def disambiguate_titles(self, source):
entities = Entity.objects.filter(source=source)
inferred_names = {}
for entity in entities:
inferred_name = entity.metadata['osm']['tags'].get('name') or entity.metadata['osm']['tags'].get('operator')
if not inferred_name:
continue
if not inferred_name in inferred_names:
inferred_names[inferred_name] = set()
inferred_names[inferred_name].add(entity)
for inferred_name, entities in inferred_names.items():
if len(entities) > 1:
for entity in entities:
if entity.metadata['osm']['tags'].get('addr:street'):
entity.title = u"%s, %s" % (inferred_name, entity.metadata['osm']['tags'].get('addr:street'))
continue
try:
name = reverse_geocode(entity.location[0], entity.location[1])[0]['name']
if name:
entity.title = u"%s, %s" % (inferred_name, name)
entity.save()
except:
self.output.write("Couldn't geocode for %s\n" % inferred_name)
def endElement(self, name):
if name in ('node','way') and self.valid:
try:
types = self.find_types(self.tags)
except ValueError:
self.ignore_count += 1
return
# Ignore ways that lay partly outside our bounding box
if name == 'way' and not all(id in self.node_locations for id in self.nodes):
return
# We already have these from OxPoints, so leave them alone.
if self.tags.get('amenity') == 'library' and self.tags.get('operator') == 'University of Oxford':
return
# Ignore disused and under-construction entities
if self.tags.get('life_cycle', 'in_use') != 'in_use' or self.tags.get('disused') in ('1', 'yes', 'true'):
return
try:
entity = Entity.objects.get(source=self.source, _identifiers__scheme='osm', _identifiers__value=self.id)
created = True
except Entity.DoesNotExist:
entity = Entity(source=self.source)
created = False
if not 'osm' in entity.metadata or entity.metadata['osm'].get('attrs', {}).get('timestamp', '') < self.attrs['timestamp']:
if created:
self.create_count += 1
else:
self.modify_count += 1
if name == 'node':
entity.location = Point(self.node_location, srid=4326)
entity.geometry = entity.location
elif name == 'way':
cls = LinearRing if self.nodes[0] == self.nodes[-1] else LineString
entity.geometry = cls([self.node_locations[n] for n in self.nodes], srid=4326)
min_, max_ = (float('inf'), float('inf')), (float('-inf'), float('-inf'))
for lon, lat in [self.node_locations[n] for n in self.nodes]:
min_ = min(min_[0], lon), min(min_[1], lat)
max_ = max(max_[0], lon), max(max_[1], lat)
entity.location = Point( (min_[0]+max_[0])/2 , (min_[1]+max_[1])/2 , srid=4326)
else:
raise AssertionError("There should be no other types of entity we're to deal with.")
try:
name = self.tags.get('name') or self.tags['operator']
except (KeyError, AssertionError):
try:
name = reverse_geocode(*entity.location)[0]['name']
if not name:
raise IndexError
name = u"↝ %s" % name
except IndexError:
name = u"↝ %f, %f" % (self.node_location[1], self.node_location[0])
entity.title = name
entity.metadata['osm'] = {
'attrs': dict(self.attrs),
'tags': self.tags
}
entity.primary_type = self.entity_types[types[0]]
if 'addr:postcode' in self.tags:
entity.post_code = self.tags['addr:postcode'].replace(' ', '')
else:
entity.post_code = ""
entity.save(identifiers={'osm': self.id})
entity.all_types = [self.entity_types[et] for et in types]
entity.update_all_types_completion()
else:
self.unchanged_count += 1
