67
66
relpath = relpath.encode('utf-8')
68
67
# After quoting and encoding, the path should be perfectly
69
68
# safe as a plain ASCII string, str() just enforces this
70
return str(urllib.quote(relpath))
69
return str(urllib.quote(relpath, safe='/~'))
73
72
def file_relpath(base, path):
74
73
"""Compute just the relative sub-portion of a url
76
75
This assumes that both paths are already fully specified file:// URLs.
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)
77
if len(base) < MIN_ABS_FILEURL_LENGTH:
78
raise ValueError('Length of base (%r) must equal or'
79
' exceed the platform minimum url length (which is %d)' %
80
(base, MIN_ABS_FILEURL_LENGTH))
82
81
base = local_path_from_url(base)
83
82
path = local_path_from_url(path)
84
83
return escape(osutils.relpath(base, path))
117
121
join('http://foo', 'bar') => 'http://foo/bar'
118
122
join('http://foo', 'bar', '../baz') => 'http://foo/baz'
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
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:]
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
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',
131
path = base.split('/')
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('/')
142
for chunk in arg.split('/'):
147
# Don't pop off the host portion
150
raise errors.InvalidURLJoin('Cannot go above root',
156
178
return '/'.join(path)
157
return scheme + '://' + '/'.join(path)
160
181
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
161
182
def _posix_local_path_from_url(url):
162
183
"""Convert a url like file:///path/to/foo into /path/to/foo"""
163
if not url.startswith('file:///'):
164
raise errors.InvalidURL(url, 'local urls must start with file:///')
184
file_localhost_prefix = 'file://localhost/'
185
if url.startswith(file_localhost_prefix):
186
path = url[len(file_localhost_prefix) - 1:]
187
elif not url.startswith('file:///'):
188
raise errors.InvalidURL(
189
url, 'local urls must start with file:/// or file://localhost/')
191
path = url[len('file://'):]
165
192
# We only strip off 2 slashes
166
return unescape(url[len('file://'):])
193
return unescape(path)
169
196
def _posix_local_path_to_url(path):
207
239
This also handles transforming escaping unicode characters, etc.
209
# importing directly from ntpath allows us to test this
241
# importing directly from ntpath allows us to test this
210
242
# on non-win32 platform
211
243
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
212
244
# which actually strips trailing space characters.
213
# The worst part is that under linux ntpath.abspath has different
245
# The worst part is that on linux ntpath.abspath has different
214
246
# semantics, since 'nt' is not an available module.
215
250
win32_path = osutils._win32_abspath(path)
216
251
# check for UNC path \\HOST\path
217
252
if win32_path.startswith('//'):
218
253
return 'file:' + escape(win32_path)
219
return 'file:///' + win32_path[0].upper() + ':' + escape(win32_path[2:])
254
return ('file:///' + str(win32_path[0].upper()) + ':' +
255
escape(win32_path[2:]))
222
258
local_path_to_url = _posix_local_path_to_url
231
267
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
234
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
270
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,}):(//)?(?P<path>.*)$')
271
_url_hex_escapes_re = re.compile(r'(%[0-9a-fA-F]{2})')
274
def _unescape_safe_chars(matchobj):
275
"""re.sub callback to convert hex-escapes to plain characters (if safe).
277
e.g. '%7E' will be converted to '~'.
279
hex_digits = matchobj.group(0)[1:]
280
char = chr(int(hex_digits, 16))
281
if char in _url_dont_escape_characters:
284
return matchobj.group(0).upper()
237
287
def normalize_url(url):
238
288
"""Make sure that a path string is in fully normalized URL form.
240
This handles URLs which have unicode characters, spaces,
290
This handles URLs which have unicode characters, spaces,
241
291
special characters, etc.
243
293
It has two basic modes of operation, depending on whether the
253
303
:param url: Either a hybrid URL or a local path
254
304
:return: A normalized URL which only includes 7-bit ASCII characters.
256
m = _url_scheme_re.match(url)
306
scheme_end, path_start = _find_scheme_and_separator(url)
307
if scheme_end is None:
258
308
return local_path_to_url(url)
309
prefix = url[:path_start]
310
path = url[path_start:]
259
311
if not isinstance(url, unicode):
261
313
if c not in _url_safe_characters:
262
314
raise errors.InvalidURL(url, 'URLs can only contain specific'
263
315
' safe characters (not %r)' % c)
316
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
317
return str(prefix + ''.join(path))
265
319
# We have a unicode (hybrid) url
266
scheme = m.group('scheme')
267
path = list(m.group('path'))
320
path_chars = list(path)
269
for i in xrange(len(path)):
270
if path[i] not in _url_safe_characters:
271
chars = path[i].encode('utf-8')
272
path[i] = ''.join(['%%%02X' % ord(c) for c in path[i].encode('utf-8')])
273
return str(scheme + '://' + ''.join(path))
322
for i in xrange(len(path_chars)):
323
if path_chars[i] not in _url_safe_characters:
324
chars = path_chars[i].encode('utf-8')
325
path_chars[i] = ''.join(
326
['%%%02X' % ord(c) for c in path_chars[i].encode('utf-8')])
327
path = ''.join(path_chars)
328
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
329
return str(prefix + path)
276
332
def relative_url(base, other):
372
433
return url_base + head, tail
436
def split_segment_parameters_raw(url):
437
"""Split the subsegment of the last segment of a URL.
439
:param url: A relative or absolute URL
440
:return: (url, subsegments)
442
(parent_url, child_dir) = split(url)
443
subsegments = child_dir.split(",")
444
if len(subsegments) == 1:
446
return (join(parent_url, subsegments[0]), subsegments[1:])
449
def split_segment_parameters(url):
450
"""Split the segment parameters of the last segment of a URL.
452
:param url: A relative or absolute URL
453
:return: (url, segment_parameters)
455
(base_url, subsegments) = split_segment_parameters_raw(url)
457
for subsegment in subsegments:
458
(key, value) = subsegment.split("=", 1)
459
parameters[key] = value
460
return (base_url, parameters)
463
def join_segment_parameters_raw(base, *subsegments):
464
"""Create a new URL by adding subsegments to an existing one.
466
This adds the specified subsegments to the last path in the specified
467
base URL. The subsegments should be bytestrings.
469
:note: You probably want to use join_segment_parameters instead.
473
for subsegment in subsegments:
474
if type(subsegment) is not str:
475
raise TypeError("Subsegment %r is not a bytestring" % subsegment)
476
if "," in subsegment:
477
raise errors.InvalidURLJoin(", exists in subsegments",
479
return ",".join((base,) + subsegments)
482
def join_segment_parameters(url, parameters):
483
"""Create a new URL by adding segment parameters to an existing one.
485
The parameters of the last segment in the URL will be updated; if a
486
parameter with the same key already exists it will be overwritten.
488
:param url: A URL, as string
489
:param parameters: Dictionary of parameters, keys and values as bytestrings
491
(base, existing_parameters) = split_segment_parameters(url)
493
new_parameters.update(existing_parameters)
494
for key, value in parameters.iteritems():
495
if type(key) is not str:
496
raise TypeError("parameter key %r is not a bytestring" % key)
497
if type(value) is not str:
498
raise TypeError("parameter value %r for %s is not a bytestring" %
501
raise errors.InvalidURLJoin("= exists in parameter key", url,
503
new_parameters[key] = value
504
return join_segment_parameters_raw(base,
505
*["%s=%s" % item for item in sorted(new_parameters.items())])
375
508
def _win32_strip_local_trailing_slash(url):
376
509
"""Strip slashes after the drive letter"""
377
510
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
449
582
# These are characters that if escaped, should stay that way
450
583
_no_decode_chars = ';/?:@&=+$,#'
451
584
_no_decode_ords = [ord(c) for c in _no_decode_chars]
452
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
585
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
453
586
+ ['%02X' % o for o in _no_decode_ords])
454
587
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
455
588
+ [('%02X' % o, chr(o)) for o in range(256)]))
456
589
#These entries get mapped to themselves
457
590
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
592
# These characters shouldn't be percent-encoded, and it's always safe to
593
# unencode them if they are.
594
_url_dont_escape_characters = set(
595
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
596
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
597
"0123456789" # Numbers
598
"-._~" # Unreserved characters
459
601
# These characters should not be escaped
460
602
_url_safe_characters = set(
461
603
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
472
614
This will turn file:// urls into local paths, and try to decode
473
615
any portions of a http:// style url that it can.
475
Any sections of the URL which can't be represented in the encoding or
617
Any sections of the URL which can't be represented in the encoding or
476
618
need to stay as escapes are left alone.
478
620
:param url: A 7-bit ASCII URL
479
621
:param encoding: The final output encoding
481
:return: A unicode string which can be safely encoded into the
623
:return: A unicode string which can be safely encoded into the
482
624
specified encoding.
484
assert encoding is not None, 'you cannot specify None for the display encoding.'
627
raise ValueError('you cannot specify None for the display encoding')
485
628
if url.startswith('file://'):
487
630
path = local_path_from_url(url)
521
664
# Otherwise take the url decoded one
523
666
return u'/'.join(res)
669
def derive_to_location(from_location):
670
"""Derive a TO_LOCATION given a FROM_LOCATION.
672
The normal case is a FROM_LOCATION of http://foo/bar => bar.
673
The Right Thing for some logical destinations may differ though
674
because no / may be present at all. In that case, the result is
675
the full name without the scheme indicator, e.g. lp:foo-bar => foo-bar.
676
This latter case also applies when a Windows drive
677
is used without a path, e.g. c:foo-bar => foo-bar.
678
If no /, path separator or : is found, the from_location is returned.
680
if from_location.find("/") >= 0 or from_location.find(os.sep) >= 0:
681
return os.path.basename(from_location.rstrip("/\\"))
683
sep = from_location.find(":")
685
return from_location[sep+1:]
690
def _is_absolute(url):
691
return (osutils.pathjoin('/foo', url) == url)
694
def rebase_url(url, old_base, new_base):
695
"""Convert a relative path from an old base URL to a new base URL.
697
The result will be a relative path.
698
Absolute paths and full URLs are returned unaltered.
700
scheme, separator = _find_scheme_and_separator(url)
701
if scheme is not None:
703
if _is_absolute(url):
705
old_parsed = urlparse.urlparse(old_base)
706
new_parsed = urlparse.urlparse(new_base)
707
if (old_parsed[:2]) != (new_parsed[:2]):
708
raise errors.InvalidRebaseURLs(old_base, new_base)
709
return determine_relative_path(new_parsed[2],
710
join(old_parsed[2], url))
713
def determine_relative_path(from_path, to_path):
714
"""Determine a relative path from from_path to to_path."""
715
from_segments = osutils.splitpath(from_path)
716
to_segments = osutils.splitpath(to_path)
718
for count, (from_element, to_element) in enumerate(zip(from_segments,
720
if from_element != to_element:
724
unique_from = from_segments[count:]
725
unique_to = to_segments[count:]
726
segments = (['..'] * len(unique_from) + unique_to)
727
if len(segments) == 0:
729
return osutils.pathjoin(*segments)
734
"""Extract the server address, the credentials and the path from the url.
736
user, password, host and path should be quoted if they contain reserved
739
:param url: an quoted url
741
:return: (scheme, user, password, host, port, path) tuple, all fields
744
if isinstance(url, unicode):
745
raise errors.InvalidURL('should be ascii:\n%r' % url)
746
url = url.encode('utf-8')
747
(scheme, netloc, path, params,
748
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
749
user = password = host = port = None
751
user, host = netloc.rsplit('@', 1)
753
user, password = user.split(':', 1)
754
password = urllib.unquote(password)
755
user = urllib.unquote(user)
759
if ':' in host and not (host[0] == '[' and host[-1] == ']'): #there *is* port
760
host, port = host.rsplit(':',1)
764
raise errors.InvalidURL('invalid port number %s in url:\n%s' %
766
if host != "" and host[0] == '[' and host[-1] == ']': #IPv6
769
host = urllib.unquote(host)
770
path = urllib.unquote(path)
772
return (scheme, user, password, host, port, path)
added
removed
bzrlib/urlutils.py
