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
# We strip off all 3 slashes
185
win32_url = url[len('file:///'):]
186
if (win32_url[0] not in ('abcdefghijklmnopqrstuvwxyz'
187
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
188
or win32_url[1] not in '|:'
189
or win32_url[2] != '/'):
190
raise errors.InvalidURL(url, 'Win32 file urls start with'
191
' file:///x:/, where x is a valid drive letter')
192
return win32_url[0].upper() + u':' + unescape(win32_url[2:])
195
def _win32_local_path_to_url(path):
196
"""Convert a local path like ./foo into a URL like file:///C:/path/to/foo
198
This also handles transforming escaping unicode characters, etc.
200
# importing directly from ntpath allows us to test this
201
# on non-win32 platform
202
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
203
# which actually strips trailing space characters.
204
# The worst part is that under linux ntpath.abspath has different
205
# semantics, since 'nt' is not an available module.
206
win32_path = osutils._nt_normpath(
207
osutils._win32_abspath(path)).replace('\\', '/')
208
return 'file:///' + win32_path[0].upper() + ':' + escape(win32_path[2:])
211
local_path_to_url = _posix_local_path_to_url
212
local_path_from_url = _posix_local_path_from_url
213
MIN_ABS_FILEURL_LENGTH = len('file:///')
214
WIN32_MIN_ABS_FILEURL_LENGTH = len('file:///C:/')
216
if sys.platform == 'win32':
217
local_path_to_url = _win32_local_path_to_url
218
local_path_from_url = _win32_local_path_from_url
220
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
223
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
226
def normalize_url(url):
227
"""Make sure that a path string is in fully normalized URL form.
229
This handles URLs which have unicode characters, spaces,
230
special characters, etc.
232
It has two basic modes of operation, depending on whether the
233
supplied string starts with a url specifier (scheme://) or not.
234
If it does not have a specifier it is considered a local path,
235
and will be converted into a file:/// url. Non-ascii characters
236
will be encoded using utf-8.
237
If it does have a url specifier, it will be treated as a "hybrid"
238
URL. Basically, a URL that should have URL special characters already
239
escaped (like +?&# etc), but may have unicode characters, etc
240
which would not be valid in a real URL.
242
:param url: Either a hybrid URL or a local path
243
:return: A normalized URL which only includes 7-bit ASCII characters.
245
m = _url_scheme_re.match(url)
247
return local_path_to_url(url)
248
if not isinstance(url, unicode):
250
if c not in _url_safe_characters:
251
raise errors.InvalidURL(url, 'URLs can only contain specific'
252
' safe characters (not %r)' % c)
254
# We have a unicode (hybrid) url
255
scheme = m.group('scheme')
256
path = list(m.group('path'))
258
for i in xrange(len(path)):
259
if path[i] not in _url_safe_characters:
260
chars = path[i].encode('utf-8')
261
path[i] = ''.join(['%%%02X' % ord(c) for c in path[i].encode('utf-8')])
262
return scheme + '://' + ''.join(path)
265
def relative_url(base, other):
266
"""Return a path to other from base.
268
If other is unrelated to base, return other. Else return a relative path.
269
This assumes no symlinks as part of the url.
271
dummy, base_first_slash = _find_scheme_and_separator(base)
272
if base_first_slash is None:
275
dummy, other_first_slash = _find_scheme_and_separator(other)
276
if other_first_slash is None:
279
# this takes care of differing schemes or hosts
280
base_scheme = base[:base_first_slash]
281
other_scheme = other[:other_first_slash]
282
if base_scheme != other_scheme:
285
base_path = base[base_first_slash+1:]
286
other_path = other[other_first_slash+1:]
288
if base_path.endswith('/'):
289
base_path = base_path[:-1]
291
base_sections = base_path.split('/')
292
other_sections = other_path.split('/')
294
if base_sections == ['']:
296
if other_sections == ['']:
300
for b, o in zip(base_sections, other_sections):
303
output_sections.append(b)
305
match_len = len(output_sections)
306
output_sections = ['..' for x in base_sections[match_len:]]
307
output_sections.extend(other_sections[match_len:])
309
return "/".join(output_sections) or "."
312
def _win32_extract_drive_letter(url_base, path):
313
"""On win32 the drive letter needs to be added to the url base."""
314
# Strip off the drive letter
315
# path is currently /C:/foo
316
if len(path) < 3 or path[2] not in ':|' or path[3] != '/':
317
raise errors.InvalidURL(url_base + path,
318
'win32 file:/// paths need a drive letter')
319
url_base += path[0:3] # file:// + /C:
320
path = path[3:] # /foo
321
return url_base, path
324
def split(url, exclude_trailing_slash=True):
325
"""Split a URL into its parent directory and a child directory.
327
:param url: A relative or absolute URL
328
:param exclude_trailing_slash: Strip off a final '/' if it is part
329
of the path (but not if it is part of the protocol specification)
331
:return: (parent_url, child_dir). child_dir may be the empty string if we're at
334
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
336
if first_path_slash is None:
337
# We have either a relative path, or no separating slash
338
if scheme_loc is None:
340
if exclude_trailing_slash and url.endswith('/'):
342
return _posix_split(url)
344
# Scheme with no path
347
# We have a fully defined path
348
url_base = url[:first_path_slash] # http://host, file://
349
path = url[first_path_slash:] # /file/foo
351
if sys.platform == 'win32' and url.startswith('file:///'):
352
# Strip off the drive letter
353
# url_base is currently file://
354
# path is currently /C:/foo
355
url_base, path = _win32_extract_drive_letter(url_base, path)
356
# now it should be file:///C: and /foo
358
if exclude_trailing_slash and len(path) > 1 and path.endswith('/'):
360
head, tail = _posix_split(path)
361
return url_base + head, tail
364
def _win32_strip_local_trailing_slash(url):
365
"""Strip slashes after the drive letter"""
366
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
372
def strip_trailing_slash(url):
373
"""Strip trailing slash, except for root paths.
375
The definition of 'root path' is platform-dependent.
376
This assumes that all URLs are valid netloc urls, such that they
379
It searches for ://, and then refuses to remove the next '/'.
380
It can also handle relative paths
382
path/to/foo => path/to/foo
383
path/to/foo/ => path/to/foo
384
http://host/path/ => http://host/path
385
http://host/path => http://host/path
386
http://host/ => http://host/
388
file:///foo/ => file:///foo
389
# This is unique on win32 platforms, and is the only URL
390
# format which does it differently.
391
file:///c|/ => file:///c:/
393
if not url.endswith('/'):
396
if sys.platform == 'win32' and url.startswith('file:///'):
397
return _win32_strip_local_trailing_slash(url)
399
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
400
if scheme_loc is None:
401
# This is a relative path, as it has no scheme
402
# so just chop off the last character
405
if first_path_slash is None or first_path_slash == len(url)-1:
406
# Don't chop off anything if the only slash is the path
414
"""Unescape relpath from url format.
416
This returns a Unicode path from a URL
418
# jam 20060427 URLs are supposed to be ASCII only strings
419
# If they are passed in as unicode, urllib.unquote
420
# will return a UNICODE string, which actually contains
421
# utf-8 bytes. So we have to ensure that they are
422
# plain ASCII strings, or the final .decode will
423
# try to encode the UNICODE => ASCII, and then decode
427
except UnicodeError, e:
428
raise errors.InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
430
unquoted = urllib.unquote(url)
432
unicode_path = unquoted.decode('utf-8')
433
except UnicodeError, e:
434
raise errors.InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
438
# These are characters that if escaped, should stay that way
439
_no_decode_chars = ';/?:@&=+$,#'
440
_no_decode_ords = [ord(c) for c in _no_decode_chars]
441
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
442
+ ['%02X' % o for o in _no_decode_ords])
443
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
444
+ [('%02X' % o, chr(o)) for o in range(256)]))
445
#These entries get mapped to themselves
446
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
448
# These characters should not be escaped
449
_url_safe_characters = set('abcdefghijklmnopqrstuvwxyz'
450
'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
455
def unescape_for_display(url, encoding):
456
"""Decode what you can for a URL, so that we get a nice looking path.
458
This will turn file:// urls into local paths, and try to decode
459
any portions of a http:// style url that it can.
461
Any sections of the URL which can't be represented in the encoding or
462
need to stay as escapes are left alone.
464
:param url: A 7-bit ASCII URL
465
:param encoding: The final output encoding
467
:return: A unicode string which can be safely encoded into the
470
assert encoding is not None, 'you cannot specify None for the display encoding.'
471
if url.startswith('file://'):
473
path = local_path_from_url(url)
474
path.encode(encoding)
479
# Split into sections to try to decode utf-8
481
for i in xrange(1, len(res)):
482
escaped_chunks = res[i].split('%')
483
for j in xrange(1, len(escaped_chunks)):
484
item = escaped_chunks[j]
486
escaped_chunks[j] = _hex_display_map[item[:2]] + item[2:]
488
# Put back the percent symbol
489
escaped_chunks[j] = '%' + item
490
except UnicodeDecodeError:
491
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
492
unescaped = ''.join(escaped_chunks)
494
decoded = unescaped.decode('utf-8')
495
except UnicodeDecodeError:
496
# If this path segment cannot be properly utf-8 decoded
497
# after doing unescaping we will just leave it alone
501
decoded.encode(encoding)
502
except UnicodeEncodeError:
503
# If this chunk cannot be encoded in the local
504
# encoding, then we should leave it alone
507
# Otherwise take the url decoded one
509
return u'/'.join(res)
added
removed
bzrlib/urlutils.py
