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 must be equal or'
79
' exceed the platform minimum url length (which is %d)' %
80
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):
125
132
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.
128
m = _url_scheme_re.match(arg)
147
match = _url_scheme_re.match(arg)
131
scheme = m.group('scheme')
132
path = m.group('path').split('/')
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.
134
for chunk in arg.split('/'):
139
# Don't pop off the host portion
142
raise errors.InvalidURLJoin('Cannot go above root',
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:
147
179
if scheme is None:
148
180
return '/'.join(path)
149
181
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)
152
217
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
153
218
def _posix_local_path_from_url(url):
154
219
"""Convert a url like file:///path/to/foo into /path/to/foo"""
164
229
This also handles transforming escaping unicode characters, etc.
166
# importing directly from posixpath allows us to test this
231
# importing directly from posixpath allows us to test this
167
232
# on non-posix platforms
168
233
return 'file://' + escape(_posix_normpath(
169
bzrlib.osutils._posix_abspath(path)))
234
osutils._posix_abspath(path)))
172
237
def _win32_local_path_from_url(url):
173
238
"""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:///')
239
if not url.startswith('file://'):
240
raise errors.InvalidURL(url, 'local urls must start with file:///, '
241
'UNC path urls must start with file://')
176
242
# We strip off all 3 slashes
177
win32_url = url[len('file:///'):]
178
if (win32_url[0] not in ('abcdefghijklmnopqrstuvwxyz'
243
win32_url = url[len('file:'):]
244
# check for UNC path: //HOST/path
245
if not win32_url.startswith('///'):
246
if (win32_url[2] == '/'
247
or win32_url[3] in '|:'):
248
raise errors.InvalidURL(url, 'Win32 UNC path urls'
249
' have form file://HOST/path')
250
return unescape(win32_url)
252
# allow empty paths so we can serve all roots
253
if win32_url == '///':
256
# usual local path with drive letter
257
if (win32_url[3] not in ('abcdefghijklmnopqrstuvwxyz'
179
258
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
180
or win32_url[1] not in '|:'
181
or win32_url[2] != '/'):
259
or win32_url[4] not in '|:'
260
or win32_url[5] != '/'):
182
261
raise errors.InvalidURL(url, 'Win32 file urls start with'
183
262
' file:///x:/, where x is a valid drive letter')
184
return win32_url[0].upper() + u':' + unescape(win32_url[2:])
263
return win32_url[3].upper() + u':' + unescape(win32_url[5:])
187
266
def _win32_local_path_to_url(path):
190
269
This also handles transforming escaping unicode characters, etc.
192
# importing directly from ntpath allows us to test this
271
# importing directly from ntpath allows us to test this
193
272
# on non-win32 platform
194
273
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
195
274
# which actually strips trailing space characters.
196
275
# The worst part is that under linux ntpath.abspath has different
197
276
# 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:])
280
win32_path = osutils._win32_abspath(path)
281
# check for UNC path \\HOST\path
282
if win32_path.startswith('//'):
283
return 'file:' + escape(win32_path)
284
return ('file:///' + str(win32_path[0].upper()) + ':' +
285
escape(win32_path[2:]))
203
288
local_path_to_url = _posix_local_path_to_url
215
300
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
301
_url_hex_escapes_re = re.compile(r'(%[0-9a-fA-F]{2})')
304
def _unescape_safe_chars(matchobj):
305
"""re.sub callback to convert hex-escapes to plain characters (if safe).
307
e.g. '%7E' will be converted to '~'.
309
hex_digits = matchobj.group(0)[1:]
310
char = chr(int(hex_digits, 16))
311
if char in _url_dont_escape_characters:
314
return matchobj.group(0).upper()
218
317
def normalize_url(url):
219
318
"""Make sure that a path string is in fully normalized URL form.
221
This handles URLs which have unicode characters, spaces,
320
This handles URLs which have unicode characters, spaces,
222
321
special characters, etc.
224
323
It has two basic modes of operation, depending on whether the
237
336
m = _url_scheme_re.match(url)
239
338
return local_path_to_url(url)
339
scheme = m.group('scheme')
340
path = m.group('path')
240
341
if not isinstance(url, unicode):
242
343
if c not in _url_safe_characters:
243
344
raise errors.InvalidURL(url, 'URLs can only contain specific'
244
345
' safe characters (not %r)' % c)
346
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
347
return str(scheme + '://' + ''.join(path))
246
349
# We have a unicode (hybrid) url
247
scheme = m.group('scheme')
248
path = list(m.group('path'))
350
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)
352
for i in xrange(len(path_chars)):
353
if path_chars[i] not in _url_safe_characters:
354
chars = path_chars[i].encode('utf-8')
355
path_chars[i] = ''.join(
356
['%%%02X' % ord(c) for c in path_chars[i].encode('utf-8')])
357
path = ''.join(path_chars)
358
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
359
return str(scheme + '://' + path)
257
362
def relative_url(base, other):
430
540
# These are characters that if escaped, should stay that way
431
541
_no_decode_chars = ';/?:@&=+$,#'
432
542
_no_decode_ords = [ord(c) for c in _no_decode_chars]
433
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
543
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
434
544
+ ['%02X' % o for o in _no_decode_ords])
435
545
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
436
546
+ [('%02X' % o, chr(o)) for o in range(256)]))
437
547
#These entries get mapped to themselves
438
548
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
550
# These characters shouldn't be percent-encoded, and it's always safe to
551
# unencode them if they are.
552
_url_dont_escape_characters = set(
553
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
554
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
555
"0123456789" # Numbers
556
"-._~" # Unreserved characters
440
559
# These characters should not be escaped
441
_url_safe_characters = set('abcdefghijklmnopqrstuvwxyz'
442
'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
560
_url_safe_characters = set(
561
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
562
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
563
"0123456789" # Numbers
564
"_.-!~*'()" # Unreserved characters
565
"/;?:@&=+$," # Reserved characters
566
"%#" # Extra reserved characters
447
569
def unescape_for_display(url, encoding):
448
570
"""Decode what you can for a URL, so that we get a nice looking path.
450
572
This will turn file:// urls into local paths, and try to decode
451
573
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
575
Any sections of the URL which can't be represented in the encoding or
454
576
need to stay as escapes are left alone.
456
578
:param url: A 7-bit ASCII URL
457
579
:param encoding: The final output encoding
459
:return: A unicode string which can be safely encoded into the
581
:return: A unicode string which can be safely encoded into the
460
582
specified encoding.
462
assert encoding is not None, 'you cannot specify None for the display encoding.'
585
raise ValueError('you cannot specify None for the display encoding')
463
586
if url.startswith('file://'):
465
588
path = local_path_from_url(url)
499
622
# Otherwise take the url decoded one
501
624
return u'/'.join(res)
627
def derive_to_location(from_location):
628
"""Derive a TO_LOCATION given a FROM_LOCATION.
630
The normal case is a FROM_LOCATION of http://foo/bar => bar.
631
The Right Thing for some logical destinations may differ though
632
because no / may be present at all. In that case, the result is
633
the full name without the scheme indicator, e.g. lp:foo-bar => foo-bar.
634
This latter case also applies when a Windows drive
635
is used without a path, e.g. c:foo-bar => foo-bar.
636
If no /, path separator or : is found, the from_location is returned.
638
if from_location.find("/") >= 0 or from_location.find(os.sep) >= 0:
639
return os.path.basename(from_location.rstrip("/\\"))
641
sep = from_location.find(":")
643
return from_location[sep+1:]
648
def _is_absolute(url):
649
return (osutils.pathjoin('/foo', url) == url)
652
def rebase_url(url, old_base, new_base):
653
"""Convert a relative path from an old base URL to a new base URL.
655
The result will be a relative path.
656
Absolute paths and full URLs are returned unaltered.
658
scheme, separator = _find_scheme_and_separator(url)
659
if scheme is not None:
661
if _is_absolute(url):
663
old_parsed = urlparse.urlparse(old_base)
664
new_parsed = urlparse.urlparse(new_base)
665
if (old_parsed[:2]) != (new_parsed[:2]):
666
raise errors.InvalidRebaseURLs(old_base, new_base)
667
return determine_relative_path(new_parsed[2],
668
join(old_parsed[2], url))
671
def determine_relative_path(from_path, to_path):
672
"""Determine a relative path from from_path to to_path."""
673
from_segments = osutils.splitpath(from_path)
674
to_segments = osutils.splitpath(to_path)
676
for count, (from_element, to_element) in enumerate(zip(from_segments,
678
if from_element != to_element:
682
unique_from = from_segments[count:]
683
unique_to = to_segments[count:]
684
segments = (['..'] * len(unique_from) + unique_to)
685
if len(segments) == 0:
687
return osutils.pathjoin(*segments)
692
"""Extract the server address, the credentials and the path from the url.
694
user, password, host and path should be quoted if they contain reserved
697
:param url: an quoted url
699
:return: (scheme, user, password, host, port, path) tuple, all fields
702
if isinstance(url, unicode):
703
raise errors.InvalidURL('should be ascii:\n%r' % url)
704
url = url.encode('utf-8')
705
(scheme, netloc, path, params,
706
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
707
user = password = host = port = None
709
user, host = netloc.rsplit('@', 1)
711
user, password = user.split(':', 1)
712
password = urllib.unquote(password)
713
user = urllib.unquote(user)
717
if ':' in host and not (host[0] == '[' and host[-1] == ']'): #there *is* port
718
host, port = host.rsplit(':',1)
722
raise errors.InvalidURL('invalid port number %s in url:\n%s' %
724
if host != "" and host[0] == '[' and host[-1] == ']': #IPv6
727
host = urllib.unquote(host)
728
path = urllib.unquote(path)
730
return (scheme, user, password, host, port, path)