def make_get_query(update_def):
"""
Given an update_def, return the sparql query needed to produce a spreadsheet of the data to be managed.
See do_get
:return: a sparql query string
"""
from vivopump import add_qualifiers
front_query = 'SELECT ?uri ?' + ' ?'.join(update_def['column_defs'].keys()) + '\nWHERE {\n ' + \
update_def['entity_def']['entity_sparql'] + '\n'
# Fake recursion here to depth 3. Could be replaced by real recursion to arbitrary path length
middle_query = ""
for name, path in update_def['column_defs'].items():
middle_query += ' OPTIONAL { ?uri <' + str(path[0]['predicate']['ref']) + '> ?'
if len(path) == 1:
middle_query += name + ' . ' + add_qualifiers(path) + ' }\n'
else:
middle_query += path[0]['object']['name'] + ' . ?' +\
path[0]['object']['name'] + ' <' + str(path[1]['predicate']['ref']) + '> ?'
if len(path) == 2:
middle_query += name + ' . ' + add_qualifiers(path) + ' }\n'
else:
middle_query += path[1]['object']['name'] + ' . ?' +\
path[1]['object']['name'] + ' <' + str(path[2]['predicate']['ref']) + '> ?'
if len(path) == 3:
middle_query += name + ' . ' + add_qualifiers(path) + ' }\n'
else:
raise PathLengthException('Path length >3 not supported in do_get')
back_query = '}\nORDER BY ?' + update_def['entity_def']['order_by']
return front_query + middle_query + back_query
def get_step_triples(update_graph, uri, step_def, debug=True):
"""
Return the triples matching the criteria defined in the current step of an update
:param update_graph: the update graph
:param uri: uri of the entity currently the subject of an update
:param step_def: step definition from update_def
:return: Graph containing one or more triples that match the criteria for the step
"""
from rdflib import Graph
from vivopump import vivo_query, add_qualifiers, make_rdf_term
if 'qualifier' not in step_def['object']:
g = update_graph.triples((uri, step_def['predicate']['ref'], None))
else:
q = 'select (?' + step_def['object']['name'] +' as ?o) where { <' + str(uri) + '> <' + \
str(step_def['predicate']['ref']) + '> ?' + step_def['object']['name'] + ' .\n' + \
add_qualifiers([step_def]) + ' }\n'
if debug:
print "\nStep Triples Query\n", q
result_set = vivo_query(q)
g = Graph()
for binding in result_set['results']['bindings']:
o = make_rdf_term(binding['o'])
g.add((uri, step_def['predicate']['ref'], o))
if debug:
print "Step Triples", len(g)
return g
def get_step_triples(update_graph, uri, step_def, debug=True):
"""
Return the triples matching the criteria defined in the current step of an update
:param update_graph: the update graph
:param uri: uri of the entity currently the subject of an update
:param step_def: step definition from update_def
:return: Graph containing one or more triples that match the criteria for the step
"""
from rdflib import Graph
from vivopump import vivo_query, add_qualifiers, make_rdf_term
if 'qualifier' not in step_def['object']:
g = update_graph.triples((uri, step_def['predicate']['ref'], None))
else:
q = 'select (?' + step_def['object']['name'] +' as ?o) where { <' + str(uri) + '> <' + \
str(step_def['predicate']['ref']) + '> ?' + step_def['object']['name'] + ' .\n' + \
add_qualifiers([step_def]) + ' }\n'
if debug:
print "\nStep Triples Query\n", q
result_set = vivo_query(q)
g = Graph()
for binding in result_set['results']['bindings']:
o = make_rdf_term(binding['o'])
g.add((uri, step_def['predicate']['ref'], o))
if debug:
print "Step Triples", len(g)
return g
def make_get_query(update_def):
"""
Given an update_def, return the sparql query needed to produce a spreadsheet of the data to be managed.
See do_get
:return: a sparql query string
"""
from vivopump import add_qualifiers
front_query = 'SELECT ?uri ?' + ' ?'.join(update_def['column_defs'].keys()) + '\nWHERE {\n ' + \
update_def['entity_def']['entity_sparql'] + '\n'
# Fake recursion here to depth 3. Could be replaced by real recursion to arbitrary path length
middle_query = ""
for name, path in update_def['column_defs'].items():
middle_query += ' OPTIONAL { ?uri <' + str(
path[0]['predicate']['ref']) + '> ?'
if len(path) == 1:
middle_query += name + ' . ' + add_qualifiers(path) + ' }\n'
else:
middle_query += path[0]['object']['name'] + ' . ?' +\
path[0]['object']['name'] + ' <' + str(path[1]['predicate']['ref']) + '> ?'
if len(path) == 2:
middle_query += name + ' . ' + add_qualifiers(path) + ' }\n'
else:
middle_query += path[1]['object']['name'] + ' . ?' +\
path[1]['object']['name'] + ' <' + str(path[2]['predicate']['ref']) + '> ?'
if len(path) == 3:
middle_query += name + ' . ' + add_qualifiers(
path) + ' }\n'
else:
raise PathLengthException(
'Path length >3 not supported in do_get')
back_query = '}\nORDER BY ?' + update_def['entity_def']['order_by']
return front_query + middle_query + back_query
def _get_step_triples(self, uri, step_def):
"""
Return the triples matching the criteria defined in the current step of an update
:param uri: uri of the entity currently the subject of an update
:param step_def: step definition from update_def
:return: Graph containing zero or more triples that match the criteria for the step
"""
from rdflib import Graph, RDF
from vivopump import add_qualifiers, vivo_query, make_rdf_term
def step_graph(uris, pred, otype=None, graph=self.update_graph):
"""
Given a list of uri, a pred and a type, return a graph of the update_graph triples satisfying
uri pred any <- these are the returned triples
any a type
:param uris: list of uris.
:param pred: the predicate to use in selecting triples for the step_graph
:param otype: the object type to use. default in None, and no type selection will be done.
:param graph: default is update_graph. Closure sieve requires original_graph
:return: graph
"""
sg = Graph()
for suri in uris:
for obj in graph.objects(suri, pred):
if otype is None:
sg.add((suri, pred, obj))
elif (obj, RDF.type, otype) in self.update_graph:
sg.add((suri, pred, obj))
return sg
def sieve_triples(sgc, column_name):
"""
Given a step graph of triples from a closure (sgc), and the current column_name,
select the triples from the closure graph that have a path from the entity_uri to
one or more objects in the closure. If there is no path, return an empty graph.
:param sgc: the step closure graph to be "sieved"
:param column_name: the name of the column to use
:return: the sieved closure graph
"""
print "\nBeginning Closure Graph for", column_name
for (s, p, o) in sgc.triples((None, None, None)):
print s, p, o
if len(sgc) == 0:
return sgc # Nothing to sieve
else:
pred = self.update_def['column_defs'][column_name][0]['predicate']['ref']
otype = self.update_def['column_defs'][column_name][0]['object'].get('type', None)
sg = step_graph([self.entity_uri], pred, otype, graph=self.original_graph)
if len(sg) == 0 or len(self.update_def['column_defs'][column_name]) == 1:
return sg
print "step 0 graph"
for (s, p, o) in sg.triples((None, None, None)):
print s, p, o
for step in self.update_def['column_defs'][column_name][1:]:
sg = step_graph([y for y in sg.objects(None, None)], step['predicate']['ref'],
step['object'].get('type', None), graph=self.original_graph)
print "next step graph"
for (s, p, o) in sg.triples((None, None, None)):
print s, p, o
if len(sg) == 0:
return sg # column path is empty, so nothing in the closure can match
# Wait for it .... Here's the sieve. Return triples in the closure graph that have
# objects on the column graph
sgr = Graph()
for (sgcs, sgcp, sgco) in sgc.triples((None, None, None)):
if sgco in sg.objects(None, None):
sgr.add((sgcs, sgcp, sgco))
print "reduced step graph"
for (s, p, o) in sgr.triples((None, None, None)):
print s, p, o
return sgr
if 'qualifier' not in step_def['object']:
g = step_graph([uri], step_def['predicate']['ref'], step_def['object'].get('type', None))
# print "\nStep_triples for", step_def['column_name'], [uri],
# step_def['predicate']['ref'], step_def['object'].get('type', None)
for (s, p, o) in g.triples((None, None, None)):
print unicode(s), unicode(p), unicode(o)
# If the step_def is in a closure, and its the last step in the closure, then the
# closure triples must be sieved against the objects defined by the column.
if step_def['closure'] and step_def['last']:
g = sieve_triples(g, step_def['column_name'])
else:
# Handle non-specific predicates qualified by SPARQL (a rare case for VIVO-ISF)
q = 'select (?' + step_def['object']['name'] + ' as ?o) where { <' + str(uri) + '> <' + \
str(step_def['predicate']['ref']) + '> ?' + step_def['object']['name'] + ' . \n' + \
add_qualifiers([step_def]) + ' }\n'
logger.debug(u"Qualified Step Triples Query {}".format(q))
result_set = vivo_query(q, self.query_parms) # SLOW
g = Graph()
for binding in result_set['results']['bindings']:
o = make_rdf_term(binding['o'])
g.add((uri, step_def['predicate']['ref'], o))
logger.debug(u"Step Triples {}".format(g.serialize(format='nt')))
return g
