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):
132
125
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)
128
m = _url_scheme_re.match(arg)
150
scheme = match.group('scheme')
131
scheme = m.group('scheme')
151
132
# this skips .. normalisation, making http://host/../../..
152
133
# be rather strange.
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.
134
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:
136
for chunk in arg.split('/'):
141
# Don't pop off the host portion
144
raise errors.InvalidURLJoin('Cannot go above root',
179
149
if scheme is None:
180
150
return '/'.join(path)
181
151
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
154
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
218
155
def _posix_local_path_from_url(url):
219
156
"""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://'):]
157
if not url.startswith('file:///'):
158
raise errors.InvalidURL(url, 'local urls must start with file:///')
228
159
# We only strip off 2 slashes
229
return unescape(path)
160
return unescape(url[len('file://'):])
232
163
def _posix_local_path_to_url(path):
235
166
This also handles transforming escaping unicode characters, etc.
237
# importing directly from posixpath allows us to test this
168
# importing directly from posixpath allows us to test this
238
169
# on non-posix platforms
239
170
return 'file://' + escape(_posix_normpath(
240
osutils._posix_abspath(path)))
171
bzrlib.osutils._posix_abspath(path)))
243
174
def _win32_local_path_from_url(url):
244
175
"""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://')
176
if not url.startswith('file:///'):
177
raise errors.InvalidURL(url, 'local urls must start with file:///')
248
178
# 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'
179
win32_url = url[len('file:///'):]
180
if (win32_url[0] not in ('abcdefghijklmnopqrstuvwxyz'
264
181
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
265
or win32_url[4] not in '|:'
266
or win32_url[5] != '/'):
182
or win32_url[1] not in '|:'
183
or win32_url[2] != '/'):
267
184
raise errors.InvalidURL(url, 'Win32 file urls start with'
268
185
' file:///x:/, where x is a valid drive letter')
269
return win32_url[3].upper() + u':' + unescape(win32_url[5:])
186
return win32_url[0].upper() + u':' + unescape(win32_url[2:])
272
189
def _win32_local_path_to_url(path):
275
192
This also handles transforming escaping unicode characters, etc.
277
# importing directly from ntpath allows us to test this
194
# importing directly from ntpath allows us to test this
278
195
# on non-win32 platform
279
196
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
280
197
# which actually strips trailing space characters.
281
# The worst part is that on linux ntpath.abspath has different
198
# The worst part is that under linux ntpath.abspath has different
282
199
# 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:]))
200
win32_path = bzrlib.osutils._nt_normpath(
201
bzrlib.osutils._win32_abspath(path)).replace('\\', '/')
202
return 'file:///' + win32_path[0].upper() + ':' + escape(win32_path[2:])
294
205
local_path_to_url = _posix_local_path_to_url
306
217
_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
220
def normalize_url(url):
324
221
"""Make sure that a path string is in fully normalized URL form.
326
This handles URLs which have unicode characters, spaces,
223
This handles URLs which have unicode characters, spaces,
327
224
special characters, etc.
329
226
It has two basic modes of operation, depending on whether the
342
239
m = _url_scheme_re.match(url)
344
241
return local_path_to_url(url)
345
scheme = m.group('scheme')
346
path = m.group('path')
347
242
if not isinstance(url, unicode):
349
244
if c not in _url_safe_characters:
350
245
raise errors.InvalidURL(url, 'URLs can only contain specific'
351
246
' safe characters (not %r)' % c)
352
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
353
return str(scheme + '://' + ''.join(path))
355
248
# We have a unicode (hybrid) url
356
path_chars = list(path)
249
scheme = m.group('scheme')
250
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)
252
for i in xrange(len(path)):
253
if path[i] not in _url_safe_characters:
254
chars = path[i].encode('utf-8')
255
path[i] = ''.join(['%%%02X' % ord(c) for c in path[i].encode('utf-8')])
256
return scheme + '://' + ''.join(path)
368
259
def relative_url(base, other):
469
355
return url_base + head, tail
472
def split_segment_parameters_raw(url):
473
"""Split the subsegment of the last segment of a URL.
475
:param url: A relative or absolute URL
476
:return: (url, subsegments)
478
(parent_url, child_dir) = split(url)
479
subsegments = child_dir.split(",")
480
if len(subsegments) == 1:
482
return (join(parent_url, subsegments[0]), subsegments[1:])
485
def split_segment_parameters(url):
486
"""Split the segment parameters of the last segment of a URL.
488
:param url: A relative or absolute URL
489
:return: (url, segment_parameters)
491
(base_url, subsegments) = split_segment_parameters_raw(url)
493
for subsegment in subsegments:
494
(key, value) = subsegment.split("=", 1)
495
parameters[key] = value
496
return (base_url, parameters)
499
def join_segment_parameters_raw(base, *subsegments):
500
"""Create a new URL by adding subsegments to an existing one.
502
This adds the specified subsegments to the last path in the specified
503
base URL. The subsegments should be bytestrings.
505
:note: You probably want to use join_segment_parameters instead.
509
for subsegment in subsegments:
510
if type(subsegment) is not str:
511
raise TypeError("Subsegment %r is not a bytestring" % subsegment)
512
if "," in subsegment:
513
raise errors.InvalidURLJoin(", exists in subsegments",
515
return ",".join((base,) + subsegments)
518
def join_segment_parameters(url, parameters):
519
"""Create a new URL by adding segment parameters to an existing one.
521
The parameters of the last segment in the URL will be updated; if a
522
parameter with the same key already exists it will be overwritten.
524
:param url: A URL, as string
525
:param parameters: Dictionary of parameters, keys and values as bytestrings
527
(base, existing_parameters) = split_segment_parameters(url)
529
new_parameters.update(existing_parameters)
530
for key, value in parameters.iteritems():
531
if type(key) is not str:
532
raise TypeError("parameter key %r is not a bytestring" % key)
533
if type(value) is not str:
534
raise TypeError("parameter value %r for %s is not a bytestring" %
537
raise errors.InvalidURLJoin("= exists in parameter key", url,
539
new_parameters[key] = value
540
return join_segment_parameters_raw(base,
541
*["%s=%s" % item for item in sorted(new_parameters.items())])
544
358
def _win32_strip_local_trailing_slash(url):
545
359
"""Strip slashes after the drive letter"""
546
360
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
618
432
# These are characters that if escaped, should stay that way
619
433
_no_decode_chars = ';/?:@&=+$,#'
620
434
_no_decode_ords = [ord(c) for c in _no_decode_chars]
621
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
435
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
622
436
+ ['%02X' % o for o in _no_decode_ords])
623
437
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
624
438
+ [('%02X' % o, chr(o)) for o in range(256)]))
625
439
#These entries get mapped to themselves
626
440
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
628
# These characters shouldn't be percent-encoded, and it's always safe to
629
# unencode them if they are.
630
_url_dont_escape_characters = set(
631
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
632
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
633
"0123456789" # Numbers
634
"-._~" # Unreserved characters
637
442
# These characters should not be escaped
638
_url_safe_characters = set(
639
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
640
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
641
"0123456789" # Numbers
642
"_.-!~*'()" # Unreserved characters
643
"/;?:@&=+$," # Reserved characters
644
"%#" # Extra reserved characters
443
_url_safe_characters = set('abcdefghijklmnopqrstuvwxyz'
444
'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
647
449
def unescape_for_display(url, encoding):
648
450
"""Decode what you can for a URL, so that we get a nice looking path.
650
452
This will turn file:// urls into local paths, and try to decode
651
453
any portions of a http:// style url that it can.
653
Any sections of the URL which can't be represented in the encoding or
455
Any sections of the URL which can't be represented in the encoding or
654
456
need to stay as escapes are left alone.
656
458
:param url: A 7-bit ASCII URL
657
459
:param encoding: The final output encoding
659
:return: A unicode string which can be safely encoded into the
461
:return: A unicode string which can be safely encoded into the
660
462
specified encoding.
663
raise ValueError('you cannot specify None for the display encoding')
464
assert encoding is not None, 'you cannot specify None for the display encoding.'
664
465
if url.startswith('file://'):
666
467
path = local_path_from_url(url)
700
501
# Otherwise take the url decoded one
702
503
return u'/'.join(res)
705
def derive_to_location(from_location):
706
"""Derive a TO_LOCATION given a FROM_LOCATION.
708
The normal case is a FROM_LOCATION of http://foo/bar => bar.
709
The Right Thing for some logical destinations may differ though
710
because no / may be present at all. In that case, the result is
711
the full name without the scheme indicator, e.g. lp:foo-bar => foo-bar.
712
This latter case also applies when a Windows drive
713
is used without a path, e.g. c:foo-bar => foo-bar.
714
If no /, path separator or : is found, the from_location is returned.
716
if from_location.find("/") >= 0 or from_location.find(os.sep) >= 0:
717
return os.path.basename(from_location.rstrip("/\\"))
719
sep = from_location.find(":")
721
return from_location[sep+1:]
726
def _is_absolute(url):
727
return (osutils.pathjoin('/foo', url) == url)
730
def rebase_url(url, old_base, new_base):
731
"""Convert a relative path from an old base URL to a new base URL.
733
The result will be a relative path.
734
Absolute paths and full URLs are returned unaltered.
736
scheme, separator = _find_scheme_and_separator(url)
737
if scheme is not None:
739
if _is_absolute(url):
741
old_parsed = urlparse.urlparse(old_base)
742
new_parsed = urlparse.urlparse(new_base)
743
if (old_parsed[:2]) != (new_parsed[:2]):
744
raise errors.InvalidRebaseURLs(old_base, new_base)
745
return determine_relative_path(new_parsed[2],
746
join(old_parsed[2], url))
749
def determine_relative_path(from_path, to_path):
750
"""Determine a relative path from from_path to to_path."""
751
from_segments = osutils.splitpath(from_path)
752
to_segments = osutils.splitpath(to_path)
754
for count, (from_element, to_element) in enumerate(zip(from_segments,
756
if from_element != to_element:
760
unique_from = from_segments[count:]
761
unique_to = to_segments[count:]
762
segments = (['..'] * len(unique_from) + unique_to)
763
if len(segments) == 0:
765
return osutils.pathjoin(*segments)
770
"""Extract the server address, the credentials and the path from the url.
772
user, password, host and path should be quoted if they contain reserved
775
:param url: an quoted url
777
:return: (scheme, user, password, host, port, path) tuple, all fields
780
if isinstance(url, unicode):
781
raise errors.InvalidURL('should be ascii:\n%r' % url)
782
url = url.encode('utf-8')
783
(scheme, netloc, path, params,
784
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
785
user = password = host = port = None
787
user, host = netloc.rsplit('@', 1)
789
user, password = user.split(':', 1)
790
password = urllib.unquote(password)
791
user = urllib.unquote(user)
795
if ':' in host and not (host[0] == '[' and host[-1] == ']'): #there *is* port
796
host, port = host.rsplit(':',1)
800
raise errors.InvalidURL('invalid port number %s in url:\n%s' %
802
if host != "" and host[0] == '[' and host[-1] == ']': #IPv6
805
host = urllib.unquote(host)
806
path = urllib.unquote(path)
808
return (scheme, user, password, host, port, path)
added
removed
bzrlib/urlutils.py
