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('/')
134
m = _url_scheme_re.match(arg)
137
scheme = m.group('scheme')
138
# this skips .. normalisation, making http://host/../../..
140
path = m.group('path').split('/')
142
for chunk in arg.split('/'):
147
# Don't pop off the host portion
150
raise errors.InvalidURLJoin('Cannot go above root',
156
return '/'.join(path)
157
return scheme + '://' + '/'.join(path)
160
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
161
def _posix_local_path_from_url(url):
162
"""Convert a url like file:///path/to/foo into /path/to/foo"""
163
if not url.startswith('file:///'):
164
raise errors.InvalidURL(url, 'local urls must start with file:///')
165
# We only strip off 2 slashes
166
return unescape(url[len('file://'):])
169
def _posix_local_path_to_url(path):
170
"""Convert a local path like ./foo into a URL like file:///path/to/foo
172
This also handles transforming escaping unicode characters, etc.
174
# importing directly from posixpath allows us to test this
175
# on non-posix platforms
176
return 'file://' + escape(_posix_normpath(
177
osutils._posix_abspath(path)))
180
def _win32_local_path_from_url(url):
181
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
182
if not url.startswith('file://'):
183
raise errors.InvalidURL(url, 'local urls must start with file:///, '
184
'UNC path urls must start with file://')
185
# We strip off all 3 slashes
186
win32_url = url[len('file:'):]
187
# check for UNC path: //HOST/path
188
if not win32_url.startswith('///'):
189
if (win32_url[2] == '/'
190
or win32_url[3] in '|:'):
191
raise errors.InvalidURL(url, 'Win32 UNC path urls'
192
' have form file://HOST/path')
193
return unescape(win32_url)
194
# usual local path with drive letter
195
if (win32_url[3] not in ('abcdefghijklmnopqrstuvwxyz'
196
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
197
or win32_url[4] not in '|:'
198
or win32_url[5] != '/'):
199
raise errors.InvalidURL(url, 'Win32 file urls start with'
200
' file:///x:/, where x is a valid drive letter')
201
return win32_url[3].upper() + u':' + unescape(win32_url[5:])
204
def _win32_local_path_to_url(path):
205
"""Convert a local path like ./foo into a URL like file:///C:/path/to/foo
207
This also handles transforming escaping unicode characters, etc.
209
# importing directly from ntpath allows us to test this
210
# on non-win32 platform
211
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
212
# which actually strips trailing space characters.
213
# The worst part is that under linux ntpath.abspath has different
214
# semantics, since 'nt' is not an available module.
215
win32_path = osutils._nt_normpath(
216
osutils._win32_abspath(path)).replace('\\', '/')
217
# check for UNC path \\HOST\path
218
if win32_path.startswith('//'):
219
return 'file:' + escape(win32_path)
220
return 'file:///' + win32_path[0].upper() + ':' + escape(win32_path[2:])
223
local_path_to_url = _posix_local_path_to_url
224
local_path_from_url = _posix_local_path_from_url
225
MIN_ABS_FILEURL_LENGTH = len('file:///')
226
WIN32_MIN_ABS_FILEURL_LENGTH = len('file:///C:/')
228
if sys.platform == 'win32':
229
local_path_to_url = _win32_local_path_to_url
230
local_path_from_url = _win32_local_path_from_url
232
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
235
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
238
def normalize_url(url):
239
"""Make sure that a path string is in fully normalized URL form.
241
This handles URLs which have unicode characters, spaces,
242
special characters, etc.
244
It has two basic modes of operation, depending on whether the
245
supplied string starts with a url specifier (scheme://) or not.
246
If it does not have a specifier it is considered a local path,
247
and will be converted into a file:/// url. Non-ascii characters
248
will be encoded using utf-8.
249
If it does have a url specifier, it will be treated as a "hybrid"
250
URL. Basically, a URL that should have URL special characters already
251
escaped (like +?&# etc), but may have unicode characters, etc
252
which would not be valid in a real URL.
254
:param url: Either a hybrid URL or a local path
255
:return: A normalized URL which only includes 7-bit ASCII characters.
257
m = _url_scheme_re.match(url)
259
return local_path_to_url(url)
260
if not isinstance(url, unicode):
262
if c not in _url_safe_characters:
263
raise errors.InvalidURL(url, 'URLs can only contain specific'
264
' safe characters (not %r)' % c)
266
# We have a unicode (hybrid) url
267
scheme = m.group('scheme')
268
path = list(m.group('path'))
270
for i in xrange(len(path)):
271
if path[i] not in _url_safe_characters:
272
chars = path[i].encode('utf-8')
273
path[i] = ''.join(['%%%02X' % ord(c) for c in path[i].encode('utf-8')])
274
return str(scheme + '://' + ''.join(path))
277
def relative_url(base, other):
278
"""Return a path to other from base.
280
If other is unrelated to base, return other. Else return a relative path.
281
This assumes no symlinks as part of the url.
283
dummy, base_first_slash = _find_scheme_and_separator(base)
284
if base_first_slash is None:
287
dummy, other_first_slash = _find_scheme_and_separator(other)
288
if other_first_slash is None:
291
# this takes care of differing schemes or hosts
292
base_scheme = base[:base_first_slash]
293
other_scheme = other[:other_first_slash]
294
if base_scheme != other_scheme:
297
base_path = base[base_first_slash+1:]
298
other_path = other[other_first_slash+1:]
300
if base_path.endswith('/'):
301
base_path = base_path[:-1]
303
base_sections = base_path.split('/')
304
other_sections = other_path.split('/')
306
if base_sections == ['']:
308
if other_sections == ['']:
312
for b, o in zip(base_sections, other_sections):
315
output_sections.append(b)
317
match_len = len(output_sections)
318
output_sections = ['..' for x in base_sections[match_len:]]
319
output_sections.extend(other_sections[match_len:])
321
return "/".join(output_sections) or "."
324
def _win32_extract_drive_letter(url_base, path):
325
"""On win32 the drive letter needs to be added to the url base."""
326
# Strip off the drive letter
327
# path is currently /C:/foo
328
if len(path) < 3 or path[2] not in ':|' or path[3] != '/':
329
raise errors.InvalidURL(url_base + path,
330
'win32 file:/// paths need a drive letter')
331
url_base += path[0:3] # file:// + /C:
332
path = path[3:] # /foo
333
return url_base, path
336
def split(url, exclude_trailing_slash=True):
337
"""Split a URL into its parent directory and a child directory.
339
:param url: A relative or absolute URL
340
:param exclude_trailing_slash: Strip off a final '/' if it is part
341
of the path (but not if it is part of the protocol specification)
343
:return: (parent_url, child_dir). child_dir may be the empty string if we're at
346
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
348
if first_path_slash is None:
349
# We have either a relative path, or no separating slash
350
if scheme_loc is None:
352
if exclude_trailing_slash and url.endswith('/'):
354
return _posix_split(url)
356
# Scheme with no path
359
# We have a fully defined path
360
url_base = url[:first_path_slash] # http://host, file://
361
path = url[first_path_slash:] # /file/foo
363
if sys.platform == 'win32' and url.startswith('file:///'):
364
# Strip off the drive letter
365
# url_base is currently file://
366
# path is currently /C:/foo
367
url_base, path = _win32_extract_drive_letter(url_base, path)
368
# now it should be file:///C: and /foo
370
if exclude_trailing_slash and len(path) > 1 and path.endswith('/'):
372
head, tail = _posix_split(path)
373
return url_base + head, tail
376
def _win32_strip_local_trailing_slash(url):
377
"""Strip slashes after the drive letter"""
378
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
384
def strip_trailing_slash(url):
385
"""Strip trailing slash, except for root paths.
387
The definition of 'root path' is platform-dependent.
388
This assumes that all URLs are valid netloc urls, such that they
391
It searches for ://, and then refuses to remove the next '/'.
392
It can also handle relative paths
394
path/to/foo => path/to/foo
395
path/to/foo/ => path/to/foo
396
http://host/path/ => http://host/path
397
http://host/path => http://host/path
398
http://host/ => http://host/
400
file:///foo/ => file:///foo
401
# This is unique on win32 platforms, and is the only URL
402
# format which does it differently.
403
file:///c|/ => file:///c:/
405
if not url.endswith('/'):
408
if sys.platform == 'win32' and url.startswith('file://'):
409
return _win32_strip_local_trailing_slash(url)
411
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
412
if scheme_loc is None:
413
# This is a relative path, as it has no scheme
414
# so just chop off the last character
417
if first_path_slash is None or first_path_slash == len(url)-1:
418
# Don't chop off anything if the only slash is the path
426
"""Unescape relpath from url format.
428
This returns a Unicode path from a URL
430
# jam 20060427 URLs are supposed to be ASCII only strings
431
# If they are passed in as unicode, urllib.unquote
432
# will return a UNICODE string, which actually contains
433
# utf-8 bytes. So we have to ensure that they are
434
# plain ASCII strings, or the final .decode will
435
# try to encode the UNICODE => ASCII, and then decode
439
except UnicodeError, e:
440
raise errors.InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
442
unquoted = urllib.unquote(url)
444
unicode_path = unquoted.decode('utf-8')
445
except UnicodeError, e:
446
raise errors.InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
450
# These are characters that if escaped, should stay that way
451
_no_decode_chars = ';/?:@&=+$,#'
452
_no_decode_ords = [ord(c) for c in _no_decode_chars]
453
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
454
+ ['%02X' % o for o in _no_decode_ords])
455
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
456
+ [('%02X' % o, chr(o)) for o in range(256)]))
457
#These entries get mapped to themselves
458
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
460
# These characters should not be escaped
461
_url_safe_characters = set(
462
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
463
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
464
"0123456789" # Numbers
465
"_.-!~*'()" # Unreserved characters
466
"/;?:@&=+$," # Reserved characters
467
"%#" # Extra reserved characters
470
def unescape_for_display(url, encoding):
471
"""Decode what you can for a URL, so that we get a nice looking path.
473
This will turn file:// urls into local paths, and try to decode
474
any portions of a http:// style url that it can.
476
Any sections of the URL which can't be represented in the encoding or
477
need to stay as escapes are left alone.
479
:param url: A 7-bit ASCII URL
480
:param encoding: The final output encoding
482
:return: A unicode string which can be safely encoded into the
485
assert encoding is not None, 'you cannot specify None for the display encoding.'
486
if url.startswith('file://'):
488
path = local_path_from_url(url)
489
path.encode(encoding)
494
# Split into sections to try to decode utf-8
496
for i in xrange(1, len(res)):
497
escaped_chunks = res[i].split('%')
498
for j in xrange(1, len(escaped_chunks)):
499
item = escaped_chunks[j]
501
escaped_chunks[j] = _hex_display_map[item[:2]] + item[2:]
503
# Put back the percent symbol
504
escaped_chunks[j] = '%' + item
505
except UnicodeDecodeError:
506
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
507
unescaped = ''.join(escaped_chunks)
509
decoded = unescaped.decode('utf-8')
510
except UnicodeDecodeError:
511
# If this path segment cannot be properly utf-8 decoded
512
# after doing unescaping we will just leave it alone
516
decoded.encode(encoding)
517
except UnicodeEncodeError:
518
# If this chunk cannot be encoded in the local
519
# encoding, then we should leave it alone
522
# Otherwise take the url decoded one
524
return u'/'.join(res)
added
removed
bzrlib/urlutils.py
