def _get_dbpedia_nl_record(self, dbpedia_nl_identifier):
if self.debug:
self.log.info("getting.. %s " % (dbpedia_nl_identifier))
if dbpedia_nl_identifier.lower().find("http") > -1:
if dbpedia_nl_identifier.find('http:/') > -1:
dbpedia_nl_identifier = dbpedia_nl_identifier.split(
'/')[-1].strip()
elif dbpedia_nl_identifier.find('%2F') > -1:
dbpedia_nl_identifier = dbpedia_nl_identifier.split(
'%2F')[-1].strip()
if dbpedia_nl_identifier.lower().startswith(
"dbp:") or dbpedia_nl_identifier.lower().startswith(
"dbpedia_nl:"):
dbpedia_nl_identifier = dbpedia_nl_identifier.split(':')[0]
if dbpedia_nl_identifier[0].islower():
dbpedia_nl_identifier = dbpedia_nl_identifier.title()
if dbpedia_nl_identifier.find(' ') > -1:
dbpedia_nl_identifier = dbpedia_nl_identifier.replace(" ", "_")
if dbpedia_nl_identifier.find('%20') > -1:
dbpedia_nl_identifier = dbpedia_nl_identifier.replace("%20", "_")
if dbpedia_nl_identifier.find('+') > -1 or dbpedia_nl_identifier.find(
'_') > -1:
q = ""
for item in dbpedia_nl_identifier.split('+'):
q += "+OR+prefLabel:" + item
for item in dbpedia_nl_identifier.split('_'):
q += "+OR+prefLabel:" + item
url = self.DBPEDIA_NL_URL % (dbpedia_nl_identifier.strip(), q)
else:
url = self.DBPEDIA_NL_URL % (dbpedia_nl_identifier.strip(), "+OR+"
+ dbpedia_nl_identifier.strip())
data = self.get(url)
if not data:
if self.debug:
self.log.info("No DBPedia_nl data for: %s @ %s (via %s)" %
(dbpedia_nl_identifier, url, self.backend))
return (False)
else:
if type(data) == str:
data = simplejson.loads(data)
if data["response"]["numFound"] > 0:
identifier = data["response"]["docs"][0]["id"].split('/')[-1]
record = DBPedia([identifier],
backend=self.backend,
log_path=self.log_path,
debug=self.debug)
record.execute()
self[record.keys()[0]] = record[record.keys()[0]]
s = self[record.keys()[0]].pop("sameAs")[0]
rec = sameAs([s],
backend=self.backend,
log_path=self.log_path,
debug=self.debug)
rec.execute()
self[record.keys()[0]]["same"] = rec.values()[0]["same"]
def __init__(self):
self.dbpedia = DBPedia()
self.synonyms_base = SynonymsBase()
self.tokenizer = Tokenizer()
