def indexing_reindex_item (item):
full_url = aacore.utils.full_site_url(item.get_absolute_url())
if hasattr(item, 'get_rdf_as_stream'):
# in a way the URL here is the resource.url directly ?!
stream = item.get_rdf_as_stream()
context = RDF.Node(full_url)
RDF_MODEL.context_remove_statements(context=context)
RDF_MODEL.add_statements(stream, context=context)
else:
# # FOR DEBUGGING
# resp = utils.direct_get_response(item.get_absolute_url())
# page = resp.content
# rdfaparser = RDF.Parser("rdfa")
# furl = utils.full_site_url(item.get_absolute_url())
# s = rdfaparser.parse_string_as_stream(page, furl)
# print "RDFA:", (len(list(s))), "triples"
aacore.utils.parse_localurl_into_model(RDF_MODEL, full_url, format="rdfa", baseuri=full_url, context=full_url)
def index(self):
"""
Inspects the resource and store the information found in the RDF store.
"""
request = requests.get(self.url, prefetch=False)
# Indexes the content with the appropriate agents (sniffers)
for sniffer in sniffers:
sniffer = sniffer(request=request, model=self.dummy_model)
string = sniffer.sniff() if sniffer.test() else None
if string:
parser = RDF.Parser(name=sniffer.syntax)
parser.parse_string_into_model(self.dummy_model, string.encode("utf-8"), self.url)
# Replaces from the RDF model the existing statements with the new ones
RDF_MODEL.remove_statements_with_context(RDF.Node(self.url))
RDF_MODEL.add_statements(self.dummy_model.as_stream(), RDF.Node(self.url))
RDF_MODEL.sync()