60
61
return split(url, exclude_trailing_slash=exclude_trailing_slash)[0]
64
# Private copies of quote and unquote, copied from Python's
65
# urllib module because urllib unconditionally imports socket, which imports
68
always_safe = ('ABCDEFGHIJKLMNOPQRSTUVWXYZ'
69
'abcdefghijklmnopqrstuvwxyz'
72
for i, c in zip(xrange(256), str(bytearray(xrange(256)))):
73
_safe_map[c] = c if (i < 128 and c in always_safe) else '%{0:02X}'.format(i)
77
def quote(s, safe='/'):
78
"""quote('abc def') -> 'abc%20def'
80
Each part of a URL, e.g. the path info, the query, etc., has a
81
different set of reserved characters that must be quoted.
83
RFC 2396 Uniform Resource Identifiers (URI): Generic Syntax lists
84
the following reserved characters.
86
reserved = ";" | "/" | "?" | ":" | "@" | "&" | "=" | "+" |
89
Each of these characters is reserved in some component of a URL,
90
but not necessarily in all of them.
92
By default, the quote function is intended for quoting the path
93
section of a URL. Thus, it will not encode '/'. This character
94
is reserved, but in typical usage the quote function is being
95
called on a path where the existing slash characters are used as
101
raise TypeError('None object cannot be quoted')
103
cachekey = (safe, always_safe)
105
(quoter, safe) = _safe_quoters[cachekey]
107
safe_map = _safe_map.copy()
108
safe_map.update([(c, c) for c in safe])
109
quoter = safe_map.__getitem__
110
safe = always_safe + safe
111
_safe_quoters[cachekey] = (quoter, safe)
112
if not s.rstrip(safe):
114
return ''.join(map(quoter, s))
117
_hexdig = '0123456789ABCDEFabcdef'
118
_hextochr = dict((a + b, chr(int(a + b, 16)))
119
for a in _hexdig for b in _hexdig)
122
"""unquote('abc%20def') -> 'abc def'."""
130
s += _hextochr[item[:2]] + item[2:]
133
except UnicodeDecodeError:
134
s += unichr(int(item[:2], 16)) + item[2:]
63
138
def escape(relpath):
64
139
"""Escape relpath to be a valid url."""
65
140
if isinstance(relpath, unicode):
66
141
relpath = relpath.encode('utf-8')
67
142
# After quoting and encoding, the path should be perfectly
68
143
# safe as a plain ASCII string, str() just enforces this
69
return str(urllib.quote(relpath, safe='/~'))
144
return str(quote(relpath, safe='/~'))
72
147
def file_relpath(base, path):
121
match = _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
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.
201
scheme_end, path_start = _find_scheme_and_separator(base)
202
if scheme_end is None and path_start is None:
204
elif path_start is None:
205
path_start = len(base)
206
path = base[path_start:]
147
match = _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.
208
arg_scheme_end, arg_path_start = _find_scheme_and_separator(arg)
209
if arg_scheme_end is None and arg_path_start is None:
211
elif arg_path_start is None:
212
arg_path_start = len(arg)
213
if arg_scheme_end is not None:
215
path = arg[arg_path_start:]
216
scheme_end = arg_scheme_end
217
path_start = arg_path_start
167
path = '/'.join(path)
168
219
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:
180
return '/'.join(path)
181
return scheme + '://' + '/'.join(path)
220
return base[:path_start] + path
184
223
def joinpath(base, *args):
463
510
return url_base + head, tail
513
def split_segment_parameters_raw(url):
514
"""Split the subsegment of the last segment of a URL.
516
:param url: A relative or absolute URL
517
:return: (url, subsegments)
519
# GZ 2011-11-18: Dodgy removing the terminal slash like this, function
520
# operates on urls not url+segments, and Transport classes
521
# should not be blindly adding slashes in the first place.
522
lurl = strip_trailing_slash(url)
523
# Segments begin at first comma after last forward slash, if one exists
524
segment_start = lurl.find(",", lurl.rfind("/")+1)
525
if segment_start == -1:
527
return (lurl[:segment_start], lurl[segment_start+1:].split(","))
530
def split_segment_parameters(url):
531
"""Split the segment parameters of the last segment of a URL.
533
:param url: A relative or absolute URL
534
:return: (url, segment_parameters)
536
(base_url, subsegments) = split_segment_parameters_raw(url)
538
for subsegment in subsegments:
539
(key, value) = subsegment.split("=", 1)
540
parameters[key] = value
541
return (base_url, parameters)
544
def join_segment_parameters_raw(base, *subsegments):
545
"""Create a new URL by adding subsegments to an existing one.
547
This adds the specified subsegments to the last path in the specified
548
base URL. The subsegments should be bytestrings.
550
:note: You probably want to use join_segment_parameters instead.
554
for subsegment in subsegments:
555
if type(subsegment) is not str:
556
raise TypeError("Subsegment %r is not a bytestring" % subsegment)
557
if "," in subsegment:
558
raise errors.InvalidURLJoin(", exists in subsegments",
560
return ",".join((base,) + subsegments)
563
def join_segment_parameters(url, parameters):
564
"""Create a new URL by adding segment parameters to an existing one.
566
The parameters of the last segment in the URL will be updated; if a
567
parameter with the same key already exists it will be overwritten.
569
:param url: A URL, as string
570
:param parameters: Dictionary of parameters, keys and values as bytestrings
572
(base, existing_parameters) = split_segment_parameters(url)
574
new_parameters.update(existing_parameters)
575
for key, value in parameters.iteritems():
576
if type(key) is not str:
577
raise TypeError("parameter key %r is not a bytestring" % key)
578
if type(value) is not str:
579
raise TypeError("parameter value %r for %s is not a bytestring" %
582
raise errors.InvalidURLJoin("= exists in parameter key", url,
584
new_parameters[key] = value
585
return join_segment_parameters_raw(base,
586
*["%s=%s" % item for item in sorted(new_parameters.items())])
466
589
def _win32_strip_local_trailing_slash(url):
467
590
"""Strip slashes after the drive letter"""
468
591
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
687
810
return osutils.pathjoin(*segments)
816
def __init__(self, scheme, quoted_user, quoted_password, quoted_host,
819
self.quoted_host = quoted_host
820
self.host = unquote(self.quoted_host)
821
self.quoted_user = quoted_user
822
if self.quoted_user is not None:
823
self.user = unquote(self.quoted_user)
826
self.quoted_password = quoted_password
827
if self.quoted_password is not None:
828
self.password = unquote(self.quoted_password)
832
self.quoted_path = _url_hex_escapes_re.sub(_unescape_safe_chars, quoted_path)
833
self.path = unquote(self.quoted_path)
835
def __eq__(self, other):
836
return (isinstance(other, self.__class__) and
837
self.scheme == other.scheme and
838
self.host == other.host and
839
self.user == other.user and
840
self.password == other.password and
841
self.path == other.path)
844
return "<%s(%r, %r, %r, %r, %r, %r)>" % (
845
self.__class__.__name__,
846
self.scheme, self.quoted_user, self.quoted_password,
847
self.quoted_host, self.port, self.quoted_path)
850
def from_string(cls, url):
851
"""Create a URL object from a string.
853
:param url: URL as bytestring
855
if isinstance(url, unicode):
856
raise errors.InvalidURL('should be ascii:\n%r' % url)
857
url = url.encode('utf-8')
858
(scheme, netloc, path, params,
859
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
860
user = password = host = port = None
862
user, host = netloc.rsplit('@', 1)
864
user, password = user.split(':', 1)
868
if ':' in host and not (host[0] == '[' and host[-1] == ']'):
870
host, port = host.rsplit(':',1)
874
raise errors.InvalidURL('invalid port number %s in url:\n%s' %
876
if host != "" and host[0] == '[' and host[-1] == ']': #IPv6
879
return cls(scheme, user, password, host, port, path)
882
netloc = self.quoted_host
884
netloc = "[%s]" % netloc
885
if self.quoted_user is not None:
886
# Note that we don't put the password back even if we
887
# have one so that it doesn't get accidentally
889
netloc = '%s@%s' % (self.quoted_user, netloc)
890
if self.port is not None:
891
netloc = '%s:%d' % (netloc, self.port)
892
return urlparse.urlunparse(
893
(self.scheme, netloc, self.quoted_path, None, None, None))
896
def _combine_paths(base_path, relpath):
897
"""Transform a Transport-relative path to a remote absolute path.
899
This does not handle substitution of ~ but does handle '..' and '.'
904
t._combine_paths('/home/sarah', 'project/foo')
905
=> '/home/sarah/project/foo'
906
t._combine_paths('/home/sarah', '../../etc')
908
t._combine_paths('/home/sarah', '/etc')
911
:param base_path: base path
912
:param relpath: relative url string for relative part of remote path.
913
:return: urlencoded string for final path.
915
if not isinstance(relpath, str):
916
raise errors.InvalidURL(relpath)
917
relpath = _url_hex_escapes_re.sub(_unescape_safe_chars, relpath)
918
if relpath.startswith('/'):
921
base_parts = base_path.split('/')
922
if len(base_parts) > 0 and base_parts[-1] == '':
923
base_parts = base_parts[:-1]
924
for p in relpath.split('/'):
926
if len(base_parts) == 0:
927
# In most filesystems, a request for the parent
928
# of root, just returns root.
935
path = '/'.join(base_parts)
936
if not path.startswith('/'):
940
def clone(self, offset=None):
941
"""Return a new URL for a path relative to this URL.
943
:param offset: A relative path, already urlencoded
944
:return: `URL` instance
946
if offset is not None:
947
relative = unescape(offset).encode('utf-8')
948
path = self._combine_paths(self.path, relative)
949
path = quote(path, safe="/~")
951
path = self.quoted_path
952
return self.__class__(self.scheme, self.quoted_user,
953
self.quoted_password, self.quoted_host, self.port,
691
957
def parse_url(url):
692
958
"""Extract the server address, the credentials and the path from the url.
697
963
:param url: an quoted url
699
964
: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)
967
parsed_url = URL.from_string(url)
968
return (parsed_url.scheme, parsed_url.user, parsed_url.password,
969
parsed_url.host, parsed_url.port, parsed_url.path)
added
removed
bzrlib/urlutils.py
