def add_temporal_relation(g):
"""
Adds a temporal argmax relation in the graph, that is only the latest/earliest entity is returned as the answer.
:param g: a graph with a non-empty edgeSet
:return: a list of suggested graphs
>>> add_temporal_relation({'edgeSet': [{'right':[2]}, {'right':[8]}], 'entities': [], 'tokens':['who', 'president']}) == \
[{'edgeSet': [{'right':[2]}, {'right':[8]}, {'type':'time', 'argmax':'time'}], 'entities': [], 'tokens':['who', 'president']}]
True
>>> add_temporal_relation({'edgeSet': [{'right':[2]}, {'right':[8], 'argmin':'time'}], 'entities': []})
[]
>>> add_temporal_relation({'edgeSet': [{'right':[2]}, {'type':'time'}], 'entities': []})
[]
>>> add_temporal_relation({'edgeSet': [{'kbID': 'P161v','type': 'direct'}],'tokens': ['where','was','<e>','assassinated','?']})
[]
"""
if len(g.get('edgeSet', [])) == 0 or graph.graph_has_temporal(g):
return []
new_graphs = []
consider_types = set()
if any(t in argmax_time_markers for t in g.get('tokens', [])):
consider_types.add('argmax')
if any(t in argmin_time_markers for t in g.get('tokens', [])):
consider_types.add('argmin')
for t in consider_types.intersection(ARG_TYPES):
new_g = graph.copy_graph(g)
new_edge = {'type': 'time', t: 'time', 'kbID': "P585v"}
new_g['edgeSet'].append(new_edge)
new_graphs.append(new_g)
return new_graphs
def last_relation_numeric(g):
"""
Adds a numeric restriction to the last relation in the graph.
:param g: a graph with a non-empty edgeSet
:return: a list of suggested graphs
>>> last_relation_numeric({'edgeSet': [{'right':[2], 'type':'direct'}, {'right':[8], 'type':'direct'}], 'entities': [{'linkings':[("Q37876", "Natalie Portman")], 'tokens':["Portman"], 'type':'PERSON'}, {'linkings': [(None, ['2012'])], 'type': 'CD', 'tokens': ['2012']}]}) == \
[{'edgeSet': [{'right':[2], 'type':'direct'}, {'right':[8], 'type':'direct', 'num': ['2012']}], 'entities': [{'linkings':[("Q37876", "Natalie Portman")], 'tokens':["Portman"], 'type':'PERSON'}]}]
True
>>> last_relation_numeric({'edgeSet': [{'right':[2]}, {'right':[8], 'argmin':'time'}], 'entities': [{'linkings': [(None, ['2012'])], 'type': 'CD', 'tokens': ['2012']}]})
[]
>>> last_relation_numeric({'edgeSet': [{'right':[2]}, {'right':[8], 'num':'2009'}], 'entities': [{'linkings': [(None, ['2012'])], 'type': 'CD', 'tokens': ['2012']}]})
[]
>>> last_relation_numeric({'edgeSet': [{'right':[2]}], 'entities': []})
[]
>>> last_relation_numeric({'edgeSet': [{'right':[2]}], 'entities': [{'linkings':[("Q37876", "Natalie Portman")], 'tokens':["Portman"], 'type':'PERSON'}]})
[]
"""
if len(g.get('edgeSet', [])) == 0 or graph.graph_has_temporal(g):
return []
if len(g.get('entities', [])) == 0 or not any(
e.get("type") == 'CD' for e in g['entities'] if len(e) > 1):
return []
entities = copy.copy(g.get('entities', []))
cd_entities = [e.get("tokens") for e in entities if e.get("type") == 'CD']
if len(cd_entities) == 0:
return []
cd_entity = cd_entities[0]
entities = [e for e in entities if e.get("tokens") != cd_entity]
new_g = graph.copy_graph(g)
new_g['entities'] = entities
edge_to_modify = graph.get_graph_last_edge(
new_g, filter_out_types={'iclass', 'v-structure', 'class'})
if len(edge_to_modify) > 0:
edge_to_modify['num'] = cd_entity
return [new_g]
return []
def last_relation_hop(g):
"""
Takes a graph with an existing relation and an intermediate variable by performing a hop-up for the second entity.
:param g: a graph with an non-empty edgeSet
:return: a list of suggested graphs
>>> last_relation_hop({'edgeSet': [], 'entities': [[4, 5, 6]]})
[]
>>> last_relation_hop({'edgeSet': [{'right':[4,5,6]}], 'entities': []}) ==\
[{'edgeSet': [{'right':[4,5,6], 'hopUp': None}], 'entities': []}, {'edgeSet': [{'right':[4,5,6], 'hopDown': None}], 'entities': []}]
True
>>> last_relation_hop({'edgeSet': [{'right':[4,5,6], 'hopUp': None}], 'entities': []})
[]
>>> last_relation_hop({'edgeSet': [{'right':[4,5,6], 'kbID': "P31v", "type": "direct"}], 'entities': []})
[]
>>> last_relation_hop({'edgeSet': [{'right':["Bahama"], "rightkbID":"Q6754", 'type':'direct'}], 'entities': []}) ==\
[{'edgeSet': [{'right':["Bahama"], "rightkbID":"Q6754", 'hopUp': None, 'type':'direct'}], 'entities': []}, {'edgeSet': [{'right':["Bahama"], "rightkbID":"Q6754", 'hopDown': None, 'type':'direct'}], 'entities': []}]
True
>>> last_relation_hop({'edgeSet': [{'right':[4,5,6], 'argmax':'time'}], 'entities': []})
[]
>>> last_relation_hop({'edgeSet': [{'right':[4,5,6], 'num':['2012']}], 'entities': []})
[]
"""
if len(g.get('edgeSet', [])) == 0 or any(
hop in g['edgeSet'][-1]
for hop in {'hopUp', 'hopDown'}) or graph.graph_has_temporal(g):
return []
new_graphs = []
for hop in HOP_TYPES:
new_g = graph.copy_graph(g)
edge_to_modify = graph.get_graph_last_edge(
new_g, filter_out_types={'iclass', 'class', 'v-structure', 'time'})
if len(edge_to_modify) > 0 and 'kbID' not in edge_to_modify:
edge_to_modify[hop] = None
new_graphs.append(new_g)
return new_graphs
def graph_to_query(g,
ask=False,
return_var_values=False,
limit=GLOBAL_RESULT_LIMIT):
"""
Convert graph to a sparql query.
:param ask: if the a simple existence of the graph should be checked instead of returning variable values.
:param g: a graph as a dictionary with non-empty edgeSet
:param return_var_values: if True the denotations for free variables will be returned
:param limit: limit on the result list size
:return: a sparql query
>>> g = {'edgeSet': [{'kbID': 'P35v', 'type': 'reverse', 'rightkbID': 'Q155', 'right': [5], 'argmax':'time'}], 'entities': []}
>>> len(query_wikidata(graph_to_query(g, return_var_values = True)))
1
>>> g = {'edgeSet': [{'kbID': 'P35v', 'type': 'reverse', 'rightkbID': 'Q155', 'right': [5]}], 'entities': []}
>>> len(query_wikidata(graph_to_query(g, return_var_values = True)))
6
>>> g = {'edgeSet': [{'right': ["Missouri"]}], 'entities': [[4]], 'tokens': ['who', 'are', 'the', 'current', 'senator', 'from', 'missouri', '?']}
>>> len(query_wikidata(graph_to_query(g, return_var_values = False)))
152
"""
variables = []
order_by = []
query = "{"
if graph.graph_has_temporal(g):
if any(edge.get('type') == 'time' for edge in g.get('edgeSet', [])):
query += sparql_temporal_values_v
else:
query += sparql_temporal_values_q
edge_set = [
e for e in g.get('edgeSet', [])
if e.get('type') != "iclass" or "canonical_right" not in e
] # implicit class edges are ignored as long as they are resolved already
for i, edge in enumerate(edge_set):
local_variables = []
if 'type' in edge:
sparql_relation_inst = sparql_relation[edge['type']]
else:
sparql_relation_inst = sparql_relation_complex
if 'kbID' in edge and len(edge['kbID']) > 1:
if edge.get('type') == 'v-structure':
sparql_relation_inst = sparql_relation_inst.replace(
"VALUES ?rv { e:P161v e:P453q e:P175q}", "")
sparql_relation_inst = sparql_relation_inst.replace(
"?rv", "e:" + edge['kbID'])
elif edge.get('type') == 'iclass':
sparql_relation_inst = sparql_relation_inst.replace(
"VALUES (?pt ?rt) { (e:P106s e:P106v) (e:P31s e:P31v) }",
"")
sparql_relation_inst = sparql_relation_inst.replace(
"?rt", "e:" + edge['kbID'])
sparql_relation_inst = sparql_relation_inst.replace(
"?pt", "e:" + edge['kbID'][:-1] + "s")
else:
if wdaccess_p.get(
'expand.transitive.relations',
False) and edge['kbID'] == "P131v" and edge.get(
'type') == "reverse":
sparql_relation_inst = sparql_relation_inst.replace(
"?p ?m . ?m ?rr", "(e:P131s/e:P131v)+")
else:
sparql_relation_inst = re.sub(r"\?r[dr]",
"e:" + edge['kbID'],
sparql_relation_inst)
elif edge.get('type') not in {'time', 'iclass'}:
sparql_relation_inst = sparql_relation_inst.replace(
"?r", "?r" + str(i))
local_variables.extend(
["?r{}{}".format(i, t[0])
for t in ['direct', 'reverse']] if 'type' not in
edge else ["?r{}{}".format(i, edge['type'][0])])
# for v in local_variables:
# sparql_relation_inst += sparql_relation_filter.replace("%relationvar%", v)
if edge.get('type') == "iclass":
if "canonical_right" not in edge:
local_variables.append("?topic")
if "kbID" not in edge:
local_variables.append("?rt")
if 'hopUp' in edge or 'hopDown' in edge:
hop = 'hopDown' if 'hopDown' in edge else 'hopUp'
sparql_hop = sparql_entity_specify if 'hopDown' in edge else sparql_entity_abstract
sparql_relation_inst = sparql_relation_inst.replace("?e2", "?e3")
sparql_relation_inst = sparql_relation_inst.replace(
"%restriction%", sparql_hop + " %restriction%")
if edge[hop]:
sparql_relation_inst = sparql_relation_inst.replace(
"?hopv", "e:" + edge[hop])
sparql_relation_inst = sparql_relation_inst.replace(
"?hops", "e:" + edge[hop][:-1] + "s")
else:
sparql_relation_inst = sparql_hopup_values + sparql_relation_inst
local_variables.append("?hop{}v".format(i))
if any(arg_type in edge for arg_type in ['argmax', 'argmin']):
sparql_relation_inst = sparql_relation_inst.replace(
"%restriction%", sparql_restriction_time_argmax)
sparql_relation_inst = sparql_relation_inst.replace(
"FILTER (YEAR(?n) = ?yearvalue)", "")
# sparql_relation_inst = sparql_relation_inst.replace("?n", "?n" + str(i))
# sparql_relation_inst = sparql_relation_inst.replace("?a", "?a" + str(i))
if return_var_values:
order_by.append("{}({})".format(
"DESC" if 'argmax' in edge else "ASC", "?n"))
limit = 1
elif 'num' in edge:
sparql_relation_inst = sparql_relation_inst.replace(
"%restriction%", sparql_restriction_time_argmax)
sparql_relation_inst = sparql_relation_inst.replace(
"?yearvalue", " ".join(edge['num']))
else:
sparql_relation_inst = sparql_relation_inst.replace(
"%restriction%", "")
sparql_relation_inst = sparql_relation_inst.replace(
"?p", "?p" + str(i))
sparql_relation_inst = sparql_relation_inst.replace(
"?m", "?m" + str(i))
sparql_relation_inst = sparql_relation_inst.replace(
"?j", "?j" + str(i))
sparql_relation_inst = sparql_relation_inst.replace(
"?e3", "?e3" + str(i))
sparql_relation_inst = sparql_relation_inst.replace(
"?hop", "?hop" + str(i))
if 'leftkbID' in edge and ask:
sparql_relation_inst = sparql_relation_inst.replace(
"?e1", "e:" + edge['leftkbID'])
if 'rightkbID' in edge:
sparql_relation_inst = sparql_relation_inst.replace(
"?e2", "e:" + edge['rightkbID'])
elif 'right' in edge:
sparql_relation_inst = sparql_relation_inst.replace(
"?e2", "?e2" + str(i))
right_label = " ".join(edge['right'])
sparql_entity_label_inst = sparql_entity_label.replace(
"VALUES ?labelright { %entitylabels }", "")
sparql_entity_label_inst = sparql_entity_label_inst.replace(
"%linkbyname%", "")
sparql_entity_label_inst = sparql_entity_label_inst.replace(
"?e2", "?e2" + str(i))
sparql_entity_label_inst = sparql_entity_label_inst.replace(
"?labelright", "\"{}\"@en".format(right_label))
local_variables.append("?e2" + str(i))
query += sparql_entity_label_inst
sparql_relation_inst = sparql_relation_inst.replace(
"_:m", "_:m" + str(i))
sparql_relation_inst = sparql_relation_inst.replace(
"_:s", "_:s" + str(i))
if edge.get('type') != "class" or all(
e.get('type') != 'v-structure' for e in g.get('edgeSet', [])):
query += sparql_relation_inst
variables.extend(local_variables)
# if not local_variables or return_var_values:
# local_variables.append('?e1')
# query = query.replace("%queryvariables%", " ".join(local_variables))
query = sparql_prefix + (sparql_select if not ask else sparql_ask) + query
# if return_var_values and all('leftkbID' not in edge for edge in g.get('edgeSet', [])) and not graph.graph_question_is_temporal(g):
# query += sparql_filter_main_entity
if return_var_values and not ask:
variables.append("?e1")
# query += "BIND (xsd:integer(SUBSTR(STR(?e1), 33)) AS ?eid)"
# order_by.append("?eid")
query += "}"
query = query.replace("%queryvariables%", " ".join(variables))
if order_by and not ask:
order_by_pattern = sparql_close_order.format(" ".join(order_by))
query += order_by_pattern
if not ask:
query += sparql_close.format(limit)
logger.debug("Querying with variables: {}".format(variables))
return query