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'
120
m = _url_scheme_re.match(base)
123
scheme = m.group('scheme')
124
path = m.group('path').split('/')
125
if path[-1:] == ['']:
126
# Strip off a trailing slash
127
# This helps both when we are at the root, and when
128
# 'base' has an extra slash at the end
131
path = base.split('/')
133
if scheme is not None and len(path) >= 1:
139
m = _url_scheme_re.match(arg)
142
scheme = m.group('scheme')
143
# this skips .. normalisation, making http://host/../../..
145
path = m.group('path').split('/')
146
# set the host and path according to new absolute URL, discarding
147
# any previous values.
148
# XXX: duplicates mess from earlier in this function. This URL
149
# manipulation code needs some cleaning up.
150
if scheme is not None and len(path) >= 1:
156
path = '/'.join(path)
157
path = joinpath(path, arg)
158
path = path.split('/')
160
# Remove the leading slash from the path, so long as it isn't also the
161
# trailing slash, which we want to keep if present.
162
if path and path[0] == '' and len(path) > 1:
167
return '/'.join(path)
168
return scheme + '://' + '/'.join(path)
171
def joinpath(base, *args):
172
"""Join URL path segments to a URL path segment.
174
This is somewhat like osutils.joinpath, but intended for URLs.
176
XXX: this duplicates some normalisation logic, and also duplicates a lot of
177
path handling logic that already exists in some Transport implementations.
178
We really should try to have exactly one place in the code base responsible
179
for combining paths of URLs.
184
base = base.rstrip('/')
185
path = base.split('/')
187
if arg.startswith('/'):
189
for chunk in arg.split('/'):
194
raise errors.InvalidURLJoin('Cannot go above root',
202
return '/'.join(path)
205
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
206
def _posix_local_path_from_url(url):
207
"""Convert a url like file:///path/to/foo into /path/to/foo"""
208
if not url.startswith('file:///'):
209
raise errors.InvalidURL(url, 'local urls must start with file:///')
210
# We only strip off 2 slashes
211
return unescape(url[len('file://'):])
214
def _posix_local_path_to_url(path):
215
"""Convert a local path like ./foo into a URL like file:///path/to/foo
217
This also handles transforming escaping unicode characters, etc.
219
# importing directly from posixpath allows us to test this
220
# on non-posix platforms
221
return 'file://' + escape(_posix_normpath(
222
osutils._posix_abspath(path)))
225
def _win32_local_path_from_url(url):
226
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
227
if not url.startswith('file://'):
228
raise errors.InvalidURL(url, 'local urls must start with file:///, '
229
'UNC path urls must start with file://')
230
# We strip off all 3 slashes
231
win32_url = url[len('file:'):]
232
# check for UNC path: //HOST/path
233
if not win32_url.startswith('///'):
234
if (win32_url[2] == '/'
235
or win32_url[3] in '|:'):
236
raise errors.InvalidURL(url, 'Win32 UNC path urls'
237
' have form file://HOST/path')
238
return unescape(win32_url)
239
# usual local path with drive letter
240
if (win32_url[3] not in ('abcdefghijklmnopqrstuvwxyz'
241
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
242
or win32_url[4] not in '|:'
243
or win32_url[5] != '/'):
244
raise errors.InvalidURL(url, 'Win32 file urls start with'
245
' file:///x:/, where x is a valid drive letter')
246
return win32_url[3].upper() + u':' + unescape(win32_url[5:])
249
def _win32_local_path_to_url(path):
250
"""Convert a local path like ./foo into a URL like file:///C:/path/to/foo
252
This also handles transforming escaping unicode characters, etc.
254
# importing directly from ntpath allows us to test this
255
# on non-win32 platform
256
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
257
# which actually strips trailing space characters.
258
# The worst part is that under linux ntpath.abspath has different
259
# semantics, since 'nt' is not an available module.
260
win32_path = osutils._win32_abspath(path)
261
# check for UNC path \\HOST\path
262
if win32_path.startswith('//'):
263
return 'file:' + escape(win32_path)
264
return 'file:///' + win32_path[0].upper() + ':' + escape(win32_path[2:])
267
local_path_to_url = _posix_local_path_to_url
268
local_path_from_url = _posix_local_path_from_url
269
MIN_ABS_FILEURL_LENGTH = len('file:///')
270
WIN32_MIN_ABS_FILEURL_LENGTH = len('file:///C:/')
272
if sys.platform == 'win32':
273
local_path_to_url = _win32_local_path_to_url
274
local_path_from_url = _win32_local_path_from_url
276
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
279
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
282
def normalize_url(url):
283
"""Make sure that a path string is in fully normalized URL form.
285
This handles URLs which have unicode characters, spaces,
286
special characters, etc.
288
It has two basic modes of operation, depending on whether the
289
supplied string starts with a url specifier (scheme://) or not.
290
If it does not have a specifier it is considered a local path,
291
and will be converted into a file:/// url. Non-ascii characters
292
will be encoded using utf-8.
293
If it does have a url specifier, it will be treated as a "hybrid"
294
URL. Basically, a URL that should have URL special characters already
295
escaped (like +?&# etc), but may have unicode characters, etc
296
which would not be valid in a real URL.
298
:param url: Either a hybrid URL or a local path
299
:return: A normalized URL which only includes 7-bit ASCII characters.
301
m = _url_scheme_re.match(url)
303
return local_path_to_url(url)
304
if not isinstance(url, unicode):
306
if c not in _url_safe_characters:
307
raise errors.InvalidURL(url, 'URLs can only contain specific'
308
' safe characters (not %r)' % c)
310
# We have a unicode (hybrid) url
311
scheme = m.group('scheme')
312
path = list(m.group('path'))
314
for i in xrange(len(path)):
315
if path[i] not in _url_safe_characters:
316
chars = path[i].encode('utf-8')
317
path[i] = ''.join(['%%%02X' % ord(c) for c in path[i].encode('utf-8')])
318
return str(scheme + '://' + ''.join(path))
321
def relative_url(base, other):
322
"""Return a path to other from base.
324
If other is unrelated to base, return other. Else return a relative path.
325
This assumes no symlinks as part of the url.
327
dummy, base_first_slash = _find_scheme_and_separator(base)
328
if base_first_slash is None:
331
dummy, other_first_slash = _find_scheme_and_separator(other)
332
if other_first_slash is None:
335
# this takes care of differing schemes or hosts
336
base_scheme = base[:base_first_slash]
337
other_scheme = other[:other_first_slash]
338
if base_scheme != other_scheme:
341
base_path = base[base_first_slash+1:]
342
other_path = other[other_first_slash+1:]
344
if base_path.endswith('/'):
345
base_path = base_path[:-1]
347
base_sections = base_path.split('/')
348
other_sections = other_path.split('/')
350
if base_sections == ['']:
352
if other_sections == ['']:
356
for b, o in zip(base_sections, other_sections):
359
output_sections.append(b)
361
match_len = len(output_sections)
362
output_sections = ['..' for x in base_sections[match_len:]]
363
output_sections.extend(other_sections[match_len:])
365
return "/".join(output_sections) or "."
368
def _win32_extract_drive_letter(url_base, path):
369
"""On win32 the drive letter needs to be added to the url base."""
370
# Strip off the drive letter
371
# path is currently /C:/foo
372
if len(path) < 3 or path[2] not in ':|' or path[3] != '/':
373
raise errors.InvalidURL(url_base + path,
374
'win32 file:/// paths need a drive letter')
375
url_base += path[0:3] # file:// + /C:
376
path = path[3:] # /foo
377
return url_base, path
380
def split(url, exclude_trailing_slash=True):
381
"""Split a URL into its parent directory and a child directory.
383
:param url: A relative or absolute URL
384
:param exclude_trailing_slash: Strip off a final '/' if it is part
385
of the path (but not if it is part of the protocol specification)
387
:return: (parent_url, child_dir). child_dir may be the empty string if we're at
390
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
392
if first_path_slash is None:
393
# We have either a relative path, or no separating slash
394
if scheme_loc is None:
396
if exclude_trailing_slash and url.endswith('/'):
398
return _posix_split(url)
400
# Scheme with no path
403
# We have a fully defined path
404
url_base = url[:first_path_slash] # http://host, file://
405
path = url[first_path_slash:] # /file/foo
407
if sys.platform == 'win32' and url.startswith('file:///'):
408
# Strip off the drive letter
409
# url_base is currently file://
410
# path is currently /C:/foo
411
url_base, path = _win32_extract_drive_letter(url_base, path)
412
# now it should be file:///C: and /foo
414
if exclude_trailing_slash and len(path) > 1 and path.endswith('/'):
416
head, tail = _posix_split(path)
417
return url_base + head, tail
420
def _win32_strip_local_trailing_slash(url):
421
"""Strip slashes after the drive letter"""
422
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
428
def strip_trailing_slash(url):
429
"""Strip trailing slash, except for root paths.
431
The definition of 'root path' is platform-dependent.
432
This assumes that all URLs are valid netloc urls, such that they
435
It searches for ://, and then refuses to remove the next '/'.
436
It can also handle relative paths
438
path/to/foo => path/to/foo
439
path/to/foo/ => path/to/foo
440
http://host/path/ => http://host/path
441
http://host/path => http://host/path
442
http://host/ => http://host/
444
file:///foo/ => file:///foo
445
# This is unique on win32 platforms, and is the only URL
446
# format which does it differently.
447
file:///c|/ => file:///c:/
449
if not url.endswith('/'):
452
if sys.platform == 'win32' and url.startswith('file://'):
453
return _win32_strip_local_trailing_slash(url)
455
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
456
if scheme_loc is None:
457
# This is a relative path, as it has no scheme
458
# so just chop off the last character
461
if first_path_slash is None or first_path_slash == len(url)-1:
462
# Don't chop off anything if the only slash is the path
470
"""Unescape relpath from url format.
472
This returns a Unicode path from a URL
474
# jam 20060427 URLs are supposed to be ASCII only strings
475
# If they are passed in as unicode, urllib.unquote
476
# will return a UNICODE string, which actually contains
477
# utf-8 bytes. So we have to ensure that they are
478
# plain ASCII strings, or the final .decode will
479
# try to encode the UNICODE => ASCII, and then decode
483
except UnicodeError, e:
484
raise errors.InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
486
unquoted = urllib.unquote(url)
488
unicode_path = unquoted.decode('utf-8')
489
except UnicodeError, e:
490
raise errors.InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
494
# These are characters that if escaped, should stay that way
495
_no_decode_chars = ';/?:@&=+$,#'
496
_no_decode_ords = [ord(c) for c in _no_decode_chars]
497
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
498
+ ['%02X' % o for o in _no_decode_ords])
499
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
500
+ [('%02X' % o, chr(o)) for o in range(256)]))
501
#These entries get mapped to themselves
502
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
504
# These characters should not be escaped
505
_url_safe_characters = set(
506
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
507
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
508
"0123456789" # Numbers
509
"_.-!~*'()" # Unreserved characters
510
"/;?:@&=+$," # Reserved characters
511
"%#" # Extra reserved characters
514
def unescape_for_display(url, encoding):
515
"""Decode what you can for a URL, so that we get a nice looking path.
517
This will turn file:// urls into local paths, and try to decode
518
any portions of a http:// style url that it can.
520
Any sections of the URL which can't be represented in the encoding or
521
need to stay as escapes are left alone.
523
:param url: A 7-bit ASCII URL
524
:param encoding: The final output encoding
526
:return: A unicode string which can be safely encoded into the
529
assert encoding is not None, 'you cannot specify None for the display encoding.'
530
if url.startswith('file://'):
532
path = local_path_from_url(url)
533
path.encode(encoding)
538
# Split into sections to try to decode utf-8
540
for i in xrange(1, len(res)):
541
escaped_chunks = res[i].split('%')
542
for j in xrange(1, len(escaped_chunks)):
543
item = escaped_chunks[j]
545
escaped_chunks[j] = _hex_display_map[item[:2]] + item[2:]
547
# Put back the percent symbol
548
escaped_chunks[j] = '%' + item
549
except UnicodeDecodeError:
550
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
551
unescaped = ''.join(escaped_chunks)
553
decoded = unescaped.decode('utf-8')
554
except UnicodeDecodeError:
555
# If this path segment cannot be properly utf-8 decoded
556
# after doing unescaping we will just leave it alone
560
decoded.encode(encoding)
561
except UnicodeEncodeError:
562
# If this chunk cannot be encoded in the local
563
# encoding, then we should leave it alone
566
# Otherwise take the url decoded one
568
return u'/'.join(res)
added
removed
bzrlib/urlutils.py
