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."""
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
path = m.group('path').split('/')
140
for chunk in arg.split('/'):
145
# Don't pop off the host portion
148
raise errors.InvalidURLJoin('Cannot go above root',
154
return '/'.join(path)
155
return scheme + '://' + '/'.join(path)
158
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
159
def _posix_local_path_from_url(url):
160
"""Convert a url like file:///path/to/foo into /path/to/foo"""
161
if not url.startswith('file:///'):
162
raise errors.InvalidURL(url, 'local urls must start with file:///')
163
# We only strip off 2 slashes
164
return unescape(url[len('file://'):])
167
def _posix_local_path_to_url(path):
168
"""Convert a local path like ./foo into a URL like file:///path/to/foo
170
This also handles transforming escaping unicode characters, etc.
172
# importing directly from posixpath allows us to test this
173
# on non-posix platforms
174
return 'file://' + escape(_posix_normpath(
175
osutils._posix_abspath(path)))
178
def _win32_local_path_from_url(url):
179
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
180
if not url.startswith('file:///'):
181
raise errors.InvalidURL(url, 'local urls must start with file:///')
182
# We strip off all 3 slashes
183
win32_url = url[len('file:///'):]
184
if (win32_url[0] not in ('abcdefghijklmnopqrstuvwxyz'
185
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
186
or win32_url[1] not in '|:'
187
or win32_url[2] != '/'):
188
raise errors.InvalidURL(url, 'Win32 file urls start with'
189
' file:///x:/, where x is a valid drive letter')
190
return win32_url[0].upper() + u':' + unescape(win32_url[2:])
193
def _win32_local_path_to_url(path):
194
"""Convert a local path like ./foo into a URL like file:///C:/path/to/foo
196
This also handles transforming escaping unicode characters, etc.
198
# importing directly from ntpath allows us to test this
199
# on non-win32 platform
200
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
201
# which actually strips trailing space characters.
202
# The worst part is that under linux ntpath.abspath has different
203
# semantics, since 'nt' is not an available module.
204
win32_path = osutils._nt_normpath(
205
osutils._win32_abspath(path)).replace('\\', '/')
206
return 'file:///' + win32_path[0].upper() + ':' + escape(win32_path[2:])
209
local_path_to_url = _posix_local_path_to_url
210
local_path_from_url = _posix_local_path_from_url
211
MIN_ABS_FILEURL_LENGTH = len('file:///')
212
WIN32_MIN_ABS_FILEURL_LENGTH = len('file:///C:/')
214
if sys.platform == 'win32':
215
local_path_to_url = _win32_local_path_to_url
216
local_path_from_url = _win32_local_path_from_url
218
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
221
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
224
def normalize_url(url):
225
"""Make sure that a path string is in fully normalized URL form.
227
This handles URLs which have unicode characters, spaces,
228
special characters, etc.
230
It has two basic modes of operation, depending on whether the
231
supplied string starts with a url specifier (scheme://) or not.
232
If it does not have a specifier it is considered a local path,
233
and will be converted into a file:/// url. Non-ascii characters
234
will be encoded using utf-8.
235
If it does have a url specifier, it will be treated as a "hybrid"
236
URL. Basically, a URL that should have URL special characters already
237
escaped (like +?&# etc), but may have unicode characters, etc
238
which would not be valid in a real URL.
240
:param url: Either a hybrid URL or a local path
241
:return: A normalized URL which only includes 7-bit ASCII characters.
243
m = _url_scheme_re.match(url)
245
return local_path_to_url(url)
246
if not isinstance(url, unicode):
248
if c not in _url_safe_characters:
249
raise errors.InvalidURL(url, 'URLs can only contain specific'
250
' safe characters (not %r)' % c)
252
# We have a unicode (hybrid) url
253
scheme = m.group('scheme')
254
path = list(m.group('path'))
256
for i in xrange(len(path)):
257
if path[i] not in _url_safe_characters:
258
chars = path[i].encode('utf-8')
259
path[i] = ''.join(['%%%02X' % ord(c) for c in path[i].encode('utf-8')])
260
return scheme + '://' + ''.join(path)
263
def relative_url(base, other):
264
"""Return a path to other from base.
266
If other is unrelated to base, return other. Else return a relative path.
267
This assumes no symlinks as part of the url.
269
dummy, base_first_slash = _find_scheme_and_separator(base)
270
if base_first_slash is None:
273
dummy, other_first_slash = _find_scheme_and_separator(other)
274
if other_first_slash is None:
277
# this takes care of differing schemes or hosts
278
base_scheme = base[:base_first_slash]
279
other_scheme = other[:other_first_slash]
280
if base_scheme != other_scheme:
283
base_path = base[base_first_slash+1:]
284
other_path = other[other_first_slash+1:]
286
if base_path.endswith('/'):
287
base_path = base_path[:-1]
289
base_sections = base_path.split('/')
290
other_sections = other_path.split('/')
292
if base_sections == ['']:
294
if other_sections == ['']:
298
for b, o in zip(base_sections, other_sections):
301
output_sections.append(b)
303
match_len = len(output_sections)
304
output_sections = ['..' for x in base_sections[match_len:]]
305
output_sections.extend(other_sections[match_len:])
307
return "/".join(output_sections) or "."
310
def _win32_extract_drive_letter(url_base, path):
311
"""On win32 the drive letter needs to be added to the url base."""
312
# Strip off the drive letter
313
# path is currently /C:/foo
314
if len(path) < 3 or path[2] not in ':|' or path[3] != '/':
315
raise errors.InvalidURL(url_base + path,
316
'win32 file:/// paths need a drive letter')
317
url_base += path[0:3] # file:// + /C:
318
path = path[3:] # /foo
319
return url_base, path
322
def split(url, exclude_trailing_slash=True):
323
"""Split a URL into its parent directory and a child directory.
325
:param url: A relative or absolute URL
326
:param exclude_trailing_slash: Strip off a final '/' if it is part
327
of the path (but not if it is part of the protocol specification)
329
:return: (parent_url, child_dir). child_dir may be the empty string if we're at
332
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
334
if first_path_slash is None:
335
# We have either a relative path, or no separating slash
336
if scheme_loc is None:
338
if exclude_trailing_slash and url.endswith('/'):
340
return _posix_split(url)
342
# Scheme with no path
345
# We have a fully defined path
346
url_base = url[:first_path_slash] # http://host, file://
347
path = url[first_path_slash:] # /file/foo
349
if sys.platform == 'win32' and url.startswith('file:///'):
350
# Strip off the drive letter
351
# url_base is currently file://
352
# path is currently /C:/foo
353
url_base, path = _win32_extract_drive_letter(url_base, path)
354
# now it should be file:///C: and /foo
356
if exclude_trailing_slash and len(path) > 1 and path.endswith('/'):
358
head, tail = _posix_split(path)
359
return url_base + head, tail
362
def _win32_strip_local_trailing_slash(url):
363
"""Strip slashes after the drive letter"""
364
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
370
def strip_trailing_slash(url):
371
"""Strip trailing slash, except for root paths.
373
The definition of 'root path' is platform-dependent.
374
This assumes that all URLs are valid netloc urls, such that they
377
It searches for ://, and then refuses to remove the next '/'.
378
It can also handle relative paths
380
path/to/foo => path/to/foo
381
path/to/foo/ => path/to/foo
382
http://host/path/ => http://host/path
383
http://host/path => http://host/path
384
http://host/ => http://host/
386
file:///foo/ => file:///foo
387
# This is unique on win32 platforms, and is the only URL
388
# format which does it differently.
389
file:///c|/ => file:///c:/
391
if not url.endswith('/'):
394
if sys.platform == 'win32' and url.startswith('file:///'):
395
return _win32_strip_local_trailing_slash(url)
397
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
398
if scheme_loc is None:
399
# This is a relative path, as it has no scheme
400
# so just chop off the last character
403
if first_path_slash is None or first_path_slash == len(url)-1:
404
# Don't chop off anything if the only slash is the path
412
"""Unescape relpath from url format.
414
This returns a Unicode path from a URL
416
# jam 20060427 URLs are supposed to be ASCII only strings
417
# If they are passed in as unicode, urllib.unquote
418
# will return a UNICODE string, which actually contains
419
# utf-8 bytes. So we have to ensure that they are
420
# plain ASCII strings, or the final .decode will
421
# try to encode the UNICODE => ASCII, and then decode
425
except UnicodeError, e:
426
raise errors.InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
428
unquoted = urllib.unquote(url)
430
unicode_path = unquoted.decode('utf-8')
431
except UnicodeError, e:
432
raise errors.InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
436
# These are characters that if escaped, should stay that way
437
_no_decode_chars = ';/?:@&=+$,#'
438
_no_decode_ords = [ord(c) for c in _no_decode_chars]
439
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
440
+ ['%02X' % o for o in _no_decode_ords])
441
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
442
+ [('%02X' % o, chr(o)) for o in range(256)]))
443
#These entries get mapped to themselves
444
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
446
# These characters should not be escaped
447
_url_safe_characters = set('abcdefghijklmnopqrstuvwxyz'
448
'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
453
def unescape_for_display(url, encoding):
454
"""Decode what you can for a URL, so that we get a nice looking path.
456
This will turn file:// urls into local paths, and try to decode
457
any portions of a http:// style url that it can.
459
Any sections of the URL which can't be represented in the encoding or
460
need to stay as escapes are left alone.
462
:param url: A 7-bit ASCII URL
463
:param encoding: The final output encoding
465
:return: A unicode string which can be safely encoded into the
468
assert encoding is not None, 'you cannot specify None for the display encoding.'
469
if url.startswith('file://'):
471
path = local_path_from_url(url)
472
path.encode(encoding)
477
# Split into sections to try to decode utf-8
479
for i in xrange(1, len(res)):
480
escaped_chunks = res[i].split('%')
481
for j in xrange(1, len(escaped_chunks)):
482
item = escaped_chunks[j]
484
escaped_chunks[j] = _hex_display_map[item[:2]] + item[2:]
486
# Put back the percent symbol
487
escaped_chunks[j] = '%' + item
488
except UnicodeDecodeError:
489
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
490
unescaped = ''.join(escaped_chunks)
492
decoded = unescaped.decode('utf-8')
493
except UnicodeDecodeError:
494
# If this path segment cannot be properly utf-8 decoded
495
# after doing unescaping we will just leave it alone
499
decoded.encode(encoding)
500
except UnicodeEncodeError:
501
# If this chunk cannot be encoded in the local
502
# encoding, then we should leave it alone
505
# Otherwise take the url decoded one
507
return u'/'.join(res)
added
removed
bzrlib/urlutils.py
