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+len(scheme)+3
107
def join(base, *args):
108
"""Create a URL by joining sections.
110
This will normalize '..', assuming that paths are absolute
111
(it assumes no symlinks in either path)
113
If any of *args is an absolute URL, it will be treated correctly.
115
join('http://foo', 'http://bar') => 'http://bar'
116
join('http://foo', 'bar') => 'http://foo/bar'
117
join('http://foo', 'bar', '../baz') => 'http://foo/baz'
121
match = _url_scheme_re.match(base)
124
scheme = match.group('scheme')
125
path = match.group('path').split('/')
126
if path[-1:] == ['']:
127
# Strip off a trailing slash
128
# This helps both when we are at the root, and when
129
# 'base' has an extra slash at the end
132
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)
150
scheme = match.group('scheme')
151
# this skips .. normalisation, making http://host/../../..
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.
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:
180
return '/'.join(path)
181
return scheme + '://' + '/'.join(path)
184
def joinpath(base, *args):
185
"""Join URL path segments to a URL path segment.
187
This is somewhat like osutils.joinpath, but intended for URLs.
189
XXX: this duplicates some normalisation logic, and also duplicates a lot of
190
path handling logic that already exists in some Transport implementations.
191
We really should try to have exactly one place in the code base responsible
192
for combining paths of URLs.
194
path = base.split('/')
195
if len(path) > 1 and path[-1] == '':
196
#If the path ends in a trailing /, remove it.
199
if arg.startswith('/'):
201
for chunk in arg.split('/'):
206
raise errors.InvalidURLJoin('Cannot go above root',
214
return '/'.join(path)
217
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
218
def _posix_local_path_from_url(url):
219
"""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://'):]
228
# We only strip off 2 slashes
229
return unescape(path)
232
def _posix_local_path_to_url(path):
233
"""Convert a local path like ./foo into a URL like file:///path/to/foo
235
This also handles transforming escaping unicode characters, etc.
237
# importing directly from posixpath allows us to test this
238
# on non-posix platforms
239
return 'file://' + escape(_posix_normpath(
240
osutils._posix_abspath(path)))
243
def _win32_local_path_from_url(url):
244
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
245
if not url.startswith('file://'):
246
raise errors.InvalidURL(url, 'local urls must start with file:///, '
247
'UNC path urls must start with file://')
248
# We strip off all 3 slashes
249
win32_url = url[len('file:'):]
250
# check for UNC path: //HOST/path
251
if not win32_url.startswith('///'):
252
if (win32_url[2] == '/'
253
or win32_url[3] in '|:'):
254
raise errors.InvalidURL(url, 'Win32 UNC path urls'
255
' have form file://HOST/path')
256
return unescape(win32_url)
258
# allow empty paths so we can serve all roots
259
if win32_url == '///':
262
# usual local path with drive letter
263
if (win32_url[3] not in ('abcdefghijklmnopqrstuvwxyz'
264
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
265
or win32_url[4] not in '|:'
266
or win32_url[5] != '/'):
267
raise errors.InvalidURL(url, 'Win32 file urls start with'
268
' file:///x:/, where x is a valid drive letter')
269
return win32_url[3].upper() + u':' + unescape(win32_url[5:])
272
def _win32_local_path_to_url(path):
273
"""Convert a local path like ./foo into a URL like file:///C:/path/to/foo
275
This also handles transforming escaping unicode characters, etc.
277
# importing directly from ntpath allows us to test this
278
# on non-win32 platform
279
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
280
# which actually strips trailing space characters.
281
# The worst part is that on linux ntpath.abspath has different
282
# semantics, since 'nt' is not an available module.
286
win32_path = osutils._win32_abspath(path)
287
# check for UNC path \\HOST\path
288
if win32_path.startswith('//'):
289
return 'file:' + escape(win32_path)
290
return ('file:///' + str(win32_path[0].upper()) + ':' +
291
escape(win32_path[2:]))
294
local_path_to_url = _posix_local_path_to_url
295
local_path_from_url = _posix_local_path_from_url
296
MIN_ABS_FILEURL_LENGTH = len('file:///')
297
WIN32_MIN_ABS_FILEURL_LENGTH = len('file:///C:/')
299
if sys.platform == 'win32':
300
local_path_to_url = _win32_local_path_to_url
301
local_path_from_url = _win32_local_path_from_url
303
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
306
_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
def normalize_url(url):
324
"""Make sure that a path string is in fully normalized URL form.
326
This handles URLs which have unicode characters, spaces,
327
special characters, etc.
329
It has two basic modes of operation, depending on whether the
330
supplied string starts with a url specifier (scheme://) or not.
331
If it does not have a specifier it is considered a local path,
332
and will be converted into a file:/// url. Non-ascii characters
333
will be encoded using utf-8.
334
If it does have a url specifier, it will be treated as a "hybrid"
335
URL. Basically, a URL that should have URL special characters already
336
escaped (like +?&# etc), but may have unicode characters, etc
337
which would not be valid in a real URL.
339
:param url: Either a hybrid URL or a local path
340
:return: A normalized URL which only includes 7-bit ASCII characters.
342
m = _url_scheme_re.match(url)
344
return local_path_to_url(url)
345
scheme = m.group('scheme')
346
path = m.group('path')
347
if not isinstance(url, unicode):
349
if c not in _url_safe_characters:
350
raise errors.InvalidURL(url, 'URLs can only contain specific'
351
' safe characters (not %r)' % c)
352
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
353
return str(scheme + '://' + ''.join(path))
355
# We have a unicode (hybrid) url
356
path_chars = list(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)
368
def relative_url(base, other):
369
"""Return a path to other from base.
371
If other is unrelated to base, return other. Else return a relative path.
372
This assumes no symlinks as part of the url.
374
dummy, base_first_slash = _find_scheme_and_separator(base)
375
if base_first_slash is None:
378
dummy, other_first_slash = _find_scheme_and_separator(other)
379
if other_first_slash is None:
382
# this takes care of differing schemes or hosts
383
base_scheme = base[:base_first_slash]
384
other_scheme = other[:other_first_slash]
385
if base_scheme != other_scheme:
387
elif sys.platform == 'win32' and base_scheme == 'file://':
388
base_drive = base[base_first_slash+1:base_first_slash+3]
389
other_drive = other[other_first_slash+1:other_first_slash+3]
390
if base_drive != other_drive:
393
base_path = base[base_first_slash+1:]
394
other_path = other[other_first_slash+1:]
396
if base_path.endswith('/'):
397
base_path = base_path[:-1]
399
base_sections = base_path.split('/')
400
other_sections = other_path.split('/')
402
if base_sections == ['']:
404
if other_sections == ['']:
408
for b, o in zip(base_sections, other_sections):
411
output_sections.append(b)
413
match_len = len(output_sections)
414
output_sections = ['..' for x in base_sections[match_len:]]
415
output_sections.extend(other_sections[match_len:])
417
return "/".join(output_sections) or "."
420
def _win32_extract_drive_letter(url_base, path):
421
"""On win32 the drive letter needs to be added to the url base."""
422
# Strip off the drive letter
423
# path is currently /C:/foo
424
if len(path) < 3 or path[2] not in ':|' or path[3] != '/':
425
raise errors.InvalidURL(url_base + path,
426
'win32 file:/// paths need a drive letter')
427
url_base += path[0:3] # file:// + /C:
428
path = path[3:] # /foo
429
return url_base, path
432
def split(url, exclude_trailing_slash=True):
433
"""Split a URL into its parent directory and a child directory.
435
:param url: A relative or absolute URL
436
:param exclude_trailing_slash: Strip off a final '/' if it is part
437
of the path (but not if it is part of the protocol specification)
439
:return: (parent_url, child_dir). child_dir may be the empty string if we're at
442
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
444
if first_path_slash is None:
445
# We have either a relative path, or no separating slash
446
if scheme_loc is None:
448
if exclude_trailing_slash and url.endswith('/'):
450
return _posix_split(url)
452
# Scheme with no path
455
# We have a fully defined path
456
url_base = url[:first_path_slash] # http://host, file://
457
path = url[first_path_slash:] # /file/foo
459
if sys.platform == 'win32' and url.startswith('file:///'):
460
# Strip off the drive letter
461
# url_base is currently file://
462
# path is currently /C:/foo
463
url_base, path = _win32_extract_drive_letter(url_base, path)
464
# now it should be file:///C: and /foo
466
if exclude_trailing_slash and len(path) > 1 and path.endswith('/'):
468
head, tail = _posix_split(path)
469
return url_base + head, tail
472
def split_segment_parameters_raw(url):
473
"""Split the subsegment of the last segment of a URL.
475
:param url: A relative or absolute URL
476
:return: (url, subsegments)
478
(parent_url, child_dir) = split(url)
479
subsegments = child_dir.split(",")
480
if len(subsegments) == 1:
482
return (join(parent_url, subsegments[0]), subsegments[1:])
485
def split_segment_parameters(url):
486
"""Split the segment parameters of the last segment of a URL.
488
:param url: A relative or absolute URL
489
:return: (url, segment_parameters)
491
(base_url, subsegments) = split_segment_parameters_raw(url)
493
for subsegment in subsegments:
494
(key, value) = subsegment.split("=", 1)
495
parameters[key] = value
496
return (base_url, parameters)
499
def join_segment_parameters_raw(base, *subsegments):
500
"""Create a new URL by adding subsegments to an existing one.
502
This adds the specified subsegments to the last path in the specified
503
base URL. The subsegments should be bytestrings.
505
:note: You probably want to use join_segment_parameters instead.
509
for subsegment in subsegments:
510
if type(subsegment) is not str:
511
raise TypeError("Subsegment %r is not a bytestring" % subsegment)
512
if "," in subsegment:
513
raise errors.InvalidURLJoin(", exists in subsegments",
515
return ",".join((base,) + subsegments)
518
def join_segment_parameters(url, parameters):
519
"""Create a new URL by adding segment parameters to an existing one.
521
The parameters of the last segment in the URL will be updated; if a
522
parameter with the same key already exists it will be overwritten.
524
:param url: A URL, as string
525
:param parameters: Dictionary of parameters, keys and values as bytestrings
527
(base, existing_parameters) = split_segment_parameters(url)
529
new_parameters.update(existing_parameters)
530
for key, value in parameters.iteritems():
531
if type(key) is not str:
532
raise TypeError("parameter key %r is not a bytestring" % key)
533
if type(value) is not str:
534
raise TypeError("parameter value %r for %s is not a bytestring" %
537
raise errors.InvalidURLJoin("= exists in parameter key", url,
539
new_parameters[key] = value
540
return join_segment_parameters_raw(base,
541
*["%s=%s" % item for item in sorted(new_parameters.items())])
544
def _win32_strip_local_trailing_slash(url):
545
"""Strip slashes after the drive letter"""
546
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
552
def strip_trailing_slash(url):
553
"""Strip trailing slash, except for root paths.
555
The definition of 'root path' is platform-dependent.
556
This assumes that all URLs are valid netloc urls, such that they
559
It searches for ://, and then refuses to remove the next '/'.
560
It can also handle relative paths
562
path/to/foo => path/to/foo
563
path/to/foo/ => path/to/foo
564
http://host/path/ => http://host/path
565
http://host/path => http://host/path
566
http://host/ => http://host/
568
file:///foo/ => file:///foo
569
# This is unique on win32 platforms, and is the only URL
570
# format which does it differently.
571
file:///c|/ => file:///c:/
573
if not url.endswith('/'):
576
if sys.platform == 'win32' and url.startswith('file://'):
577
return _win32_strip_local_trailing_slash(url)
579
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
580
if scheme_loc is None:
581
# This is a relative path, as it has no scheme
582
# so just chop off the last character
585
if first_path_slash is None or first_path_slash == len(url)-1:
586
# Don't chop off anything if the only slash is the path
594
"""Unescape relpath from url format.
596
This returns a Unicode path from a URL
598
# jam 20060427 URLs are supposed to be ASCII only strings
599
# If they are passed in as unicode, urllib.unquote
600
# will return a UNICODE string, which actually contains
601
# utf-8 bytes. So we have to ensure that they are
602
# plain ASCII strings, or the final .decode will
603
# try to encode the UNICODE => ASCII, and then decode
607
except UnicodeError, e:
608
raise errors.InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
610
unquoted = urllib.unquote(url)
612
unicode_path = unquoted.decode('utf-8')
613
except UnicodeError, e:
614
raise errors.InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
618
# These are characters that if escaped, should stay that way
619
_no_decode_chars = ';/?:@&=+$,#'
620
_no_decode_ords = [ord(c) for c in _no_decode_chars]
621
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
622
+ ['%02X' % o for o in _no_decode_ords])
623
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
624
+ [('%02X' % o, chr(o)) for o in range(256)]))
625
#These entries get mapped to themselves
626
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
628
# These characters shouldn't be percent-encoded, and it's always safe to
629
# unencode them if they are.
630
_url_dont_escape_characters = set(
631
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
632
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
633
"0123456789" # Numbers
634
"-._~" # Unreserved characters
637
# These characters should not be escaped
638
_url_safe_characters = set(
639
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
640
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
641
"0123456789" # Numbers
642
"_.-!~*'()" # Unreserved characters
643
"/;?:@&=+$," # Reserved characters
644
"%#" # Extra reserved characters
647
def unescape_for_display(url, encoding):
648
"""Decode what you can for a URL, so that we get a nice looking path.
650
This will turn file:// urls into local paths, and try to decode
651
any portions of a http:// style url that it can.
653
Any sections of the URL which can't be represented in the encoding or
654
need to stay as escapes are left alone.
656
:param url: A 7-bit ASCII URL
657
:param encoding: The final output encoding
659
:return: A unicode string which can be safely encoded into the
663
raise ValueError('you cannot specify None for the display encoding')
664
if url.startswith('file://'):
666
path = local_path_from_url(url)
667
path.encode(encoding)
672
# Split into sections to try to decode utf-8
674
for i in xrange(1, len(res)):
675
escaped_chunks = res[i].split('%')
676
for j in xrange(1, len(escaped_chunks)):
677
item = escaped_chunks[j]
679
escaped_chunks[j] = _hex_display_map[item[:2]] + item[2:]
681
# Put back the percent symbol
682
escaped_chunks[j] = '%' + item
683
except UnicodeDecodeError:
684
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
685
unescaped = ''.join(escaped_chunks)
687
decoded = unescaped.decode('utf-8')
688
except UnicodeDecodeError:
689
# If this path segment cannot be properly utf-8 decoded
690
# after doing unescaping we will just leave it alone
694
decoded.encode(encoding)
695
except UnicodeEncodeError:
696
# If this chunk cannot be encoded in the local
697
# encoding, then we should leave it alone
700
# Otherwise take the url decoded one
702
return u'/'.join(res)
705
def derive_to_location(from_location):
706
"""Derive a TO_LOCATION given a FROM_LOCATION.
708
The normal case is a FROM_LOCATION of http://foo/bar => bar.
709
The Right Thing for some logical destinations may differ though
710
because no / may be present at all. In that case, the result is
711
the full name without the scheme indicator, e.g. lp:foo-bar => foo-bar.
712
This latter case also applies when a Windows drive
713
is used without a path, e.g. c:foo-bar => foo-bar.
714
If no /, path separator or : is found, the from_location is returned.
716
if from_location.find("/") >= 0 or from_location.find(os.sep) >= 0:
717
return os.path.basename(from_location.rstrip("/\\"))
719
sep = from_location.find(":")
721
return from_location[sep+1:]
726
def _is_absolute(url):
727
return (osutils.pathjoin('/foo', url) == url)
730
def rebase_url(url, old_base, new_base):
731
"""Convert a relative path from an old base URL to a new base URL.
733
The result will be a relative path.
734
Absolute paths and full URLs are returned unaltered.
736
scheme, separator = _find_scheme_and_separator(url)
737
if scheme is not None:
739
if _is_absolute(url):
741
old_parsed = urlparse.urlparse(old_base)
742
new_parsed = urlparse.urlparse(new_base)
743
if (old_parsed[:2]) != (new_parsed[:2]):
744
raise errors.InvalidRebaseURLs(old_base, new_base)
745
return determine_relative_path(new_parsed[2],
746
join(old_parsed[2], url))
749
def determine_relative_path(from_path, to_path):
750
"""Determine a relative path from from_path to to_path."""
751
from_segments = osutils.splitpath(from_path)
752
to_segments = osutils.splitpath(to_path)
754
for count, (from_element, to_element) in enumerate(zip(from_segments,
756
if from_element != to_element:
760
unique_from = from_segments[count:]
761
unique_to = to_segments[count:]
762
segments = (['..'] * len(unique_from) + unique_to)
763
if len(segments) == 0:
765
return osutils.pathjoin(*segments)
770
"""Extract the server address, the credentials and the path from the url.
772
user, password, host and path should be quoted if they contain reserved
775
:param url: an quoted url
777
:return: (scheme, user, password, host, port, path) tuple, all fields
780
if isinstance(url, unicode):
781
raise errors.InvalidURL('should be ascii:\n%r' % url)
782
url = url.encode('utf-8')
783
(scheme, netloc, path, params,
784
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
785
user = password = host = port = None
787
user, host = netloc.rsplit('@', 1)
789
user, password = user.split(':', 1)
790
password = urllib.unquote(password)
791
user = urllib.unquote(user)
795
if ':' in host and not (host[0] == '[' and host[-1] == ']'): #there *is* port
796
host, port = host.rsplit(':',1)
800
raise errors.InvalidURL('invalid port number %s in url:\n%s' %
802
if host != "" and host[0] == '[' and host[-1] == ']': #IPv6
805
host = urllib.unquote(host)
806
path = urllib.unquote(path)
808
return (scheme, user, password, host, port, path)
added
removed
bzrlib/urlutils.py
