def _get_curie_prefix(uself, uri, existing_curies):
ns_count = 0
from curies import CURIES
# TODO: replace this with a once-per run update CURIES function
def get_curie_online(uri):
import requests
try:
r = requests.get("http://prefix.cc/reverse",
params={
"uri": uri,
"format": "txt"
})
if r.status_code == 200:
# primitive check to see if it really is prefix.cc replying with a text/plain response
if r.headers["Content-Type"] == "text/plain":
return r.text.split("\t")[0]
else:
return None
else:
return None
except requests.exceptions.ConnectionError:
# presumably this module can't access the internet or prefix.cc is down
return None
def get_curie_from_namespace(uri, existing_curies, ns_count):
# strip off trailing hash or slash and return last path segment
c = uri.rstrip("#/").split("/")[-1]
# prevent CURIE collision = return nsX (X int) if we already have this one
if c in existing_curies:
ns_count += 1
return "ns" + str(ns_count)
return c
# attempt to look up the well-known curie for this Namespace in http://prefix.cc dump
for k, v in CURIES.items():
if v == uri:
return k
# attempt to look up the well-known CURIE for this Namespace using http://prefix.cc online (more up-to-date)
c = get_curie_online(uri)
if c is not None:
return c
# can't find CURIE online so make up one
c = get_curie_from_namespace(uri, existing_curies, ns_count)
return c if c is not None else ""
def _get_curie_prefix(uri, ns):
ns_count = 0
from curies import CURIES, EXTRA_CURIES
# TODO: replace this with a once-per run update CURIES function
def get_curie_online(uri):
try:
r = requests.get('http://prefix.cc/reverse',
params={
'uri': uri,
'format': 'txt'
})
if r.status_code == 200:
return r.text.split('\t')[0]
else:
return None
except requests.exceptions.ConnectionError:
# presumably this module can't access the internet or prefix.cc is down
return None
def get_curie_from_namespace(uri, ns_count):
# strip off trailing hash or slash and return last path segment
c = uri.rstrip('#/').split('/')[-1]
# prevent CURIE collision = return nsX (x int) if we already have this one
for k, v in ns.items():
if c == v:
ns_count += 1
return 'ns' + str(ns_count)
return c
# attempt to look up the well-known curie for this Namespace in http://prefix.cc dump
for k, v in CURIES.items():
if v == uri:
return k
for k, v in EXTRA_CURIES.items():
if v == uri:
return k
# attempt to look up the well-known CURIE for this Namespace using http://prefix.cc online (more up-to-date)
c = get_curie_online(uri)
if c is not None:
return c
# can't fund CURIE online so make up one
c = get_curie_from_namespace(uri, ns_count)
return c if c is not None else ''
def _extract_namespaces(self):
"""
First we get the namespaces from rdflib
Then we cycle through all the URIs in the graph (all s, p & o),
create a set of them,
extract their base URIS (i.e. a non-duplicative list of them)
see if they are in the namespaces,
if not, generate their CURIE and add them to namespaces
"""
# get declared namespaces, keyed by URI
ns = {}
uri_bases = set()
for k, v in self.G.namespaces():
ns[str(v)] = k
# get other namespaces by extracting base URIs from all URIs
for s, p, o in self.G:
# exclude certain annotation URIs
# and individuals (SDO.identifier)
# exclude known annoying URIs (ORCID)
if (
p == OWL.versionIRI
or p == OWL.imports
or p == SDO.identifier
or str(o).startswith("https://orcid")
):
pass
else:
# add only URI subjects (not Blank Nodes)
if type(s) == URIRef:
uri_bases.add(self._get_namespace_from_uri(str(s)))
# predicates are always URIs so add them all
uri_bases.add(self._get_namespace_from_uri(str(p)))
# add only URI objects (not Blank Nodes or Literals), exclude emails
if type(o) == URIRef and "@" not in str(o):
uri_bases.add(self._get_namespace_from_uri(str(o)))
# for the de-duplicated URIs, if the uri_base is not in namespaces, get CURIE and add it
for uri_base in uri_bases:
if ns.get(uri_base) is None:
found = False
# try to match uri_base to stored CURIES first
if self.use_curies_stored:
from pylode.curies import CURIES
try:
ns[uri_base] = list(CURIES.keys())[list(CURIES.values()).index(uri_base)]
found = True
except ValueError:
pass
if not found:
if self.get_curies_online:
print(f"getting CURIE for {uri_base} online")
uri_prefix = self._get_curie_prefix(uri_base, [x for x in ns.values()])
ns[uri_base] = uri_prefix
found = True
# if not found:
# ns[uri_base] = "ns" + str(len(ns))
# invert the key/values in instances
for k, v in sorted(ns.items(), key=lambda x: x[1]):
if v == "": # can't use empty dict keys in Python
self.NAMESPACES[":"] = k
else:
self.NAMESPACES[v] = k
del(self.NAMESPACES["xml"]) # that bloody XML namespace has to go!