1
# Bazaar-NG -- distributed version control
3
# Copyright (C) 2006 by 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."""
22
from posixpath import split as _posix_split, normpath as _posix_normpath
27
import bzrlib.errors as errors
31
def basename(url, exclude_trailing_slash=True):
32
"""Return the last component of a URL.
34
:param url: The URL in question
35
:param exclude_trailing_slash: If the url looks like "path/to/foo/"
36
ignore the final slash and return 'foo' rather than ''
37
:return: Just the final component of the URL. This can return ''
38
if you don't exclude_trailing_slash, or if you are at the
41
return split(url, exclude_trailing_slash=exclude_trailing_slash)[1]
44
def dirname(url, exclude_trailing_slash=True):
45
"""Return the parent directory of the given path.
47
:param url: Relative or absolute URL
48
:param exclude_trailing_slash: Remove a final slash
49
(treat http://host/foo/ as http://host/foo, but
50
http://host/ stays http://host/)
51
:return: Everything in the URL except the last path chunk
53
# TODO: jam 20060502 This was named dirname to be consistent
54
# with the os functions, but maybe "parent" would be better
55
return split(url, exclude_trailing_slash=exclude_trailing_slash)[0]
59
"""Escape relpath to be a valid url."""
60
if isinstance(relpath, unicode):
61
relpath = relpath.encode('utf-8')
62
# After quoting and encoding, the path should be perfectly
63
# safe as a plain ASCII string, str() just enforces this
64
return str(urllib.quote(relpath))
67
def file_relpath(base, path):
68
"""Compute just the relative sub-portion of a url
70
This assumes that both paths are already fully specified file:// URLs.
72
assert len(base) >= MIN_ABS_FILEURL_LENGTH, ('Length of base must be equal or'
73
' exceed the platform minimum url length (which is %d)' %
74
MIN_ABS_FILEURL_LENGTH)
76
base = local_path_from_url(base)
77
path = local_path_from_url(path)
78
return escape(bzrlib.osutils.relpath(base, path))
81
def _find_scheme_and_separator(url):
82
"""Find the scheme separator (://) and the first path separator
84
This is just a helper functions for other path utilities.
85
It could probably be replaced by urlparse
87
m = _url_scheme_re.match(url)
91
scheme = m.group('scheme')
92
path = m.group('path')
94
# Find the path separating slash
95
# (first slash after the ://)
96
first_path_slash = path.find('/')
97
if first_path_slash == -1:
98
return len(scheme), None
99
return len(scheme), first_path_slash+len(scheme)+3
102
def join(base, *args):
103
"""Create a URL by joining sections.
105
This will normalize '..', assuming that paths are absolute
106
(it assumes no symlinks in either path)
108
If any of *args is an absolute URL, it will be treated correctly.
110
join('http://foo', 'http://bar') => 'http://bar'
111
join('http://foo', 'bar') => 'http://foo/bar'
112
join('http://foo', 'bar', '../baz') => 'http://foo/baz'
114
m = _url_scheme_re.match(base)
117
scheme = m.group('scheme')
118
path = m.group('path').split('/')
119
if path[-1:] == ['']:
120
# Strip off a trailing slash
121
# This helps both when we are at the root, and when
122
# 'base' has an extra slash at the end
125
path = base.split('/')
128
m = _url_scheme_re.match(arg)
131
scheme = m.group('scheme')
132
path = m.group('path').split('/')
134
for chunk in arg.split('/'):
139
# Don't pop off the host portion
142
raise errors.InvalidURLJoin('Cannot go above root',
148
return '/'.join(path)
149
return scheme + '://' + '/'.join(path)
152
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
153
def _posix_local_path_from_url(url):
154
"""Convert a url like file:///path/to/foo into /path/to/foo"""
155
if not url.startswith('file:///'):
156
raise errors.InvalidURL(url, 'local urls must start with file:///')
157
# We only strip off 2 slashes
158
return unescape(url[len('file://'):])
161
def _posix_local_path_to_url(path):
162
"""Convert a local path like ./foo into a URL like file:///path/to/foo
164
This also handles transforming escaping unicode characters, etc.
166
# importing directly from posixpath allows us to test this
167
# on non-posix platforms
168
return 'file://' + escape(_posix_normpath(
169
bzrlib.osutils._posix_abspath(path)))
172
def _win32_local_path_from_url(url):
173
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
174
if not url.startswith('file:///'):
175
raise errors.InvalidURL(url, 'local urls must start with file:///')
176
# We strip off all 3 slashes
177
win32_url = url[len('file:///'):]
178
if (win32_url[0] not in ('abcdefghijklmnopqrstuvwxyz'
179
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
180
or win32_url[1] not in '|:'
181
or win32_url[2] != '/'):
182
raise errors.InvalidURL(url, 'Win32 file urls start with'
183
' file:///x:/, where x is a valid drive letter')
184
# Preferentially using .lower() because os.getcwd() returns
185
# paths with lowercase drive letters, and that helps
186
# bzrlib.osutils.relpath() work correctly
187
return win32_url[0].lower() + u':' + unescape(win32_url[2:])
190
def _win32_local_path_to_url(path):
191
"""Convert a local path like ./foo into a URL like file:///C:/path/to/foo
193
This also handles transforming escaping unicode characters, etc.
195
# importing directly from ntpath allows us to test this
196
# on non-win32 platform
197
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
198
# which actually strips trailing space characters.
199
# The worst part is that under linux ntpath.abspath has different
200
# semantics, since 'nt' is not an available module.
201
win32_path = bzrlib.osutils._nt_normpath(
202
bzrlib.osutils._win32_abspath(path)).replace('\\', '/')
203
return 'file:///' + win32_path[0].lower() + ':' + escape(win32_path[2:])
206
local_path_to_url = _posix_local_path_to_url
207
local_path_from_url = _posix_local_path_from_url
208
MIN_ABS_FILEURL_LENGTH = len('file:///')
209
WIN32_MIN_ABS_FILEURL_LENGTH = len('file:///C:/')
211
if sys.platform == 'win32':
212
local_path_to_url = _win32_local_path_to_url
213
local_path_from_url = _win32_local_path_from_url
215
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
218
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
221
def normalize_url(url):
222
"""Make sure that a path string is in fully normalized URL form.
224
This handles URLs which have unicode characters, spaces,
225
special characters, etc.
227
It has two basic modes of operation, depending on whether the
228
supplied string starts with a url specifier (scheme://) or not.
229
If it does not have a specifier it is considered a local path,
230
and will be converted into a file:/// url. Non-ascii characters
231
will be encoded using utf-8.
232
If it does have a url specifier, it will be treated as a "hybrid"
233
URL. Basically, a URL that should have URL special characters already
234
escaped (like +?&# etc), but may have unicode characters, etc
235
which would not be valid in a real URL.
237
:param url: Either a hybrid URL or a local path
238
:return: A normalized URL which only includes 7-bit ASCII characters.
240
m = _url_scheme_re.match(url)
242
return local_path_to_url(url)
243
if not isinstance(url, unicode):
245
if c not in _url_safe_characters:
246
raise errors.InvalidURL(url, 'URLs can only contain specific'
247
' safe characters (not %r)' % c)
249
# We have a unicode (hybrid) url
250
scheme = m.group('scheme')
251
path = list(m.group('path'))
253
for i in xrange(len(path)):
254
if path[i] not in _url_safe_characters:
255
chars = path[i].encode('utf-8')
256
path[i] = ''.join(['%%%02X' % ord(c) for c in path[i].encode('utf-8')])
257
return scheme + '://' + ''.join(path)
260
def relative_url(base, other):
261
"""Return a path to other from base.
263
If other is unrelated to base, return other. Else return a relative path.
264
This assumes no symlinks as part of the url.
266
dummy, base_first_slash = _find_scheme_and_separator(base)
267
if base_first_slash is None:
270
dummy, other_first_slash = _find_scheme_and_separator(other)
271
if other_first_slash is None:
274
# this takes care of differing schemes or hosts
275
base_scheme = base[:base_first_slash]
276
other_scheme = other[:other_first_slash]
277
if base_scheme != other_scheme:
280
base_path = base[base_first_slash+1:]
281
other_path = other[other_first_slash+1:]
283
if base_path.endswith('/'):
284
base_path = base_path[:-1]
286
base_sections = base_path.split('/')
287
other_sections = other_path.split('/')
289
if base_sections == ['']:
291
if other_sections == ['']:
295
for b, o in zip(base_sections, other_sections):
298
output_sections.append(b)
300
match_len = len(output_sections)
301
output_sections = ['..' for x in base_sections[match_len:]]
302
output_sections.extend(other_sections[match_len:])
304
return "/".join(output_sections) or "."
307
def _win32_extract_drive_letter(url_base, path):
308
"""On win32 the drive letter needs to be added to the url base."""
309
# Strip off the drive letter
310
# path is currently /C:/foo
311
if len(path) < 3 or path[2] not in ':|' or path[3] != '/':
312
raise errors.InvalidURL(url_base + path,
313
'win32 file:/// paths need a drive letter')
314
url_base += path[0:3] # file:// + /C:
315
path = path[3:] # /foo
316
return url_base, path
319
def split(url, exclude_trailing_slash=True):
320
"""Split a URL into its parent directory and a child directory.
322
:param url: A relative or absolute URL
323
:param exclude_trailing_slash: Strip off a final '/' if it is part
324
of the path (but not if it is part of the protocol specification)
326
:return: (parent_url, child_dir). child_dir may be the empty string if we're at
329
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
331
if first_path_slash is None:
332
# We have either a relative path, or no separating slash
333
if scheme_loc is None:
335
if exclude_trailing_slash and url.endswith('/'):
337
return _posix_split(url)
339
# Scheme with no path
342
# We have a fully defined path
343
url_base = url[:first_path_slash] # http://host, file://
344
path = url[first_path_slash:] # /file/foo
346
if sys.platform == 'win32' and url.startswith('file:///'):
347
# Strip off the drive letter
348
# url_base is currently file://
349
# path is currently /C:/foo
350
url_base, path = _win32_extract_drive_letter(url_base, path)
351
# now it should be file:///C: and /foo
353
if exclude_trailing_slash and len(path) > 1 and path.endswith('/'):
355
head, tail = _posix_split(path)
356
return url_base + head, tail
359
def _win32_strip_local_trailing_slash(url):
360
"""Strip slashes after the drive letter"""
361
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
367
def strip_trailing_slash(url):
368
"""Strip trailing slash, except for root paths.
370
The definition of 'root path' is platform-dependent.
371
This assumes that all URLs are valid netloc urls, such that they
374
It searches for ://, and then refuses to remove the next '/'.
375
It can also handle relative paths
377
path/to/foo => path/to/foo
378
path/to/foo/ => path/to/foo
379
http://host/path/ => http://host/path
380
http://host/path => http://host/path
381
http://host/ => http://host/
383
file:///foo/ => file:///foo
384
# This is unique on win32 platforms, and is the only URL
385
# format which does it differently.
386
file:///c|/ => file:///c:/
388
if not url.endswith('/'):
391
if sys.platform == 'win32' and url.startswith('file:///'):
392
return _win32_strip_local_trailing_slash(url)
394
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
395
if scheme_loc is None:
396
# This is a relative path, as it has no scheme
397
# so just chop off the last character
400
if first_path_slash is None or first_path_slash == len(url)-1:
401
# Don't chop off anything if the only slash is the path
409
"""Unescape relpath from url format.
411
This returns a Unicode path from a URL
413
# jam 20060427 URLs are supposed to be ASCII only strings
414
# If they are passed in as unicode, urllib.unquote
415
# will return a UNICODE string, which actually contains
416
# utf-8 bytes. So we have to ensure that they are
417
# plain ASCII strings, or the final .decode will
418
# try to encode the UNICODE => ASCII, and then decode
422
except UnicodeError, e:
423
raise errors.InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
425
unquoted = urllib.unquote(url)
427
unicode_path = unquoted.decode('utf-8')
428
except UnicodeError, e:
429
raise errors.InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
433
# These are characters that if escaped, should stay that way
434
_no_decode_chars = ';/?:@&=+$,#'
435
_no_decode_ords = [ord(c) for c in _no_decode_chars]
436
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
437
+ ['%02X' % o for o in _no_decode_ords])
438
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
439
+ [('%02X' % o, chr(o)) for o in range(256)]))
440
#These entries get mapped to themselves
441
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
443
# These characters should not be escaped
444
_url_safe_characters = set('abcdefghijklmnopqrstuvwxyz'
445
'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
450
def unescape_for_display(url, encoding):
451
"""Decode what you can for a URL, so that we get a nice looking path.
453
This will turn file:// urls into local paths, and try to decode
454
any portions of a http:// style url that it can.
456
Any sections of the URL which can't be represented in the encoding or
457
need to stay as escapes are left alone.
459
:param url: A 7-bit ASCII URL
460
:param encoding: The final output encoding
462
:return: A unicode string which can be safely encoded into the
465
assert encoding is not None, 'you cannot specify None for the display encoding.'
466
if url.startswith('file://'):
468
path = local_path_from_url(url)
469
path.encode(encoding)
474
# Split into sections to try to decode utf-8
476
for i in xrange(1, len(res)):
477
escaped_chunks = res[i].split('%')
478
for j in xrange(1, len(escaped_chunks)):
479
item = escaped_chunks[j]
481
escaped_chunks[j] = _hex_display_map[item[:2]] + item[2:]
483
# Put back the percent symbol
484
escaped_chunks[j] = '%' + item
485
except UnicodeDecodeError:
486
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
487
unescaped = ''.join(escaped_chunks)
489
decoded = unescaped.decode('utf-8')
490
except UnicodeDecodeError:
491
# If this path segment cannot be properly utf-8 decoded
492
# after doing unescaping we will just leave it alone
496
decoded.encode(encoding)
497
except UnicodeEncodeError:
498
# If this chunk cannot be encoded in the local
499
# encoding, then we should leave it alone
502
# Otherwise take the url decoded one
504
return u'/'.join(res)
added
removed
bzrlib/urlutils.py
