def parse(self, uri):
u"""parses a URI in the form of
<scheme>://<username>:<password>@<host>:<port>/<path>?<query>#<fragment>
"""
if uri == b'*':
self.path = u'*'
return
if not uri:
raise InvalidURI(u'empty URI')
uri, _, fragment = uri.partition(b'#')
uri, _, query_string = uri.partition(b'?')
scheme, scheme_exists, uri = uri.rpartition(b'://')
if not scheme_exists and uri.startswith(b'//'):
uri = uri[2:]
authority, _, path = uri.partition(b'/')
path = b'/%s' % path
userinfo, _, hostport = authority.rpartition(b'@')
username, _, password = userinfo.partition(b':')
if hostport.endswith(b']') and hostport.startswith(b'['):
host, port = hostport, b''
else:
host, _, port = hostport.rpartition(b':')
unquote = self.unquote
path = u'/'.join([unquote(seq).replace(u'/', u'%2f') for seq in path.split(b'/')])
if path.startswith(u'//'):
raise InvalidURI(u'Invalid path: must not start with "//"')
try:
scheme = scheme.decode('ascii')
except UnicodeDecodeError:
raise InvalidURI(u'Invalid scheme: must be ascii')
self.tuple = (
scheme,
unquote(username),
unquote(password),
unquote(host),
port,
path,
QueryString.encode(QueryString.decode(query_string)),
unquote(fragment)
)
def query(self, query): self.query_string = QueryString.encode(query, self.encoding)
def query(self): return tuple(QueryString.decode(self.query_string, self.encoding))
def parse(self, uri):
r"""Parses a well formed absolute or relative URI.
foo://example.com:8042/over/there?name=ferret#nose
\_/ \______________/\_________/ \_________/ \__/
| | | | |
scheme authority path query fragment
| _____________________|__
/ \ / \
urn:example:animal:ferret:nose
https://username:password@[::1]:8090/some/path?query#fragment
<scheme>://<username>:<password>@<host>:<port>/<path>?<query>#<fragment>
[<scheme>:][//[<username>[:<password>]@][<host>][:<port>]/]<path>[?<query>][#<fragment>]
"""
if isinstance(uri, Unicode):
try:
uri = uri.encode('ascii')
except UnicodeEncodeError:
raise TypeError('URI must be ASCII bytes.')
if type(self) is URI and b':' in uri:
self.scheme = uri.split(b':', 1)[0].lower()
if type(self) is not URI:
return self.parse(uri)
if uri and uri.strip(b'0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ!"#$%&\'()*+,-./:;<=>?@[\\]^_`{|}~'):
raise InvalidURI(_(u'Invalid URI: must consist of printable ASCII characters without whitespace.'))
uri, __, fragment = uri.partition(b'#')
uri, __, query_string = uri.partition(b'?')
scheme, authority_exists, uri = uri.rpartition(b'://')
if not authority_exists and uri.startswith(b'//'):
uri = uri[2:]
authority_exists = True
if not authority_exists and b':' in uri:
scheme, __, uri = uri.partition(b':')
authority, path = b'', uri
if authority_exists:
authority, __, path = uri.partition(b'/')
path = b'%s%s' % (__, path)
userinfo, __, hostport = authority.rpartition(b'@')
username, __, password = userinfo.partition(b':')
if ':' in hostport and not hostport.endswith(b']'):
host, __, port = hostport.rpartition(b':')
else:
host, port = hostport, b''
unquote = self.unquote
path = u'/'.join([unquote(seq).replace(u'/', u'%2f') for seq in path.split(b'/')])
try:
scheme = scheme.decode('ascii').lower()
except UnicodeDecodeError:
raise InvalidURI(_(u'Invalid scheme: must be ASCII.'))
if scheme and scheme.strip(u'abcdefghijklmnopqrstuvwxyz0123456789.-+'):
raise InvalidURI(_(u'Invalid scheme: must only contain alphanumeric letters or plus, dash, dot.'))
if query_string:
query_string = QueryString.encode(QueryString.decode(query_string, self.encoding), self.encoding)
self.tuple = (
scheme,
unquote(username),
unquote(password),
self._unquote_host(host),
port,
path,
query_string.decode(self.encoding),
unquote(fragment)
)
def query(self, query): self.query_string = QueryString.encode(query, self.encoding)
def query(self): return tuple(QueryString.decode(self.query_string, self.encoding))
def parse(self, uri):
r"""Parses a well formed absolute or relative URI.
foo://example.com:8042/over/there?name=ferret#nose
\_/ \______________/\_________/ \_________/ \__/
| | | | |
scheme authority path query fragment
| _____________________|__
/ \ / \
urn:example:animal:ferret:nose
https://username:password@[::1]:8090/some/path?query#fragment
<scheme>://<username>:<password>@<host>:<port>/<path>?<query>#<fragment>
[<scheme>:][//[<username>[:<password>]@][<host>][:<port>]/]<path>[?<query>][#<fragment>]
"""
if isinstance(uri, Unicode):
try:
uri = uri.encode('ascii')
except UnicodeEncodeError:
raise TypeError('URI must be ASCII bytes.')
if type(self) is URI and b':' in uri:
self.scheme = uri.split(b':', 1)[0].lower()
if type(self) is not URI:
return self.parse(uri)
if uri and uri.strip(b'0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ!"#$%&\'()*+,-./:;<=>?@[\\]^_`{|}~'):
raise InvalidURI(_(u'Invalid URI: must consist of printable ASCII characters without whitespace.'))
uri, __, fragment = uri.partition(b'#')
uri, __, query_string = uri.partition(b'?')
scheme, authority_exists, uri = uri.rpartition(b'://')
if not authority_exists and uri.startswith(b'//'):
uri = uri[2:]
authority_exists = True
if not authority_exists and b':' in uri:
scheme, __, uri = uri.partition(b':')
authority, path = b'', uri
if authority_exists:
authority, __, path = uri.partition(b'/')
path = b'%s%s' % (__, path)
userinfo, __, hostport = authority.rpartition(b'@')
username, __, password = userinfo.partition(b':')
if b':' in hostport and not hostport.endswith(b']'):
host, __, port = hostport.rpartition(b':')
else:
host, port = hostport, b''
unquote = self.unquote
path = u'/'.join([unquote(seq).replace(u'/', u'%2f') for seq in path.split(b'/')])
try:
scheme = scheme.decode('ascii').lower()
except UnicodeDecodeError:
raise InvalidURI(_(u'Invalid scheme: must be ASCII.'))
if scheme and scheme.strip(u'abcdefghijklmnopqrstuvwxyz0123456789.-+'):
raise InvalidURI(_(u'Invalid scheme: must only contain alphanumeric letters or plus, dash, dot.'))
if query_string:
query_string = QueryString.encode(QueryString.decode(query_string, self.encoding), self.encoding)
self.tuple = (
scheme,
unquote(username),
unquote(password),
self._unquote_host(host),
port,
path,
query_string.decode(self.encoding),
unquote(fragment)
)
