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."""
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
31
36
def basename(url, exclude_trailing_slash=True):
61
66
relpath = relpath.encode('utf-8')
62
67
# After quoting and encoding, the path should be perfectly
63
68
# safe as a plain ASCII string, str() just enforces this
64
return str(urllib.quote(relpath))
69
return str(urllib.quote(relpath, safe='/~'))
67
72
def file_relpath(base, path):
68
73
"""Compute just the relative sub-portion of a url
70
75
This assumes that both paths are already fully specified file:// URLs.
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)
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))
76
81
base = local_path_from_url(base)
77
82
path = local_path_from_url(path)
78
return escape(bzrlib.osutils.relpath(base, path))
83
return escape(osutils.relpath(base, path))
81
86
def _find_scheme_and_separator(url):
111
121
join('http://foo', 'bar') => 'http://foo/bar'
112
122
join('http://foo', 'bar', '../baz') => 'http://foo/baz'
114
m = _url_scheme_re.match(base)
117
scheme = m.group('scheme')
118
path = m.group('path').split('/')
119
if path[-1:] == ['']:
120
# Strip off a trailing slash
121
# This helps both when we are at the root, and when
122
# 'base' has an extra slash at the end
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:]
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
144
path = joinpath(path, arg)
145
return base[:path_start] + path
148
def joinpath(base, *args):
149
"""Join URL path segments to a URL path segment.
151
This is somewhat like osutils.joinpath, but intended for URLs.
153
XXX: this duplicates some normalisation logic, and also duplicates a lot of
154
path handling logic that already exists in some Transport implementations.
155
We really should try to have exactly one place in the code base responsible
156
for combining paths of URLs.
158
path = base.split('/')
159
if len(path) > 1 and path[-1] == '':
160
#If the path ends in a trailing /, remove it.
163
if arg.startswith('/'):
165
for chunk in arg.split('/'):
170
raise errors.InvalidURLJoin('Cannot go above root',
125
path = base.split('/')
128
m = _url_scheme_re.match(arg)
131
scheme = m.group('scheme')
132
path = m.group('path').split('/')
134
for chunk in arg.split('/'):
139
# Don't pop off the host portion
142
raise errors.InvalidURLJoin('Cannot go above root',
148
178
return '/'.join(path)
149
return scheme + '://' + '/'.join(path)
152
181
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
153
182
def _posix_local_path_from_url(url):
154
183
"""Convert a url like file:///path/to/foo into /path/to/foo"""
155
if not url.startswith('file:///'):
156
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://'):]
157
192
# We only strip off 2 slashes
158
return unescape(url[len('file://'):])
193
return unescape(path)
161
196
def _posix_local_path_to_url(path):
164
199
This also handles transforming escaping unicode characters, etc.
166
# importing directly from posixpath allows us to test this
201
# importing directly from posixpath allows us to test this
167
202
# on non-posix platforms
168
203
return 'file://' + escape(_posix_normpath(
169
bzrlib.osutils._posix_abspath(path)))
204
osutils._posix_abspath(path)))
172
207
def _win32_local_path_from_url(url):
173
208
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
174
if not url.startswith('file:///'):
175
raise errors.InvalidURL(url, 'local urls must start with file:///')
209
if not url.startswith('file://'):
210
raise errors.InvalidURL(url, 'local urls must start with file:///, '
211
'UNC path urls must start with file://')
176
212
# We strip off all 3 slashes
177
win32_url = url[len('file:///'):]
178
if (win32_url[0] not in ('abcdefghijklmnopqrstuvwxyz'
179
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
180
or win32_url[1] not in '|:'
181
or win32_url[2] != '/'):
213
win32_url = url[len('file:'):]
214
# check for UNC path: //HOST/path
215
if not win32_url.startswith('///'):
216
if (win32_url[2] == '/'
217
or win32_url[3] in '|:'):
218
raise errors.InvalidURL(url, 'Win32 UNC path urls'
219
' have form file://HOST/path')
220
return unescape(win32_url)
222
# allow empty paths so we can serve all roots
223
if win32_url == '///':
226
# usual local path with drive letter
227
if (len(win32_url) < 6
228
or win32_url[3] not in ('abcdefghijklmnopqrstuvwxyz'
229
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
230
or win32_url[4] not in '|:'
231
or win32_url[5] != '/'):
182
232
raise errors.InvalidURL(url, 'Win32 file urls start with'
183
233
' file:///x:/, where x is a valid drive letter')
184
return win32_url[0].upper() + u':' + unescape(win32_url[2:])
234
return win32_url[3].upper() + u':' + unescape(win32_url[5:])
187
237
def _win32_local_path_to_url(path):
190
240
This also handles transforming escaping unicode characters, etc.
192
# importing directly from ntpath allows us to test this
242
# importing directly from ntpath allows us to test this
193
243
# on non-win32 platform
194
244
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
195
245
# which actually strips trailing space characters.
196
# The worst part is that under linux ntpath.abspath has different
246
# The worst part is that on linux ntpath.abspath has different
197
247
# semantics, since 'nt' is not an available module.
198
win32_path = bzrlib.osutils._nt_normpath(
199
bzrlib.osutils._win32_abspath(path)).replace('\\', '/')
200
return 'file:///' + win32_path[0].upper() + ':' + escape(win32_path[2:])
251
win32_path = osutils._win32_abspath(path)
252
# check for UNC path \\HOST\path
253
if win32_path.startswith('//'):
254
return 'file:' + escape(win32_path)
255
return ('file:///' + str(win32_path[0].upper()) + ':' +
256
escape(win32_path[2:]))
203
259
local_path_to_url = _posix_local_path_to_url
212
268
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
215
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
271
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,}):(//)?(?P<path>.*)$')
272
_url_hex_escapes_re = re.compile(r'(%[0-9a-fA-F]{2})')
275
def _unescape_safe_chars(matchobj):
276
"""re.sub callback to convert hex-escapes to plain characters (if safe).
278
e.g. '%7E' will be converted to '~'.
280
hex_digits = matchobj.group(0)[1:]
281
char = chr(int(hex_digits, 16))
282
if char in _url_dont_escape_characters:
285
return matchobj.group(0).upper()
218
288
def normalize_url(url):
219
289
"""Make sure that a path string is in fully normalized URL form.
221
This handles URLs which have unicode characters, spaces,
291
This handles URLs which have unicode characters, spaces,
222
292
special characters, etc.
224
294
It has two basic modes of operation, depending on whether the
234
304
:param url: Either a hybrid URL or a local path
235
305
:return: A normalized URL which only includes 7-bit ASCII characters.
237
m = _url_scheme_re.match(url)
307
scheme_end, path_start = _find_scheme_and_separator(url)
308
if scheme_end is None:
239
309
return local_path_to_url(url)
310
prefix = url[:path_start]
311
path = url[path_start:]
240
312
if not isinstance(url, unicode):
242
314
if c not in _url_safe_characters:
243
315
raise errors.InvalidURL(url, 'URLs can only contain specific'
244
316
' safe characters (not %r)' % c)
317
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
318
return str(prefix + ''.join(path))
246
320
# We have a unicode (hybrid) url
247
scheme = m.group('scheme')
248
path = list(m.group('path'))
321
path_chars = list(path)
250
for i in xrange(len(path)):
251
if path[i] not in _url_safe_characters:
252
chars = path[i].encode('utf-8')
253
path[i] = ''.join(['%%%02X' % ord(c) for c in path[i].encode('utf-8')])
254
return scheme + '://' + ''.join(path)
323
for i in xrange(len(path_chars)):
324
if path_chars[i] not in _url_safe_characters:
325
chars = path_chars[i].encode('utf-8')
326
path_chars[i] = ''.join(
327
['%%%02X' % ord(c) for c in path_chars[i].encode('utf-8')])
328
path = ''.join(path_chars)
329
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
330
return str(prefix + path)
257
333
def relative_url(base, other):
353
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())])
356
509
def _win32_strip_local_trailing_slash(url):
357
510
"""Strip slashes after the drive letter"""
358
511
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
430
583
# These are characters that if escaped, should stay that way
431
584
_no_decode_chars = ';/?:@&=+$,#'
432
585
_no_decode_ords = [ord(c) for c in _no_decode_chars]
433
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
586
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
434
587
+ ['%02X' % o for o in _no_decode_ords])
435
588
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
436
589
+ [('%02X' % o, chr(o)) for o in range(256)]))
437
590
#These entries get mapped to themselves
438
591
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
593
# These characters shouldn't be percent-encoded, and it's always safe to
594
# unencode them if they are.
595
_url_dont_escape_characters = set(
596
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
597
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
598
"0123456789" # Numbers
599
"-._~" # Unreserved characters
440
602
# These characters should not be escaped
441
_url_safe_characters = set('abcdefghijklmnopqrstuvwxyz'
442
'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
603
_url_safe_characters = set(
604
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
605
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
606
"0123456789" # Numbers
607
"_.-!~*'()" # Unreserved characters
608
"/;?:@&=+$," # Reserved characters
609
"%#" # Extra reserved characters
447
612
def unescape_for_display(url, encoding):
448
613
"""Decode what you can for a URL, so that we get a nice looking path.
450
615
This will turn file:// urls into local paths, and try to decode
451
616
any portions of a http:// style url that it can.
453
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
454
619
need to stay as escapes are left alone.
456
621
:param url: A 7-bit ASCII URL
457
622
:param encoding: The final output encoding
459
:return: A unicode string which can be safely encoded into the
624
:return: A unicode string which can be safely encoded into the
460
625
specified encoding.
462
assert encoding is not None, 'you cannot specify None for the display encoding.'
628
raise ValueError('you cannot specify None for the display encoding')
463
629
if url.startswith('file://'):
465
631
path = local_path_from_url(url)
499
665
# Otherwise take the url decoded one
501
667
return u'/'.join(res)
670
def derive_to_location(from_location):
671
"""Derive a TO_LOCATION given a FROM_LOCATION.
673
The normal case is a FROM_LOCATION of http://foo/bar => bar.
674
The Right Thing for some logical destinations may differ though
675
because no / may be present at all. In that case, the result is
676
the full name without the scheme indicator, e.g. lp:foo-bar => foo-bar.
677
This latter case also applies when a Windows drive
678
is used without a path, e.g. c:foo-bar => foo-bar.
679
If no /, path separator or : is found, the from_location is returned.
681
if from_location.find("/") >= 0 or from_location.find(os.sep) >= 0:
682
return os.path.basename(from_location.rstrip("/\\"))
684
sep = from_location.find(":")
686
return from_location[sep+1:]
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)