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
def joinpath(base, *args):
161
"""Join URL path segments to a URL path segment.
163
This is somewhat like osutils.joinpath, but intended for URLs.
165
XXX: this duplicates some normalisation logic, and also duplicates a lot of
166
path handling logic that already exists in some Transport implementations.
167
We really should try to have exactly one place in the code base responsible
168
for combining paths of URLs.
173
path = base.split('/')
175
if arg.startswith('/'):
177
for chunk in arg.split('/'):
182
raise errors.InvalidURLJoin('Cannot go above root',
190
return '/'.join(path)
193
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
194
def _posix_local_path_from_url(url):
195
"""Convert a url like file:///path/to/foo into /path/to/foo"""
196
if not url.startswith('file:///'):
197
raise errors.InvalidURL(url, 'local urls must start with file:///')
198
# We only strip off 2 slashes
199
return unescape(url[len('file://'):])
202
def _posix_local_path_to_url(path):
203
"""Convert a local path like ./foo into a URL like file:///path/to/foo
205
This also handles transforming escaping unicode characters, etc.
207
# importing directly from posixpath allows us to test this
208
# on non-posix platforms
209
return 'file://' + escape(_posix_normpath(
210
osutils._posix_abspath(path)))
213
def _win32_local_path_from_url(url):
214
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
215
if not url.startswith('file://'):
216
raise errors.InvalidURL(url, 'local urls must start with file:///, '
217
'UNC path urls must start with file://')
218
# We strip off all 3 slashes
219
win32_url = url[len('file:'):]
220
# check for UNC path: //HOST/path
221
if not win32_url.startswith('///'):
222
if (win32_url[2] == '/'
223
or win32_url[3] in '|:'):
224
raise errors.InvalidURL(url, 'Win32 UNC path urls'
225
' have form file://HOST/path')
226
return unescape(win32_url)
227
# usual local path with drive letter
228
if (win32_url[3] not in ('abcdefghijklmnopqrstuvwxyz'
229
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
230
or win32_url[4] not in '|:'
231
or win32_url[5] != '/'):
232
raise errors.InvalidURL(url, 'Win32 file urls start with'
233
' file:///x:/, where x is a valid drive letter')
234
return win32_url[3].upper() + u':' + unescape(win32_url[5:])
237
def _win32_local_path_to_url(path):
238
"""Convert a local path like ./foo into a URL like file:///C:/path/to/foo
240
This also handles transforming escaping unicode characters, etc.
242
# importing directly from ntpath allows us to test this
243
# on non-win32 platform
244
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
245
# which actually strips trailing space characters.
246
# The worst part is that under linux ntpath.abspath has different
247
# semantics, since 'nt' is not an available module.
248
win32_path = osutils._nt_normpath(
249
osutils._win32_abspath(path)).replace('\\', '/')
250
# check for UNC path \\HOST\path
251
if win32_path.startswith('//'):
252
return 'file:' + escape(win32_path)
253
return 'file:///' + win32_path[0].upper() + ':' + escape(win32_path[2:])
256
local_path_to_url = _posix_local_path_to_url
257
local_path_from_url = _posix_local_path_from_url
258
MIN_ABS_FILEURL_LENGTH = len('file:///')
259
WIN32_MIN_ABS_FILEURL_LENGTH = len('file:///C:/')
261
if sys.platform == 'win32':
262
local_path_to_url = _win32_local_path_to_url
263
local_path_from_url = _win32_local_path_from_url
265
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
268
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
271
def normalize_url(url):
272
"""Make sure that a path string is in fully normalized URL form.
274
This handles URLs which have unicode characters, spaces,
275
special characters, etc.
277
It has two basic modes of operation, depending on whether the
278
supplied string starts with a url specifier (scheme://) or not.
279
If it does not have a specifier it is considered a local path,
280
and will be converted into a file:/// url. Non-ascii characters
281
will be encoded using utf-8.
282
If it does have a url specifier, it will be treated as a "hybrid"
283
URL. Basically, a URL that should have URL special characters already
284
escaped (like +?&# etc), but may have unicode characters, etc
285
which would not be valid in a real URL.
287
:param url: Either a hybrid URL or a local path
288
:return: A normalized URL which only includes 7-bit ASCII characters.
290
m = _url_scheme_re.match(url)
292
return local_path_to_url(url)
293
if not isinstance(url, unicode):
295
if c not in _url_safe_characters:
296
raise errors.InvalidURL(url, 'URLs can only contain specific'
297
' safe characters (not %r)' % c)
299
# We have a unicode (hybrid) url
300
scheme = m.group('scheme')
301
path = list(m.group('path'))
303
for i in xrange(len(path)):
304
if path[i] not in _url_safe_characters:
305
chars = path[i].encode('utf-8')
306
path[i] = ''.join(['%%%02X' % ord(c) for c in path[i].encode('utf-8')])
307
return str(scheme + '://' + ''.join(path))
310
def relative_url(base, other):
311
"""Return a path to other from base.
313
If other is unrelated to base, return other. Else return a relative path.
314
This assumes no symlinks as part of the url.
316
dummy, base_first_slash = _find_scheme_and_separator(base)
317
if base_first_slash is None:
320
dummy, other_first_slash = _find_scheme_and_separator(other)
321
if other_first_slash is None:
324
# this takes care of differing schemes or hosts
325
base_scheme = base[:base_first_slash]
326
other_scheme = other[:other_first_slash]
327
if base_scheme != other_scheme:
330
base_path = base[base_first_slash+1:]
331
other_path = other[other_first_slash+1:]
333
if base_path.endswith('/'):
334
base_path = base_path[:-1]
336
base_sections = base_path.split('/')
337
other_sections = other_path.split('/')
339
if base_sections == ['']:
341
if other_sections == ['']:
345
for b, o in zip(base_sections, other_sections):
348
output_sections.append(b)
350
match_len = len(output_sections)
351
output_sections = ['..' for x in base_sections[match_len:]]
352
output_sections.extend(other_sections[match_len:])
354
return "/".join(output_sections) or "."
357
def _win32_extract_drive_letter(url_base, path):
358
"""On win32 the drive letter needs to be added to the url base."""
359
# Strip off the drive letter
360
# path is currently /C:/foo
361
if len(path) < 3 or path[2] not in ':|' or path[3] != '/':
362
raise errors.InvalidURL(url_base + path,
363
'win32 file:/// paths need a drive letter')
364
url_base += path[0:3] # file:// + /C:
365
path = path[3:] # /foo
366
return url_base, path
369
def split(url, exclude_trailing_slash=True):
370
"""Split a URL into its parent directory and a child directory.
372
:param url: A relative or absolute URL
373
:param exclude_trailing_slash: Strip off a final '/' if it is part
374
of the path (but not if it is part of the protocol specification)
376
:return: (parent_url, child_dir). child_dir may be the empty string if we're at
379
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
381
if first_path_slash is None:
382
# We have either a relative path, or no separating slash
383
if scheme_loc is None:
385
if exclude_trailing_slash and url.endswith('/'):
387
return _posix_split(url)
389
# Scheme with no path
392
# We have a fully defined path
393
url_base = url[:first_path_slash] # http://host, file://
394
path = url[first_path_slash:] # /file/foo
396
if sys.platform == 'win32' and url.startswith('file:///'):
397
# Strip off the drive letter
398
# url_base is currently file://
399
# path is currently /C:/foo
400
url_base, path = _win32_extract_drive_letter(url_base, path)
401
# now it should be file:///C: and /foo
403
if exclude_trailing_slash and len(path) > 1 and path.endswith('/'):
405
head, tail = _posix_split(path)
406
return url_base + head, tail
409
def _win32_strip_local_trailing_slash(url):
410
"""Strip slashes after the drive letter"""
411
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
417
def strip_trailing_slash(url):
418
"""Strip trailing slash, except for root paths.
420
The definition of 'root path' is platform-dependent.
421
This assumes that all URLs are valid netloc urls, such that they
424
It searches for ://, and then refuses to remove the next '/'.
425
It can also handle relative paths
427
path/to/foo => path/to/foo
428
path/to/foo/ => path/to/foo
429
http://host/path/ => http://host/path
430
http://host/path => http://host/path
431
http://host/ => http://host/
433
file:///foo/ => file:///foo
434
# This is unique on win32 platforms, and is the only URL
435
# format which does it differently.
436
file:///c|/ => file:///c:/
438
if not url.endswith('/'):
441
if sys.platform == 'win32' and url.startswith('file:///'):
442
return _win32_strip_local_trailing_slash(url)
444
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
445
if scheme_loc is None:
446
# This is a relative path, as it has no scheme
447
# so just chop off the last character
450
if first_path_slash is None or first_path_slash == len(url)-1:
451
# Don't chop off anything if the only slash is the path
459
"""Unescape relpath from url format.
461
This returns a Unicode path from a URL
463
# jam 20060427 URLs are supposed to be ASCII only strings
464
# If they are passed in as unicode, urllib.unquote
465
# will return a UNICODE string, which actually contains
466
# utf-8 bytes. So we have to ensure that they are
467
# plain ASCII strings, or the final .decode will
468
# try to encode the UNICODE => ASCII, and then decode
472
except UnicodeError, e:
473
raise errors.InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
475
unquoted = urllib.unquote(url)
477
unicode_path = unquoted.decode('utf-8')
478
except UnicodeError, e:
479
raise errors.InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
483
# These are characters that if escaped, should stay that way
484
_no_decode_chars = ';/?:@&=+$,#'
485
_no_decode_ords = [ord(c) for c in _no_decode_chars]
486
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
487
+ ['%02X' % o for o in _no_decode_ords])
488
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
489
+ [('%02X' % o, chr(o)) for o in range(256)]))
490
#These entries get mapped to themselves
491
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
493
# These characters should not be escaped
494
_url_safe_characters = set(
495
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
496
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
497
"0123456789" # Numbers
498
"_.-!~*'()" # Unreserved characters
499
"/;?:@&=+$," # Reserved characters
500
"%#" # Extra reserved characters
503
def unescape_for_display(url, encoding):
504
"""Decode what you can for a URL, so that we get a nice looking path.
506
This will turn file:// urls into local paths, and try to decode
507
any portions of a http:// style url that it can.
509
Any sections of the URL which can't be represented in the encoding or
510
need to stay as escapes are left alone.
512
:param url: A 7-bit ASCII URL
513
:param encoding: The final output encoding
515
:return: A unicode string which can be safely encoded into the
518
assert encoding is not None, 'you cannot specify None for the display encoding.'
519
if url.startswith('file://'):
521
path = local_path_from_url(url)
522
path.encode(encoding)
527
# Split into sections to try to decode utf-8
529
for i in xrange(1, len(res)):
530
escaped_chunks = res[i].split('%')
531
for j in xrange(1, len(escaped_chunks)):
532
item = escaped_chunks[j]
534
escaped_chunks[j] = _hex_display_map[item[:2]] + item[2:]
536
# Put back the percent symbol
537
escaped_chunks[j] = '%' + item
538
except UnicodeDecodeError:
539
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
540
unescaped = ''.join(escaped_chunks)
542
decoded = unescaped.decode('utf-8')
543
except UnicodeDecodeError:
544
# If this path segment cannot be properly utf-8 decoded
545
# after doing unescaping we will just leave it alone
549
decoded.encode(encoding)
550
except UnicodeEncodeError:
551
# If this chunk cannot be encoded in the local
552
# encoding, then we should leave it alone
555
# Otherwise take the url decoded one
557
return u'/'.join(res)
added
removed
bzrlib/urlutils.py
