13
15
# You should have received a copy of the GNU General Public License
14
16
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
19
"""A collection of function for handling URL operations."""
22
from posixpath import split as _posix_split, normpath as _posix_normpath
23
from bzrlib.lazy_import import lazy_import
24
lazy_import(globals(), """
25
from posixpath import split as _posix_split, normpath as _posix_normpath
27
import bzrlib.errors as errors
36
31
def basename(url, exclude_trailing_slash=True):
66
61
relpath = relpath.encode('utf-8')
67
62
# After quoting and encoding, the path should be perfectly
68
63
# safe as a plain ASCII string, str() just enforces this
69
return str(urllib.quote(relpath, safe='/~'))
64
return str(urllib.quote(relpath))
72
67
def file_relpath(base, path):
73
68
"""Compute just the relative sub-portion of a url
75
70
This assumes that both paths are already fully specified file:// URLs.
77
if len(base) < MIN_ABS_FILEURL_LENGTH:
78
raise ValueError('Length of base (%r) must equal or'
79
' exceed the platform minimum url length (which is %d)' %
80
(base, MIN_ABS_FILEURL_LENGTH))
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)
81
76
base = local_path_from_url(base)
82
77
path = local_path_from_url(path)
83
return escape(osutils.relpath(base, path))
78
return escape(bzrlib.osutils.relpath(base, path))
86
81
def _find_scheme_and_separator(url):
116
111
join('http://foo', 'bar') => 'http://foo/bar'
117
112
join('http://foo', 'bar', '../baz') => 'http://foo/baz'
121
match = _url_scheme_re.match(base)
114
m = _url_scheme_re.match(base)
124
scheme = match.group('scheme')
125
path = match.group('path').split('/')
126
if path[-1:] == ['']:
127
# Strip off a trailing slash
128
# This helps both when we are at the root, and when
129
# 'base' has an extra slash at the end
117
scheme = m.group('scheme')
118
path = m.group('path').split('/')
132
120
path = base.split('/')
134
if scheme is not None and len(path) >= 1:
136
# the path should be represented as an abs path.
137
# we know this must be absolute because of the presence of a URL scheme.
139
path = [''] + path[1:]
141
# create an empty host, but dont alter the path - this might be a
142
# relative url fragment.
147
match = _url_scheme_re.match(arg)
123
m = _url_scheme_re.match(arg)
150
scheme = match.group('scheme')
151
# this skips .. normalisation, making http://host/../../..
153
path = match.group('path').split('/')
154
# set the host and path according to new absolute URL, discarding
155
# any previous values.
156
# XXX: duplicates mess from earlier in this function. This URL
157
# manipulation code needs some cleaning up.
158
if scheme is not None and len(path) >= 1:
161
# url scheme implies absolute path.
164
# no url scheme we take the path as is.
126
scheme = m.group('scheme')
127
path = m.group('path').split('/')
167
path = '/'.join(path)
168
path = joinpath(path, arg)
169
path = path.split('/')
170
if remove_root and path[0:1] == ['']:
173
# Remove the leading slash from the path, so long as it isn't also the
174
# trailing slash, which we want to keep if present.
175
if path and path[0] == '' and len(path) > 1:
129
for chunk in arg.split('/'):
134
# Don't pop off the host portion
137
raise errors.InvalidURLJoin('Cannot go above root',
179
142
if scheme is None:
180
143
return '/'.join(path)
181
144
return scheme + '://' + '/'.join(path)
184
def joinpath(base, *args):
185
"""Join URL path segments to a URL path segment.
187
This is somewhat like osutils.joinpath, but intended for URLs.
189
XXX: this duplicates some normalisation logic, and also duplicates a lot of
190
path handling logic that already exists in some Transport implementations.
191
We really should try to have exactly one place in the code base responsible
192
for combining paths of URLs.
194
path = base.split('/')
195
if len(path) > 1 and path[-1] == '':
196
#If the path ends in a trailing /, remove it.
199
if arg.startswith('/'):
201
for chunk in arg.split('/'):
206
raise errors.InvalidURLJoin('Cannot go above root',
214
return '/'.join(path)
217
147
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
218
148
def _posix_local_path_from_url(url):
219
149
"""Convert a url like file:///path/to/foo into /path/to/foo"""
220
file_localhost_prefix = 'file://localhost/'
221
if url.startswith(file_localhost_prefix):
222
path = url[len(file_localhost_prefix) - 1:]
223
elif not url.startswith('file:///'):
224
raise errors.InvalidURL(
225
url, 'local urls must start with file:/// or file://localhost/')
227
path = url[len('file://'):]
150
if not url.startswith('file:///'):
151
raise errors.InvalidURL(url, 'local urls must start with file:///')
228
152
# We only strip off 2 slashes
229
return unescape(path)
153
return unescape(url[len('file://'):])
232
156
def _posix_local_path_to_url(path):
235
159
This also handles transforming escaping unicode characters, etc.
237
# importing directly from posixpath allows us to test this
161
# importing directly from posixpath allows us to test this
238
162
# on non-posix platforms
239
163
return 'file://' + escape(_posix_normpath(
240
osutils._posix_abspath(path)))
164
bzrlib.osutils._posix_abspath(path)))
243
167
def _win32_local_path_from_url(url):
244
168
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
245
if not url.startswith('file://'):
246
raise errors.InvalidURL(url, 'local urls must start with file:///, '
247
'UNC path urls must start with file://')
169
if not url.startswith('file:///'):
170
raise errors.InvalidURL(url, 'local urls must start with file:///')
248
171
# We strip off all 3 slashes
249
win32_url = url[len('file:'):]
250
# check for UNC path: //HOST/path
251
if not win32_url.startswith('///'):
252
if (win32_url[2] == '/'
253
or win32_url[3] in '|:'):
254
raise errors.InvalidURL(url, 'Win32 UNC path urls'
255
' have form file://HOST/path')
256
return unescape(win32_url)
258
# allow empty paths so we can serve all roots
259
if win32_url == '///':
262
# usual local path with drive letter
263
if (win32_url[3] not in ('abcdefghijklmnopqrstuvwxyz'
172
win32_url = url[len('file:///'):]
173
if (win32_url[0] not in ('abcdefghijklmnopqrstuvwxyz'
264
174
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
265
or win32_url[4] not in '|:'
266
or win32_url[5] != '/'):
175
or win32_url[1] not in '|:'
176
or win32_url[2] != '/'):
267
177
raise errors.InvalidURL(url, 'Win32 file urls start with'
268
178
' file:///x:/, where x is a valid drive letter')
269
return win32_url[3].upper() + u':' + unescape(win32_url[5:])
179
# Preferentially using .lower() because os.getcwd() returns
180
# paths with lowercase drive letters, and that helps
181
# bzrlib.osutils.relpath() work correctly
182
return win32_url[0].lower() + u':' + unescape(win32_url[2:])
272
185
def _win32_local_path_to_url(path):
275
188
This also handles transforming escaping unicode characters, etc.
277
# importing directly from ntpath allows us to test this
190
# importing directly from ntpath allows us to test this
278
191
# on non-win32 platform
279
192
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
280
193
# which actually strips trailing space characters.
281
194
# The worst part is that under linux ntpath.abspath has different
282
195
# semantics, since 'nt' is not an available module.
286
win32_path = osutils._win32_abspath(path)
287
# check for UNC path \\HOST\path
288
if win32_path.startswith('//'):
289
return 'file:' + escape(win32_path)
290
return ('file:///' + str(win32_path[0].upper()) + ':' +
291
escape(win32_path[2:]))
196
win32_path = bzrlib.osutils._nt_normpath(
197
bzrlib.osutils._win32_abspath(path)).replace('\\', '/')
198
return 'file:///' + win32_path[0].lower() + ':' + escape(win32_path[2:])
294
201
local_path_to_url = _posix_local_path_to_url
295
202
local_path_from_url = _posix_local_path_from_url
296
203
MIN_ABS_FILEURL_LENGTH = len('file:///')
297
WIN32_MIN_ABS_FILEURL_LENGTH = len('file:///C:/')
299
205
if sys.platform == 'win32':
300
206
local_path_to_url = _win32_local_path_to_url
301
207
local_path_from_url = _win32_local_path_from_url
303
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
209
MIN_ABS_FILEURL_LENGTH = len('file:///C:/')
306
212
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
307
_url_hex_escapes_re = re.compile(r'(%[0-9a-fA-F]{2})')
310
def _unescape_safe_chars(matchobj):
311
"""re.sub callback to convert hex-escapes to plain characters (if safe).
313
e.g. '%7E' will be converted to '~'.
315
hex_digits = matchobj.group(0)[1:]
316
char = chr(int(hex_digits, 16))
317
if char in _url_dont_escape_characters:
320
return matchobj.group(0).upper()
323
215
def normalize_url(url):
324
216
"""Make sure that a path string is in fully normalized URL form.
326
This handles URLs which have unicode characters, spaces,
218
This handles URLs which have unicode characters, spaces,
327
219
special characters, etc.
329
221
It has two basic modes of operation, depending on whether the
342
234
m = _url_scheme_re.match(url)
344
236
return local_path_to_url(url)
345
scheme = m.group('scheme')
346
path = m.group('path')
347
237
if not isinstance(url, unicode):
349
239
if c not in _url_safe_characters:
350
240
raise errors.InvalidURL(url, 'URLs can only contain specific'
351
241
' safe characters (not %r)' % c)
352
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
353
return str(scheme + '://' + ''.join(path))
355
243
# We have a unicode (hybrid) url
356
path_chars = list(path)
244
scheme = m.group('scheme')
245
path = list(m.group('path'))
358
for i in xrange(len(path_chars)):
359
if path_chars[i] not in _url_safe_characters:
360
chars = path_chars[i].encode('utf-8')
361
path_chars[i] = ''.join(
362
['%%%02X' % ord(c) for c in path_chars[i].encode('utf-8')])
363
path = ''.join(path_chars)
364
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
365
return str(scheme + '://' + path)
247
for i in xrange(len(path)):
248
if path[i] not in _url_safe_characters:
249
chars = path[i].encode('utf-8')
250
path[i] = ''.join(['%%%02X' % ord(c) for c in path[i].encode('utf-8')])
251
return scheme + '://' + ''.join(path)
368
254
def relative_url(base, other):
417
298
return "/".join(output_sections) or "."
420
def _win32_extract_drive_letter(url_base, path):
421
"""On win32 the drive letter needs to be added to the url base."""
422
# Strip off the drive letter
423
# path is currently /C:/foo
424
if len(path) < 3 or path[2] not in ':|' or path[3] != '/':
425
raise errors.InvalidURL(url_base + path,
426
'win32 file:/// paths need a drive letter')
427
url_base += path[0:3] # file:// + /C:
428
path = path[3:] # /foo
429
return url_base, path
432
301
def split(url, exclude_trailing_slash=True):
433
302
"""Split a URL into its parent directory and a child directory.
459
328
if sys.platform == 'win32' and url.startswith('file:///'):
460
329
# Strip off the drive letter
461
# url_base is currently file://
462
330
# path is currently /C:/foo
463
url_base, path = _win32_extract_drive_letter(url_base, path)
464
# now it should be file:///C: and /foo
331
if path[2:3] not in ':|' or path[3:4] not in '\\/':
332
raise errors.InvalidURL(url,
333
'win32 file:/// paths need a drive letter')
334
url_base += path[0:3] # file:// + /c:
335
path = path[3:] # /foo
466
337
if exclude_trailing_slash and len(path) > 1 and path.endswith('/'):
546
414
# These are characters that if escaped, should stay that way
547
415
_no_decode_chars = ';/?:@&=+$,#'
548
416
_no_decode_ords = [ord(c) for c in _no_decode_chars]
549
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
417
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
550
418
+ ['%02X' % o for o in _no_decode_ords])
551
419
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
552
420
+ [('%02X' % o, chr(o)) for o in range(256)]))
553
421
#These entries get mapped to themselves
554
422
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
556
# These characters shouldn't be percent-encoded, and it's always safe to
557
# unencode them if they are.
558
_url_dont_escape_characters = set(
559
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
560
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
561
"0123456789" # Numbers
562
"-._~" # Unreserved characters
565
424
# These characters should not be escaped
566
_url_safe_characters = set(
567
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
568
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
569
"0123456789" # Numbers
570
"_.-!~*'()" # Unreserved characters
571
"/;?:@&=+$," # Reserved characters
572
"%#" # Extra reserved characters
425
_url_safe_characters = set('abcdefghijklmnopqrstuvwxyz'
426
'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
575
431
def unescape_for_display(url, encoding):
576
432
"""Decode what you can for a URL, so that we get a nice looking path.
578
434
This will turn file:// urls into local paths, and try to decode
579
435
any portions of a http:// style url that it can.
581
Any sections of the URL which can't be represented in the encoding or
437
Any sections of the URL which can't be represented in the encoding or
582
438
need to stay as escapes are left alone.
584
440
:param url: A 7-bit ASCII URL
585
441
:param encoding: The final output encoding
587
:return: A unicode string which can be safely encoded into the
443
:return: A unicode string which can be safely encoded into the
588
444
specified encoding.
591
raise ValueError('you cannot specify None for the display encoding')
446
assert encoding is not None, 'you cannot specify None for the display encoding.'
592
447
if url.startswith('file://'):
594
449
path = local_path_from_url(url)
628
483
# Otherwise take the url decoded one
630
485
return u'/'.join(res)
633
def derive_to_location(from_location):
634
"""Derive a TO_LOCATION given a FROM_LOCATION.
636
The normal case is a FROM_LOCATION of http://foo/bar => bar.
637
The Right Thing for some logical destinations may differ though
638
because no / may be present at all. In that case, the result is
639
the full name without the scheme indicator, e.g. lp:foo-bar => foo-bar.
640
This latter case also applies when a Windows drive
641
is used without a path, e.g. c:foo-bar => foo-bar.
642
If no /, path separator or : is found, the from_location is returned.
644
if from_location.find("/") >= 0 or from_location.find(os.sep) >= 0:
645
return os.path.basename(from_location.rstrip("/\\"))
647
sep = from_location.find(":")
649
return from_location[sep+1:]
654
def _is_absolute(url):
655
return (osutils.pathjoin('/foo', url) == url)
658
def rebase_url(url, old_base, new_base):
659
"""Convert a relative path from an old base URL to a new base URL.
661
The result will be a relative path.
662
Absolute paths and full URLs are returned unaltered.
664
scheme, separator = _find_scheme_and_separator(url)
665
if scheme is not None:
667
if _is_absolute(url):
669
old_parsed = urlparse.urlparse(old_base)
670
new_parsed = urlparse.urlparse(new_base)
671
if (old_parsed[:2]) != (new_parsed[:2]):
672
raise errors.InvalidRebaseURLs(old_base, new_base)
673
return determine_relative_path(new_parsed[2],
674
join(old_parsed[2], url))
677
def determine_relative_path(from_path, to_path):
678
"""Determine a relative path from from_path to to_path."""
679
from_segments = osutils.splitpath(from_path)
680
to_segments = osutils.splitpath(to_path)
682
for count, (from_element, to_element) in enumerate(zip(from_segments,
684
if from_element != to_element:
688
unique_from = from_segments[count:]
689
unique_to = to_segments[count:]
690
segments = (['..'] * len(unique_from) + unique_to)
691
if len(segments) == 0:
693
return osutils.pathjoin(*segments)
698
"""Extract the server address, the credentials and the path from the url.
700
user, password, host and path should be quoted if they contain reserved
703
:param url: an quoted url
705
:return: (scheme, user, password, host, port, path) tuple, all fields
708
if isinstance(url, unicode):
709
raise errors.InvalidURL('should be ascii:\n%r' % url)
710
url = url.encode('utf-8')
711
(scheme, netloc, path, params,
712
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
713
user = password = host = port = None
715
user, host = netloc.rsplit('@', 1)
717
user, password = user.split(':', 1)
718
password = urllib.unquote(password)
719
user = urllib.unquote(user)
723
if ':' in host and not (host[0] == '[' and host[-1] == ']'): #there *is* port
724
host, port = host.rsplit(':',1)
728
raise errors.InvalidURL('invalid port number %s in url:\n%s' %
730
if host != "" and host[0] == '[' and host[-1] == ']': #IPv6
733
host = urllib.unquote(host)
734
path = urllib.unquote(path)
736
return (scheme, user, password, host, port, path)
added
removed
bzrlib/urlutils.py
