):
693
456
"""Decode what you can for a URL, so that we get a nice looking path.
745
507
# Otherwise take the url decoded one
747
509
return u'/'.join(res)
750
def derive_to_location(from_location):
751
"""Derive a TO_LOCATION given a FROM_LOCATION.
753
The normal case is a FROM_LOCATION of http://foo/bar => bar.
754
The Right Thing for some logical destinations may differ though
755
because no / may be present at all. In that case, the result is
756
the full name without the scheme indicator, e.g. lp:foo-bar => foo-bar.
757
This latter case also applies when a Windows drive
758
is used without a path, e.g. c:foo-bar => foo-bar.
759
If no /, path separator or : is found, the from_location is returned.
761
if from_location.find("/") >= 0 or from_location.find(os.sep) >= 0:
762
return os.path.basename(from_location.rstrip("/\\"))
764
sep = from_location.find(":")
766
return from_location[sep+1:]
771
def _is_absolute(url):
772
return (osutils.pathjoin('/foo', url) == url)
775
def rebase_url(url, old_base, new_base):
776
"""Convert a relative path from an old base URL to a new base URL.
778
The result will be a relative path.
779
Absolute paths and full URLs are returned unaltered.
781
scheme, separator = _find_scheme_and_separator(url)
782
if scheme is not None:
784
if _is_absolute(url):
786
old_parsed = urlparse.urlparse(old_base)
787
new_parsed = urlparse.urlparse(new_base)
788
if (old_parsed[:2]) != (new_parsed[:2]):
789
raise errors.InvalidRebaseURLs(old_base, new_base)
790
return determine_relative_path(new_parsed[2],
791
join(old_parsed[2], url))
794
def determine_relative_path(from_path, to_path):
795
"""Determine a relative path from from_path to to_path."""
796
from_segments = osutils.splitpath(from_path)
797
to_segments = osutils.splitpath(to_path)
799
for count, (from_element, to_element) in enumerate(zip(from_segments,
801
if from_element != to_element:
805
unique_from = from_segments[count:]
806
unique_to = to_segments[count:]
807
segments = (['..'] * len(unique_from) + unique_to)
808
if len(segments) == 0:
810
return osutils.pathjoin(*segments)
816
def __init__(self, scheme, quoted_user, quoted_password, quoted_host,
819
self.quoted_host = quoted_host
820
self.host = unquote(self.quoted_host)
821
self.quoted_user = quoted_user
822
if self.quoted_user is not None:
823
self.user = unquote(self.quoted_user)
826
self.quoted_password = quoted_password
827
if self.quoted_password is not None:
828
self.password = unquote(self.quoted_password)
832
self.quoted_path = _url_hex_escapes_re.sub(_unescape_safe_chars, quoted_path)
833
self.path = unquote(self.quoted_path)
835
def __eq__(self, other):
836
return (isinstance(other, self.__class__) and
837
self.scheme == other.scheme and
838
self.host == other.host and
839
self.user == other.user and
840
self.password == other.password and
841
self.path == other.path)
844
return "<%s(%r, %r, %r, %r, %r, %r)>" % (
845
self.__class__.__name__,
846
self.scheme, self.quoted_user, self.quoted_password,
847
self.quoted_host, self.port, self.quoted_path)
850
def from_string(cls, url):
851
"""Create a URL object from a string.
853
:param url: URL as bytestring
855
if isinstance(url, unicode):
856
raise errors.InvalidURL('should be ascii:\n%r' % url)
857
url = url.encode('utf-8')
858
(scheme, netloc, path, params,
859
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
860
user = password = host = port = None
862
user, host = netloc.rsplit('@', 1)
864
user, password = user.split(':', 1)
868
if ':' in host and not (host[0] == '[' and host[-1] == ']'):
870
host, port = host.rsplit(':',1)
874
raise errors.InvalidURL('invalid port number %s in url:\n%s' %
876
if host != "" and host[0] == '[' and host[-1] == ']': #IPv6
879
return cls(scheme, user, password, host, port, path)
882
netloc = self.quoted_host
884
netloc = "[%s]" % netloc
885
if self.quoted_user is not None:
886
# Note that we don't put the password back even if we
887
# have one so that it doesn't get accidentally
889
netloc = '%s@%s' % (self.quoted_user, netloc)
890
if self.port is not None:
891
netloc = '%s:%d' % (netloc, self.port)
892
return urlparse.urlunparse(
893
(self.scheme, netloc, self.quoted_path, None, None, None))
896
def _combine_paths(base_path, relpath):
897
"""Transform a Transport-relative path to a remote absolute path.
899
This does not handle substitution of ~ but does handle '..' and '.'
904
t._combine_paths('/home/sarah', 'project/foo')
905
=> '/home/sarah/project/foo'
906
t._combine_paths('/home/sarah', '../../etc')
908
t._combine_paths('/home/sarah', '/etc')
911
:param base_path: base path
912
:param relpath: relative url string for relative part of remote path.
913
:return: urlencoded string for final path.
915
if not isinstance(relpath, str):
916
raise errors.InvalidURL(relpath)
917
relpath = _url_hex_escapes_re.sub(_unescape_safe_chars, relpath)
918
if relpath.startswith('/'):
921
base_parts = base_path.split('/')
922
if len(base_parts) > 0 and base_parts[-1] == '':
923
base_parts = base_parts[:-1]
924
for p in relpath.split('/'):
926
if len(base_parts) == 0:
927
# In most filesystems, a request for the parent
928
# of root, just returns root.
935
path = '/'.join(base_parts)
936
if not path.startswith('/'):
940
def clone(self, offset=None):
941
"""Return a new URL for a path relative to this URL.
943
:param offset: A relative path, already urlencoded
944
:return: `URL` instance
946
if offset is not None:
947
relative = unescape(offset).encode('utf-8')
948
path = self._combine_paths(self.path, relative)
949
path = quote(path, safe="/~")
951
path = self.quoted_path
952
return self.__class__(self.scheme, self.quoted_user,
953
self.quoted_password, self.quoted_host, self.port,
958
"""Extract the server address, the credentials and the path from the url.
960
user, password, host and path should be quoted if they contain reserved
963
:param url: an quoted url
964
:return: (scheme, user, password, host, port, path) tuple, all fields
967
parsed_url = URL.from_string(url)
968
return (parsed_url.scheme, parsed_url.user, parsed_url.password,
969
parsed_url.host, parsed_url.port, parsed_url.path)
added
removed
bzrlib/urlutils.py
