def get_pred_objs(self, star):
"""
Returns a key value of predicate:object in a BGP/star-shaped subquery
:param star: list of triple patterns
:return: list of predicates
"""
preds = {
utils.getUri(tr.predicate, self.prefixes)[1:-1]:
(utils.getUri(tr.theobject, self.prefixes)
if tr.theobject.constant else tr.theobject.name)
for tr in star if tr.predicate.constant
}
return preds
def decompose_bgp(self, stars, bgp_preds):
bgpstars = {}
mtres = {}
relevant_mts = {}
starnames = sorted(list(stars.keys()))
for s in starnames:
spred = self.get_pred_objs(stars[s])
bgpstars[s] = {}
bgpstars[s]['triples'] = sorted(stars[s])
bgpstars[s]['predicates'] = spred
types = self.checkRDFTypeStatemnt(stars[s])
if len(types) > 0:
rdfmts = types
else:
rdfmts = self.config.find_rdfmt_by_preds(spred)
bgpstars[s]['rdfmts'] = list(rdfmts.keys())
mtres[s] = bgpstars[s]['rdfmts']
relevant_mts.update(rdfmts)
star_conn = self.getStarsConnections(stars)
mt_conn = self.getMTsConnection(mtres, bgp_preds, relevant_mts)
res = self.prune(star_conn, mt_conn, mtres, stars, relevant_mts)
for s in res:
bgpstars[s]['rdfmts'] = res[s]
for s in res:
datasources = {}
for m in res[s]:
for d in self.config.metadata[m].datasources:
dspreds = self.config.metadata[m].datasources[d]
preds = list(
set(bgpstars[s]['predicates']).intersection(dspreds))
if len(preds) > 0:
datasources.setdefault(d,
{}).setdefault(m,
[]).extend(preds)
else:
datasources.setdefault(d, {}).setdefault(
m, []).extend(dspreds)
if len(bgpstars[s]['predicates']) == 0:
preds = {
tr.predicate.name:
(utils.getUri(tr.theobject, self.prefixes) if
tr.theobject.constant else tr.theobject.name)
for tr in stars[s]
}
bgpstars[s]['predicates'] = preds
if len(datasources) == 0:
return [], [], []
bgpstars[s]['datasources'] = datasources
return bgpstars, star_conn, mt_conn
def get_preds(self, star):
"""
Returns a set of predicates in a BGP/star-shaped subquery
:param star: list of triple patterns
:return: list of predicates
"""
preds = [utils.getUri(tr.predicate, self.prefixes)[1:-1]
for tr in star if tr.predicate.constant]
return preds
def checkRDFTypeStatemnt(self, ltr):
types = self.getRDFTypeStatement(ltr)
typemols = {}
for t in types:
tt = utils.getUri(t.theobject, self.prefixes)[1:-1]
if tt in self.config.metadata:
mt = self.config.metadata[tt]
typemols[tt] = mt
if len(types) > 0 and len(typemols) == 0:
return {}
return typemols
def prune(self, star_conn, res_conn, selectedmolecules, stars, relevant_mts):
newselected = {}
res = {}
counter = 0
for s in selectedmolecules:
if len(selectedmolecules[s]) == 1:
newselected[s] = list(selectedmolecules[s])
res[s] = list(selectedmolecules[s])
counter += 1
else:
newselected[s] = []
res[s] = []
if counter == len(selectedmolecules):
return res
# check predicate level connections
newfilteredonly = {}
for s in res:
sc = [c for c in star_conn if s in star_conn[c]['SO']]
for c in sc:
connectingtp = [utils.getUri(tp.predicate, self.prefixes)[1:-1]
for tp in stars[s] if tp.theobject.name == c]
connectingtp = list(set(connectingtp))
sm = selectedmolecules[s]
for m in sm:
srange = [p for r in relevant_mts[m].predicates
for p in relevant_mts[m].predicates[r].ranges
if relevant_mts[m].predicates[r].predicate in connectingtp]
srange = list(set(srange).intersection(selectedmolecules[c]))
# if len(srange) == 0:
# selectedmolecules[s].remove(m)
if c in newfilteredonly:
newfilteredonly[c].extend(srange)
else:
newfilteredonly[c] = srange
newfilteredonly[c] = list(set(newfilteredonly[c]))
already_checked = []
for s in res:
sc = [c for c in star_conn if s in star_conn[c]['SO']]
for c in sc:
if s + c in already_checked or c + s in already_checked:
continue
already_checked.extend([s + c, c + s])
if c in newfilteredonly:
c_newfilter = newfilteredonly[c].copy()
else:
c_newfilter = selectedmolecules[c].copy()
newfilteredonly[c] = selectedmolecules[c].copy()
if s in newfilteredonly:
s_newfilter = newfilteredonly[s].copy()
else:
s_newfilter = selectedmolecules[s].copy()
newfilteredonly[s] = selectedmolecules[s].copy()
for m in s_newfilter:
con = res_conn[m]
if len(con) == 0:
continue
new_res = list(set(con).intersection(c_newfilter))
if len(new_res) == 0:
newfilteredonly[s].remove(m)
# for m in c_newfilter:
# con = res_conn[m]
# if len(con) == 0:
# continue
# new_res = list(set(con).intersection(s_newfilter))
# if len(new_res) == 0:
# newfilteredonly[c].remove(m)
for s in newfilteredonly:
res[s] = list(set(newfilteredonly[s]))
for s in res:
if len(res[s]) == 0:
res[s] = selectedmolecules[s]
res[s] = list(set(res[s]))
return res