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
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('/')
120
path = base.split('/')
123
m = _url_scheme_re.match(arg)
126
scheme = m.group('scheme')
127
path = m.group('path').split('/')
129
for chunk in arg.split('/'):
134
# Don't pop off the host portion
137
raise errors.InvalidURLJoin('Cannot go above root',
142
return '/'.join(path)
143
return scheme + '://' + '/'.join(path)
146
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
147
def _posix_local_path_from_url(url):
148
"""Convert a url like file:///path/to/foo into /path/to/foo"""
149
if not url.startswith('file:///'):
150
raise errors.InvalidURL(url, 'local urls must start with file:///')
151
# We only strip off 2 slashes
152
return unescape(url[len('file://'):])
155
def _posix_local_path_to_url(path):
156
"""Convert a local path like ./foo into a URL like file:///path/to/foo
158
This also handles transforming escaping unicode characters, etc.
160
# importing directly from posixpath allows us to test this
161
# on non-posix platforms
162
from posixpath import normpath
163
return 'file://' + escape(normpath(bzrlib.osutils._posix_abspath(path)))
166
def _win32_local_path_from_url(url):
167
"""Convert a url like file:///C|/path/to/foo into C:/path/to/foo"""
168
if not url.startswith('file:///'):
169
raise errors.InvalidURL(url, 'local urls must start with file:///')
170
# We strip off all 3 slashes
171
win32_url = url[len('file:///'):]
172
if (win32_url[0] not in 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ'
173
or win32_url[1] not in '|:'
174
or win32_url[2] != '/'):
175
raise errors.InvalidURL(url, 'Win32 file urls start with file:///X|/, where X is a valid drive letter')
176
# TODO: jam 20060426, we could .upper() or .lower() the drive letter
177
# for better consistency.
178
return win32_url[0].upper() + u':' + unescape(win32_url[2:])
181
def _win32_local_path_to_url(path):
182
"""Convert a local path like ./foo into a URL like file:///C|/path/to/foo
184
This also handles transforming escaping unicode characters, etc.
186
# importing directly from ntpath allows us to test this
187
# on non-win32 platforms
188
# TODO: jam 20060426 consider moving this import outside of the function
189
win32_path = bzrlib.osutils._nt_normpath(
190
bzrlib.osutils._win32_abspath(path)).replace('\\', '/')
191
return 'file:///' + win32_path[0].upper() + '|' + escape(win32_path[2:])
194
local_path_to_url = _posix_local_path_to_url
195
local_path_from_url = _posix_local_path_from_url
196
MIN_ABS_FILEURL_LENGTH = len('file:///')
198
if sys.platform == 'win32':
199
local_path_to_url = _win32_local_path_to_url
200
local_path_from_url = _win32_local_path_from_url
202
MIN_ABS_FILEURL_LENGTH = len('file:///C|/')
205
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
208
def normalize_url(url):
209
"""Make sure that a path string is in fully normalized URL form.
211
This handles URLs which have unicode characters, spaces,
212
special characters, etc.
214
It has two basic modes of operation, depending on whether the
215
supplied string starts with a url specifier (scheme://) or not.
216
If it does not have a specifier it is considered a local path,
217
and will be converted into a file:/// url. Non-ascii characters
218
will be encoded using utf-8.
219
If it does have a url specifier, it will be treated as a "hybrid"
220
URL. Basically, a URL that should have URL special characters already
221
escaped (like +?&# etc), but may have unicode characters, etc
222
which would not be valid in a real URL.
224
:param url: Either a hybrid URL or a local path
225
:return: A normalized URL which only includes 7-bit ASCII characters.
227
m = _url_scheme_re.match(url)
229
return local_path_to_url(url)
230
if not isinstance(url, unicode):
231
# TODO: jam 20060510 We need to test for ascii characters that
232
# shouldn't be allowed in URLs
234
if c not in _url_safe_characters:
235
raise errors.InvalidURL(url, 'URLs can only contain specific'
236
' safe characters (not %r)' % c)
238
# We have a unicode (hybrid) url
239
scheme = m.group('scheme')
240
path = list(m.group('path'))
242
for i in xrange(len(path)):
243
if path[i] not in _url_safe_characters:
244
chars = path[i].encode('utf-8')
245
path[i] = ''.join(['%%%02X' % ord(c) for c in path[i].encode('utf-8')])
246
return scheme + '://' + ''.join(path)
249
def split(url, exclude_trailing_slash=True):
250
"""Split a URL into its parent directory and a child directory.
252
:param url: A relative or absolute URL
253
:param exclude_trailing_slash: Strip off a final '/' if it is part
254
of the path (but not if it is part of the protocol specification)
256
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
258
if first_path_slash is None:
259
# We have either a relative path, or no separating slash
260
if scheme_loc is None:
262
if exclude_trailing_slash and url.endswith('/'):
264
return _posix_split(url)
266
# Scheme with no path
269
# We have a fully defined path
270
url_base = url[:first_path_slash] # http://host, file://
271
path = url[first_path_slash:] # /file/foo
273
if sys.platform == 'win32' and url.startswith('file:///'):
274
# Strip off the drive letter
275
if path[2:3] not in '\\/':
276
raise errors.InvalidURL(url,
277
'win32 file:/// paths need a drive letter')
278
url_base += path[1:4] # file:///C|/
281
if exclude_trailing_slash and len(path) > 1 and path.endswith('/'):
283
head, tail = _posix_split(path)
284
return url_base + head, tail
287
def strip_trailing_slash(url):
288
"""Strip trailing slash, except for root paths.
290
The definition of 'root path' is platform-dependent.
291
This assumes that all URLs are valid netloc urls, such that they
294
It searches for ://, and then refuses to remove the next '/'.
295
It can also handle relative paths
297
path/to/foo => path/to/foo
298
path/to/foo/ => path/to/foo
299
http://host/path/ => http://host/path
300
http://host/path => http://host/path
301
http://host/ => http://host/
303
file:///foo/ => file:///foo
304
# This is unique on win32 platforms, and is the only URL
305
# format which does it differently.
306
file:///C|/ => file:///C|/
308
if not url.endswith('/'):
311
if sys.platform == 'win32' and url.startswith('file:///'):
312
# This gets handled specially, because the 'top-level'
313
# of a win32 path is actually the drive letter
314
if len(url) > MIN_ABS_FILEURL_LENGTH:
318
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
319
if scheme_loc is None:
320
# This is a relative path, as it has no scheme
321
# so just chop off the last character
324
if first_path_slash is None or first_path_slash == len(url)-1:
325
# Don't chop off anything if the only slash is the path
333
"""Unescape relpath from url format.
335
This returns a Unicode path from a URL
337
# jam 20060427 URLs are supposed to be ASCII only strings
338
# If they are passed in as unicode, urllib.unquote
339
# will return a UNICODE string, which actually contains
340
# utf-8 bytes. So we have to ensure that they are
341
# plain ASCII strings, or the final .decode will
342
# try to encode the UNICODE => ASCII, and then decode
346
except UnicodeError, e:
347
raise errors.InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
348
unquoted = urllib.unquote(url)
350
unicode_path = unquoted.decode('utf-8')
351
except UnicodeError, e:
352
raise errors.InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
356
# These are characters that if escaped, should stay that way
357
_no_decode_chars = ';/?:@&=+$,#'
358
_no_decode_ords = [ord(c) for c in _no_decode_chars]
359
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
360
+ ['%02X' % o for o in _no_decode_ords])
361
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
362
+ [('%02X' % o, chr(o)) for o in range(256)]))
363
#These entries get mapped to themselves
364
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
366
# These characters should not be escaped
367
_url_safe_characters = set('abcdefghijklmnopqrstuvwxyz'
368
'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
373
def unescape_for_display(url, encoding):
374
"""Decode what you can for a URL, so that we get a nice looking path.
376
This will turn file:// urls into local paths, and try to decode
377
any portions of a http:// style url that it can.
379
Any sections of the URL which can't be represented in the encoding or
380
need to stay as escapes are left alone.
382
:param url: A 7-bit ASCII URL
383
:param encoding: The final output encoding
385
:return: A unicode string which can be safely encoded into the
388
if url.startswith('file://'):
390
path = local_path_from_url(url)
391
path.encode(encoding)
396
# Split into sections to try to decode utf-8
398
for i in xrange(1, len(res)):
399
escaped_chunks = res[i].split('%')
400
for j in xrange(1, len(escaped_chunks)):
401
item = escaped_chunks[j]
403
escaped_chunks[j] = _hex_display_map[item[:2]] + item[2:]
405
# Put back the percent symbol
406
escaped_chunks[j] = '%' + item
407
except UnicodeDecodeError:
408
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
409
unescaped = ''.join(escaped_chunks)
411
decoded = unescaped.decode('utf-8')
412
except UnicodeDecodeError:
413
# If this path segment cannot be properly utf-8 decoded
414
# after doing unescaping we will just leave it alone
418
decoded.encode(encoding)
419
except UnicodeEncodeError:
420
# If this chunk cannot be encoded in the local
421
# encoding, then we should leave it alone
424
# Otherwise take the url decoded one
426
return u'/'.join(res)
added
removed
bzrlib/urlutils.py
