15
13
# You should have received a copy of the GNU General Public License
16
14
# along with this program; if not, write to the Free Software
17
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19
17
"""A collection of function for handling URL operations."""
67
66
relpath = relpath.encode('utf-8')
68
67
# After quoting and encoding, the path should be perfectly
69
68
# safe as a plain ASCII string, str() just enforces this
70
return str(urllib.quote(relpath))
69
return str(urllib.quote(relpath, safe='/~'))
73
72
def file_relpath(base, path):
74
73
"""Compute just the relative sub-portion of a url
76
75
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)
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))
82
81
base = local_path_from_url(base)
83
82
path = local_path_from_url(path)
84
83
return escape(osutils.relpath(base, path))
122
match = _url_scheme_re.match(base)
125
scheme = match.group('scheme')
126
path = match.group('path').split('/')
127
if path[-1:] == ['']:
128
# Strip off a trailing slash
129
# This helps both when we are at the root, and when
130
# 'base' has an extra slash at the end
133
path = base.split('/')
135
if scheme is not None and len(path) >= 1:
137
# the path should be represented as an abs path.
138
# we know this must be absolute because of the presence of a URL scheme.
140
path = [''] + path[1:]
142
# create an empty host, but dont alter the path - this might be a
143
# relative url fragment.
126
scheme_end, path_start = _find_scheme_and_separator(base)
127
if scheme_end is None and path_start is None:
129
elif path_start is None:
130
path_start = len(base)
131
path = base[path_start:]
148
match = _url_scheme_re.match(arg)
151
scheme = match.group('scheme')
152
# this skips .. normalisation, making http://host/../../..
154
path = match.group('path').split('/')
155
# set the host and path according to new absolute URL, discarding
156
# any previous values.
157
# XXX: duplicates mess from earlier in this function. This URL
158
# manipulation code needs some cleaning up.
159
if scheme is not None and len(path) >= 1:
162
# url scheme implies absolute path.
165
# no url scheme we take the path as is.
133
arg_scheme_end, arg_path_start = _find_scheme_and_separator(arg)
134
if arg_scheme_end is None and arg_path_start is None:
136
elif arg_path_start is None:
137
arg_path_start = len(arg)
138
if arg_scheme_end is not None:
140
path = arg[arg_path_start:]
141
scheme_end = arg_scheme_end
142
path_start = arg_path_start
168
path = '/'.join(path)
169
144
path = joinpath(path, arg)
170
path = path.split('/')
171
if remove_root and path[0:1] == ['']:
174
# Remove the leading slash from the path, so long as it isn't also the
175
# trailing slash, which we want to keep if present.
176
if path and path[0] == '' and len(path) > 1:
181
return '/'.join(path)
182
return scheme + '://' + '/'.join(path)
145
return base[:path_start] + path
185
148
def joinpath(base, *args):
186
149
"""Join URL path segments to a URL path segment.
188
151
This is somewhat like osutils.joinpath, but intended for URLs.
190
153
XXX: this duplicates some normalisation logic, and also duplicates a lot of
218
181
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
219
182
def _posix_local_path_from_url(url):
220
183
"""Convert a url like file:///path/to/foo into /path/to/foo"""
221
if not url.startswith('file:///'):
222
raise errors.InvalidURL(url, 'local urls must start with file:///')
184
file_localhost_prefix = 'file://localhost/'
185
if url.startswith(file_localhost_prefix):
186
path = url[len(file_localhost_prefix) - 1:]
187
elif not url.startswith('file:///'):
188
raise errors.InvalidURL(
189
url, 'local urls must start with file:/// or file://localhost/')
191
path = url[len('file://'):]
223
192
# We only strip off 2 slashes
224
return unescape(url[len('file://'):])
193
return unescape(path)
227
196
def _posix_local_path_to_url(path):
249
218
raise errors.InvalidURL(url, 'Win32 UNC path urls'
250
219
' have form file://HOST/path')
251
220
return unescape(win32_url)
222
# allow empty paths so we can serve all roots
223
if win32_url == '///':
252
226
# usual local path with drive letter
253
if (win32_url[3] not in ('abcdefghijklmnopqrstuvwxyz'
254
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
227
if (len(win32_url) < 6
228
or win32_url[3] not in ('abcdefghijklmnopqrstuvwxyz'
229
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
255
230
or win32_url[4] not in '|:'
256
231
or win32_url[5] != '/'):
257
232
raise errors.InvalidURL(url, 'Win32 file urls start with'
265
240
This also handles transforming escaping unicode characters, etc.
267
# importing directly from ntpath allows us to test this
242
# importing directly from ntpath allows us to test this
268
243
# on non-win32 platform
269
244
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
270
245
# which actually strips trailing space characters.
271
# The worst part is that under linux ntpath.abspath has different
246
# The worst part is that on linux ntpath.abspath has different
272
247
# semantics, since 'nt' is not an available module.
273
251
win32_path = osutils._win32_abspath(path)
274
252
# check for UNC path \\HOST\path
275
253
if win32_path.startswith('//'):
276
254
return 'file:' + escape(win32_path)
277
return 'file:///' + win32_path[0].upper() + ':' + escape(win32_path[2:])
255
return ('file:///' + str(win32_path[0].upper()) + ':' +
256
escape(win32_path[2:]))
280
259
local_path_to_url = _posix_local_path_to_url
289
268
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
292
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
271
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,}):(//)?(?P<path>.*)$')
293
272
_url_hex_escapes_re = re.compile(r'(%[0-9a-fA-F]{2})')
296
275
def _unescape_safe_chars(matchobj):
297
276
"""re.sub callback to convert hex-escapes to plain characters (if safe).
299
278
e.g. '%7E' will be converted to '~'.
301
280
hex_digits = matchobj.group(0)[1:]
325
304
:param url: Either a hybrid URL or a local path
326
305
:return: A normalized URL which only includes 7-bit ASCII characters.
328
m = _url_scheme_re.match(url)
307
scheme_end, path_start = _find_scheme_and_separator(url)
308
if scheme_end is None:
330
309
return local_path_to_url(url)
331
scheme = m.group('scheme')
332
path = m.group('path')
310
prefix = url[:path_start]
311
path = url[path_start:]
333
312
if not isinstance(url, unicode):
335
314
if c not in _url_safe_characters:
336
315
raise errors.InvalidURL(url, 'URLs can only contain specific'
337
316
' safe characters (not %r)' % c)
338
317
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
339
return str(scheme + '://' + ''.join(path))
318
return str(prefix + ''.join(path))
341
320
# We have a unicode (hybrid) url
342
321
path_chars = list(path)
455
434
return url_base + head, tail
437
def split_segment_parameters_raw(url):
438
"""Split the subsegment of the last segment of a URL.
440
:param url: A relative or absolute URL
441
:return: (url, subsegments)
443
(parent_url, child_dir) = split(url)
444
subsegments = child_dir.split(",")
445
if len(subsegments) == 1:
447
return (join(parent_url, subsegments[0]), subsegments[1:])
450
def split_segment_parameters(url):
451
"""Split the segment parameters of the last segment of a URL.
453
:param url: A relative or absolute URL
454
:return: (url, segment_parameters)
456
(base_url, subsegments) = split_segment_parameters_raw(url)
458
for subsegment in subsegments:
459
(key, value) = subsegment.split("=", 1)
460
parameters[key] = value
461
return (base_url, parameters)
464
def join_segment_parameters_raw(base, *subsegments):
465
"""Create a new URL by adding subsegments to an existing one.
467
This adds the specified subsegments to the last path in the specified
468
base URL. The subsegments should be bytestrings.
470
:note: You probably want to use join_segment_parameters instead.
474
for subsegment in subsegments:
475
if type(subsegment) is not str:
476
raise TypeError("Subsegment %r is not a bytestring" % subsegment)
477
if "," in subsegment:
478
raise errors.InvalidURLJoin(", exists in subsegments",
480
return ",".join((base,) + subsegments)
483
def join_segment_parameters(url, parameters):
484
"""Create a new URL by adding segment parameters to an existing one.
486
The parameters of the last segment in the URL will be updated; if a
487
parameter with the same key already exists it will be overwritten.
489
:param url: A URL, as string
490
:param parameters: Dictionary of parameters, keys and values as bytestrings
492
(base, existing_parameters) = split_segment_parameters(url)
494
new_parameters.update(existing_parameters)
495
for key, value in parameters.iteritems():
496
if type(key) is not str:
497
raise TypeError("parameter key %r is not a bytestring" % key)
498
if type(value) is not str:
499
raise TypeError("parameter value %r for %s is not a bytestring" %
502
raise errors.InvalidURLJoin("= exists in parameter key", url,
504
new_parameters[key] = value
505
return join_segment_parameters_raw(base,
506
*["%s=%s" % item for item in sorted(new_parameters.items())])
458
509
def _win32_strip_local_trailing_slash(url):
459
510
"""Strip slashes after the drive letter"""
460
511
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
532
583
# These are characters that if escaped, should stay that way
533
584
_no_decode_chars = ';/?:@&=+$,#'
534
585
_no_decode_ords = [ord(c) for c in _no_decode_chars]
535
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
586
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
536
587
+ ['%02X' % o for o in _no_decode_ords])
537
588
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
538
589
+ [('%02X' % o, chr(o)) for o in range(256)]))
564
615
This will turn file:// urls into local paths, and try to decode
565
616
any portions of a http:// style url that it can.
567
Any sections of the URL which can't be represented in the encoding or
618
Any sections of the URL which can't be represented in the encoding or
568
619
need to stay as escapes are left alone.
570
621
:param url: A 7-bit ASCII URL
571
622
:param encoding: The final output encoding
573
:return: A unicode string which can be safely encoded into the
624
:return: A unicode string which can be safely encoded into the
574
625
specified encoding.
576
assert encoding is not None, 'you cannot specify None for the display encoding.'
628
raise ValueError('you cannot specify None for the display encoding')
577
629
if url.startswith('file://'):
579
631
path = local_path_from_url(url)
634
686
return from_location[sep+1:]
636
688
return from_location
691
def _is_absolute(url):
692
return (osutils.pathjoin('/foo', url) == url)
695
def rebase_url(url, old_base, new_base):
696
"""Convert a relative path from an old base URL to a new base URL.
698
The result will be a relative path.
699
Absolute paths and full URLs are returned unaltered.
701
scheme, separator = _find_scheme_and_separator(url)
702
if scheme is not None:
704
if _is_absolute(url):
706
old_parsed = urlparse.urlparse(old_base)
707
new_parsed = urlparse.urlparse(new_base)
708
if (old_parsed[:2]) != (new_parsed[:2]):
709
raise errors.InvalidRebaseURLs(old_base, new_base)
710
return determine_relative_path(new_parsed[2],
711
join(old_parsed[2], url))
714
def determine_relative_path(from_path, to_path):
715
"""Determine a relative path from from_path to to_path."""
716
from_segments = osutils.splitpath(from_path)
717
to_segments = osutils.splitpath(to_path)
719
for count, (from_element, to_element) in enumerate(zip(from_segments,
721
if from_element != to_element:
725
unique_from = from_segments[count:]
726
unique_to = to_segments[count:]
727
segments = (['..'] * len(unique_from) + unique_to)
728
if len(segments) == 0:
730
return osutils.pathjoin(*segments)
735
"""Extract the server address, the credentials and the path from the url.
737
user, password, host and path should be quoted if they contain reserved
740
:param url: an quoted url
742
:return: (scheme, user, password, host, port, path) tuple, all fields
745
if isinstance(url, unicode):
746
raise errors.InvalidURL('should be ascii:\n%r' % url)
747
url = url.encode('utf-8')
748
(scheme, netloc, path, params,
749
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
750
user = password = host = port = None
752
user, host = netloc.rsplit('@', 1)
754
user, password = user.split(':', 1)
755
password = urllib.unquote(password)
756
user = urllib.unquote(user)
760
if ':' in host and not (host[0] == '[' and host[-1] == ']'): #there *is* port
761
host, port = host.rsplit(':',1)
765
raise errors.InvalidURL('invalid port number %s in url:\n%s' %
767
if host != "" and host[0] == '[' and host[-1] == ']': #IPv6
770
host = urllib.unquote(host)
771
path = urllib.unquote(path)
773
return (scheme, user, password, host, port, path)