66
67
relpath = relpath.encode('utf-8')
67
68
# After quoting and encoding, the path should be perfectly
68
69
# safe as a plain ASCII string, str() just enforces this
69
return str(urllib.quote(relpath, safe='/~'))
70
return str(urllib.quote(relpath))
72
73
def file_relpath(base, path):
73
74
"""Compute just the relative sub-portion of a url
75
76
This assumes that both paths are already fully specified file:// URLs.
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))
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)
81
82
base = local_path_from_url(base)
82
83
path = local_path_from_url(path)
83
84
return escape(osutils.relpath(base, path))
132
131
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.
147
match = _url_scheme_re.match(arg)
134
m = _url_scheme_re.match(arg)
150
scheme = match.group('scheme')
137
scheme = m.group('scheme')
151
138
# this skips .. normalisation, making http://host/../../..
152
139
# be rather strange.
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.
140
path = m.group('path').split('/')
167
path = '/'.join(path)
168
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:
142
for chunk in arg.split('/'):
147
# Don't pop off the host portion
150
raise errors.InvalidURLJoin('Cannot go above root',
179
155
if scheme is None:
180
156
return '/'.join(path)
217
193
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
218
194
def _posix_local_path_from_url(url):
219
195
"""Convert a url like file:///path/to/foo into /path/to/foo"""
220
file_localhost_prefix = 'file://localhost/'
221
if url.startswith(file_localhost_prefix):
222
path = url[len(file_localhost_prefix) - 1:]
223
elif not url.startswith('file:///'):
224
raise errors.InvalidURL(
225
url, 'local urls must start with file:/// or file://localhost/')
227
path = url[len('file://'):]
196
if not url.startswith('file:///'):
197
raise errors.InvalidURL(url, 'local urls must start with file:///')
228
198
# We only strip off 2 slashes
229
return unescape(path)
199
return unescape(url[len('file://'):])
232
202
def _posix_local_path_to_url(path):
275
240
This also handles transforming escaping unicode characters, etc.
277
# importing directly from ntpath allows us to test this
242
# importing directly from ntpath allows us to test this
278
243
# on non-win32 platform
279
244
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
280
245
# which actually strips trailing space characters.
281
246
# The worst part is that under linux ntpath.abspath has different
282
247
# semantics, since 'nt' is not an available module.
286
win32_path = osutils._win32_abspath(path)
248
win32_path = osutils._nt_normpath(
249
osutils._win32_abspath(path)).replace('\\', '/')
287
250
# check for UNC path \\HOST\path
288
251
if win32_path.startswith('//'):
289
252
return 'file:' + escape(win32_path)
290
return ('file:///' + str(win32_path[0].upper()) + ':' +
291
escape(win32_path[2:]))
253
return 'file:///' + win32_path[0].upper() + ':' + escape(win32_path[2:])
294
256
local_path_to_url = _posix_local_path_to_url
306
268
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
307
_url_hex_escapes_re = re.compile(r'(%[0-9a-fA-F]{2})')
310
def _unescape_safe_chars(matchobj):
311
"""re.sub callback to convert hex-escapes to plain characters (if safe).
313
e.g. '%7E' will be converted to '~'.
315
hex_digits = matchobj.group(0)[1:]
316
char = chr(int(hex_digits, 16))
317
if char in _url_dont_escape_characters:
320
return matchobj.group(0).upper()
323
271
def normalize_url(url):
324
272
"""Make sure that a path string is in fully normalized URL form.
326
This handles URLs which have unicode characters, spaces,
274
This handles URLs which have unicode characters, spaces,
327
275
special characters, etc.
329
277
It has two basic modes of operation, depending on whether the
342
290
m = _url_scheme_re.match(url)
344
292
return local_path_to_url(url)
345
scheme = m.group('scheme')
346
path = m.group('path')
347
293
if not isinstance(url, unicode):
349
295
if c not in _url_safe_characters:
350
296
raise errors.InvalidURL(url, 'URLs can only contain specific'
351
297
' safe characters (not %r)' % c)
352
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
353
return str(scheme + '://' + ''.join(path))
355
299
# We have a unicode (hybrid) url
356
path_chars = list(path)
300
scheme = m.group('scheme')
301
path = list(m.group('path'))
358
for i in xrange(len(path_chars)):
359
if path_chars[i] not in _url_safe_characters:
360
chars = path_chars[i].encode('utf-8')
361
path_chars[i] = ''.join(
362
['%%%02X' % ord(c) for c in path_chars[i].encode('utf-8')])
363
path = ''.join(path_chars)
364
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
365
return str(scheme + '://' + path)
303
for i in xrange(len(path)):
304
if path[i] not in _url_safe_characters:
305
chars = path[i].encode('utf-8')
306
path[i] = ''.join(['%%%02X' % ord(c) for c in path[i].encode('utf-8')])
307
return str(scheme + '://' + ''.join(path))
368
310
def relative_url(base, other):
546
483
# These are characters that if escaped, should stay that way
547
484
_no_decode_chars = ';/?:@&=+$,#'
548
485
_no_decode_ords = [ord(c) for c in _no_decode_chars]
549
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
486
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
550
487
+ ['%02X' % o for o in _no_decode_ords])
551
488
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
552
489
+ [('%02X' % o, chr(o)) for o in range(256)]))
553
490
#These entries get mapped to themselves
554
491
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
556
# These characters shouldn't be percent-encoded, and it's always safe to
557
# unencode them if they are.
558
_url_dont_escape_characters = set(
559
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
560
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
561
"0123456789" # Numbers
562
"-._~" # Unreserved characters
565
493
# These characters should not be escaped
566
494
_url_safe_characters = set(
567
495
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
578
506
This will turn file:// urls into local paths, and try to decode
579
507
any portions of a http:// style url that it can.
581
Any sections of the URL which can't be represented in the encoding or
509
Any sections of the URL which can't be represented in the encoding or
582
510
need to stay as escapes are left alone.
584
512
:param url: A 7-bit ASCII URL
585
513
:param encoding: The final output encoding
587
:return: A unicode string which can be safely encoded into the
515
:return: A unicode string which can be safely encoded into the
588
516
specified encoding.
591
raise ValueError('you cannot specify None for the display encoding')
518
assert encoding is not None, 'you cannot specify None for the display encoding.'
592
519
if url.startswith('file://'):
594
521
path = local_path_from_url(url)
628
555
# Otherwise take the url decoded one
630
557
return u'/'.join(res)
633
def derive_to_location(from_location):
634
"""Derive a TO_LOCATION given a FROM_LOCATION.
636
The normal case is a FROM_LOCATION of http://foo/bar => bar.
637
The Right Thing for some logical destinations may differ though
638
because no / may be present at all. In that case, the result is
639
the full name without the scheme indicator, e.g. lp:foo-bar => foo-bar.
640
This latter case also applies when a Windows drive
641
is used without a path, e.g. c:foo-bar => foo-bar.
642
If no /, path separator or : is found, the from_location is returned.
644
if from_location.find("/") >= 0 or from_location.find(os.sep) >= 0:
645
return os.path.basename(from_location.rstrip("/\\"))
647
sep = from_location.find(":")
649
return from_location[sep+1:]
654
def _is_absolute(url):
655
return (osutils.pathjoin('/foo', url) == url)
658
def rebase_url(url, old_base, new_base):
659
"""Convert a relative path from an old base URL to a new base URL.
661
The result will be a relative path.
662
Absolute paths and full URLs are returned unaltered.
664
scheme, separator = _find_scheme_and_separator(url)
665
if scheme is not None:
667
if _is_absolute(url):
669
old_parsed = urlparse.urlparse(old_base)
670
new_parsed = urlparse.urlparse(new_base)
671
if (old_parsed[:2]) != (new_parsed[:2]):
672
raise errors.InvalidRebaseURLs(old_base, new_base)
673
return determine_relative_path(new_parsed[2],
674
join(old_parsed[2], url))
677
def determine_relative_path(from_path, to_path):
678
"""Determine a relative path from from_path to to_path."""
679
from_segments = osutils.splitpath(from_path)
680
to_segments = osutils.splitpath(to_path)
682
for count, (from_element, to_element) in enumerate(zip(from_segments,
684
if from_element != to_element:
688
unique_from = from_segments[count:]
689
unique_to = to_segments[count:]
690
segments = (['..'] * len(unique_from) + unique_to)
691
if len(segments) == 0:
693
return osutils.pathjoin(*segments)
698
"""Extract the server address, the credentials and the path from the url.
700
user, password, host and path should be quoted if they contain reserved
703
:param url: an quoted url
705
: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)
added
removed
bzrlib/urlutils.py
