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):
121
111
join('http://foo', 'bar') => 'http://foo/bar'
122
112
join('http://foo', 'bar', '../baz') => 'http://foo/baz'
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:]
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
125
path = base.split('/')
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
128
m = _url_scheme_re.match(arg)
131
scheme = m.group('scheme')
132
path = m.group('path').split('/')
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',
134
for chunk in arg.split('/'):
139
# Don't pop off the host portion
142
raise errors.InvalidURLJoin('Cannot go above root',
178
148
return '/'.join(path)
149
return scheme + '://' + '/'.join(path)
181
152
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
182
153
def _posix_local_path_from_url(url):
183
154
"""Convert a url like file:///path/to/foo into /path/to/foo"""
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://'):]
155
if not url.startswith('file:///'):
156
raise errors.InvalidURL(url, 'local urls must start with file:///')
192
157
# We only strip off 2 slashes
193
return unescape(path)
158
return unescape(url[len('file://'):])
196
161
def _posix_local_path_to_url(path):
199
164
This also handles transforming escaping unicode characters, etc.
201
# importing directly from posixpath allows us to test this
166
# importing directly from posixpath allows us to test this
202
167
# on non-posix platforms
203
168
return 'file://' + escape(_posix_normpath(
204
osutils._posix_abspath(path)))
169
bzrlib.osutils._posix_abspath(path)))
207
172
def _win32_local_path_from_url(url):
208
173
"""Convert a url like file:///C:/path/to/foo into C:/path/to/foo"""
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://')
174
if not url.startswith('file:///'):
175
raise errors.InvalidURL(url, 'local urls must start with file:///')
212
176
# We strip off all 3 slashes
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 (win32_url[3] not in ('abcdefghijklmnopqrstuvwxyz'
177
win32_url = url[len('file:///'):]
178
if (win32_url[0] not in ('abcdefghijklmnopqrstuvwxyz'
228
179
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
229
or win32_url[4] not in '|:'
230
or win32_url[5] != '/'):
180
or win32_url[1] not in '|:'
181
or win32_url[2] != '/'):
231
182
raise errors.InvalidURL(url, 'Win32 file urls start with'
232
183
' file:///x:/, where x is a valid drive letter')
233
return win32_url[3].upper() + u':' + unescape(win32_url[5:])
184
return win32_url[0].upper() + u':' + unescape(win32_url[2:])
236
187
def _win32_local_path_to_url(path):
239
190
This also handles transforming escaping unicode characters, etc.
241
# importing directly from ntpath allows us to test this
192
# importing directly from ntpath allows us to test this
242
193
# on non-win32 platform
243
194
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
244
195
# which actually strips trailing space characters.
245
# The worst part is that on linux ntpath.abspath has different
196
# The worst part is that under linux ntpath.abspath has different
246
197
# semantics, since 'nt' is not an available module.
250
win32_path = osutils._win32_abspath(path)
251
# check for UNC path \\HOST\path
252
if win32_path.startswith('//'):
253
return 'file:' + escape(win32_path)
254
return ('file:///' + str(win32_path[0].upper()) + ':' +
255
escape(win32_path[2:]))
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:])
258
203
local_path_to_url = _posix_local_path_to_url
267
212
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
270
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,}):(//)?(?P<path>.*)$')
271
_url_hex_escapes_re = re.compile(r'(%[0-9a-fA-F]{2})')
274
def _unescape_safe_chars(matchobj):
275
"""re.sub callback to convert hex-escapes to plain characters (if safe).
277
e.g. '%7E' will be converted to '~'.
279
hex_digits = matchobj.group(0)[1:]
280
char = chr(int(hex_digits, 16))
281
if char in _url_dont_escape_characters:
284
return matchobj.group(0).upper()
215
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
287
218
def normalize_url(url):
288
219
"""Make sure that a path string is in fully normalized URL form.
290
This handles URLs which have unicode characters, spaces,
221
This handles URLs which have unicode characters, spaces,
291
222
special characters, etc.
293
224
It has two basic modes of operation, depending on whether the
303
234
:param url: Either a hybrid URL or a local path
304
235
:return: A normalized URL which only includes 7-bit ASCII characters.
306
scheme_end, path_start = _find_scheme_and_separator(url)
307
if scheme_end is None:
237
m = _url_scheme_re.match(url)
308
239
return local_path_to_url(url)
309
prefix = url[:path_start]
310
path = url[path_start:]
311
240
if not isinstance(url, unicode):
313
242
if c not in _url_safe_characters:
314
243
raise errors.InvalidURL(url, 'URLs can only contain specific'
315
244
' safe characters (not %r)' % c)
316
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
317
return str(prefix + ''.join(path))
319
246
# We have a unicode (hybrid) url
320
path_chars = list(path)
247
scheme = m.group('scheme')
248
path = list(m.group('path'))
322
for i in xrange(len(path_chars)):
323
if path_chars[i] not in _url_safe_characters:
324
chars = path_chars[i].encode('utf-8')
325
path_chars[i] = ''.join(
326
['%%%02X' % ord(c) for c in path_chars[i].encode('utf-8')])
327
path = ''.join(path_chars)
328
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
329
return str(prefix + 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)
332
257
def relative_url(base, other):
433
353
return url_base + head, tail
436
def split_segment_parameters_raw(url):
437
"""Split the subsegment of the last segment of a URL.
439
:param url: A relative or absolute URL
440
:return: (url, subsegments)
442
(parent_url, child_dir) = split(url)
443
subsegments = child_dir.split(",")
444
if len(subsegments) == 1:
446
return (join(parent_url, subsegments[0]), subsegments[1:])
449
def split_segment_parameters(url):
450
"""Split the segment parameters of the last segment of a URL.
452
:param url: A relative or absolute URL
453
:return: (url, segment_parameters)
455
(base_url, subsegments) = split_segment_parameters_raw(url)
457
for subsegment in subsegments:
458
(key, value) = subsegment.split("=", 1)
459
parameters[key] = value
460
return (base_url, parameters)
463
def join_segment_parameters_raw(base, *subsegments):
464
"""Create a new URL by adding subsegments to an existing one.
466
This adds the specified subsegments to the last path in the specified
467
base URL. The subsegments should be bytestrings.
469
:note: You probably want to use join_segment_parameters instead.
473
for subsegment in subsegments:
474
if type(subsegment) is not str:
475
raise TypeError("Subsegment %r is not a bytestring" % subsegment)
476
if "," in subsegment:
477
raise errors.InvalidURLJoin(", exists in subsegments",
479
return ",".join((base,) + subsegments)
482
def join_segment_parameters(url, parameters):
483
"""Create a new URL by adding segment parameters to an existing one.
485
The parameters of the last segment in the URL will be updated; if a
486
parameter with the same key already exists it will be overwritten.
488
:param url: A URL, as string
489
:param parameters: Dictionary of parameters, keys and values as bytestrings
491
(base, existing_parameters) = split_segment_parameters(url)
493
new_parameters.update(existing_parameters)
494
for key, value in parameters.iteritems():
495
if type(key) is not str:
496
raise TypeError("parameter key %r is not a bytestring" % key)
497
if type(value) is not str:
498
raise TypeError("parameter value %r for %s is not a bytestring" %
501
raise errors.InvalidURLJoin("= exists in parameter key", url,
503
new_parameters[key] = value
504
return join_segment_parameters_raw(base,
505
*["%s=%s" % item for item in sorted(new_parameters.items())])
508
356
def _win32_strip_local_trailing_slash(url):
509
357
"""Strip slashes after the drive letter"""
510
358
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
582
430
# These are characters that if escaped, should stay that way
583
431
_no_decode_chars = ';/?:@&=+$,#'
584
432
_no_decode_ords = [ord(c) for c in _no_decode_chars]
585
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
433
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
586
434
+ ['%02X' % o for o in _no_decode_ords])
587
435
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
588
436
+ [('%02X' % o, chr(o)) for o in range(256)]))
589
437
#These entries get mapped to themselves
590
438
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
592
# These characters shouldn't be percent-encoded, and it's always safe to
593
# unencode them if they are.
594
_url_dont_escape_characters = set(
595
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
596
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
597
"0123456789" # Numbers
598
"-._~" # Unreserved characters
601
440
# These characters should not be escaped
602
_url_safe_characters = set(
603
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
604
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
605
"0123456789" # Numbers
606
"_.-!~*'()" # Unreserved characters
607
"/;?:@&=+$," # Reserved characters
608
"%#" # Extra reserved characters
441
_url_safe_characters = set('abcdefghijklmnopqrstuvwxyz'
442
'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
611
447
def unescape_for_display(url, encoding):
612
448
"""Decode what you can for a URL, so that we get a nice looking path.
614
450
This will turn file:// urls into local paths, and try to decode
615
451
any portions of a http:// style url that it can.
617
Any sections of the URL which can't be represented in the encoding or
453
Any sections of the URL which can't be represented in the encoding or
618
454
need to stay as escapes are left alone.
620
456
:param url: A 7-bit ASCII URL
621
457
:param encoding: The final output encoding
623
:return: A unicode string which can be safely encoded into the
459
:return: A unicode string which can be safely encoded into the
624
460
specified encoding.
627
raise ValueError('you cannot specify None for the display encoding')
462
assert encoding is not None, 'you cannot specify None for the display encoding.'
628
463
if url.startswith('file://'):
630
465
path = local_path_from_url(url)
664
499
# Otherwise take the url decoded one
666
501
return u'/'.join(res)
669
def derive_to_location(from_location):
670
"""Derive a TO_LOCATION given a FROM_LOCATION.
672
The normal case is a FROM_LOCATION of http://foo/bar => bar.
673
The Right Thing for some logical destinations may differ though
674
because no / may be present at all. In that case, the result is
675
the full name without the scheme indicator, e.g. lp:foo-bar => foo-bar.
676
This latter case also applies when a Windows drive
677
is used without a path, e.g. c:foo-bar => foo-bar.
678
If no /, path separator or : is found, the from_location is returned.
680
if from_location.find("/") >= 0 or from_location.find(os.sep) >= 0:
681
return os.path.basename(from_location.rstrip("/\\"))
683
sep = from_location.find(":")
685
return from_location[sep+1:]
690
def _is_absolute(url):
691
return (osutils.pathjoin('/foo', url) == url)
694
def rebase_url(url, old_base, new_base):
695
"""Convert a relative path from an old base URL to a new base URL.
697
The result will be a relative path.
698
Absolute paths and full URLs are returned unaltered.
700
scheme, separator = _find_scheme_and_separator(url)
701
if scheme is not None:
703
if _is_absolute(url):
705
old_parsed = urlparse.urlparse(old_base)
706
new_parsed = urlparse.urlparse(new_base)
707
if (old_parsed[:2]) != (new_parsed[:2]):
708
raise errors.InvalidRebaseURLs(old_base, new_base)
709
return determine_relative_path(new_parsed[2],
710
join(old_parsed[2], url))
713
def determine_relative_path(from_path, to_path):
714
"""Determine a relative path from from_path to to_path."""
715
from_segments = osutils.splitpath(from_path)
716
to_segments = osutils.splitpath(to_path)
718
for count, (from_element, to_element) in enumerate(zip(from_segments,
720
if from_element != to_element:
724
unique_from = from_segments[count:]
725
unique_to = to_segments[count:]
726
segments = (['..'] * len(unique_from) + unique_to)
727
if len(segments) == 0:
729
return osutils.pathjoin(*segments)
734
"""Extract the server address, the credentials and the path from the url.
736
user, password, host and path should be quoted if they contain reserved
739
:param url: an quoted url
741
:return: (scheme, user, password, host, port, path) tuple, all fields
744
if isinstance(url, unicode):
745
raise errors.InvalidURL('should be ascii:\n%r' % url)
746
url = url.encode('utf-8')
747
(scheme, netloc, path, params,
748
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
749
user = password = host = port = None
751
user, host = netloc.rsplit('@', 1)
753
user, password = user.split(':', 1)
754
password = urllib.unquote(password)
755
user = urllib.unquote(user)
759
if ':' in host and not (host[0] == '[' and host[-1] == ']'): #there *is* port
760
host, port = host.rsplit(':',1)
764
raise errors.InvalidURL('invalid port number %s in url:\n%s' %
766
if host != "" and host[0] == '[' and host[-1] == ']': #IPv6
769
host = urllib.unquote(host)
770
path = urllib.unquote(path)
772
return (scheme, user, password, host, port, path)