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
if path and path[0] == '':
165
return '/'.join(path)
166
return scheme + '://' + '/'.join(path)
169
def joinpath(base, *args):
170
"""Join URL path segments to a URL path segment.
172
This is somewhat like osutils.joinpath, but intended for URLs.
174
XXX: this duplicates some normalisation logic, and also duplicates a lot of
175
path handling logic that already exists in some Transport implementations.
176
We really should try to have exactly one place in the code base responsible
177
for combining paths of URLs.
182
base = base.rstrip('/')
183
path = base.split('/')
185
if arg.startswith('/'):
187
for chunk in arg.split('/'):
192
raise errors.InvalidURLJoin('Cannot go above root',
200
return '/'.join(path)
203
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
204
def _posix_local_path_from_url(url):
205
"""Convert a url like file:///path/to/foo into /path/to/foo"""
206
if not url.startswith('file:///'):
207
raise errors.InvalidURL(url, 'local urls must start with file:///')
208
# We only strip off 2 slashes
209
return unescape(url[len('file://'):])
212
def _posix_local_path_to_url(path):
213
"""Convert a local path like ./foo into a URL like file:///path/to/foo
215
This also handles transforming escaping unicode characters, etc.
217
# importing directly from posixpath allows us to test this
218
# on non-posix platforms
219
return 'file://' + escape(_posix_normpath(
220
osutils._posix_abspath(path)))
223
def _win32_local_path_from_url(url):
224
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
225
if not url.startswith('file://'):
226
raise errors.InvalidURL(url, 'local urls must start with file:///, '
227
'UNC path urls must start with file://')
228
# We strip off all 3 slashes
229
win32_url = url[len('file:'):]
230
# check for UNC path: //HOST/path
231
if not win32_url.startswith('///'):
232
if (win32_url[2] == '/'
233
or win32_url[3] in '|:'):
234
raise errors.InvalidURL(url, 'Win32 UNC path urls'
235
' have form file://HOST/path')
236
return unescape(win32_url)
237
# usual local path with drive letter
238
if (win32_url[3] not in ('abcdefghijklmnopqrstuvwxyz'
239
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
240
or win32_url[4] not in '|:'
241
or win32_url[5] != '/'):
242
raise errors.InvalidURL(url, 'Win32 file urls start with'
243
' file:///x:/, where x is a valid drive letter')
244
return win32_url[3].upper() + u':' + unescape(win32_url[5:])
247
def _win32_local_path_to_url(path):
248
"""Convert a local path like ./foo into a URL like file:///C:/path/to/foo
250
This also handles transforming escaping unicode characters, etc.
252
# importing directly from ntpath allows us to test this
253
# on non-win32 platform
254
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
255
# which actually strips trailing space characters.
256
# The worst part is that under linux ntpath.abspath has different
257
# semantics, since 'nt' is not an available module.
258
win32_path = osutils._win32_abspath(path)
259
# check for UNC path \\HOST\path
260
if win32_path.startswith('//'):
261
return 'file:' + escape(win32_path)
262
return 'file:///' + win32_path[0].upper() + ':' + escape(win32_path[2:])
265
local_path_to_url = _posix_local_path_to_url
266
local_path_from_url = _posix_local_path_from_url
267
MIN_ABS_FILEURL_LENGTH = len('file:///')
268
WIN32_MIN_ABS_FILEURL_LENGTH = len('file:///C:/')
270
if sys.platform == 'win32':
271
local_path_to_url = _win32_local_path_to_url
272
local_path_from_url = _win32_local_path_from_url
274
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
277
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
280
def normalize_url(url):
281
"""Make sure that a path string is in fully normalized URL form.
283
This handles URLs which have unicode characters, spaces,
284
special characters, etc.
286
It has two basic modes of operation, depending on whether the
287
supplied string starts with a url specifier (scheme://) or not.
288
If it does not have a specifier it is considered a local path,
289
and will be converted into a file:/// url. Non-ascii characters
290
will be encoded using utf-8.
291
If it does have a url specifier, it will be treated as a "hybrid"
292
URL. Basically, a URL that should have URL special characters already
293
escaped (like +?&# etc), but may have unicode characters, etc
294
which would not be valid in a real URL.
296
:param url: Either a hybrid URL or a local path
297
:return: A normalized URL which only includes 7-bit ASCII characters.
299
m = _url_scheme_re.match(url)
301
return local_path_to_url(url)
302
if not isinstance(url, unicode):
304
if c not in _url_safe_characters:
305
raise errors.InvalidURL(url, 'URLs can only contain specific'
306
' safe characters (not %r)' % c)
308
# We have a unicode (hybrid) url
309
scheme = m.group('scheme')
310
path = list(m.group('path'))
312
for i in xrange(len(path)):
313
if path[i] not in _url_safe_characters:
314
chars = path[i].encode('utf-8')
315
path[i] = ''.join(['%%%02X' % ord(c) for c in path[i].encode('utf-8')])
316
return str(scheme + '://' + ''.join(path))
319
def relative_url(base, other):
320
"""Return a path to other from base.
322
If other is unrelated to base, return other. Else return a relative path.
323
This assumes no symlinks as part of the url.
325
dummy, base_first_slash = _find_scheme_and_separator(base)
326
if base_first_slash is None:
329
dummy, other_first_slash = _find_scheme_and_separator(other)
330
if other_first_slash is None:
333
# this takes care of differing schemes or hosts
334
base_scheme = base[:base_first_slash]
335
other_scheme = other[:other_first_slash]
336
if base_scheme != other_scheme:
339
base_path = base[base_first_slash+1:]
340
other_path = other[other_first_slash+1:]
342
if base_path.endswith('/'):
343
base_path = base_path[:-1]
345
base_sections = base_path.split('/')
346
other_sections = other_path.split('/')
348
if base_sections == ['']:
350
if other_sections == ['']:
354
for b, o in zip(base_sections, other_sections):
357
output_sections.append(b)
359
match_len = len(output_sections)
360
output_sections = ['..' for x in base_sections[match_len:]]
361
output_sections.extend(other_sections[match_len:])
363
return "/".join(output_sections) or "."
366
def _win32_extract_drive_letter(url_base, path):
367
"""On win32 the drive letter needs to be added to the url base."""
368
# Strip off the drive letter
369
# path is currently /C:/foo
370
if len(path) < 3 or path[2] not in ':|' or path[3] != '/':
371
raise errors.InvalidURL(url_base + path,
372
'win32 file:/// paths need a drive letter')
373
url_base += path[0:3] # file:// + /C:
374
path = path[3:] # /foo
375
return url_base, path
378
def split(url, exclude_trailing_slash=True):
379
"""Split a URL into its parent directory and a child directory.
381
:param url: A relative or absolute URL
382
:param exclude_trailing_slash: Strip off a final '/' if it is part
383
of the path (but not if it is part of the protocol specification)
385
:return: (parent_url, child_dir). child_dir may be the empty string if we're at
388
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
390
if first_path_slash is None:
391
# We have either a relative path, or no separating slash
392
if scheme_loc is None:
394
if exclude_trailing_slash and url.endswith('/'):
396
return _posix_split(url)
398
# Scheme with no path
401
# We have a fully defined path
402
url_base = url[:first_path_slash] # http://host, file://
403
path = url[first_path_slash:] # /file/foo
405
if sys.platform == 'win32' and url.startswith('file:///'):
406
# Strip off the drive letter
407
# url_base is currently file://
408
# path is currently /C:/foo
409
url_base, path = _win32_extract_drive_letter(url_base, path)
410
# now it should be file:///C: and /foo
412
if exclude_trailing_slash and len(path) > 1 and path.endswith('/'):
414
head, tail = _posix_split(path)
415
return url_base + head, tail
418
def _win32_strip_local_trailing_slash(url):
419
"""Strip slashes after the drive letter"""
420
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
426
def strip_trailing_slash(url):
427
"""Strip trailing slash, except for root paths.
429
The definition of 'root path' is platform-dependent.
430
This assumes that all URLs are valid netloc urls, such that they
433
It searches for ://, and then refuses to remove the next '/'.
434
It can also handle relative paths
436
path/to/foo => path/to/foo
437
path/to/foo/ => path/to/foo
438
http://host/path/ => http://host/path
439
http://host/path => http://host/path
440
http://host/ => http://host/
442
file:///foo/ => file:///foo
443
# This is unique on win32 platforms, and is the only URL
444
# format which does it differently.
445
file:///c|/ => file:///c:/
447
if not url.endswith('/'):
450
if sys.platform == 'win32' and url.startswith('file://'):
451
return _win32_strip_local_trailing_slash(url)
453
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
454
if scheme_loc is None:
455
# This is a relative path, as it has no scheme
456
# so just chop off the last character
459
if first_path_slash is None or first_path_slash == len(url)-1:
460
# Don't chop off anything if the only slash is the path
468
"""Unescape relpath from url format.
470
This returns a Unicode path from a URL
472
# jam 20060427 URLs are supposed to be ASCII only strings
473
# If they are passed in as unicode, urllib.unquote
474
# will return a UNICODE string, which actually contains
475
# utf-8 bytes. So we have to ensure that they are
476
# plain ASCII strings, or the final .decode will
477
# try to encode the UNICODE => ASCII, and then decode
481
except UnicodeError, e:
482
raise errors.InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
484
unquoted = urllib.unquote(url)
486
unicode_path = unquoted.decode('utf-8')
487
except UnicodeError, e:
488
raise errors.InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
492
# These are characters that if escaped, should stay that way
493
_no_decode_chars = ';/?:@&=+$,#'
494
_no_decode_ords = [ord(c) for c in _no_decode_chars]
495
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
496
+ ['%02X' % o for o in _no_decode_ords])
497
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
498
+ [('%02X' % o, chr(o)) for o in range(256)]))
499
#These entries get mapped to themselves
500
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
502
# These characters should not be escaped
503
_url_safe_characters = set(
504
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
505
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
506
"0123456789" # Numbers
507
"_.-!~*'()" # Unreserved characters
508
"/;?:@&=+$," # Reserved characters
509
"%#" # Extra reserved characters
512
def unescape_for_display(url, encoding):
513
"""Decode what you can for a URL, so that we get a nice looking path.
515
This will turn file:// urls into local paths, and try to decode
516
any portions of a http:// style url that it can.
518
Any sections of the URL which can't be represented in the encoding or
519
need to stay as escapes are left alone.
521
:param url: A 7-bit ASCII URL
522
:param encoding: The final output encoding
524
:return: A unicode string which can be safely encoded into the
527
assert encoding is not None, 'you cannot specify None for the display encoding.'
528
if url.startswith('file://'):
530
path = local_path_from_url(url)
531
path.encode(encoding)
536
# Split into sections to try to decode utf-8
538
for i in xrange(1, len(res)):
539
escaped_chunks = res[i].split('%')
540
for j in xrange(1, len(escaped_chunks)):
541
item = escaped_chunks[j]
543
escaped_chunks[j] = _hex_display_map[item[:2]] + item[2:]
545
# Put back the percent symbol
546
escaped_chunks[j] = '%' + item
547
except UnicodeDecodeError:
548
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
549
unescaped = ''.join(escaped_chunks)
551
decoded = unescaped.decode('utf-8')
552
except UnicodeDecodeError:
553
# If this path segment cannot be properly utf-8 decoded
554
# after doing unescaping we will just leave it alone
558
decoded.encode(encoding)
559
except UnicodeEncodeError:
560
# If this chunk cannot be encoded in the local
561
# encoding, then we should leave it alone
564
# Otherwise take the url decoded one
566
return u'/'.join(res)
added
removed
bzrlib/urlutils.py
