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
37
def basename(url, exclude_trailing_slash=True):
38
"""Return the last component of a URL.
40
:param url: The URL in question
41
:param exclude_trailing_slash: If the url looks like "path/to/foo/"
42
ignore the final slash and return 'foo' rather than ''
43
:return: Just the final component of the URL. This can return ''
44
if you don't exclude_trailing_slash, or if you are at the
47
return split(url, exclude_trailing_slash=exclude_trailing_slash)[1]
50
def dirname(url, exclude_trailing_slash=True):
51
"""Return the parent directory of the given path.
53
:param url: Relative or absolute URL
54
:param exclude_trailing_slash: Remove a final slash
55
(treat http://host/foo/ as http://host/foo, but
56
http://host/ stays http://host/)
57
:return: Everything in the URL except the last path chunk
59
# TODO: jam 20060502 This was named dirname to be consistent
60
# with the os functions, but maybe "parent" would be better
61
return split(url, exclude_trailing_slash=exclude_trailing_slash)[0]
65
"""Escape relpath to be a valid url."""
66
if isinstance(relpath, unicode):
67
relpath = relpath.encode('utf-8')
68
# After quoting and encoding, the path should be perfectly
69
# safe as a plain ASCII string, str() just enforces this
70
return str(urllib.quote(relpath))
73
def file_relpath(base, path):
74
"""Compute just the relative sub-portion of a url
76
This assumes that both paths are already fully specified file:// URLs.
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)
82
base = local_path_from_url(base)
83
path = local_path_from_url(path)
84
return escape(osutils.relpath(base, path))
87
def _find_scheme_and_separator(url):
88
"""Find the scheme separator (://) and the first path separator
90
This is just a helper functions for other path utilities.
91
It could probably be replaced by urlparse
93
m = _url_scheme_re.match(url)
97
scheme = m.group('scheme')
98
path = m.group('path')
100
# Find the path separating slash
101
# (first slash after the ://)
102
first_path_slash = path.find('/')
103
if first_path_slash == -1:
104
return len(scheme), None
105
return len(scheme), first_path_slash+len(scheme)+3
108
def join(base, *args):
109
"""Create a URL by joining sections.
111
This will normalize '..', assuming that paths are absolute
112
(it assumes no symlinks in either path)
114
If any of *args is an absolute URL, it will be treated correctly.
116
join('http://foo', 'http://bar') => 'http://bar'
117
join('http://foo', 'bar') => 'http://foo/bar'
118
join('http://foo', 'bar', '../baz') => 'http://foo/baz'
122
match = _url_scheme_re.match(base)
125
scheme = match.group('scheme')
126
path = match.group('path').split('/')
127
if path[-1:] == ['']:
128
# Strip off a trailing slash
129
# This helps both when we are at the root, and when
130
# 'base' has an extra slash at the end
133
path = base.split('/')
135
if scheme is not None and len(path) >= 1:
137
# the path should be represented as an abs path.
138
# we know this must be absolute because of the presence of a URL scheme.
140
path = [''] + path[1:]
142
# create an empty host, but dont alter the path - this might be a
143
# relative url fragment.
148
match = _url_scheme_re.match(arg)
151
scheme = match.group('scheme')
152
# this skips .. normalisation, making http://host/../../..
154
path = match.group('path').split('/')
155
# set the host and path according to new absolute URL, discarding
156
# any previous values.
157
# XXX: duplicates mess from earlier in this function. This URL
158
# manipulation code needs some cleaning up.
159
if scheme is not None and len(path) >= 1:
162
# url scheme implies absolute path.
165
# no url scheme we take the path as is.
168
path = '/'.join(path)
169
path = joinpath(path, arg)
170
path = path.split('/')
171
if remove_root and path[0:1] == ['']:
174
# Remove the leading slash from the path, so long as it isn't also the
175
# trailing slash, which we want to keep if present.
176
if path and path[0] == '' and len(path) > 1:
181
return '/'.join(path)
182
return scheme + '://' + '/'.join(path)
185
def joinpath(base, *args):
186
"""Join URL path segments to a URL path segment.
188
This is somewhat like osutils.joinpath, but intended for URLs.
190
XXX: this duplicates some normalisation logic, and also duplicates a lot of
191
path handling logic that already exists in some Transport implementations.
192
We really should try to have exactly one place in the code base responsible
193
for combining paths of URLs.
195
path = base.split('/')
196
if len(path) > 1 and path[-1] == '':
197
#If the path ends in a trailing /, remove it.
200
if arg.startswith('/'):
202
for chunk in arg.split('/'):
207
raise errors.InvalidURLJoin('Cannot go above root',
215
return '/'.join(path)
218
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
219
def _posix_local_path_from_url(url):
220
"""Convert a url like file:///path/to/foo into /path/to/foo"""
221
if not url.startswith('file:///'):
222
raise errors.InvalidURL(url, 'local urls must start with file:///')
223
# We only strip off 2 slashes
224
return unescape(url[len('file://'):])
227
def _posix_local_path_to_url(path):
228
"""Convert a local path like ./foo into a URL like file:///path/to/foo
230
This also handles transforming escaping unicode characters, etc.
232
# importing directly from posixpath allows us to test this
233
# on non-posix platforms
234
return 'file://' + escape(_posix_normpath(
235
osutils._posix_abspath(path)))
238
def _win32_local_path_from_url(url):
239
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
240
if not url.startswith('file://'):
241
raise errors.InvalidURL(url, 'local urls must start with file:///, '
242
'UNC path urls must start with file://')
243
# We strip off all 3 slashes
244
win32_url = url[len('file:'):]
245
# check for UNC path: //HOST/path
246
if not win32_url.startswith('///'):
247
if (win32_url[2] == '/'
248
or win32_url[3] in '|:'):
249
raise errors.InvalidURL(url, 'Win32 UNC path urls'
250
' have form file://HOST/path')
251
return unescape(win32_url)
252
# usual local path with drive letter
253
if (win32_url[3] not in ('abcdefghijklmnopqrstuvwxyz'
254
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
255
or win32_url[4] not in '|:'
256
or win32_url[5] != '/'):
257
raise errors.InvalidURL(url, 'Win32 file urls start with'
258
' file:///x:/, where x is a valid drive letter')
259
return win32_url[3].upper() + u':' + unescape(win32_url[5:])
262
def _win32_local_path_to_url(path):
263
"""Convert a local path like ./foo into a URL like file:///C:/path/to/foo
265
This also handles transforming escaping unicode characters, etc.
267
# importing directly from ntpath allows us to test this
268
# on non-win32 platform
269
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
270
# which actually strips trailing space characters.
271
# The worst part is that under linux ntpath.abspath has different
272
# semantics, since 'nt' is not an available module.
273
win32_path = osutils._win32_abspath(path)
274
# check for UNC path \\HOST\path
275
if win32_path.startswith('//'):
276
return 'file:' + escape(win32_path)
277
return 'file:///' + win32_path[0].upper() + ':' + escape(win32_path[2:])
280
local_path_to_url = _posix_local_path_to_url
281
local_path_from_url = _posix_local_path_from_url
282
MIN_ABS_FILEURL_LENGTH = len('file:///')
283
WIN32_MIN_ABS_FILEURL_LENGTH = len('file:///C:/')
285
if sys.platform == 'win32':
286
local_path_to_url = _win32_local_path_to_url
287
local_path_from_url = _win32_local_path_from_url
289
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
292
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
293
_url_hex_escapes_re = re.compile(r'(%[0-9a-fA-F]{2})')
296
def _unescape_safe_chars(matchobj):
297
"""re.sub callback to convert hex-escapes to plain characters (if safe).
299
e.g. '%7E' will be converted to '~'.
301
hex_digits = matchobj.group(0)[1:]
302
char = chr(int(hex_digits, 16))
303
if char in _url_dont_escape_characters:
306
return matchobj.group(0).upper()
309
def normalize_url(url):
310
"""Make sure that a path string is in fully normalized URL form.
312
This handles URLs which have unicode characters, spaces,
313
special characters, etc.
315
It has two basic modes of operation, depending on whether the
316
supplied string starts with a url specifier (scheme://) or not.
317
If it does not have a specifier it is considered a local path,
318
and will be converted into a file:/// url. Non-ascii characters
319
will be encoded using utf-8.
320
If it does have a url specifier, it will be treated as a "hybrid"
321
URL. Basically, a URL that should have URL special characters already
322
escaped (like +?&# etc), but may have unicode characters, etc
323
which would not be valid in a real URL.
325
:param url: Either a hybrid URL or a local path
326
:return: A normalized URL which only includes 7-bit ASCII characters.
328
m = _url_scheme_re.match(url)
330
return local_path_to_url(url)
331
scheme = m.group('scheme')
332
path = m.group('path')
333
if not isinstance(url, unicode):
335
if c not in _url_safe_characters:
336
raise errors.InvalidURL(url, 'URLs can only contain specific'
337
' safe characters (not %r)' % c)
338
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
339
return str(scheme + '://' + ''.join(path))
341
# We have a unicode (hybrid) url
342
path_chars = list(path)
344
for i in xrange(len(path_chars)):
345
if path_chars[i] not in _url_safe_characters:
346
chars = path_chars[i].encode('utf-8')
347
path_chars[i] = ''.join(
348
['%%%02X' % ord(c) for c in path_chars[i].encode('utf-8')])
349
path = ''.join(path_chars)
350
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
351
return str(scheme + '://' + path)
354
def relative_url(base, other):
355
"""Return a path to other from base.
357
If other is unrelated to base, return other. Else return a relative path.
358
This assumes no symlinks as part of the url.
360
dummy, base_first_slash = _find_scheme_and_separator(base)
361
if base_first_slash is None:
364
dummy, other_first_slash = _find_scheme_and_separator(other)
365
if other_first_slash is None:
368
# this takes care of differing schemes or hosts
369
base_scheme = base[:base_first_slash]
370
other_scheme = other[:other_first_slash]
371
if base_scheme != other_scheme:
374
base_path = base[base_first_slash+1:]
375
other_path = other[other_first_slash+1:]
377
if base_path.endswith('/'):
378
base_path = base_path[:-1]
380
base_sections = base_path.split('/')
381
other_sections = other_path.split('/')
383
if base_sections == ['']:
385
if other_sections == ['']:
389
for b, o in zip(base_sections, other_sections):
392
output_sections.append(b)
394
match_len = len(output_sections)
395
output_sections = ['..' for x in base_sections[match_len:]]
396
output_sections.extend(other_sections[match_len:])
398
return "/".join(output_sections) or "."
401
def _win32_extract_drive_letter(url_base, path):
402
"""On win32 the drive letter needs to be added to the url base."""
403
# Strip off the drive letter
404
# path is currently /C:/foo
405
if len(path) < 3 or path[2] not in ':|' or path[3] != '/':
406
raise errors.InvalidURL(url_base + path,
407
'win32 file:/// paths need a drive letter')
408
url_base += path[0:3] # file:// + /C:
409
path = path[3:] # /foo
410
return url_base, path
413
def split(url, exclude_trailing_slash=True):
414
"""Split a URL into its parent directory and a child directory.
416
:param url: A relative or absolute URL
417
:param exclude_trailing_slash: Strip off a final '/' if it is part
418
of the path (but not if it is part of the protocol specification)
420
:return: (parent_url, child_dir). child_dir may be the empty string if we're at
423
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
425
if first_path_slash is None:
426
# We have either a relative path, or no separating slash
427
if scheme_loc is None:
429
if exclude_trailing_slash and url.endswith('/'):
431
return _posix_split(url)
433
# Scheme with no path
436
# We have a fully defined path
437
url_base = url[:first_path_slash] # http://host, file://
438
path = url[first_path_slash:] # /file/foo
440
if sys.platform == 'win32' and url.startswith('file:///'):
441
# Strip off the drive letter
442
# url_base is currently file://
443
# path is currently /C:/foo
444
url_base, path = _win32_extract_drive_letter(url_base, path)
445
# now it should be file:///C: and /foo
447
if exclude_trailing_slash and len(path) > 1 and path.endswith('/'):
449
head, tail = _posix_split(path)
450
return url_base + head, tail
453
def _win32_strip_local_trailing_slash(url):
454
"""Strip slashes after the drive letter"""
455
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
461
def strip_trailing_slash(url):
462
"""Strip trailing slash, except for root paths.
464
The definition of 'root path' is platform-dependent.
465
This assumes that all URLs are valid netloc urls, such that they
468
It searches for ://, and then refuses to remove the next '/'.
469
It can also handle relative paths
471
path/to/foo => path/to/foo
472
path/to/foo/ => path/to/foo
473
http://host/path/ => http://host/path
474
http://host/path => http://host/path
475
http://host/ => http://host/
477
file:///foo/ => file:///foo
478
# This is unique on win32 platforms, and is the only URL
479
# format which does it differently.
480
file:///c|/ => file:///c:/
482
if not url.endswith('/'):
485
if sys.platform == 'win32' and url.startswith('file://'):
486
return _win32_strip_local_trailing_slash(url)
488
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
489
if scheme_loc is None:
490
# This is a relative path, as it has no scheme
491
# so just chop off the last character
494
if first_path_slash is None or first_path_slash == len(url)-1:
495
# Don't chop off anything if the only slash is the path
503
"""Unescape relpath from url format.
505
This returns a Unicode path from a URL
507
# jam 20060427 URLs are supposed to be ASCII only strings
508
# If they are passed in as unicode, urllib.unquote
509
# will return a UNICODE string, which actually contains
510
# utf-8 bytes. So we have to ensure that they are
511
# plain ASCII strings, or the final .decode will
512
# try to encode the UNICODE => ASCII, and then decode
516
except UnicodeError, e:
517
raise errors.InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
519
unquoted = urllib.unquote(url)
521
unicode_path = unquoted.decode('utf-8')
522
except UnicodeError, e:
523
raise errors.InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
527
# These are characters that if escaped, should stay that way
528
_no_decode_chars = ';/?:@&=+$,#'
529
_no_decode_ords = [ord(c) for c in _no_decode_chars]
530
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
531
+ ['%02X' % o for o in _no_decode_ords])
532
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
533
+ [('%02X' % o, chr(o)) for o in range(256)]))
534
#These entries get mapped to themselves
535
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
537
# These characters shouldn't be percent-encoded, and it's always safe to
538
# unencode them if they are.
539
_url_dont_escape_characters = set(
540
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
541
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
542
"0123456789" # Numbers
543
"-._~" # Unreserved characters
546
# These characters should not be escaped
547
_url_safe_characters = set(
548
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
549
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
550
"0123456789" # Numbers
551
"_.-!~*'()" # Unreserved characters
552
"/;?:@&=+$," # Reserved characters
553
"%#" # Extra reserved characters
556
def unescape_for_display(url, encoding):
557
"""Decode what you can for a URL, so that we get a nice looking path.
559
This will turn file:// urls into local paths, and try to decode
560
any portions of a http:// style url that it can.
562
Any sections of the URL which can't be represented in the encoding or
563
need to stay as escapes are left alone.
565
:param url: A 7-bit ASCII URL
566
:param encoding: The final output encoding
568
:return: A unicode string which can be safely encoded into the
571
assert encoding is not None, 'you cannot specify None for the display encoding.'
572
if url.startswith('file://'):
574
path = local_path_from_url(url)
575
path.encode(encoding)
580
# Split into sections to try to decode utf-8
582
for i in xrange(1, len(res)):
583
escaped_chunks = res[i].split('%')
584
for j in xrange(1, len(escaped_chunks)):
585
item = escaped_chunks[j]
587
escaped_chunks[j] = _hex_display_map[item[:2]] + item[2:]
589
# Put back the percent symbol
590
escaped_chunks[j] = '%' + item
591
except UnicodeDecodeError:
592
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
593
unescaped = ''.join(escaped_chunks)
595
decoded = unescaped.decode('utf-8')
596
except UnicodeDecodeError:
597
# If this path segment cannot be properly utf-8 decoded
598
# after doing unescaping we will just leave it alone
602
decoded.encode(encoding)
603
except UnicodeEncodeError:
604
# If this chunk cannot be encoded in the local
605
# encoding, then we should leave it alone
608
# Otherwise take the url decoded one
610
return u'/'.join(res)
added
removed
bzrlib/urlutils.py
