1
# Copyright (C) 2006-2010 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
"""A collection of function for handling URL operations."""
23
from bzrlib.lazy_import import lazy_import
24
lazy_import(globals(), """
25
from posixpath import split as _posix_split, normpath as _posix_normpath
36
def basename(url, exclude_trailing_slash=True):
37
"""Return the last component of a URL.
39
:param url: The URL in question
40
:param exclude_trailing_slash: If the url looks like "path/to/foo/"
41
ignore the final slash and return 'foo' rather than ''
42
:return: Just the final component of the URL. This can return ''
43
if you don't exclude_trailing_slash, or if you are at the
46
return split(url, exclude_trailing_slash=exclude_trailing_slash)[1]
49
def dirname(url, exclude_trailing_slash=True):
50
"""Return the parent directory of the given path.
52
:param url: Relative or absolute URL
53
:param exclude_trailing_slash: Remove a final slash
54
(treat http://host/foo/ as http://host/foo, but
55
http://host/ stays http://host/)
56
:return: Everything in the URL except the last path chunk
58
# TODO: jam 20060502 This was named dirname to be consistent
59
# with the os functions, but maybe "parent" would be better
60
return split(url, exclude_trailing_slash=exclude_trailing_slash)[0]
64
"""Escape relpath to be a valid url."""
65
if isinstance(relpath, unicode):
66
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='/~'))
72
def file_relpath(base, path):
73
"""Compute just the relative sub-portion of a url
75
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))
81
base = local_path_from_url(base)
82
path = local_path_from_url(path)
83
return escape(osutils.relpath(base, path))
86
def _find_scheme_and_separator(url):
87
"""Find the scheme separator (://) and the first path separator
89
This is just a helper functions for other path utilities.
90
It could probably be replaced by urlparse
92
m = _url_scheme_re.match(url)
96
scheme = m.group('scheme')
97
path = m.group('path')
99
# Find the path separating slash
100
# (first slash after the ://)
101
first_path_slash = path.find('/')
102
if first_path_slash == -1:
103
return len(scheme), None
104
return len(scheme), first_path_slash+m.start('path')
108
"""Tests whether a URL is in actual fact a URL."""
109
return _url_scheme_re.match(url) is not None
112
def join(base, *args):
113
"""Create a URL by joining sections.
115
This will normalize '..', assuming that paths are absolute
116
(it assumes no symlinks in either path)
118
If any of *args is an absolute URL, it will be treated correctly.
120
join('http://foo', 'http://bar') => 'http://bar'
121
join('http://foo', 'bar') => 'http://foo/bar'
122
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:]
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',
178
return '/'.join(path)
181
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
182
def _posix_local_path_from_url(url):
183
"""Convert a url like file:///path/to/foo into /path/to/foo"""
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://'):]
192
# We only strip off 2 slashes
193
return unescape(path)
196
def _posix_local_path_to_url(path):
197
"""Convert a local path like ./foo into a URL like file:///path/to/foo
199
This also handles transforming escaping unicode characters, etc.
201
# importing directly from posixpath allows us to test this
202
# on non-posix platforms
203
return 'file://' + escape(_posix_normpath(
204
osutils._posix_abspath(path)))
207
def _win32_local_path_from_url(url):
208
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
209
if not url.startswith('file://'):
210
raise errors.InvalidURL(url, 'local urls must start with file:///, '
211
'UNC path urls must start with file://')
212
# We strip off all 3 slashes
213
win32_url = url[len('file:'):]
214
# check for UNC path: //HOST/path
215
if not win32_url.startswith('///'):
216
if (win32_url[2] == '/'
217
or win32_url[3] in '|:'):
218
raise errors.InvalidURL(url, 'Win32 UNC path urls'
219
' have form file://HOST/path')
220
return unescape(win32_url)
222
# allow empty paths so we can serve all roots
223
if win32_url == '///':
226
# usual local path with drive letter
227
if (len(win32_url) < 6
228
or win32_url[3] not in ('abcdefghijklmnopqrstuvwxyz'
229
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
230
or win32_url[4] not in '|:'
231
or win32_url[5] != '/'):
232
raise errors.InvalidURL(url, 'Win32 file urls start with'
233
' file:///x:/, where x is a valid drive letter')
234
return win32_url[3].upper() + u':' + unescape(win32_url[5:])
237
def _win32_local_path_to_url(path):
238
"""Convert a local path like ./foo into a URL like file:///C:/path/to/foo
240
This also handles transforming escaping unicode characters, etc.
242
# importing directly from ntpath allows us to test this
243
# on non-win32 platform
244
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
245
# which actually strips trailing space characters.
246
# The worst part is that on linux ntpath.abspath has different
247
# semantics, since 'nt' is not an available module.
251
win32_path = osutils._win32_abspath(path)
252
# check for UNC path \\HOST\path
253
if win32_path.startswith('//'):
254
return 'file:' + escape(win32_path)
255
return ('file:///' + str(win32_path[0].upper()) + ':' +
256
escape(win32_path[2:]))
259
local_path_to_url = _posix_local_path_to_url
260
local_path_from_url = _posix_local_path_from_url
261
MIN_ABS_FILEURL_LENGTH = len('file:///')
262
WIN32_MIN_ABS_FILEURL_LENGTH = len('file:///C:/')
264
if sys.platform == 'win32':
265
local_path_to_url = _win32_local_path_to_url
266
local_path_from_url = _win32_local_path_from_url
268
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
271
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,}):(//)?(?P<path>.*)$')
272
_url_hex_escapes_re = re.compile(r'(%[0-9a-fA-F]{2})')
275
def _unescape_safe_chars(matchobj):
276
"""re.sub callback to convert hex-escapes to plain characters (if safe).
278
e.g. '%7E' will be converted to '~'.
280
hex_digits = matchobj.group(0)[1:]
281
char = chr(int(hex_digits, 16))
282
if char in _url_dont_escape_characters:
285
return matchobj.group(0).upper()
288
def normalize_url(url):
289
"""Make sure that a path string is in fully normalized URL form.
291
This handles URLs which have unicode characters, spaces,
292
special characters, etc.
294
It has two basic modes of operation, depending on whether the
295
supplied string starts with a url specifier (scheme://) or not.
296
If it does not have a specifier it is considered a local path,
297
and will be converted into a file:/// url. Non-ascii characters
298
will be encoded using utf-8.
299
If it does have a url specifier, it will be treated as a "hybrid"
300
URL. Basically, a URL that should have URL special characters already
301
escaped (like +?&# etc), but may have unicode characters, etc
302
which would not be valid in a real URL.
304
:param url: Either a hybrid URL or a local path
305
:return: A normalized URL which only includes 7-bit ASCII characters.
307
scheme_end, path_start = _find_scheme_and_separator(url)
308
if scheme_end is None:
309
return local_path_to_url(url)
310
prefix = url[:path_start]
311
path = url[path_start:]
312
if not isinstance(url, unicode):
314
if c not in _url_safe_characters:
315
raise errors.InvalidURL(url, 'URLs can only contain specific'
316
' safe characters (not %r)' % c)
317
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
318
return str(prefix + ''.join(path))
320
# We have a unicode (hybrid) url
321
path_chars = list(path)
323
for i in xrange(len(path_chars)):
324
if path_chars[i] not in _url_safe_characters:
325
chars = path_chars[i].encode('utf-8')
326
path_chars[i] = ''.join(
327
['%%%02X' % ord(c) for c in path_chars[i].encode('utf-8')])
328
path = ''.join(path_chars)
329
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
330
return str(prefix + path)
333
def relative_url(base, other):
334
"""Return a path to other from base.
336
If other is unrelated to base, return other. Else return a relative path.
337
This assumes no symlinks as part of the url.
339
dummy, base_first_slash = _find_scheme_and_separator(base)
340
if base_first_slash is None:
343
dummy, other_first_slash = _find_scheme_and_separator(other)
344
if other_first_slash is None:
347
# this takes care of differing schemes or hosts
348
base_scheme = base[:base_first_slash]
349
other_scheme = other[:other_first_slash]
350
if base_scheme != other_scheme:
352
elif sys.platform == 'win32' and base_scheme == 'file://':
353
base_drive = base[base_first_slash+1:base_first_slash+3]
354
other_drive = other[other_first_slash+1:other_first_slash+3]
355
if base_drive != other_drive:
358
base_path = base[base_first_slash+1:]
359
other_path = other[other_first_slash+1:]
361
if base_path.endswith('/'):
362
base_path = base_path[:-1]
364
base_sections = base_path.split('/')
365
other_sections = other_path.split('/')
367
if base_sections == ['']:
369
if other_sections == ['']:
373
for b, o in zip(base_sections, other_sections):
376
output_sections.append(b)
378
match_len = len(output_sections)
379
output_sections = ['..' for x in base_sections[match_len:]]
380
output_sections.extend(other_sections[match_len:])
382
return "/".join(output_sections) or "."
385
def _win32_extract_drive_letter(url_base, path):
386
"""On win32 the drive letter needs to be added to the url base."""
387
# Strip off the drive letter
388
# path is currently /C:/foo
389
if len(path) < 3 or path[2] not in ':|' or path[3] != '/':
390
raise errors.InvalidURL(url_base + path,
391
'win32 file:/// paths need a drive letter')
392
url_base += path[0:3] # file:// + /C:
393
path = path[3:] # /foo
394
return url_base, path
397
def split(url, exclude_trailing_slash=True):
398
"""Split a URL into its parent directory and a child directory.
400
:param url: A relative or absolute URL
401
:param exclude_trailing_slash: Strip off a final '/' if it is part
402
of the path (but not if it is part of the protocol specification)
404
:return: (parent_url, child_dir). child_dir may be the empty string if we're at
407
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
409
if first_path_slash is None:
410
# We have either a relative path, or no separating slash
411
if scheme_loc is None:
413
if exclude_trailing_slash and url.endswith('/'):
415
return _posix_split(url)
417
# Scheme with no path
420
# We have a fully defined path
421
url_base = url[:first_path_slash] # http://host, file://
422
path = url[first_path_slash:] # /file/foo
424
if sys.platform == 'win32' and url.startswith('file:///'):
425
# Strip off the drive letter
426
# url_base is currently file://
427
# path is currently /C:/foo
428
url_base, path = _win32_extract_drive_letter(url_base, path)
429
# now it should be file:///C: and /foo
431
if exclude_trailing_slash and len(path) > 1 and path.endswith('/'):
433
head, tail = _posix_split(path)
434
return url_base + head, tail
437
def split_segment_parameters_raw(url):
438
"""Split the subsegment of the last segment of a URL.
440
:param url: A relative or absolute URL
441
:return: (url, subsegments)
443
(parent_url, child_dir) = split(url)
444
subsegments = child_dir.split(",")
445
if len(subsegments) == 1:
447
return (join(parent_url, subsegments[0]), subsegments[1:])
450
def split_segment_parameters(url):
451
"""Split the segment parameters of the last segment of a URL.
453
:param url: A relative or absolute URL
454
:return: (url, segment_parameters)
456
(base_url, subsegments) = split_segment_parameters_raw(url)
458
for subsegment in subsegments:
459
(key, value) = subsegment.split("=", 1)
460
parameters[key] = value
461
return (base_url, parameters)
464
def join_segment_parameters_raw(base, *subsegments):
465
"""Create a new URL by adding subsegments to an existing one.
467
This adds the specified subsegments to the last path in the specified
468
base URL. The subsegments should be bytestrings.
470
:note: You probably want to use join_segment_parameters instead.
474
for subsegment in subsegments:
475
if type(subsegment) is not str:
476
raise TypeError("Subsegment %r is not a bytestring" % subsegment)
477
if "," in subsegment:
478
raise errors.InvalidURLJoin(", exists in subsegments",
480
return ",".join((base,) + subsegments)
483
def join_segment_parameters(url, parameters):
484
"""Create a new URL by adding segment parameters to an existing one.
486
The parameters of the last segment in the URL will be updated; if a
487
parameter with the same key already exists it will be overwritten.
489
:param url: A URL, as string
490
:param parameters: Dictionary of parameters, keys and values as bytestrings
492
(base, existing_parameters) = split_segment_parameters(url)
494
new_parameters.update(existing_parameters)
495
for key, value in parameters.iteritems():
496
if type(key) is not str:
497
raise TypeError("parameter key %r is not a bytestring" % key)
498
if type(value) is not str:
499
raise TypeError("parameter value %r for %s is not a bytestring" %
502
raise errors.InvalidURLJoin("= exists in parameter key", url,
504
new_parameters[key] = value
505
return join_segment_parameters_raw(base,
506
*["%s=%s" % item for item in sorted(new_parameters.items())])
509
def _win32_strip_local_trailing_slash(url):
510
"""Strip slashes after the drive letter"""
511
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
517
def strip_trailing_slash(url):
518
"""Strip trailing slash, except for root paths.
520
The definition of 'root path' is platform-dependent.
521
This assumes that all URLs are valid netloc urls, such that they
524
It searches for ://, and then refuses to remove the next '/'.
525
It can also handle relative paths
527
path/to/foo => path/to/foo
528
path/to/foo/ => path/to/foo
529
http://host/path/ => http://host/path
530
http://host/path => http://host/path
531
http://host/ => http://host/
533
file:///foo/ => file:///foo
534
# This is unique on win32 platforms, and is the only URL
535
# format which does it differently.
536
file:///c|/ => file:///c:/
538
if not url.endswith('/'):
541
if sys.platform == 'win32' and url.startswith('file://'):
542
return _win32_strip_local_trailing_slash(url)
544
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
545
if scheme_loc is None:
546
# This is a relative path, as it has no scheme
547
# so just chop off the last character
550
if first_path_slash is None or first_path_slash == len(url)-1:
551
# Don't chop off anything if the only slash is the path
559
"""Unescape relpath from url format.
561
This returns a Unicode path from a URL
563
# jam 20060427 URLs are supposed to be ASCII only strings
564
# If they are passed in as unicode, urllib.unquote
565
# will return a UNICODE string, which actually contains
566
# utf-8 bytes. So we have to ensure that they are
567
# plain ASCII strings, or the final .decode will
568
# try to encode the UNICODE => ASCII, and then decode
572
except UnicodeError, e:
573
raise errors.InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
575
unquoted = urllib.unquote(url)
577
unicode_path = unquoted.decode('utf-8')
578
except UnicodeError, e:
579
raise errors.InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
583
# These are characters that if escaped, should stay that way
584
_no_decode_chars = ';/?:@&=+$,#'
585
_no_decode_ords = [ord(c) for c in _no_decode_chars]
586
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
587
+ ['%02X' % o for o in _no_decode_ords])
588
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
589
+ [('%02X' % o, chr(o)) for o in range(256)]))
590
#These entries get mapped to themselves
591
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
593
# These characters shouldn't be percent-encoded, and it's always safe to
594
# unencode them if they are.
595
_url_dont_escape_characters = set(
596
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
597
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
598
"0123456789" # Numbers
599
"-._~" # Unreserved characters
602
# These characters should not be escaped
603
_url_safe_characters = set(
604
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
605
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
606
"0123456789" # Numbers
607
"_.-!~*'()" # Unreserved characters
608
"/;?:@&=+$," # Reserved characters
609
"%#" # Extra reserved characters
612
def unescape_for_display(url, encoding):
613
"""Decode what you can for a URL, so that we get a nice looking path.
615
This will turn file:// urls into local paths, and try to decode
616
any portions of a http:// style url that it can.
618
Any sections of the URL which can't be represented in the encoding or
619
need to stay as escapes are left alone.
621
:param url: A 7-bit ASCII URL
622
:param encoding: The final output encoding
624
:return: A unicode string which can be safely encoded into the
628
raise ValueError('you cannot specify None for the display encoding')
629
if url.startswith('file://'):
631
path = local_path_from_url(url)
632
path.encode(encoding)
637
# Split into sections to try to decode utf-8
639
for i in xrange(1, len(res)):
640
escaped_chunks = res[i].split('%')
641
for j in xrange(1, len(escaped_chunks)):
642
item = escaped_chunks[j]
644
escaped_chunks[j] = _hex_display_map[item[:2]] + item[2:]
646
# Put back the percent symbol
647
escaped_chunks[j] = '%' + item
648
except UnicodeDecodeError:
649
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
650
unescaped = ''.join(escaped_chunks)
652
decoded = unescaped.decode('utf-8')
653
except UnicodeDecodeError:
654
# If this path segment cannot be properly utf-8 decoded
655
# after doing unescaping we will just leave it alone
659
decoded.encode(encoding)
660
except UnicodeEncodeError:
661
# If this chunk cannot be encoded in the local
662
# encoding, then we should leave it alone
665
# Otherwise take the url decoded one
667
return u'/'.join(res)
670
def derive_to_location(from_location):
671
"""Derive a TO_LOCATION given a FROM_LOCATION.
673
The normal case is a FROM_LOCATION of http://foo/bar => bar.
674
The Right Thing for some logical destinations may differ though
675
because no / may be present at all. In that case, the result is
676
the full name without the scheme indicator, e.g. lp:foo-bar => foo-bar.
677
This latter case also applies when a Windows drive
678
is used without a path, e.g. c:foo-bar => foo-bar.
679
If no /, path separator or : is found, the from_location is returned.
681
if from_location.find("/") >= 0 or from_location.find(os.sep) >= 0:
682
return os.path.basename(from_location.rstrip("/\\"))
684
sep = from_location.find(":")
686
return from_location[sep+1:]
691
def _is_absolute(url):
692
return (osutils.pathjoin('/foo', url) == url)
695
def rebase_url(url, old_base, new_base):
696
"""Convert a relative path from an old base URL to a new base URL.
698
The result will be a relative path.
699
Absolute paths and full URLs are returned unaltered.
701
scheme, separator = _find_scheme_and_separator(url)
702
if scheme is not None:
704
if _is_absolute(url):
706
old_parsed = urlparse.urlparse(old_base)
707
new_parsed = urlparse.urlparse(new_base)
708
if (old_parsed[:2]) != (new_parsed[:2]):
709
raise errors.InvalidRebaseURLs(old_base, new_base)
710
return determine_relative_path(new_parsed[2],
711
join(old_parsed[2], url))
714
def determine_relative_path(from_path, to_path):
715
"""Determine a relative path from from_path to to_path."""
716
from_segments = osutils.splitpath(from_path)
717
to_segments = osutils.splitpath(to_path)
719
for count, (from_element, to_element) in enumerate(zip(from_segments,
721
if from_element != to_element:
725
unique_from = from_segments[count:]
726
unique_to = to_segments[count:]
727
segments = (['..'] * len(unique_from) + unique_to)
728
if len(segments) == 0:
730
return osutils.pathjoin(*segments)
735
"""Extract the server address, the credentials and the path from the url.
737
user, password, host and path should be quoted if they contain reserved
740
:param url: an quoted url
742
:return: (scheme, user, password, host, port, path) tuple, all fields
745
if isinstance(url, unicode):
746
raise errors.InvalidURL('should be ascii:\n%r' % url)
747
url = url.encode('utf-8')
748
(scheme, netloc, path, params,
749
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
750
user = password = host = port = None
752
user, host = netloc.rsplit('@', 1)
754
user, password = user.split(':', 1)
755
password = urllib.unquote(password)
756
user = urllib.unquote(user)
760
if ':' in host and not (host[0] == '[' and host[-1] == ']'): #there *is* port
761
host, port = host.rsplit(':',1)
765
raise errors.InvalidURL('invalid port number %s in url:\n%s' %
767
if host != "" and host[0] == '[' and host[-1] == ']': #IPv6
770
host = urllib.unquote(host)
771
path = urllib.unquote(path)
773
return (scheme, user, password, host, port, path)
added
removed
bzrlib/urlutils.py
