def get_edge_prov_counts(qid,
pid,
o_qid,
use_subclass_sub=False,
extend_subclass_sub=False,
use_subclass_obj=False,
extend_subclass_obj=False,
biological_edge=False):
global LOGSTR
p_subj = gc.determine_p(use_subclass_sub, extend_subclass_sub)
p_obj = gc.determine_p(use_subclass_obj, extend_subclass_obj)
if biological_edge:
bio_edge = ""
else:
bio_edge = (""" ?item {p_subj} wd:{qid}.
?obj {p_obj} wd:{o_qid}.""".replace('{p_subj}', p_subj).replace(
'{qid}', qid).replace('{p_obj}', p_obj).replace('{o_qid}', o_qid))
edge_query = """
SELECT ?ref ?refLabel ?count WHERE {
{SELECT ?ref (COUNT(*) AS ?count) WHERE {
SELECT DISTINCT ?item ?obj ?ref WHERE {
{bio_edge}
?item p:{pid} [ps:{pid} ?obj;
prov:wasDerivedFrom
[pr:P248 ?ref;]
]
}} GROUP BY ?ref
}
SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}ORDER BY DESC (?count)""".replace('{pid}', pid).replace(
'{bio_edge}', bio_edge)
#print(edge_query)
try:
d = gc.execute_sparql_query(edge_query)['results']['bindings']
except:
print("***** FAILED SPARQL *****")
print("Item QID: {}\tProp PID: {} Obj QID: {}\n".format(
qid, pid, o_qid))
d = []
LOGSTR += 'Edge Reference Query:'
LOGSTR += edge_query + '\n\n'
return [(x['ref']['value'].replace("http://www.wikidata.org/entity/",
""), x['refLabel']['value'],
int(x['count']['value'])) for x in d]
def count_edges(s, p, o, s_subclass, s_expand, o_subclass, o_expand):
p_sub = determine_p(s_subclass, s_expand)
p_obj = determine_p(o_subclass, o_expand)
q_string = """
SELECT (count(distinct *) as ?count) WHERE {
?subject {p_sub} wd:{s} .
?subject wdt:{p} ?object .
?object {p_obj} wd:{o} }
""".replace('{p_sub}',
p_sub).replace('{s}',
s).replace('{p}',
p).replace('{p_obj}',
p_obj).replace('{o}', o)
d = execute_sparql_query(q_string)['results']['bindings']
return [int(x['count']['value']) for x in d][0]
def count_edges(s, p, o, s_subclass, s_expand, o_subclass, o_expand):
p_sub = gc.determine_p(s_subclass, s_expand)
p_obj = gc.determine_p(o_subclass, o_expand)
# test for reciprocal relationships that need to be collapsed
recip_rels = {
'P527': 'P361',
'P361': 'P527',
'P2176': 'P2175',
'P2175': 'P2176',
'P702': 'P688',
'P688': 'P702',
'P1343': 'P4510',
'P4510': 'P1343',
'P828': 'P1542',
'P1542': 'P828',
'P3781': 'P3780',
'P3780': 'P3781'
}
if p in recip_rels.keys():
u = """UNION
{?object wdt:""" + recip_rels[p] + """ ?subject .}"""
else:
u = ""
q_string = """
SELECT (count(distinct *) as ?count) WHERE {
?subject {p_sub} wd:{s} .
{
{?subject wdt:{p} ?object .}
{u}
}
?object {p_obj} wd:{o} }
""".replace('{p_sub}', p_sub).replace('{s}', s).replace('{p}', p).replace(
'{p_obj}', p_obj).replace('{o}', o).replace('{u}', u)
print("B1: " + q_string)
try:
d = gc.execute_sparql_query(q_string)['results']['bindings']
print("B2: " + str(d))
edge_count = [int(x['count']['value']) for x in d][0]
except:
edge_count = -1
print("B3: " + str(edge_count))
return edge_count
def count_prop(qid, prop, is_subclass, expand):
p = determine_p(is_subclass, expand)
q_string = """
SELECT (count(*) as ?count) WHERE {
?item {p} wd:{qid} .
?item wdt:{prop} [] . }
""".replace('{p}', p).replace('{qid}', qid).replace('{prop}', prop)
d = execute_sparql_query(q_string)['results']['bindings']
return [int(x['count']['value']) for x in d][0]
def count_prop(qid, prop, is_subclass, expand):
p = gc.determine_p(is_subclass, expand)
q_string = """
SELECT (count(?item) as ?count) where {
SELECT DISTINCT ?item WHERE {
?item {p} wd:{qid} .
?item wdt:{prop} [] . }}
""".replace('{p}', p).replace('{qid}', qid).replace('{prop}', prop)
print("A1: " + q_string)
try:
d = gc.execute_sparql_query(q_string)['results']['bindings']
print("A2: " + str(d))
prop_count = [int(x['count']['value']) for x in d][0]
except:
prop_count = -1
print("A3: " + str(prop_count))
return prop_count
def get_property_prov_counts(qid,
pid,
use_subclass=False,
extend_subclass=False):
global LOGSTR
p_subj = gc.determine_p(use_subclass, extend_subclass)
prop_query = """
SELECT ?ref ?refLabel ?count WHERE {
{SELECT ?ref (COUNT(*) AS ?count) WHERE {
SELECT DISTINCT ?item ?xref ?ref WHERE {
?item {p_subj} wd:{qid}.
?item p:{pid} [ps:{pid} ?xref;
prov:wasDerivedFrom
[pr:P248 ?ref;]
]
}} GROUP BY ?ref
}
SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}ORDER BY DESC (?count)""".replace('{p_subj}', p_subj).replace(
'{qid}', qid).replace('{pid}', pid)
#print(prop_query)
try:
d = gc.execute_sparql_query(prop_query)['results']['bindings']
except:
print("***** FAILED SPARQL *****")
print("Item QID: {}\tProp PID: {}\n".format(qid, pid))
d = []
LOGSTR += 'Node Property Reference Query:'
LOGSTR += prop_query + '\n\n'
return [(x['ref']['value'].replace("http://www.wikidata.org/entity/",
""), x['refLabel']['value'],
int(x['count']['value'])) for x in d]