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 (win32_url[3] not in ('abcdefghijklmnopqrstuvwxyz'
228
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
229
or win32_url[4] not in '|:'
230
or win32_url[5] != '/'):
231
raise errors.InvalidURL(url, 'Win32 file urls start with'
232
' file:///x:/, where x is a valid drive letter')
233
return win32_url[3].upper() + u':' + unescape(win32_url[5:])
236
def _win32_local_path_to_url(path):
237
"""Convert a local path like ./foo into a URL like file:///C:/path/to/foo
239
This also handles transforming escaping unicode characters, etc.
241
# importing directly from ntpath allows us to test this
242
# on non-win32 platform
243
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
244
# which actually strips trailing space characters.
245
# The worst part is that on linux ntpath.abspath has different
246
# semantics, since 'nt' is not an available module.
250
win32_path = osutils._win32_abspath(path)
251
# check for UNC path \\HOST\path
252
if win32_path.startswith('//'):
253
return 'file:' + escape(win32_path)
254
return ('file:///' + str(win32_path[0].upper()) + ':' +
255
escape(win32_path[2:]))
258
local_path_to_url = _posix_local_path_to_url
259
local_path_from_url = _posix_local_path_from_url
260
MIN_ABS_FILEURL_LENGTH = len('file:///')
261
WIN32_MIN_ABS_FILEURL_LENGTH = len('file:///C:/')
263
if sys.platform == 'win32':
264
local_path_to_url = _win32_local_path_to_url
265
local_path_from_url = _win32_local_path_from_url
267
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
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()
287
def normalize_url(url):
288
"""Make sure that a path string is in fully normalized URL form.
290
This handles URLs which have unicode characters, spaces,
291
special characters, etc.
293
It has two basic modes of operation, depending on whether the
294
supplied string starts with a url specifier (scheme://) or not.
295
If it does not have a specifier it is considered a local path,
296
and will be converted into a file:/// url. Non-ascii characters
297
will be encoded using utf-8.
298
If it does have a url specifier, it will be treated as a "hybrid"
299
URL. Basically, a URL that should have URL special characters already
300
escaped (like +?&# etc), but may have unicode characters, etc
301
which would not be valid in a real URL.
303
:param url: Either a hybrid URL or a local path
304
:return: A normalized URL which only includes 7-bit ASCII characters.
306
scheme_end, path_start = _find_scheme_and_separator(url)
307
if scheme_end is None:
308
return local_path_to_url(url)
309
prefix = url[:path_start]
310
path = url[path_start:]
311
if not isinstance(url, unicode):
313
if c not in _url_safe_characters:
314
raise errors.InvalidURL(url, 'URLs can only contain specific'
315
' safe characters (not %r)' % c)
316
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
317
return str(prefix + ''.join(path))
319
# We have a unicode (hybrid) url
320
path_chars = list(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)
332
def relative_url(base, other):
333
"""Return a path to other from base.
335
If other is unrelated to base, return other. Else return a relative path.
336
This assumes no symlinks as part of the url.
338
dummy, base_first_slash = _find_scheme_and_separator(base)
339
if base_first_slash is None:
342
dummy, other_first_slash = _find_scheme_and_separator(other)
343
if other_first_slash is None:
346
# this takes care of differing schemes or hosts
347
base_scheme = base[:base_first_slash]
348
other_scheme = other[:other_first_slash]
349
if base_scheme != other_scheme:
351
elif sys.platform == 'win32' and base_scheme == 'file://':
352
base_drive = base[base_first_slash+1:base_first_slash+3]
353
other_drive = other[other_first_slash+1:other_first_slash+3]
354
if base_drive != other_drive:
357
base_path = base[base_first_slash+1:]
358
other_path = other[other_first_slash+1:]
360
if base_path.endswith('/'):
361
base_path = base_path[:-1]
363
base_sections = base_path.split('/')
364
other_sections = other_path.split('/')
366
if base_sections == ['']:
368
if other_sections == ['']:
372
for b, o in zip(base_sections, other_sections):
375
output_sections.append(b)
377
match_len = len(output_sections)
378
output_sections = ['..' for x in base_sections[match_len:]]
379
output_sections.extend(other_sections[match_len:])
381
return "/".join(output_sections) or "."
384
def _win32_extract_drive_letter(url_base, path):
385
"""On win32 the drive letter needs to be added to the url base."""
386
# Strip off the drive letter
387
# path is currently /C:/foo
388
if len(path) < 3 or path[2] not in ':|' or path[3] != '/':
389
raise errors.InvalidURL(url_base + path,
390
'win32 file:/// paths need a drive letter')
391
url_base += path[0:3] # file:// + /C:
392
path = path[3:] # /foo
393
return url_base, path
396
def split(url, exclude_trailing_slash=True):
397
"""Split a URL into its parent directory and a child directory.
399
:param url: A relative or absolute URL
400
:param exclude_trailing_slash: Strip off a final '/' if it is part
401
of the path (but not if it is part of the protocol specification)
403
:return: (parent_url, child_dir). child_dir may be the empty string if we're at
406
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
408
if first_path_slash is None:
409
# We have either a relative path, or no separating slash
410
if scheme_loc is None:
412
if exclude_trailing_slash and url.endswith('/'):
414
return _posix_split(url)
416
# Scheme with no path
419
# We have a fully defined path
420
url_base = url[:first_path_slash] # http://host, file://
421
path = url[first_path_slash:] # /file/foo
423
if sys.platform == 'win32' and url.startswith('file:///'):
424
# Strip off the drive letter
425
# url_base is currently file://
426
# path is currently /C:/foo
427
url_base, path = _win32_extract_drive_letter(url_base, path)
428
# now it should be file:///C: and /foo
430
if exclude_trailing_slash and len(path) > 1 and path.endswith('/'):
432
head, tail = _posix_split(path)
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())])
508
def _win32_strip_local_trailing_slash(url):
509
"""Strip slashes after the drive letter"""
510
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
516
def strip_trailing_slash(url):
517
"""Strip trailing slash, except for root paths.
519
The definition of 'root path' is platform-dependent.
520
This assumes that all URLs are valid netloc urls, such that they
523
It searches for ://, and then refuses to remove the next '/'.
524
It can also handle relative paths
526
path/to/foo => path/to/foo
527
path/to/foo/ => path/to/foo
528
http://host/path/ => http://host/path
529
http://host/path => http://host/path
530
http://host/ => http://host/
532
file:///foo/ => file:///foo
533
# This is unique on win32 platforms, and is the only URL
534
# format which does it differently.
535
file:///c|/ => file:///c:/
537
if not url.endswith('/'):
540
if sys.platform == 'win32' and url.startswith('file://'):
541
return _win32_strip_local_trailing_slash(url)
543
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
544
if scheme_loc is None:
545
# This is a relative path, as it has no scheme
546
# so just chop off the last character
549
if first_path_slash is None or first_path_slash == len(url)-1:
550
# Don't chop off anything if the only slash is the path
558
"""Unescape relpath from url format.
560
This returns a Unicode path from a URL
562
# jam 20060427 URLs are supposed to be ASCII only strings
563
# If they are passed in as unicode, urllib.unquote
564
# will return a UNICODE string, which actually contains
565
# utf-8 bytes. So we have to ensure that they are
566
# plain ASCII strings, or the final .decode will
567
# try to encode the UNICODE => ASCII, and then decode
571
except UnicodeError, e:
572
raise errors.InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
574
unquoted = urllib.unquote(url)
576
unicode_path = unquoted.decode('utf-8')
577
except UnicodeError, e:
578
raise errors.InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
582
# These are characters that if escaped, should stay that way
583
_no_decode_chars = ';/?:@&=+$,#'
584
_no_decode_ords = [ord(c) for c in _no_decode_chars]
585
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
586
+ ['%02X' % o for o in _no_decode_ords])
587
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
588
+ [('%02X' % o, chr(o)) for o in range(256)]))
589
#These entries get mapped to themselves
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
601
# These characters should not be escaped
602
_url_safe_characters = set(
603
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
604
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
605
"0123456789" # Numbers
606
"_.-!~*'()" # Unreserved characters
607
"/;?:@&=+$," # Reserved characters
608
"%#" # Extra reserved characters
611
def unescape_for_display(url, encoding):
612
"""Decode what you can for a URL, so that we get a nice looking path.
614
This will turn file:// urls into local paths, and try to decode
615
any portions of a http:// style url that it can.
617
Any sections of the URL which can't be represented in the encoding or
618
need to stay as escapes are left alone.
620
:param url: A 7-bit ASCII URL
621
:param encoding: The final output encoding
623
:return: A unicode string which can be safely encoded into the
627
raise ValueError('you cannot specify None for the display encoding')
628
if url.startswith('file://'):
630
path = local_path_from_url(url)
631
path.encode(encoding)
636
# Split into sections to try to decode utf-8
638
for i in xrange(1, len(res)):
639
escaped_chunks = res[i].split('%')
640
for j in xrange(1, len(escaped_chunks)):
641
item = escaped_chunks[j]
643
escaped_chunks[j] = _hex_display_map[item[:2]] + item[2:]
645
# Put back the percent symbol
646
escaped_chunks[j] = '%' + item
647
except UnicodeDecodeError:
648
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
649
unescaped = ''.join(escaped_chunks)
651
decoded = unescaped.decode('utf-8')
652
except UnicodeDecodeError:
653
# If this path segment cannot be properly utf-8 decoded
654
# after doing unescaping we will just leave it alone
658
decoded.encode(encoding)
659
except UnicodeEncodeError:
660
# If this chunk cannot be encoded in the local
661
# encoding, then we should leave it alone
664
# Otherwise take the url decoded one
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
