1
# Bazaar -- 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
# this skips .. normalisation, making http://host/../../..
134
path = m.group('path').split('/')
136
for chunk in arg.split('/'):
141
# Don't pop off the host portion
144
raise errors.InvalidURLJoin('Cannot go above root',
150
return '/'.join(path)
151
return scheme + '://' + '/'.join(path)
154
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
155
def _posix_local_path_from_url(url):
156
"""Convert a url like file:///path/to/foo into /path/to/foo"""
157
if not url.startswith('file:///'):
158
raise errors.InvalidURL(url, 'local urls must start with file:///')
159
# We only strip off 2 slashes
160
return unescape(url[len('file://'):])
163
def _posix_local_path_to_url(path):
164
"""Convert a local path like ./foo into a URL like file:///path/to/foo
166
This also handles transforming escaping unicode characters, etc.
168
# importing directly from posixpath allows us to test this
169
# on non-posix platforms
170
return 'file://' + escape(_posix_normpath(
171
bzrlib.osutils._posix_abspath(path)))
174
def _win32_local_path_from_url(url):
175
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
176
if not url.startswith('file:///'):
177
raise errors.InvalidURL(url, 'local urls must start with file:///')
178
# We strip off all 3 slashes
179
win32_url = url[len('file:///'):]
180
if (win32_url[0] not in ('abcdefghijklmnopqrstuvwxyz'
181
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
182
or win32_url[1] not in '|:'
183
or win32_url[2] != '/'):
184
raise errors.InvalidURL(url, 'Win32 file urls start with'
185
' file:///x:/, where x is a valid drive letter')
186
return win32_url[0].upper() + u':' + unescape(win32_url[2:])
189
def _win32_local_path_to_url(path):
190
"""Convert a local path like ./foo into a URL like file:///C:/path/to/foo
192
This also handles transforming escaping unicode characters, etc.
194
# importing directly from ntpath allows us to test this
195
# on non-win32 platform
196
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
197
# which actually strips trailing space characters.
198
# The worst part is that under linux ntpath.abspath has different
199
# semantics, since 'nt' is not an available module.
200
win32_path = bzrlib.osutils._nt_normpath(
201
bzrlib.osutils._win32_abspath(path)).replace('\\', '/')
202
return 'file:///' + win32_path[0].upper() + ':' + escape(win32_path[2:])
205
local_path_to_url = _posix_local_path_to_url
206
local_path_from_url = _posix_local_path_from_url
207
MIN_ABS_FILEURL_LENGTH = len('file:///')
208
WIN32_MIN_ABS_FILEURL_LENGTH = len('file:///C:/')
210
if sys.platform == 'win32':
211
local_path_to_url = _win32_local_path_to_url
212
local_path_from_url = _win32_local_path_from_url
214
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
217
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
220
def normalize_url(url):
221
"""Make sure that a path string is in fully normalized URL form.
223
This handles URLs which have unicode characters, spaces,
224
special characters, etc.
226
It has two basic modes of operation, depending on whether the
227
supplied string starts with a url specifier (scheme://) or not.
228
If it does not have a specifier it is considered a local path,
229
and will be converted into a file:/// url. Non-ascii characters
230
will be encoded using utf-8.
231
If it does have a url specifier, it will be treated as a "hybrid"
232
URL. Basically, a URL that should have URL special characters already
233
escaped (like +?&# etc), but may have unicode characters, etc
234
which would not be valid in a real URL.
236
:param url: Either a hybrid URL or a local path
237
:return: A normalized URL which only includes 7-bit ASCII characters.
239
m = _url_scheme_re.match(url)
241
return local_path_to_url(url)
242
if not isinstance(url, unicode):
244
if c not in _url_safe_characters:
245
raise errors.InvalidURL(url, 'URLs can only contain specific'
246
' safe characters (not %r)' % c)
248
# We have a unicode (hybrid) url
249
scheme = m.group('scheme')
250
path = list(m.group('path'))
252
for i in xrange(len(path)):
253
if path[i] not in _url_safe_characters:
254
chars = path[i].encode('utf-8')
255
path[i] = ''.join(['%%%02X' % ord(c) for c in path[i].encode('utf-8')])
256
return scheme + '://' + ''.join(path)
259
def relative_url(base, other):
260
"""Return a path to other from base.
262
If other is unrelated to base, return other. Else return a relative path.
263
This assumes no symlinks as part of the url.
265
dummy, base_first_slash = _find_scheme_and_separator(base)
266
if base_first_slash is None:
269
dummy, other_first_slash = _find_scheme_and_separator(other)
270
if other_first_slash is None:
273
# this takes care of differing schemes or hosts
274
base_scheme = base[:base_first_slash]
275
other_scheme = other[:other_first_slash]
276
if base_scheme != other_scheme:
279
base_path = base[base_first_slash+1:]
280
other_path = other[other_first_slash+1:]
282
if base_path.endswith('/'):
283
base_path = base_path[:-1]
285
base_sections = base_path.split('/')
286
other_sections = other_path.split('/')
288
if base_sections == ['']:
290
if other_sections == ['']:
294
for b, o in zip(base_sections, other_sections):
297
output_sections.append(b)
299
match_len = len(output_sections)
300
output_sections = ['..' for x in base_sections[match_len:]]
301
output_sections.extend(other_sections[match_len:])
303
return "/".join(output_sections) or "."
306
def _win32_extract_drive_letter(url_base, path):
307
"""On win32 the drive letter needs to be added to the url base."""
308
# Strip off the drive letter
309
# path is currently /C:/foo
310
if len(path) < 3 or path[2] not in ':|' or path[3] != '/':
311
raise errors.InvalidURL(url_base + path,
312
'win32 file:/// paths need a drive letter')
313
url_base += path[0:3] # file:// + /C:
314
path = path[3:] # /foo
315
return url_base, path
318
def split(url, exclude_trailing_slash=True):
319
"""Split a URL into its parent directory and a child directory.
321
:param url: A relative or absolute URL
322
:param exclude_trailing_slash: Strip off a final '/' if it is part
323
of the path (but not if it is part of the protocol specification)
325
:return: (parent_url, child_dir). child_dir may be the empty string if we're at
328
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
330
if first_path_slash is None:
331
# We have either a relative path, or no separating slash
332
if scheme_loc is None:
334
if exclude_trailing_slash and url.endswith('/'):
336
return _posix_split(url)
338
# Scheme with no path
341
# We have a fully defined path
342
url_base = url[:first_path_slash] # http://host, file://
343
path = url[first_path_slash:] # /file/foo
345
if sys.platform == 'win32' and url.startswith('file:///'):
346
# Strip off the drive letter
347
# url_base is currently file://
348
# path is currently /C:/foo
349
url_base, path = _win32_extract_drive_letter(url_base, path)
350
# now it should be file:///C: and /foo
352
if exclude_trailing_slash and len(path) > 1 and path.endswith('/'):
354
head, tail = _posix_split(path)
355
return url_base + head, tail
358
def _win32_strip_local_trailing_slash(url):
359
"""Strip slashes after the drive letter"""
360
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
366
def strip_trailing_slash(url):
367
"""Strip trailing slash, except for root paths.
369
The definition of 'root path' is platform-dependent.
370
This assumes that all URLs are valid netloc urls, such that they
373
It searches for ://, and then refuses to remove the next '/'.
374
It can also handle relative paths
376
path/to/foo => path/to/foo
377
path/to/foo/ => path/to/foo
378
http://host/path/ => http://host/path
379
http://host/path => http://host/path
380
http://host/ => http://host/
382
file:///foo/ => file:///foo
383
# This is unique on win32 platforms, and is the only URL
384
# format which does it differently.
385
file:///c|/ => file:///c:/
387
if not url.endswith('/'):
390
if sys.platform == 'win32' and url.startswith('file:///'):
391
return _win32_strip_local_trailing_slash(url)
393
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
394
if scheme_loc is None:
395
# This is a relative path, as it has no scheme
396
# so just chop off the last character
399
if first_path_slash is None or first_path_slash == len(url)-1:
400
# Don't chop off anything if the only slash is the path
408
"""Unescape relpath from url format.
410
This returns a Unicode path from a URL
412
# jam 20060427 URLs are supposed to be ASCII only strings
413
# If they are passed in as unicode, urllib.unquote
414
# will return a UNICODE string, which actually contains
415
# utf-8 bytes. So we have to ensure that they are
416
# plain ASCII strings, or the final .decode will
417
# try to encode the UNICODE => ASCII, and then decode
421
except UnicodeError, e:
422
raise errors.InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
424
unquoted = urllib.unquote(url)
426
unicode_path = unquoted.decode('utf-8')
427
except UnicodeError, e:
428
raise errors.InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
432
# These are characters that if escaped, should stay that way
433
_no_decode_chars = ';/?:@&=+$,#'
434
_no_decode_ords = [ord(c) for c in _no_decode_chars]
435
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
436
+ ['%02X' % o for o in _no_decode_ords])
437
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
438
+ [('%02X' % o, chr(o)) for o in range(256)]))
439
#These entries get mapped to themselves
440
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
442
# These characters should not be escaped
443
_url_safe_characters = set('abcdefghijklmnopqrstuvwxyz'
444
'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
449
def unescape_for_display(url, encoding):
450
"""Decode what you can for a URL, so that we get a nice looking path.
452
This will turn file:// urls into local paths, and try to decode
453
any portions of a http:// style url that it can.
455
Any sections of the URL which can't be represented in the encoding or
456
need to stay as escapes are left alone.
458
:param url: A 7-bit ASCII URL
459
:param encoding: The final output encoding
461
:return: A unicode string which can be safely encoded into the
464
assert encoding is not None, 'you cannot specify None for the display encoding.'
465
if url.startswith('file://'):
467
path = local_path_from_url(url)
468
path.encode(encoding)
473
# Split into sections to try to decode utf-8
475
for i in xrange(1, len(res)):
476
escaped_chunks = res[i].split('%')
477
for j in xrange(1, len(escaped_chunks)):
478
item = escaped_chunks[j]
480
escaped_chunks[j] = _hex_display_map[item[:2]] + item[2:]
482
# Put back the percent symbol
483
escaped_chunks[j] = '%' + item
484
except UnicodeDecodeError:
485
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
486
unescaped = ''.join(escaped_chunks)
488
decoded = unescaped.decode('utf-8')
489
except UnicodeDecodeError:
490
# If this path segment cannot be properly utf-8 decoded
491
# after doing unescaping we will just leave it alone
495
decoded.encode(encoding)
496
except UnicodeEncodeError:
497
# If this chunk cannot be encoded in the local
498
# encoding, then we should leave it alone
501
# Otherwise take the url decoded one
503
return u'/'.join(res)
added
removed
bzrlib/urlutils.py
