60
59
return split(url, exclude_trailing_slash=exclude_trailing_slash)[0]
62
# Private copies of quote and unquote, copied from Python's
63
# urllib module because urllib unconditionally imports socket, which imports
66
always_safe = ('ABCDEFGHIJKLMNOPQRSTUVWXYZ'
67
'abcdefghijklmnopqrstuvwxyz'
70
for i, c in zip(xrange(256), str(bytearray(xrange(256)))):
71
_safe_map[c] = c if (i < 128 and c in always_safe) else '%{0:02X}'.format(i)
75
def quote(s, safe='/'):
76
"""quote('abc def') -> 'abc%20def'
78
Each part of a URL, e.g. the path info, the query, etc., has a
79
different set of reserved characters that must be quoted.
81
RFC 2396 Uniform Resource Identifiers (URI): Generic Syntax lists
82
the following reserved characters.
84
reserved = ";" | "/" | "?" | ":" | "@" | "&" | "=" | "+" |
87
Each of these characters is reserved in some component of a URL,
88
but not necessarily in all of them.
90
By default, the quote function is intended for quoting the path
91
section of a URL. Thus, it will not encode '/'. This character
92
is reserved, but in typical usage the quote function is being
93
called on a path where the existing slash characters are used as
99
raise TypeError('None object cannot be quoted')
101
cachekey = (safe, always_safe)
103
(quoter, safe) = _safe_quoters[cachekey]
105
safe_map = _safe_map.copy()
106
safe_map.update([(c, c) for c in safe])
107
quoter = safe_map.__getitem__
108
safe = always_safe + safe
109
_safe_quoters[cachekey] = (quoter, safe)
110
if not s.rstrip(safe):
112
return ''.join(map(quoter, s))
115
_hexdig = '0123456789ABCDEFabcdef'
116
_hextochr = dict((a + b, chr(int(a + b, 16)))
117
for a in _hexdig for b in _hexdig)
120
"""unquote('abc%20def') -> 'abc def'."""
128
s += _hextochr[item[:2]] + item[2:]
131
except UnicodeDecodeError:
132
s += unichr(int(item[:2], 16)) + item[2:]
63
136
def escape(relpath):
64
137
"""Escape relpath to be a valid url."""
65
138
if isinstance(relpath, unicode):
66
139
relpath = relpath.encode('utf-8')
67
140
# After quoting and encoding, the path should be perfectly
68
141
# safe as a plain ASCII string, str() just enforces this
69
return str(urllib.quote(relpath, safe='/~'))
142
return str(quote(relpath, safe='/~'))
72
145
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.
199
scheme_end, path_start = _find_scheme_and_separator(base)
200
if scheme_end is None and path_start is None:
202
elif path_start is None:
203
path_start = len(base)
204
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.
206
arg_scheme_end, arg_path_start = _find_scheme_and_separator(arg)
207
if arg_scheme_end is None and arg_path_start is None:
209
elif arg_path_start is None:
210
arg_path_start = len(arg)
211
if arg_scheme_end is not None:
213
path = arg[arg_path_start:]
214
scheme_end = arg_scheme_end
215
path_start = arg_path_start
167
path = '/'.join(path)
168
217
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)
218
return base[:path_start] + path
184
221
def joinpath(base, *args):
469
508
return url_base + head, tail
511
def split_segment_parameters_raw(url):
512
"""Split the subsegment of the last segment of a URL.
514
:param url: A relative or absolute URL
515
:return: (url, subsegments)
517
# GZ 2011-11-18: Dodgy removing the terminal slash like this, function
518
# operates on urls not url+segments, and Transport classes
519
# should not be blindly adding slashes in the first place.
520
lurl = strip_trailing_slash(url)
521
# Segments begin at first comma after last forward slash, if one exists
522
segment_start = lurl.find(",", lurl.rfind("/")+1)
523
if segment_start == -1:
525
return (lurl[:segment_start], lurl[segment_start+1:].split(","))
528
def split_segment_parameters(url):
529
"""Split the segment parameters of the last segment of a URL.
531
:param url: A relative or absolute URL
532
:return: (url, segment_parameters)
534
(base_url, subsegments) = split_segment_parameters_raw(url)
536
for subsegment in subsegments:
537
(key, value) = subsegment.split("=", 1)
538
parameters[key] = value
539
return (base_url, parameters)
542
def join_segment_parameters_raw(base, *subsegments):
543
"""Create a new URL by adding subsegments to an existing one.
545
This adds the specified subsegments to the last path in the specified
546
base URL. The subsegments should be bytestrings.
548
:note: You probably want to use join_segment_parameters instead.
552
for subsegment in subsegments:
553
if type(subsegment) is not str:
554
raise TypeError("Subsegment %r is not a bytestring" % subsegment)
555
if "," in subsegment:
556
raise errors.InvalidURLJoin(", exists in subsegments",
558
return ",".join((base,) + subsegments)
561
def join_segment_parameters(url, parameters):
562
"""Create a new URL by adding segment parameters to an existing one.
564
The parameters of the last segment in the URL will be updated; if a
565
parameter with the same key already exists it will be overwritten.
567
:param url: A URL, as string
568
:param parameters: Dictionary of parameters, keys and values as bytestrings
570
(base, existing_parameters) = split_segment_parameters(url)
572
new_parameters.update(existing_parameters)
573
for key, value in parameters.iteritems():
574
if type(key) is not str:
575
raise TypeError("parameter key %r is not a bytestring" % key)
576
if type(value) is not str:
577
raise TypeError("parameter value %r for %s is not a bytestring" %
580
raise errors.InvalidURLJoin("= exists in parameter key", url,
582
new_parameters[key] = value
583
return join_segment_parameters_raw(base,
584
*["%s=%s" % item for item in sorted(new_parameters.items())])
472
587
def _win32_strip_local_trailing_slash(url):
473
588
"""Strip slashes after the drive letter"""
474
589
if len(url) > WIN32_MIN_ABS_FILEURL_LENGTH:
693
808
return osutils.pathjoin(*segments)
814
def __init__(self, scheme, quoted_user, quoted_password, quoted_host,
817
self.quoted_host = quoted_host
818
self.host = unquote(self.quoted_host)
819
self.quoted_user = quoted_user
820
if self.quoted_user is not None:
821
self.user = unquote(self.quoted_user)
824
self.quoted_password = quoted_password
825
if self.quoted_password is not None:
826
self.password = unquote(self.quoted_password)
830
self.quoted_path = _url_hex_escapes_re.sub(_unescape_safe_chars, quoted_path)
831
self.path = unquote(self.quoted_path)
833
def __eq__(self, other):
834
return (isinstance(other, self.__class__) and
835
self.scheme == other.scheme and
836
self.host == other.host and
837
self.user == other.user and
838
self.password == other.password and
839
self.path == other.path)
842
return "<%s(%r, %r, %r, %r, %r, %r)>" % (
843
self.__class__.__name__,
844
self.scheme, self.quoted_user, self.quoted_password,
845
self.quoted_host, self.port, self.quoted_path)
848
def from_string(cls, url):
849
"""Create a URL object from a string.
851
:param url: URL as bytestring
853
if isinstance(url, unicode):
854
raise errors.InvalidURL('should be ascii:\n%r' % url)
855
url = url.encode('utf-8')
856
(scheme, netloc, path, params,
857
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
858
user = password = host = port = None
860
user, host = netloc.rsplit('@', 1)
862
user, password = user.split(':', 1)
866
if ':' in host and not (host[0] == '[' and host[-1] == ']'):
868
host, port = host.rsplit(':',1)
872
raise errors.InvalidURL('invalid port number %s in url:\n%s' %
874
if host != "" and host[0] == '[' and host[-1] == ']': #IPv6
877
return cls(scheme, user, password, host, port, path)
880
netloc = self.quoted_host
882
netloc = "[%s]" % netloc
883
if self.quoted_user is not None:
884
# Note that we don't put the password back even if we
885
# have one so that it doesn't get accidentally
887
netloc = '%s@%s' % (self.quoted_user, netloc)
888
if self.port is not None:
889
netloc = '%s:%d' % (netloc, self.port)
890
return urlparse.urlunparse(
891
(self.scheme, netloc, self.quoted_path, None, None, None))
894
def _combine_paths(base_path, relpath):
895
"""Transform a Transport-relative path to a remote absolute path.
897
This does not handle substitution of ~ but does handle '..' and '.'
902
t._combine_paths('/home/sarah', 'project/foo')
903
=> '/home/sarah/project/foo'
904
t._combine_paths('/home/sarah', '../../etc')
906
t._combine_paths('/home/sarah', '/etc')
909
:param base_path: base path
910
:param relpath: relative url string for relative part of remote path.
911
:return: urlencoded string for final path.
913
if not isinstance(relpath, str):
914
raise errors.InvalidURL(relpath)
915
relpath = _url_hex_escapes_re.sub(_unescape_safe_chars, relpath)
916
if relpath.startswith('/'):
919
base_parts = base_path.split('/')
920
if len(base_parts) > 0 and base_parts[-1] == '':
921
base_parts = base_parts[:-1]
922
for p in relpath.split('/'):
924
if len(base_parts) == 0:
925
# In most filesystems, a request for the parent
926
# of root, just returns root.
933
path = '/'.join(base_parts)
934
if not path.startswith('/'):
938
def clone(self, offset=None):
939
"""Return a new URL for a path relative to this URL.
941
:param offset: A relative path, already urlencoded
942
:return: `URL` instance
944
if offset is not None:
945
relative = unescape(offset).encode('utf-8')
946
path = self._combine_paths(self.path, relative)
947
path = quote(path, safe="/~")
949
path = self.quoted_path
950
return self.__class__(self.scheme, self.quoted_user,
951
self.quoted_password, self.quoted_host, self.port,
697
955
def parse_url(url):
698
956
"""Extract the server address, the credentials and the path from the url.
703
961
:param url: an quoted url
705
962
: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)
965
parsed_url = URL.from_string(url)
966
return (parsed_url.scheme, parsed_url.user, parsed_url.password,
967
parsed_url.host, parsed_url.port, parsed_url.path)
added
removed
bzrlib/urlutils.py
