« back to all changes in this revision
Viewing changes to bzrlib/urlutils.py
- browse files at revision 1752
- compare with another revision
- compare with revision 3789.1.1
- download diff
- download tarball
- view history from revision 1752
- Committer: Canonical.com Patch Queue Manager
- Date: 2006-06-08 06:38:32 UTC
- mfrom: (1685.1.81 encoding)
- Revision ID: pqm@pqm.ubuntu.com-20060608063832-74b46cf8fdd4567a
(jam,mbp,wvh) Lots of updates to unicode,url,and encoding support
- files added:
- bzrlib/tests/EncodingAdapter.py
- files modified:
- NEWS
added
removed
bzrlib/urlutils.py
1
# Bazaar-NG -- distributed version control
2
#
3
# Copyright (C) 2006 by Canonical Ltd
4
#
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.
9
#
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.
14
#
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
18
19
"""A collection of function for handling URL operations."""
20
21
import os
22
from posixpath import split as _posix_split
23
import re
24
import sys
25
import urllib
26
27
import bzrlib.errors as errors
28
import bzrlib.osutils
29
30
31
def basename(url, exclude_trailing_slash=True):
32
"""Return the last component of a URL.
33
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
39
root of the URL.
40
"""
41
return split(url, exclude_trailing_slash=exclude_trailing_slash)[1]
42
43
44
def dirname(url, exclude_trailing_slash=True):
45
"""Return the parent directory of the given path.
46
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
52
"""
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]
56
57
58
def escape(relpath):
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))
65
66
67
def file_relpath(base, path):
68
"""Compute just the relative sub-portion of a url
69
70
This assumes that both paths are already fully specified file:// URLs.
71
"""
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)
75
76
base = local_path_from_url(base)
77
path = local_path_from_url(path)
78
return escape(bzrlib.osutils.relpath(base, path))
79
80
81
def _find_scheme_and_separator(url):
82
"""Find the scheme separator (://) and the first path separator
83
84
This is just a helper functions for other path utilities.
85
It could probably be replaced by urlparse
86
"""
87
m = _url_scheme_re.match(url)
88
if not m:
89
return None, None
90
91
scheme = m.group('scheme')
92
path = m.group('path')
93
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
100
101
102
def join(base, *args):
103
"""Create a URL by joining sections.
104
105
This will normalize '..', assuming that paths are absolute
106
(it assumes no symlinks in either path)
107
108
If any of *args is an absolute URL, it will be treated correctly.
109
Example:
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'
113
"""
114
m = _url_scheme_re.match(base)
115
scheme = None
116
if m:
117
scheme = m.group('scheme')
118
path = m.group('path').split('/')
119
else:
120
path = base.split('/')
121
122
for arg in args:
123
m = _url_scheme_re.match(arg)
124
if m:
125
# Absolute URL
126
scheme = m.group('scheme')
127
path = m.group('path').split('/')
128
else:
129
for chunk in arg.split('/'):
130
if chunk == '.':
131
continue
132
elif chunk == '..':
133
if len(path) >= 2:
134
# Don't pop off the host portion
135
path.pop()
136
else:
137
raise errors.InvalidURLJoin('Cannot go above root',
138
base, args)
139
else:
140
path.append(chunk)
141
142
if scheme is None:
143
return '/'.join(path)
144
return scheme + '://' + '/'.join(path)
145
146
147
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
148
def _posix_local_path_from_url(url):
149
"""Convert a url like file:///path/to/foo into /path/to/foo"""
150
if not url.startswith('file:///'):
151
raise errors.InvalidURL(url, 'local urls must start with file:///')
152
# We only strip off 2 slashes
153
return unescape(url[len('file://'):])
154
155
156
def _posix_local_path_to_url(path):
157
"""Convert a local path like ./foo into a URL like file:///path/to/foo
158
159
This also handles transforming escaping unicode characters, etc.
160
"""
161
# importing directly from posixpath allows us to test this
162
# on non-posix platforms
163
from posixpath import normpath
164
return 'file://' + escape(normpath(bzrlib.osutils._posix_abspath(path)))
165
166
167
def _win32_local_path_from_url(url):
168
"""Convert a url like file:///C|/path/to/foo into C:/path/to/foo"""
169
if not url.startswith('file:///'):
170
raise errors.InvalidURL(url, 'local urls must start with file:///')
171
# We strip off all 3 slashes
172
win32_url = url[len('file:///'):]
173
if (win32_url[0] not in 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ'
174
or win32_url[1] not in '|:'
175
or win32_url[2] != '/'):
176
raise errors.InvalidURL(url, 'Win32 file urls start with file:///X|/, where X is a valid drive letter')
177
# TODO: jam 20060426, we could .upper() or .lower() the drive letter
178
# for better consistency.
179
return win32_url[0].upper() + u':' + unescape(win32_url[2:])
180
181
182
def _win32_local_path_to_url(path):
183
"""Convert a local path like ./foo into a URL like file:///C|/path/to/foo
184
185
This also handles transforming escaping unicode characters, etc.
186
"""
187
# importing directly from ntpath allows us to test this
188
# on non-win32 platforms
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:])
192
193
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:///')
197
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
201
202
MIN_ABS_FILEURL_LENGTH = len('file:///C|/')
203
204
205
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
206
207
208
def normalize_url(url):
209
"""Make sure that a path string is in fully normalized URL form.
210
211
This handles URLs which have unicode characters, spaces,
212
special characters, etc.
213
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.
223
224
:param url: Either a hybrid URL or a local path
225
:return: A normalized URL which only includes 7-bit ASCII characters.
226
"""
227
m = _url_scheme_re.match(url)
228
if not m:
229
return local_path_to_url(url)
230
if not isinstance(url, unicode):
231
for c in url:
232
if c not in _url_safe_characters:
233
raise errors.InvalidURL(url, 'URLs can only contain specific'
234
' safe characters (not %r)' % c)
235
return url
236
# We have a unicode (hybrid) url
237
scheme = m.group('scheme')
238
path = list(m.group('path'))
239
240
for i in xrange(len(path)):
241
if path[i] not in _url_safe_characters:
242
chars = path[i].encode('utf-8')
243
path[i] = ''.join(['%%%02X' % ord(c) for c in path[i].encode('utf-8')])
244
return scheme + '://' + ''.join(path)
245
246
247
def relative_url(base, other):
248
"""Return a path to other from base.
249
250
If other is unrelated to base, return other. Else return a relative path.
251
This assumes no symlinks as part of the url.
252
"""
253
dummy, base_first_slash = _find_scheme_and_separator(base)
254
if base_first_slash is None:
255
return other
256
257
dummy, other_first_slash = _find_scheme_and_separator(other)
258
if other_first_slash is None:
259
return other
260
261
# this takes care of differing schemes or hosts
262
base_scheme = base[:base_first_slash]
263
other_scheme = other[:other_first_slash]
264
if base_scheme != other_scheme:
265
return other
266
267
base_path = base[base_first_slash+1:]
268
other_path = other[other_first_slash+1:]
269
270
if base_path.endswith('/'):
271
base_path = base_path[:-1]
272
273
base_sections = base_path.split('/')
274
other_sections = other_path.split('/')
275
276
if base_sections == ['']:
277
base_sections = []
278
if other_sections == ['']:
279
other_sections = []
280
281
output_sections = []
282
for b, o in zip(base_sections, other_sections):
283
if b != o:
284
break
285
output_sections.append(b)
286
287
match_len = len(output_sections)
288
output_sections = ['..' for x in base_sections[match_len:]]
289
output_sections.extend(other_sections[match_len:])
290
291
return "/".join(output_sections) or "."
292
293
294
def split(url, exclude_trailing_slash=True):
295
"""Split a URL into its parent directory and a child directory.
296
297
:param url: A relative or absolute URL
298
:param exclude_trailing_slash: Strip off a final '/' if it is part
299
of the path (but not if it is part of the protocol specification)
300
301
:return: (parent_url, child_dir). child_dir may be the empty string if we're at
302
the root.
303
"""
304
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
305
306
if first_path_slash is None:
307
# We have either a relative path, or no separating slash
308
if scheme_loc is None:
309
# Relative path
310
if exclude_trailing_slash and url.endswith('/'):
311
url = url[:-1]
312
return _posix_split(url)
313
else:
314
# Scheme with no path
315
return url, ''
316
317
# We have a fully defined path
318
url_base = url[:first_path_slash] # http://host, file://
319
path = url[first_path_slash:] # /file/foo
320
321
if sys.platform == 'win32' and url.startswith('file:///'):
322
# Strip off the drive letter
323
if path[2:3] not in '\\/':
324
raise errors.InvalidURL(url,
325
'win32 file:/// paths need a drive letter')
326
url_base += path[1:4] # file:///C|/
327
path = path[3:]
328
329
if exclude_trailing_slash and len(path) > 1 and path.endswith('/'):
330
path = path[:-1]
331
head, tail = _posix_split(path)
332
return url_base + head, tail
333
334
335
def strip_trailing_slash(url):
336
"""Strip trailing slash, except for root paths.
337
338
The definition of 'root path' is platform-dependent.
339
This assumes that all URLs are valid netloc urls, such that they
340
form:
341
scheme://host/path
342
It searches for ://, and then refuses to remove the next '/'.
343
It can also handle relative paths
344
Examples:
345
path/to/foo => path/to/foo
346
path/to/foo/ => path/to/foo
347
http://host/path/ => http://host/path
348
http://host/path => http://host/path
349
http://host/ => http://host/
350
file:/// => file:///
351
file:///foo/ => file:///foo
352
# This is unique on win32 platforms, and is the only URL
353
# format which does it differently.
354
file:///C|/ => file:///C|/
355
"""
356
if not url.endswith('/'):
357
# Nothing to do
358
return url
359
if sys.platform == 'win32' and url.startswith('file:///'):
360
# This gets handled specially, because the 'top-level'
361
# of a win32 path is actually the drive letter
362
if len(url) > MIN_ABS_FILEURL_LENGTH:
363
return url[:-1]
364
else:
365
return url
366
367
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
368
if scheme_loc is None:
369
# This is a relative path, as it has no scheme
370
# so just chop off the last character
371
return url[:-1]
372
373
if first_path_slash is None or first_path_slash == len(url)-1:
374
# Don't chop off anything if the only slash is the path
375
# separating slash
376
return url
377
378
return url[:-1]
379
380
381
def unescape(url):
382
"""Unescape relpath from url format.
383
384
This returns a Unicode path from a URL
385
"""
386
# jam 20060427 URLs are supposed to be ASCII only strings
387
# If they are passed in as unicode, urllib.unquote
388
# will return a UNICODE string, which actually contains
389
# utf-8 bytes. So we have to ensure that they are
390
# plain ASCII strings, or the final .decode will
391
# try to encode the UNICODE => ASCII, and then decode
392
# it into utf-8.
393
try:
394
url = str(url)
395
except UnicodeError, e:
396
raise errors.InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
397
398
unquoted = urllib.unquote(url)
399
try:
400
unicode_path = unquoted.decode('utf-8')
401
except UnicodeError, e:
402
raise errors.InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
403
return unicode_path
404
405
406
# These are characters that if escaped, should stay that way
407
_no_decode_chars = ';/?:@&=+$,#'
408
_no_decode_ords = [ord(c) for c in _no_decode_chars]
409
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
410
+ ['%02X' % o for o in _no_decode_ords])
411
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
412
+ [('%02X' % o, chr(o)) for o in range(256)]))
413
#These entries get mapped to themselves
414
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
415
416
# These characters should not be escaped
417
_url_safe_characters = set('abcdefghijklmnopqrstuvwxyz'
418
'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
419
'0123456789' '_.-/'
420
';?:@&=+$,%#')
421
422
423
def unescape_for_display(url, encoding):
424
"""Decode what you can for a URL, so that we get a nice looking path.
425
426
This will turn file:// urls into local paths, and try to decode
427
any portions of a http:// style url that it can.
428
429
Any sections of the URL which can't be represented in the encoding or
430
need to stay as escapes are left alone.
431
432
:param url: A 7-bit ASCII URL
433
:param encoding: The final output encoding
434
435
:return: A unicode string which can be safely encoded into the
436
specified encoding.
437
"""
438
if url.startswith('file://'):
439
try:
440
path = local_path_from_url(url)
441
path.encode(encoding)
442
return path
443
except UnicodeError:
444
return url
445
446
# Split into sections to try to decode utf-8
447
res = url.split('/')
448
for i in xrange(1, len(res)):
449
escaped_chunks = res[i].split('%')
450
for j in xrange(1, len(escaped_chunks)):
451
item = escaped_chunks[j]
452
try:
453
escaped_chunks[j] = _hex_display_map[item[:2]] + item[2:]
454
except KeyError:
455
# Put back the percent symbol
456
escaped_chunks[j] = '%' + item
457
except UnicodeDecodeError:
458
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
459
unescaped = ''.join(escaped_chunks)
460
try:
461
decoded = unescaped.decode('utf-8')
462
except UnicodeDecodeError:
463
# If this path segment cannot be properly utf-8 decoded
464
# after doing unescaping we will just leave it alone
465
pass
466
else:
467
try:
468
decoded.encode(encoding)
469
except UnicodeEncodeError:
470
# If this chunk cannot be encoded in the local
471
# encoding, then we should leave it alone
472
pass
473
else:
474
# Otherwise take the url decoded one
475
res[i] = decoded
476
return u'/'.join(res)
Loggerhead is a web-based interface for Breezy
Version: 2.0.1