1
# Bazaar-NG -- distributed version control
3
# Copyright (C) 2006 by Canonical Ltd
5
# This program is free software; you can redistribute it and/or modify
6
# it under the terms of the GNU General Public License as published by
7
# the Free Software Foundation; either version 2 of the License, or
8
# (at your option) any later version.
10
# This program is distributed in the hope that it will be useful,
11
# but WITHOUT ANY WARRANTY; without even the implied warranty of
12
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13
# GNU General Public License for more details.
15
# You should have received a copy of the GNU General Public License
16
# along with this program; if not, write to the Free Software
17
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
19
"""A collection of function for handling URL operations."""
22
from posixpath import split as _posix_split
27
import bzrlib.errors as errors
31
def basename(url, exclude_trailing_slash=True):
32
"""Return the last component of a URL.
34
:param url: The URL in question
35
:param exclude_trailing_slash: If the url looks like "path/to/foo/"
36
ignore the final slash and return 'foo' rather than ''
37
:return: Just the final component of the URL. This can return ''
38
if you don't exclude_trailing_slash, or if you are at the
41
return split(url, exclude_trailing_slash=exclude_trailing_slash)[1]
44
def dirname(url, exclude_trailing_slash=True):
45
"""Return the parent directory of the given path.
47
:param url: Relative or absolute URL
48
:param exclude_trailing_slash: Remove a final slash
49
(treat http://host/foo/ as http://host/foo, but
50
http://host/ stays http://host/)
51
:return: Everything in the URL except the last path chunk
53
# TODO: jam 20060502 This was named dirname to be consistent
54
# with the os functions, but maybe "parent" would be better
55
return split(url, exclude_trailing_slash=exclude_trailing_slash)[0]
59
"""Escape relpath to be a valid url."""
60
if isinstance(relpath, unicode):
61
relpath = relpath.encode('utf-8')
62
# After quoting and encoding, the path should be perfectly
63
# safe as a plain ASCII string, str() just enforces this
64
return str(urllib.quote(relpath))
67
def file_relpath(base, path):
68
"""Compute just the relative sub-portion of a url
70
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)
76
base = local_path_from_url(base)
77
path = local_path_from_url(path)
78
return escape(bzrlib.osutils.relpath(base, path))
81
def _find_scheme_and_separator(url):
82
"""Find the scheme separator (://) and the first path separator
84
This is just a helper functions for other path utilities.
85
It could probably be replaced by urlparse
87
m = _url_scheme_re.match(url)
91
scheme = m.group('scheme')
92
path = m.group('path')
94
# Find the path separating slash
95
# (first slash after the ://)
96
first_path_slash = path.find('/')
97
if first_path_slash == -1:
98
return scheme_loc, None
99
return scheme_loc, first_path_slash+len(scheme)+3
102
# jam 20060502 Sorted to 'l' because the final target is 'local_path_from_url'
103
def _posix_local_path_from_url(url):
104
"""Convert a url like file:///path/to/foo into /path/to/foo"""
105
if not url.startswith('file:///'):
106
raise errors.InvalidURL(url, 'local urls must start with file:///')
107
# We only strip off 2 slashes
108
return unescape(url[len('file://'):])
111
def _posix_local_path_to_url(path):
112
"""Convert a local path like ./foo into a URL like file:///path/to/foo
114
This also handles transforming escaping unicode characters, etc.
116
# importing directly from posixpath allows us to test this
117
# on non-posix platforms
118
from posixpath import normpath
119
return 'file://' + escape(normpath(bzrlib.osutils._posix_abspath(path)))
122
def _win32_local_path_from_url(url):
123
"""Convert a url like file:///C|/path/to/foo into C:/path/to/foo"""
124
if not url.startswith('file:///'):
125
raise errors.InvalidURL(url, 'local urls must start with file:///')
126
# We strip off all 3 slashes
127
win32_url = url[len('file:///'):]
128
if (win32_url[0] not in 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ'
129
or win32_url[1] not in '|:'
130
or win32_url[2] != '/'):
131
raise errors.InvalidURL(url, 'Win32 file urls start with file:///X|/, where X is a valid drive letter')
132
# TODO: jam 20060426, we could .upper() or .lower() the drive letter
133
# for better consistency.
134
return win32_url[0].upper() + u':' + unescape(win32_url[2:])
137
def _win32_local_path_to_url(path):
138
"""Convert a local path like ./foo into a URL like file:///C|/path/to/foo
140
This also handles transforming escaping unicode characters, etc.
142
# importing directly from ntpath allows us to test this
143
# on non-win32 platforms
144
# TODO: jam 20060426 consider moving this import outside of the function
145
win32_path = bzrlib.osutils._nt_normpath(
146
bzrlib.osutils._win32_abspath(path)).replace('\\', '/')
147
return 'file:///' + win32_path[0].upper() + '|' + escape(win32_path[2:])
150
local_path_to_url = _posix_local_path_to_url
151
local_path_from_url = _posix_local_path_from_url
152
MIN_ABS_FILEURL_LENGTH = len('file:///')
154
if sys.platform == 'win32':
155
local_path_to_url = _win32_local_path_to_url
156
local_path_from_url = _win32_local_path_from_url
158
MIN_ABS_FILEURL_LENGTH = len('file:///C|/')
161
_url_scheme_re = re.compile(r'^(?P<scheme>[^:/]{2,})://(?P<path>.*)$')
164
def normalize_url(url):
165
"""Make sure that a path string is in fully normalized URL form.
167
This handles URLs which have unicode characters, spaces,
168
special characters, etc.
170
It has two basic modes of operation, depending on whether the
171
supplied string starts with a url specifier (scheme://) or not.
172
If it does not have a specifier it is considered a local path,
173
and will be converted into a file:/// url. Non-ascii characters
174
will be encoded using utf-8.
175
If it does have a url specifier, it will be treated as a "hybrid"
176
URL. Basically, a URL that should have URL special characters already
177
escaped (like +?&# etc), but may have unicode characters, etc
178
which would not be valid in a real URL.
180
:param url: Either a hybrid URL or a local path
181
:return: A normalized URL which only includes 7-bit ASCII characters.
183
m = _url_scheme_re.match(url)
185
return local_path_to_url(url)
186
if not isinstance(url, unicode):
187
# TODO: jam 20060510 We need to test for ascii characters that
188
# shouldn't be allowed in URLs
190
if c not in _url_safe_characters:
191
raise errors.InvalidURL(url, 'URLs can only contain specific safe characters')
193
# We have a unicode (hybrid) url
194
scheme = m.group('scheme')
195
path = list(m.group('path'))
197
for i in xrange(len(path)):
198
if path[i] not in _url_safe_characters:
199
chars = path[i].encode('utf-8')
200
path[i] = ''.join(['%%%02X' % ord(c) for c in path[i].encode('utf-8')])
201
return scheme + '://' + ''.join(path)
204
def split(url, exclude_trailing_slash=True):
205
"""Split a URL into its parent directory and a child directory.
207
:param url: A relative or absolute URL
208
:param exclude_trailing_slash: Strip off a final '/' if it is part
209
of the path (but not if it is part of the protocol specification)
211
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
213
if first_path_slash is None:
214
# We have either a relative path, or no separating slash
215
if scheme_loc is None:
217
if exclude_trailing_slash and url.endswith('/'):
219
return _posix_split(url)
221
# Scheme with no path
224
# We have a fully defined path
225
url_base = url[:first_path_slash] # http://host, file://
226
path = url[first_path_slash:] # /file/foo
228
if sys.platform == 'win32' and url.startswith('file:///'):
229
# Strip off the drive letter
230
if path[2:3] not in '\\/':
231
raise errors.InvalidURL(url,
232
'win32 file:/// paths need a drive letter')
233
url_base += path[1:4] # file:///C|/
236
if exclude_trailing_slash and len(path) > 1 and path.endswith('/'):
238
head, tail = _posix_split(path)
239
return url_base + head, tail
242
def strip_trailing_slash(url):
243
"""Strip trailing slash, except for root paths.
245
The definition of 'root path' is platform-dependent.
246
This assumes that all URLs are valid netloc urls, such that they
249
It searches for ://, and then refuses to remove the next '/'.
250
It can also handle relative paths
252
path/to/foo => path/to/foo
253
path/to/foo/ => path/to/foo
254
http://host/path/ => http://host/path
255
http://host/path => http://host/path
256
http://host/ => http://host/
258
file:///foo/ => file:///foo
259
# This is unique on win32 platforms, and is the only URL
260
# format which does it differently.
261
file:///C|/ => file:///C|/
263
if not url.endswith('/'):
266
if sys.platform == 'win32' and url.startswith('file:///'):
267
# This gets handled specially, because the 'top-level'
268
# of a win32 path is actually the drive letter
269
if len(url) > MIN_ABS_FILEURL_LENGTH:
273
scheme_loc, first_path_slash = _find_scheme_and_separator(url)
274
if scheme_loc is None:
275
# This is a relative path, as it has no scheme
276
# so just chop off the last character
279
if first_path_slash is None or first_path_slash == len(url)-1:
280
# Don't chop off anything if the only slash is the path
288
"""Unescape relpath from url format.
290
This returns a Unicode path from a URL
292
# jam 20060427 URLs are supposed to be ASCII only strings
293
# If they are passed in as unicode, urllib.unquote
294
# will return a UNICODE string, which actually contains
295
# utf-8 bytes. So we have to ensure that they are
296
# plain ASCII strings, or the final .decode will
297
# try to encode the UNICODE => ASCII, and then decode
301
except UnicodeError, e:
302
raise errors.InvalidURL(url, 'URL was not a plain ASCII url: %s' % (e,))
303
unquoted = urllib.unquote(url)
305
unicode_path = unquoted.decode('utf-8')
306
except UnicodeError, e:
307
raise errors.InvalidURL(url, 'Unable to encode the URL as utf-8: %s' % (e,))
311
# These are characters that if escaped, should stay that way
312
_no_decode_chars = ';/?:@&=+$,#'
313
_no_decode_ords = [ord(c) for c in _no_decode_chars]
314
_no_decode_hex = (['%02x' % o for o in _no_decode_ords]
315
+ ['%02X' % o for o in _no_decode_ords])
316
_hex_display_map = dict(([('%02x' % o, chr(o)) for o in range(256)]
317
+ [('%02X' % o, chr(o)) for o in range(256)]))
318
#These entries get mapped to themselves
319
_hex_display_map.update((hex,'%'+hex) for hex in _no_decode_hex)
321
# These characters should not be escaped
322
_url_safe_characters = set('abcdefghijklmnopqrstuvwxyz'
323
'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
328
def unescape_for_display(url):
329
"""Decode what you can for a URL, so that we get a nice looking path.
331
This will turn file:// urls into local paths, and try to decode
332
any portions of a http:// style url that it can.
334
if url.startswith('file://'):
335
return local_path_from_url(url)
337
# Split into sections to try to decode utf-8
339
for i in xrange(1, len(res)):
340
escaped_chunks = res[i].split('%')
341
for j in xrange(1, len(escaped_chunks)):
342
item = escaped_chunks[j]
344
escaped_chunks[j] = _hex_display_map[item[:2]] + item[2:]
346
# Put back the percent symbol
347
escaped_chunks[j] = '%' + item
348
except UnicodeDecodeError:
349
escaped_chunks[j] = unichr(int(item[:2], 16)) + item[2:]
350
unescaped = ''.join(escaped_chunks)
352
res[i] = unescaped.decode('utf-8')
353
except UnicodeDecodeError:
354
# If this path segment cannot be properly utf-8 decoded
355
# after doing unescaping we will just leave it alone