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
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):
116
111
join('http://foo', 'bar') => 'http://foo/bar'
117
112
join('http://foo', 'bar', '../baz') => 'http://foo/baz'
121
match = _url_scheme_re.match(base)
114
m = _url_scheme_re.match(base)
124
scheme = match.group('scheme')
125
path = match.group('path').split('/')
126
if path[-1:] == ['']:
127
# Strip off a trailing slash
128
# This helps both when we are at the root, and when
129
# 'base' has an extra slash at the end
117
scheme = m.group('scheme')
118
path = m.group('path').split('/')
132
120
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)
123
m = _url_scheme_re.match(arg)
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.
126
scheme = m.group('scheme')
127
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:
129
for chunk in arg.split('/'):
134
# Don't pop off the host portion
137
raise errors.InvalidURLJoin('Cannot go above root',
179
141
if scheme is None:
180
142
return '/'.join(path)
181
143
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
146
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
218
147
def _posix_local_path_from_url(url):
219
148
"""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://'):]
149
if not url.startswith('file:///'):
150
raise errors.InvalidURL(url, 'local urls must start with file:///')
228
151
# We only strip off 2 slashes
229
return unescape(path)
152
return unescape(url[len('file://'):])
232
155
def _posix_local_path_to_url(path):
235
158
This also handles transforming escaping unicode characters, etc.
237
# importing directly from posixpath allows us to test this
160
# importing directly from posixpath allows us to test this
238
161
# on non-posix platforms
239
return 'file://' + escape(_posix_normpath(
240
osutils._posix_abspath(path)))
162
from posixpath import normpath
163
return 'file://' + escape(normpath(bzrlib.osutils._posix_abspath(path)))
243
166
def _win32_local_path_from_url(url):
244
"""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://')
167
"""Convert a url like file:///C|/path/to/foo into C:/path/to/foo"""
168
if not url.startswith('file:///'):
169
raise errors.InvalidURL(url, 'local urls must start with file:///')
248
170
# 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'
264
'ABCDEFGHIJKLMNOPQRSTUVWXYZ')
265
or win32_url[4] not in '|:'
266
or win32_url[5] != '/'):
267
raise errors.InvalidURL(url, 'Win32 file urls start with'
268
' file:///x:/, where x is a valid drive letter')
269
return win32_url[3].upper() + u':' + unescape(win32_url[5:])
171
win32_url = url[len('file:///'):]
172
if (win32_url[0] not in 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ'
173
or win32_url[1] not in '|:'
174
or win32_url[2] != '/'):
175
raise errors.InvalidURL(url, 'Win32 file urls start with file:///X|/, where X is a valid drive letter')
176
# TODO: jam 20060426, we could .upper() or .lower() the drive letter
177
# for better consistency.
178
return win32_url[0].upper() + u':' + unescape(win32_url[2:])
272
181
def _win32_local_path_to_url(path):
273
"""Convert a local path like ./foo into a URL like file:///C:/path/to/foo
182
"""Convert a local path like ./foo into a URL like file:///C|/path/to/foo
275
184
This also handles transforming escaping unicode characters, etc.
277
# importing directly from ntpath allows us to test this
278
# on non-win32 platform
279
# FIXME: It turns out that on nt, ntpath.abspath uses nt._getfullpathname
280
# which actually strips trailing space characters.
281
# The worst part is that under linux ntpath.abspath has different
282
# 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:]))
186
# importing directly from ntpath allows us to test this
187
# on non-win32 platforms
188
# TODO: jam 20060426 consider moving this import outside of the function
189
win32_path = bzrlib.osutils._nt_normpath(
190
bzrlib.osutils._win32_abspath(path)).replace('\\', '/')
191
return 'file:///' + win32_path[0].upper() + '|' + escape(win32_path[2:])
294
194
local_path_to_url = _posix_local_path_to_url
295
195
local_path_from_url = _posix_local_path_from_url
296
196
MIN_ABS_FILEURL_LENGTH = len('file:///')
297
WIN32_MIN_ABS_FILEURL_LENGTH = len('file:///C:/')
299
198
if sys.platform == 'win32':
300
199
local_path_to_url = _win32_local_path_to_url
301
200
local_path_from_url = _win32_local_path_from_url
303
MIN_ABS_FILEURL_LENGTH = WIN32_MIN_ABS_FILEURL_LENGTH
202
MIN_ABS_FILEURL_LENGTH = len('file:///C|/')
306
205
_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
208
def normalize_url(url):
324
209
"""Make sure that a path string is in fully normalized URL form.
326
This handles URLs which have unicode characters, spaces,
211
This handles URLs which have unicode characters, spaces,
327
212
special characters, etc.
329
214
It has two basic modes of operation, depending on whether the
342
227
m = _url_scheme_re.match(url)
344
229
return local_path_to_url(url)
345
scheme = m.group('scheme')
346
path = m.group('path')
347
230
if not isinstance(url, unicode):
231
# TODO: jam 20060510 We need to test for ascii characters that
232
# shouldn't be allowed in URLs
349
234
if c not in _url_safe_characters:
350
235
raise errors.InvalidURL(url, 'URLs can only contain specific'
351
236
' safe characters (not %r)' % c)
352
path = _url_hex_escapes_re.sub(_unescape_safe_chars, path)
353
return str(scheme + '://' + ''.join(path))
355
238
# We have a unicode (hybrid) url
356
path_chars = list(path)
239
scheme = m.group('scheme')
240
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)
242
for i in xrange(len(path)):
243
if path[i] not in _url_safe_characters:
244
chars = path[i].encode('utf-8')
245
path[i] = ''.join(['%%%02X' % ord(c) for c in path[i].encode('utf-8')])
246
return scheme + '://' + ''.join(path)
368
249
def relative_url(base, other):
496
357
file:///foo/ => file:///foo
497
358
# This is unique on win32 platforms, and is the only URL
498
359
# format which does it differently.
499
file:///c|/ => file:///c:/
360
file:///C|/ => file:///C|/
501
362
if not url.endswith('/'):
504
if sys.platform == 'win32' and url.startswith('file://'):
505
return _win32_strip_local_trailing_slash(url)
365
if sys.platform == 'win32' and url.startswith('file:///'):
366
# This gets handled specially, because the 'top-level'
367
# of a win32 path is actually the drive letter
368
if len(url) > MIN_ABS_FILEURL_LENGTH:
507
372
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
508
373
if scheme_loc is None:
509
374
# This is a relative path, as it has no scheme
546
410
# These are characters that if escaped, should stay that way
547
411
_no_decode_chars = ';/?:@&=+$,#'
548
412
_no_decode_ords = [ord(c) for c in _no_decode_chars]
549
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
413
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
550
414
+ ['%02X' % o for o in _no_decode_ords])
551
415
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
552
416
+ [('%02X' % o, chr(o)) for o in range(256)]))
553
417
#These entries get mapped to themselves
554
418
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
556
# These characters shouldn't be percent-encoded, and it's always safe to
557
# unencode them if they are.
558
_url_dont_escape_characters = set(
559
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
560
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
561
"0123456789" # Numbers
562
"-._~" # Unreserved characters
565
420
# These characters should not be escaped
566
_url_safe_characters = set(
567
"abcdefghijklmnopqrstuvwxyz" # Lowercase alpha
568
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" # Uppercase alpha
569
"0123456789" # Numbers
570
"_.-!~*'()" # Unreserved characters
571
"/;?:@&=+$," # Reserved characters
572
"%#" # Extra reserved characters
421
_url_safe_characters = set('abcdefghijklmnopqrstuvwxyz'
422
'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
575
427
def unescape_for_display(url, encoding):
576
428
"""Decode what you can for a URL, so that we get a nice looking path.
628
478
# Otherwise take the url decoded one
630
480
return u'/'.join(res)
633
def derive_to_location(from_location):
634
"""Derive a TO_LOCATION given a FROM_LOCATION.
636
The normal case is a FROM_LOCATION of http://foo/bar => bar.
637
The Right Thing for some logical destinations may differ though
638
because no / may be present at all. In that case, the result is
639
the full name without the scheme indicator, e.g. lp:foo-bar => foo-bar.
640
This latter case also applies when a Windows drive
641
is used without a path, e.g. c:foo-bar => foo-bar.
642
If no /, path separator or : is found, the from_location is returned.
644
if from_location.find("/") >= 0 or from_location.find(os.sep) >= 0:
645
return os.path.basename(from_location.rstrip("/\\"))
647
sep = from_location.find(":")
649
return from_location[sep+1:]
654
def _is_absolute(url):
655
return (osutils.pathjoin('/foo', url) == url)
658
def rebase_url(url, old_base, new_base):
659
"""Convert a relative path from an old base URL to a new base URL.
661
The result will be a relative path.
662
Absolute paths and full URLs are returned unaltered.
664
scheme, separator = _find_scheme_and_separator(url)
665
if scheme is not None:
667
if _is_absolute(url):
669
old_parsed = urlparse.urlparse(old_base)
670
new_parsed = urlparse.urlparse(new_base)
671
if (old_parsed[:2]) != (new_parsed[:2]):
672
raise errors.InvalidRebaseURLs(old_base, new_base)
673
return determine_relative_path(new_parsed[2],
674
join(old_parsed[2], url))
677
def determine_relative_path(from_path, to_path):
678
"""Determine a relative path from from_path to to_path."""
679
from_segments = osutils.splitpath(from_path)
680
to_segments = osutils.splitpath(to_path)
682
for count, (from_element, to_element) in enumerate(zip(from_segments,
684
if from_element != to_element:
688
unique_from = from_segments[count:]
689
unique_to = to_segments[count:]
690
segments = (['..'] * len(unique_from) + unique_to)
691
if len(segments) == 0:
693
return osutils.pathjoin(*segments)
698
"""Extract the server address, the credentials and the path from the url.
700
user, password, host and path should be quoted if they contain reserved
703
:param url: an quoted url
705
:return: (scheme, user, password, host, port, path) tuple, all fields
708
if isinstance(url, unicode):
709
raise errors.InvalidURL('should be ascii:\n%r' % url)
710
url = url.encode('utf-8')
711
(scheme, netloc, path, params,
712
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
713
user = password = host = port = None
715
user, host = netloc.rsplit('@', 1)
717
user, password = user.split(':', 1)
718
password = urllib.unquote(password)
719
user = urllib.unquote(user)
723
if ':' in host and not (host[0] == '[' and host[-1] == ']'): #there *is* port
724
host, port = host.rsplit(':',1)
728
raise errors.InvalidURL('invalid port number %s in url:\n%s' %
730
if host != "" and host[0] == '[' and host[-1] == ']': #IPv6
733
host = urllib.unquote(host)
734
path = urllib.unquote(path)
736
return (scheme, user, password, host, port, path)
added
removed
bzrlib/urlutils.py
