def parse_spec(self, spec_str, all_index=False):
'''Parse a a string into a `Spec`.
:param spec_str: the spec string.
:param all_index: this spec_str is just an index (useful for file
specs)
'''
scheme, netloc, path, query, fragment = urlparse.urlsplit(spec_str)
# case where we just provide a path ...
if scheme == '':
# default scheme
scheme = CONFIG.get('dpm', 'index.default')
if scheme == 'file':
if '://' in spec_str:
path = spec_str.split('://')[1]
path = urllib.url2pathname(path)
path = path.replace('/', os.sep)
path = os.path.abspath(path)
# for file netloc is everything up to last name
if all_index:
netloc = path
path = ''
else:
netloc = os.path.join(os.path.dirname(path))
path = os.path.basename(path)
elif scheme in ('ckan', 'db', 'egg'):
# python >= 2.6.5 changes behaviour of urlsplit for novel url
# schemes to be rfc compliant
# http://bugs.python.org/issue7904
# urlparse.urlsplit(ckan://ckan) gives:
# python < 2.6.5
# SplitResult(scheme='ckan', netloc='', path='ckan', query='', fragment='')
# python >= 2.5.5
# SplitResult(scheme='ckan', netloc='ckan', path='', query='', fragment='')
if netloc != '': # python >= 2.6.5
path = netloc + path if path else netloc
netloc = ''
# after urlsplit of ckan://... have path = //... for python < 2.6.5
while path.startswith('/'):
path = path[1:]
netloc = '/'.join(path.split('/')[:-1])
path = path.split('/')[-1]
if scheme == 'ckan':
# we have a path but did not put http:// ...
if netloc and not netloc.startswith('http'):
netloc = 'http://' + netloc
elif scheme == 'db':
if netloc and not netloc.startswith('file'):
netloc = 'file://' + netloc
elif scheme == 'egg':
if path and not netloc:
netloc, path = path, ''
netloc = netloc.strip("/")
path = path.lstrip("/")
spec = Spec(scheme, netloc, path)
return spec
def parse_spec(self, spec_str, all_index=False):
'''Parse a a string into a `Spec`.
:param spec_str: the spec string.
:param all_index: this spec_str is just an index (useful for file
specs)
'''
scheme, netloc, path, query, fragment = urlparse.urlsplit(spec_str)
# case where we just provide a path ...
if scheme == '':
# default scheme
scheme = CONFIG.get('dpm', 'index.default')
if scheme == 'http':
# assume this is a ckan instance ...
# for ckan netloc = full API url, path = dataset name
scheme = 'ckan'
# ckan netloc should be full url
# TODO: what happens if CKAN instance not at base domain!
netloc = 'http://' + netloc
path = path.rstrip('/')
if path:
out = re.match('(/.+)?/dataset/(.+)$', path)
if not out:
msg = 'URL is not a path to a CKAN instance or dataset: %s' % spec_str
raise Exception(msg)
(offset, name) = out.groups()
if offset:
netloc += offset
path = name
netloc += '/api'
elif scheme == 'file':
if '://' in spec_str:
path = spec_str.split('://')[1]
path = urllib.url2pathname(path)
path = path.replace('/', os.sep)
path = os.path.abspath(path)
# for file netloc is everything up to last name
if all_index:
netloc = path
path = ''
else:
netloc = os.path.join(os.path.dirname(path))
path = os.path.basename(path)
elif scheme in ('ckan', 'egg'):
# python >= 2.6.5 changes behaviour of urlsplit for novel url
# schemes to be rfc compliant
# http://bugs.python.org/issue7904
# urlparse.urlsplit(ckan://ckan) gives:
# python < 2.6.5
# SplitResult(scheme='ckan', netloc='', path='ckan', query='', fragment='')
# python >= 2.5.5
# SplitResult(scheme='ckan', netloc='ckan', path='', query='', fragment='')
if netloc != '': # python >= 2.6.5
path = netloc + path if path else netloc
netloc = ''
# after urlsplit of ckan://... have path = //... for python < 2.6.5
while path.startswith('/'):
path = path[1:]
netloc = '/'.join(path.split('/')[:-1])
path = path.split('/')[-1]
if scheme == 'ckan':
# we have a path but did not put http:// ...
if netloc and not netloc.startswith('http'):
netloc = 'http://' + netloc
elif scheme == 'egg':
if path and not netloc:
netloc, path = path, ''
netloc = netloc.strip("/")
path = path.lstrip("/")
spec = Spec(scheme, netloc, path)
return spec
