def _interlink(g, schema, source_database, target_database, source_graph,
target_graph):
ns_tbox_source = default_namespace_of(source_graph)[0]
ns_tbox_target = default_namespace_of(target_graph)[0]
# update namespaces
_update_namespaces(g.namespace_manager,\
('rws.schema.' + source_database.lower(), ns_tbox_source+"/"),\
('rws.schema.' + target_database.lower(), ns_tbox_target+"/"))
for relation in schema.from_database(source_database):
source_def = relation['source']
target_def = relation['target']
if target_def['database'] != target_database:
continue
# define relation
source_class = _classname_from_def(source_database, source_def)
target_class = _classname_from_def(target_database, target_def)
rel = _property_from_def(source_class, target_class,
target_def['property'])
# forward link
add_type(g, rel, URIRef(OWL + "ObjectProperty"))
add_domain(g, rel, URIRef(ns_tbox_source + "/" + source_class))
add_range(g, rel, URIRef(ns_tbox_target + "/" + target_class))
add_label(
g, rel, "Relatie tussen {} and {}".format(source_class,
target_class))
def _enrich(g, database, schema, graph):
# update namespaces
ns_otl = Namespace("http://otl.rws.nl/otl#")
ns_abox = default_namespace_of(graph)[0]
ns_tbox = _generate_tbox_namespace(graph)
_update_namespaces(g.namespace_manager,\
('rws.' + database.lower(), ns_abox),
('otl', ns_otl))
for tablename in schema.from_database(database):
source_class_name = _classname_from_def(database, tablename)
# build in-memory map
otl_map = _generate_otl_map(database, tablename, schema)
source_class = URIRef(ns_tbox + source_class_name)
for reference in otl_map.keys():
source_property = URIRef(ns_tbox + source_class_name.lower() +
'_' + reference)
referenced_class = _classname_from_def(
database, otl_map[reference]['referenced_table'])
referenced_property = URIRef(ns_tbox + referenced_class.lower() +
'_id')
q = _generate_query(source_class, source_property,
referenced_property)
for source_instance, target_id in graph.query(q):
if target_id.toPython() in otl_map[reference].keys():
add_type(
g, source_instance,
URIRef(ns_otl +
otl_map[reference][target_id.toPython()]))
def _enrich(g, database, schema, graph):
ns_tbox = _generate_tbox_namespace(graph)
# update namespaces
ns_abox = default_namespace_of(graph)[0]
_update_namespaces(g.namespace_manager,\
('rws.' + database.lower(), ns_abox),
('geo', Namespace("http://www.opengis.net/ont/geosparql#")))
for entry in schema.from_database(database):
source_def = entry['source']
target_def = entry['target']
source_class = _classname_from_def(database, source_def[0])
attributes = [definition['property'] for definition in source_def]
properties = [ property_from_mapping(ns_tbox, source_class, attr)\
for attr in attributes ]
# select target references
q = _generate_query(URIRef(ns_tbox + source_class), attributes,
properties)
for binding in graph.query(q).bindings:
source_uri = binding[Variable('source_id')]
if source_uri is None or source_uri == "":
continue
values = { '['+attr+']': binding[Variable(attr)].toPython() for attr in attributes\
if Variable(attr) in binding.keys() }
g += _generate_branch(schema, ns_abox, target_def, source_uri,
values)
def _generate_tbox_namespace(graph):
ns_match = match('(?P<base>.*/linked_data/)(?P<database>[a-z]*/)',
default_namespace_of(graph)[0])
if ns_match is None or ns_match.group('database') == "" or ns_match.group(
'base') == "":
raise Exception("Unable to determine namespace")
return Namespace(
ns_match.group('base') + "schema/" + ns_match.group('database'))
def _interlink(g, schema, source_database, target_database, source_graph,
target_graph):
ns_tbox_source = _generate_tbox_namespace(source_graph)
ns_tbox_target = _generate_tbox_namespace(target_graph)
# update namespaces
ns_abox_source = default_namespace_of(source_graph)[0]
ns_abox_target = default_namespace_of(target_graph)[0]
_update_namespaces(g.namespace_manager,\
('rws.' + source_database.lower(), ns_abox_source),\
('rws.' + target_database.lower(), ns_abox_target))
for relation in schema.from_database(source_database):
source_def = relation['source']
target_def = relation['target']
if target_def['database'] != target_database:
continue
# define relation
source_class = _classname_from_def(source_database, source_def)
target_class = _classname_from_def(target_database, target_def)
rel = _property_from_def(source_class, target_class,
target_def['property'])
# select target references
q = _generate_query(
URIRef(ns_tbox_source + source_class),
property_from_mapping(ns_tbox_source, source_class,
source_def['property']))
for source_id, target_id in source_graph.query(q):
if target_id is None or target_id == "":
continue
for match_id in target_graph.subjects(predicate=property_from_mapping(ns_tbox_target,\
target_class,\
target_def['property']),
object=target_id):
add_property(g, source_id, match_id, rel)
def import_graphs(source, target, schema):
for filename in [source, target]:
if not is_readable(filename):
raise Exception("File missing or wrong permissions: {}".format(filename))
source_graph = read(source)
target_graph = read(target)
source_namespace, source_type = default_namespace_of(source_graph)
target_namespace, target_type = default_namespace_of(target_graph)
linker = _determine_type(source_type, target_type)
databases = _determine_databases(source_namespace, target_namespace)
for _,v in databases.items():
if v not in schema:
raise Exception("Database not contained in mapping: {}".format(v))
return { 'source_graph': source_graph,
'target_graph': target_graph,
'source_name': databases['source'],
'target_name': databases['target'],
'linker': linker }
def import_graph(filename, schema):
if not is_readable(filename):
raise Exception(
"File missing or wrong permissions: {}".format(filename))
graph = read(filename)
namespace, gtype = default_namespace_of(graph)
database = _determine_database(namespace)
if database not in schema.schema.keys():
raise Exception(
"Database not contained in mapping: {}".format(database))
return {'graph': graph, 'type': gtype, 'name': database}
def run(args, timestamp):
# validate input paths
if len(args.graphs) < 2:
raise Exception("Requires at least 2 input graphs")
_check_paths(args.graphs)
# validate output path
output_path = args.output
if output_path is None:
output_path = "./merge_{}".format(timestamp)
if not is_writable(output_path):
return
print("Merging graphs...")
pi = ProgressIndicator()
pi.start()
graph = multiread(args.graphs)
update_metadata(graph, default_namespace_of(graph)[0], timestamp)
pi.stop()
# write graph
print("Writing graph to disk...")
write(graph, output_path, args.serialization_format)
