def inspect(self, context: Context) -> PropertyValues:
""" get properties by calling on the handler """
xsd = getns('xsd')
if self.handler is None:
#handler_factory = context['handler_factory']
handler_args = context['handler_args'] # dict of kwargs
logging.debug("in inspect, cntext has: {0}".format(context))
logging.debug(
"in inspect, handler_args has: {0}".format(handler_args))
self.handler = handlers.factory(**handler_args)
#self.handler = handler_factory(**handler_args)
retval = set()
predicate = context['predicate']
if predicate == 'dct:temporal':
context.push(handler=self.handler, predicate=predicate)
t = Temporal()
t.add_to_graph(context)
context.pop(('handler', 'predicate'))
retval.add(t.uri)
elif predicate == 'dcat:byteSize':
value = self.handler.size
xsd = getns('xsd')
retval.add(Literal(value, datatype=xsd.integer))
elif predicate == 'logset:recordCount':
if self.handler.num_records is not None:
retval.add(
Literal(self.handler.num_records, datatype=xsd.integer))
elif predicate == 'logset:estRecordCount':
if self.handler.num_records is not None:
retval.add(
Literal(self.handler.num_records, datatype=xsd.integer))
else:
buf = list(self.handler.get_slice(limit=10))
avg = sum([len(entry) for entry in buf]) / len(buf)
guess = self.handler.size / avg
retval.add(Literal(guess, datatype=xsd.integer))
elif predicate == 'logset:estRecordsPerDay':
# get timespan from handler, convert to days, etc
# the handler stores the timespan just as text, so need to convert
t_earliest = dateutil.parser.parse(self.handler.t_earliest)
t_latest = dateutil.parser.parse(self.handler.t_latest)
days = (t_latest - t_earliest).total_seconds() / 86400
nrecs = self.get_one_value('logset:recordCount') or \
self.get_one_value('logset:estRecordCount')
nrecs = int(float(str(nrecs))) # sorry
nrecs_per_day = int(nrecs / days)
retval.add(Literal(nrecs_per_day, datatype=xsd.integer))
return retval
def truefalse(self, context: Context) -> PropertyValues:
prompt = self.prompts.get(context['predicate'])
choice = UI.truefalse(prompt)
response = 'true' if choice else 'false'
xsd = graph.getns('xsd')
logging.debug("truefalse returning " + response)
retval = set([Literal(response, datatype=xsd.boolean)])
return retval
def inspect(self, context: Context) -> PropertyValues:
retval = set()
predicate = context['predicate']
handler = context['handler']
xsd = getns('xsd')
if predicate == 'logset:startDate':
retval.add(Literal(handler.t_earliest, datatype=xsd.dateTime))
elif predicate == 'logset:endDate':
# note if it is a live source, enddate should be None
retval.add(Literal(handler.t_latest, datatype=xsd.dateTime))
return retval
def add_to_graph(self, context: Context = None):
if self._in_graph:
logging.debug("already in graph, skipping")
return
if context is None:
context = Context()
# describe my properties first, so subclasses can use them to generate a helpful uri if necessary:
# I think that to avoid loops we need to do all of the asking (the user)
# before doing any of the adding to graph:
triples = []
context.push(node=self)
for predicate in self.properties:
context.push(predicate=predicate)
# need to convert string eg foaf:name to an actual uri for adding
# to graph:
logging.debug("calling Graph.geturi on {0}".format(predicate))
pred_uri = graph.geturi(predicate)
logging.debug("calling get_values with {0}, {1}".format(
str(predicate), str(context)))
for v in self.get_values(predicate, context):
if isinstance(v, Identifier):
triples.append((self.uri, pred_uri, v))
elif isinstance(v, Node):
triples.append((self.uri, pred_uri, v.uri))
else:
# I'm pretty sure this should never happen
raise Exception("oh oh! " + str(v) + " ... " +
str(type(v)))
context.pop(('predicate', ))
context.pop(('node', ))
for triple in triples:
logging.debug("adding triple {0}".format(triple))
self.graph.add(triple)
# finally, describe me:
rdf = graph.getns('rdf')
myclass = graph.geturi(self.rdf_class)
logging.info("adding me to graph: {0}, {1}, {2}".format(
self.uri, str(rdf.type), str(myclass)))
self.graph.add((self.uri, rdf.type, myclass))
self._in_graph = True
def add_to_graph(self, context: Context = None):
super().add_to_graph(context)
dcat = getns('dcat')
self.graph.add((context['logset'].uri, dcat.distribution, self.uri))