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
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
return win32_url[0].upper() + u':' + unescape(win32_url[2:])
187
def _win32_local_path_to_url(path):
188
"""Convert a local path like ./foo into a URL like file:///C:/path/to/foo
190
This also handles transforming escaping unicode characters, etc.
192
# importing directly from ntpath allows us to test this
193
# on non-win32 platform
194
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
195
# which actually strips trailing space characters.
196
# The worst part is that under linux ntpath.abspath has different
197
# semantics, since 'nt' is not an available module.
198
win32_path = bzrlib.osutils._nt_normpath(
199
bzrlib.osutils._win32_abspath(path)).replace('\\', '/')
200
return 'file:///' + win32_path[0].upper() + ':' + escape(win32_path[2:])
203
local_path_to_url = _posix_local_path_to_url
204
local_path_from_url = _posix_local_path_from_url
205
MIN_ABS_FILEURL_LENGTH = len('file:///')
206
WIN32_MIN_ABS_FILEURL_LENGTH = len('file:///C:/')
208
if sys.platform == 'win32':
209
local_path_to_url = _win32_local_path_to_url
210
local_path_from_url = _win32_local_path_from_url
212
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
215
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
218
def normalize_url(url):
219
"""Make sure that a path string is in fully normalized URL form.
221
This handles URLs which have unicode characters, spaces,
222
special characters, etc.
224
It has two basic modes of operation, depending on whether the
225
supplied string starts with a url specifier (scheme://) or not.
226
If it does not have a specifier it is considered a local path,
227
and will be converted into a file:/// url. Non-ascii characters
228
will be encoded using utf-8.
229
If it does have a url specifier, it will be treated as a "hybrid"
230
URL. Basically, a URL that should have URL special characters already
231
escaped (like +?&# etc), but may have unicode characters, etc
232
which would not be valid in a real URL.
234
:param url: Either a hybrid URL or a local path
235
:return: A normalized URL which only includes 7-bit ASCII characters.
237
m = _url_scheme_re.match(url)
239
return local_path_to_url(url)
240
if not isinstance(url, unicode):
242
if c not in _url_safe_characters:
243
raise errors.InvalidURL(url, 'URLs can only contain specific'
244
' safe characters (not %r)' % c)
246
# We have a unicode (hybrid) url
247
scheme = m.group('scheme')
248
path = list(m.group('path'))
250
for i in xrange(len(path)):
251
if path[i] not in _url_safe_characters:
252
chars = path[i].encode('utf-8')
253
path[i] = ''.join(['%%%02X' % ord(c) for c in path[i].encode('utf-8')])
254
return scheme + '://' + ''.join(path)
257
def relative_url(base, other):
258
"""Return a path to other from base.
260
If other is unrelated to base, return other. Else return a relative path.
261
This assumes no symlinks as part of the url.
263
dummy, base_first_slash = _find_scheme_and_separator(base)
264
if base_first_slash is None:
267
dummy, other_first_slash = _find_scheme_and_separator(other)
268
if other_first_slash is None:
271
# this takes care of differing schemes or hosts
272
base_scheme = base[:base_first_slash]
273
other_scheme = other[:other_first_slash]
274
if base_scheme != other_scheme:
277
base_path = base[base_first_slash+1:]
278
other_path = other[other_first_slash+1:]
280
if base_path.endswith('/'):
281
base_path = base_path[:-1]
283
base_sections = base_path.split('/')
284
other_sections = other_path.split('/')
286
if base_sections == ['']:
288
if other_sections == ['']:
292
for b, o in zip(base_sections, other_sections):
295
output_sections.append(b)
297
match_len = len(output_sections)
298
output_sections = ['..' for x in base_sections[match_len:]]
299
output_sections.extend(other_sections[match_len:])
301
return "/".join(output_sections) or "."
304
def _win32_extract_drive_letter(url_base, path):
305
"""On win32 the drive letter needs to be added to the url base."""
306
# Strip off the drive letter
307
# path is currently /C:/foo
308
if len(path) < 3 or path[2] not in ':|' or path[3] != '/':
309
raise errors.InvalidURL(url_base + path,
310
'win32 file:/// paths need a drive letter')
311
url_base += path[0:3] # file:// + /C:
312
path = path[3:] # /foo
313
return url_base, path
316
def split(url, exclude_trailing_slash=True):
317
"""Split a URL into its parent directory and a child directory.
319
:param url: A relative or absolute URL
320
:param exclude_trailing_slash: Strip off a final '/' if it is part
321
of the path (but not if it is part of the protocol specification)
323
:return: (parent_url, child_dir). child_dir may be the empty string if we're at
326
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
328
if first_path_slash is None:
329
# We have either a relative path, or no separating slash
330
if scheme_loc is None:
332
if exclude_trailing_slash and url.endswith('/'):
334
return _posix_split(url)
336
# Scheme with no path
339
# We have a fully defined path
340
url_base = url[:first_path_slash] # http://host, file://
341
path = url[first_path_slash:] # /file/foo
343
if sys.platform == 'win32' and url.startswith('file:///'):
344
# Strip off the drive letter
345
# url_base is currently file://
346
# path is currently /C:/foo
347
url_base, path = _win32_extract_drive_letter(url_base, path)
348
# now it should be file:///C: and /foo
350
if exclude_trailing_slash and len(path) > 1 and path.endswith('/'):
352
head, tail = _posix_split(path)
353
return url_base + head, tail
356
def _win32_strip_local_trailing_slash(url):
357
"""Strip slashes after the drive letter"""
358
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
364
def strip_trailing_slash(url):
365
"""Strip trailing slash, except for root paths.
367
The definition of 'root path' is platform-dependent.
368
This assumes that all URLs are valid netloc urls, such that they
371
It searches for ://, and then refuses to remove the next '/'.
372
It can also handle relative paths
374
path/to/foo => path/to/foo
375
path/to/foo/ => path/to/foo
376
http://host/path/ => http://host/path
377
http://host/path => http://host/path
378
http://host/ => http://host/
380
file:///foo/ => file:///foo
381
# This is unique on win32 platforms, and is the only URL
382
# format which does it differently.
383
file:///c|/ => file:///c:/
385
if not url.endswith('/'):
388
if sys.platform == 'win32' and url.startswith('file:///'):
389
return _win32_strip_local_trailing_slash(url)
391
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
392
if scheme_loc is None:
393
# This is a relative path, as it has no scheme
394
# so just chop off the last character
397
if first_path_slash is None or first_path_slash == len(url)-1:
398
# Don't chop off anything if the only slash is the path
406
"""Unescape relpath from url format.
408
This returns a Unicode path from a URL
410
# jam 20060427 URLs are supposed to be ASCII only strings
411
# If they are passed in as unicode, urllib.unquote
412
# will return a UNICODE string, which actually contains
413
# utf-8 bytes. So we have to ensure that they are
414
# plain ASCII strings, or the final .decode will
415
# try to encode the UNICODE => ASCII, and then decode
419
except UnicodeError, e:
420
raise errors.InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
422
unquoted = urllib.unquote(url)
424
unicode_path = unquoted.decode('utf-8')
425
except UnicodeError, e:
426
raise errors.InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
430
# These are characters that if escaped, should stay that way
431
_no_decode_chars = ';/?:@&=+$,#'
432
_no_decode_ords = [ord(c) for c in _no_decode_chars]
433
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
434
+ ['%02X' % o for o in _no_decode_ords])
435
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
436
+ [('%02X' % o, chr(o)) for o in range(256)]))
437
#These entries get mapped to themselves
438
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
440
# These characters should not be escaped
441
_url_safe_characters = set('abcdefghijklmnopqrstuvwxyz'
442
'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
447
def unescape_for_display(url, encoding):
448
"""Decode what you can for a URL, so that we get a nice looking path.
450
This will turn file:// urls into local paths, and try to decode
451
any portions of a http:// style url that it can.
453
Any sections of the URL which can't be represented in the encoding or
454
need to stay as escapes are left alone.
456
:param url: A 7-bit ASCII URL
457
:param encoding: The final output encoding
459
:return: A unicode string which can be safely encoded into the
462
assert encoding is not None, 'you cannot specify None for the display encoding.'
463
if url.startswith('file://'):
465
path = local_path_from_url(url)
466
path.encode(encoding)
471
# Split into sections to try to decode utf-8
473
for i in xrange(1, len(res)):
474
escaped_chunks = res[i].split('%')
475
for j in xrange(1, len(escaped_chunks)):
476
item = escaped_chunks[j]
478
escaped_chunks[j] = _hex_display_map[item[:2]] + item[2:]
480
# Put back the percent symbol
481
escaped_chunks[j] = '%' + item
482
except UnicodeDecodeError:
483
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
484
unescaped = ''.join(escaped_chunks)
486
decoded = unescaped.decode('utf-8')
487
except UnicodeDecodeError:
488
# If this path segment cannot be properly utf-8 decoded
489
# after doing unescaping we will just leave it alone
493
decoded.encode(encoding)
494
except UnicodeEncodeError:
495
# If this chunk cannot be encoded in the local
496
# encoding, then we should leave it alone
499
# Otherwise take the url decoded one
501
return u'/'.join(res)
added
removed
bzrlib/urlutils.py
