1
# Bazaar -- distributed version control
3
# Copyright (C) 2006 Canonical Ltd
5
# This program is free software; you can redistribute it and/or modify
6
# it under the terms of the GNU General Public License as published by
7
# the Free Software Foundation; either version 2 of the License, or
8
# (at your option) any later version.
10
# This program is distributed in the hope that it will be useful,
11
# but WITHOUT ANY WARRANTY; without even the implied warranty of
12
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13
# GNU General Public License for more details.
15
# You should have received a copy of the GNU General Public License
16
# along with this program; if not, write to the Free Software
17
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
19
"""A collection of function for handling URL operations."""
25
from bzrlib.lazy_import import lazy_import
26
lazy_import(globals(), """
27
from posixpath import split as _posix_split, normpath as _posix_normpath
38
def basename(url, exclude_trailing_slash=True):
39
"""Return the last component of a URL.
41
:param url: The URL in question
42
:param exclude_trailing_slash: If the url looks like "path/to/foo/"
43
ignore the final slash and return 'foo' rather than ''
44
:return: Just the final component of the URL. This can return ''
45
if you don't exclude_trailing_slash, or if you are at the
48
return split(url, exclude_trailing_slash=exclude_trailing_slash)[1]
51
def dirname(url, exclude_trailing_slash=True):
52
"""Return the parent directory of the given path.
54
:param url: Relative or absolute URL
55
:param exclude_trailing_slash: Remove a final slash
56
(treat http://host/foo/ as http://host/foo, but
57
http://host/ stays http://host/)
58
:return: Everything in the URL except the last path chunk
60
# TODO: jam 20060502 This was named dirname to be consistent
61
# with the os functions, but maybe "parent" would be better
62
return split(url, exclude_trailing_slash=exclude_trailing_slash)[0]
66
"""Escape relpath to be a valid url."""
67
if isinstance(relpath, unicode):
68
relpath = relpath.encode('utf-8')
69
# After quoting and encoding, the path should be perfectly
70
# safe as a plain ASCII string, str() just enforces this
71
return str(urllib.quote(relpath))
74
def file_relpath(base, path):
75
"""Compute just the relative sub-portion of a url
77
This assumes that both paths are already fully specified file:// URLs.
79
if len(base) < MIN_ABS_FILEURL_LENGTH:
80
raise ValueError('Length of base must be equal or'
81
' exceed the platform minimum url length (which is %d)' %
82
MIN_ABS_FILEURL_LENGTH)
83
base = local_path_from_url(base)
84
path = local_path_from_url(path)
85
return escape(osutils.relpath(base, path))
88
def _find_scheme_and_separator(url):
89
"""Find the scheme separator (://) and the first path separator
91
This is just a helper functions for other path utilities.
92
It could probably be replaced by urlparse
94
m = _url_scheme_re.match(url)
98
scheme = m.group('scheme')
99
path = m.group('path')
101
# Find the path separating slash
102
# (first slash after the ://)
103
first_path_slash = path.find('/')
104
if first_path_slash == -1:
105
return len(scheme), None
106
return len(scheme), first_path_slash+len(scheme)+3
109
def join(base, *args):
110
"""Create a URL by joining sections.
112
This will normalize '..', assuming that paths are absolute
113
(it assumes no symlinks in either path)
115
If any of *args is an absolute URL, it will be treated correctly.
117
join('http://foo', 'http://bar') => 'http://bar'
118
join('http://foo', 'bar') => 'http://foo/bar'
119
join('http://foo', 'bar', '../baz') => 'http://foo/baz'
123
match = _url_scheme_re.match(base)
126
scheme = match.group('scheme')
127
path = match.group('path').split('/')
128
if path[-1:] == ['']:
129
# Strip off a trailing slash
130
# This helps both when we are at the root, and when
131
# 'base' has an extra slash at the end
134
path = base.split('/')
136
if scheme is not None and len(path) >= 1:
138
# the path should be represented as an abs path.
139
# we know this must be absolute because of the presence of a URL scheme.
141
path = [''] + path[1:]
143
# create an empty host, but dont alter the path - this might be a
144
# relative url fragment.
149
match = _url_scheme_re.match(arg)
152
scheme = match.group('scheme')
153
# this skips .. normalisation, making http://host/../../..
155
path = match.group('path').split('/')
156
# set the host and path according to new absolute URL, discarding
157
# any previous values.
158
# XXX: duplicates mess from earlier in this function. This URL
159
# manipulation code needs some cleaning up.
160
if scheme is not None and len(path) >= 1:
163
# url scheme implies absolute path.
166
# no url scheme we take the path as is.
169
path = '/'.join(path)
170
path = joinpath(path, arg)
171
path = path.split('/')
172
if remove_root and path[0:1] == ['']:
175
# Remove the leading slash from the path, so long as it isn't also the
176
# trailing slash, which we want to keep if present.
177
if path and path[0] == '' and len(path) > 1:
182
return '/'.join(path)
183
return scheme + '://' + '/'.join(path)
186
def joinpath(base, *args):
187
"""Join URL path segments to a URL path segment.
189
This is somewhat like osutils.joinpath, but intended for URLs.
191
XXX: this duplicates some normalisation logic, and also duplicates a lot of
192
path handling logic that already exists in some Transport implementations.
193
We really should try to have exactly one place in the code base responsible
194
for combining paths of URLs.
196
path = base.split('/')
197
if len(path) > 1 and path[-1] == '':
198
#If the path ends in a trailing /, remove it.
201
if arg.startswith('/'):
203
for chunk in arg.split('/'):
208
raise errors.InvalidURLJoin('Cannot go above root',
216
return '/'.join(path)
219
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
220
def _posix_local_path_from_url(url):
221
"""Convert a url like file:///path/to/foo into /path/to/foo"""
222
if not url.startswith('file:///'):
223
raise errors.InvalidURL(url, 'local urls must start with file:///')
224
# We only strip off 2 slashes
225
return unescape(url[len('file://'):])
228
def _posix_local_path_to_url(path):
229
"""Convert a local path like ./foo into a URL like file:///path/to/foo
231
This also handles transforming escaping unicode characters, etc.
233
# importing directly from posixpath allows us to test this
234
# on non-posix platforms
235
return 'file://' + escape(_posix_normpath(
236
osutils._posix_abspath(path)))
239
def _win32_local_path_from_url(url):
240
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
241
if not url.startswith('file://'):
242
raise errors.InvalidURL(url, 'local urls must start with file:///, '
243
'UNC path urls must start with file://')
244
# We strip off all 3 slashes
245
win32_url = url[len('file:'):]
246
# check for UNC path: //HOST/path
247
if not win32_url.startswith('///'):
248
if (win32_url[2] == '/'
249
or win32_url[3] in '|:'):
250
raise errors.InvalidURL(url, 'Win32 UNC path urls'
251
' have form file://HOST/path')
252
return unescape(win32_url)
254
# allow empty paths so we can serve all roots
255
if win32_url == '///':
258
# usual local path with drive letter
259
if (win32_url[3] not in ('abcdefghijklmnopqrstuvwxyz'
260
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
261
or win32_url[4] not in '|:'
262
or win32_url[5] != '/'):
263
raise errors.InvalidURL(url, 'Win32 file urls start with'
264
' file:///x:/, where x is a valid drive letter')
265
return win32_url[3].upper() + u':' + unescape(win32_url[5:])
268
def _win32_local_path_to_url(path):
269
"""Convert a local path like ./foo into a URL like file:///C:/path/to/foo
271
This also handles transforming escaping unicode characters, etc.
273
# importing directly from ntpath allows us to test this
274
# on non-win32 platform
275
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
276
# which actually strips trailing space characters.
277
# The worst part is that under linux ntpath.abspath has different
278
# semantics, since 'nt' is not an available module.
282
win32_path = osutils._win32_abspath(path)
283
# check for UNC path \\HOST\path
284
if win32_path.startswith('//'):
285
return 'file:' + escape(win32_path)
286
return ('file:///' + str(win32_path[0].upper()) + ':' +
287
escape(win32_path[2:]))
290
local_path_to_url = _posix_local_path_to_url
291
local_path_from_url = _posix_local_path_from_url
292
MIN_ABS_FILEURL_LENGTH = len('file:///')
293
WIN32_MIN_ABS_FILEURL_LENGTH = len('file:///C:/')
295
if sys.platform == 'win32':
296
local_path_to_url = _win32_local_path_to_url
297
local_path_from_url = _win32_local_path_from_url
299
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
302
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
303
_url_hex_escapes_re = re.compile(r'(%[0-9a-fA-F]{2})')
306
def _unescape_safe_chars(matchobj):
307
"""re.sub callback to convert hex-escapes to plain characters (if safe).
309
e.g. '%7E' will be converted to '~'.
311
hex_digits = matchobj.group(0)[1:]
312
char = chr(int(hex_digits, 16))
313
if char in _url_dont_escape_characters:
316
return matchobj.group(0).upper()
319
def normalize_url(url):
320
"""Make sure that a path string is in fully normalized URL form.
322
This handles URLs which have unicode characters, spaces,
323
special characters, etc.
325
It has two basic modes of operation, depending on whether the
326
supplied string starts with a url specifier (scheme://) or not.
327
If it does not have a specifier it is considered a local path,
328
and will be converted into a file:/// url. Non-ascii characters
329
will be encoded using utf-8.
330
If it does have a url specifier, it will be treated as a "hybrid"
331
URL. Basically, a URL that should have URL special characters already
332
escaped (like +?&# etc), but may have unicode characters, etc
333
which would not be valid in a real URL.
335
:param url: Either a hybrid URL or a local path
336
:return: A normalized URL which only includes 7-bit ASCII characters.
338
m = _url_scheme_re.match(url)
340
return local_path_to_url(url)
341
scheme = m.group('scheme')
342
path = m.group('path')
343
if not isinstance(url, unicode):
345
if c not in _url_safe_characters:
346
raise errors.InvalidURL(url, 'URLs can only contain specific'
347
' safe characters (not %r)' % c)
348
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
349
return str(scheme + '://' + ''.join(path))
351
# We have a unicode (hybrid) url
352
path_chars = list(path)
354
for i in xrange(len(path_chars)):
355
if path_chars[i] not in _url_safe_characters:
356
chars = path_chars[i].encode('utf-8')
357
path_chars[i] = ''.join(
358
['%%%02X' % ord(c) for c in path_chars[i].encode('utf-8')])
359
path = ''.join(path_chars)
360
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
361
return str(scheme + '://' + path)
364
def relative_url(base, other):
365
"""Return a path to other from base.
367
If other is unrelated to base, return other. Else return a relative path.
368
This assumes no symlinks as part of the url.
370
dummy, base_first_slash = _find_scheme_and_separator(base)
371
if base_first_slash is None:
374
dummy, other_first_slash = _find_scheme_and_separator(other)
375
if other_first_slash is None:
378
# this takes care of differing schemes or hosts
379
base_scheme = base[:base_first_slash]
380
other_scheme = other[:other_first_slash]
381
if base_scheme != other_scheme:
383
elif sys.platform == 'win32' and base_scheme == 'file://':
384
base_drive = base[base_first_slash+1:base_first_slash+3]
385
other_drive = other[other_first_slash+1:other_first_slash+3]
386
if base_drive != other_drive:
389
base_path = base[base_first_slash+1:]
390
other_path = other[other_first_slash+1:]
392
if base_path.endswith('/'):
393
base_path = base_path[:-1]
395
base_sections = base_path.split('/')
396
other_sections = other_path.split('/')
398
if base_sections == ['']:
400
if other_sections == ['']:
404
for b, o in zip(base_sections, other_sections):
407
output_sections.append(b)
409
match_len = len(output_sections)
410
output_sections = ['..' for x in base_sections[match_len:]]
411
output_sections.extend(other_sections[match_len:])
413
return "/".join(output_sections) or "."
416
def _win32_extract_drive_letter(url_base, path):
417
"""On win32 the drive letter needs to be added to the url base."""
418
# Strip off the drive letter
419
# path is currently /C:/foo
420
if len(path) < 3 or path[2] not in ':|' or path[3] != '/':
421
raise errors.InvalidURL(url_base + path,
422
'win32 file:/// paths need a drive letter')
423
url_base += path[0:3] # file:// + /C:
424
path = path[3:] # /foo
425
return url_base, path
428
def split(url, exclude_trailing_slash=True):
429
"""Split a URL into its parent directory and a child directory.
431
:param url: A relative or absolute URL
432
:param exclude_trailing_slash: Strip off a final '/' if it is part
433
of the path (but not if it is part of the protocol specification)
435
:return: (parent_url, child_dir). child_dir may be the empty string if we're at
438
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
440
if first_path_slash is None:
441
# We have either a relative path, or no separating slash
442
if scheme_loc is None:
444
if exclude_trailing_slash and url.endswith('/'):
446
return _posix_split(url)
448
# Scheme with no path
451
# We have a fully defined path
452
url_base = url[:first_path_slash] # http://host, file://
453
path = url[first_path_slash:] # /file/foo
455
if sys.platform == 'win32' and url.startswith('file:///'):
456
# Strip off the drive letter
457
# url_base is currently file://
458
# path is currently /C:/foo
459
url_base, path = _win32_extract_drive_letter(url_base, path)
460
# now it should be file:///C: and /foo
462
if exclude_trailing_slash and len(path) > 1 and path.endswith('/'):
464
head, tail = _posix_split(path)
465
return url_base + head, tail
468
def _win32_strip_local_trailing_slash(url):
469
"""Strip slashes after the drive letter"""
470
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
476
def strip_trailing_slash(url):
477
"""Strip trailing slash, except for root paths.
479
The definition of 'root path' is platform-dependent.
480
This assumes that all URLs are valid netloc urls, such that they
483
It searches for ://, and then refuses to remove the next '/'.
484
It can also handle relative paths
486
path/to/foo => path/to/foo
487
path/to/foo/ => path/to/foo
488
http://host/path/ => http://host/path
489
http://host/path => http://host/path
490
http://host/ => http://host/
492
file:///foo/ => file:///foo
493
# This is unique on win32 platforms, and is the only URL
494
# format which does it differently.
495
file:///c|/ => file:///c:/
497
if not url.endswith('/'):
500
if sys.platform == 'win32' and url.startswith('file://'):
501
return _win32_strip_local_trailing_slash(url)
503
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
504
if scheme_loc is None:
505
# This is a relative path, as it has no scheme
506
# so just chop off the last character
509
if first_path_slash is None or first_path_slash == len(url)-1:
510
# Don't chop off anything if the only slash is the path
518
"""Unescape relpath from url format.
520
This returns a Unicode path from a URL
522
# jam 20060427 URLs are supposed to be ASCII only strings
523
# If they are passed in as unicode, urllib.unquote
524
# will return a UNICODE string, which actually contains
525
# utf-8 bytes. So we have to ensure that they are
526
# plain ASCII strings, or the final .decode will
527
# try to encode the UNICODE => ASCII, and then decode
531
except UnicodeError, e:
532
raise errors.InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
534
unquoted = urllib.unquote(url)
536
unicode_path = unquoted.decode('utf-8')
537
except UnicodeError, e:
538
raise errors.InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
542
# These are characters that if escaped, should stay that way
543
_no_decode_chars = ';/?:@&=+$,#'
544
_no_decode_ords = [ord(c) for c in _no_decode_chars]
545
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
546
+ ['%02X' % o for o in _no_decode_ords])
547
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
548
+ [('%02X' % o, chr(o)) for o in range(256)]))
549
#These entries get mapped to themselves
550
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
552
# These characters shouldn't be percent-encoded, and it's always safe to
553
# unencode them if they are.
554
_url_dont_escape_characters = set(
555
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
556
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
557
"0123456789" # Numbers
558
"-._~" # Unreserved characters
561
# These characters should not be escaped
562
_url_safe_characters = set(
563
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
564
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
565
"0123456789" # Numbers
566
"_.-!~*'()" # Unreserved characters
567
"/;?:@&=+$," # Reserved characters
568
"%#" # Extra reserved characters
571
def unescape_for_display(url, encoding):
572
"""Decode what you can for a URL, so that we get a nice looking path.
574
This will turn file:// urls into local paths, and try to decode
575
any portions of a http:// style url that it can.
577
Any sections of the URL which can't be represented in the encoding or
578
need to stay as escapes are left alone.
580
:param url: A 7-bit ASCII URL
581
:param encoding: The final output encoding
583
:return: A unicode string which can be safely encoded into the
587
raise ValueError('you cannot specify None for the display encoding')
588
if url.startswith('file://'):
590
path = local_path_from_url(url)
591
path.encode(encoding)
596
# Split into sections to try to decode utf-8
598
for i in xrange(1, len(res)):
599
escaped_chunks = res[i].split('%')
600
for j in xrange(1, len(escaped_chunks)):
601
item = escaped_chunks[j]
603
escaped_chunks[j] = _hex_display_map[item[:2]] + item[2:]
605
# Put back the percent symbol
606
escaped_chunks[j] = '%' + item
607
except UnicodeDecodeError:
608
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
609
unescaped = ''.join(escaped_chunks)
611
decoded = unescaped.decode('utf-8')
612
except UnicodeDecodeError:
613
# If this path segment cannot be properly utf-8 decoded
614
# after doing unescaping we will just leave it alone
618
decoded.encode(encoding)
619
except UnicodeEncodeError:
620
# If this chunk cannot be encoded in the local
621
# encoding, then we should leave it alone
624
# Otherwise take the url decoded one
626
return u'/'.join(res)
629
def derive_to_location(from_location):
630
"""Derive a TO_LOCATION given a FROM_LOCATION.
632
The normal case is a FROM_LOCATION of http://foo/bar => bar.
633
The Right Thing for some logical destinations may differ though
634
because no / may be present at all. In that case, the result is
635
the full name without the scheme indicator, e.g. lp:foo-bar => foo-bar.
636
This latter case also applies when a Windows drive
637
is used without a path, e.g. c:foo-bar => foo-bar.
638
If no /, path separator or : is found, the from_location is returned.
640
if from_location.find("/") >= 0 or from_location.find(os.sep) >= 0:
641
return os.path.basename(from_location.rstrip("/\\"))
643
sep = from_location.find(":")
645
return from_location[sep+1:]
650
def _is_absolute(url):
651
return (osutils.pathjoin('/foo', url) == url)
654
def rebase_url(url, old_base, new_base):
655
"""Convert a relative path from an old base URL to a new base URL.
657
The result will be a relative path.
658
Absolute paths and full URLs are returned unaltered.
660
scheme, separator = _find_scheme_and_separator(url)
661
if scheme is not None:
663
if _is_absolute(url):
665
old_parsed = urlparse.urlparse(old_base)
666
new_parsed = urlparse.urlparse(new_base)
667
if (old_parsed[:2]) != (new_parsed[:2]):
668
raise errors.InvalidRebaseURLs(old_base, new_base)
669
return determine_relative_path(new_parsed[2],
670
join(old_parsed[2], url))
673
def determine_relative_path(from_path, to_path):
674
"""Determine a relative path from from_path to to_path."""
675
from_segments = osutils.splitpath(from_path)
676
to_segments = osutils.splitpath(to_path)
678
for count, (from_element, to_element) in enumerate(zip(from_segments,
680
if from_element != to_element:
684
unique_from = from_segments[count:]
685
unique_to = to_segments[count:]
686
segments = (['..'] * len(unique_from) + unique_to)
687
if len(segments) == 0:
689
return osutils.pathjoin(*segments)
added
removed
bzrlib/urlutils.py
