def start_server():
signal.signal(signal.SIGINT, signal_exit_handler)
backed_off_search_process = None
if Config.get('CTI')['knowledge_base']['on_miss_backoff']:
backed_off_search_process = Process(target=backed_off_search)
backed_off_search_process.start()
config = Config.get('server')
server.run(host=config['host'], port=config['port'], threaded=True)
if Config.get('CTI')['knowledge_base']['on_miss_backoff']:
backed_off_search_process.join()
def __get_contexts(self, titles):
contexts = []
misses = []
for title, context in self.__contexts_cache.get(titles).items():
if context is not None:
contexts.append(context)
else:
misses.append(title)
if self.__on_miss_backoff is True:
for title in misses:
SQLiteDict.storage(
Config.get('CTI')['backed_off_search']
['storage_name'])[title] = title
else:
new_contexts = self.__search_engine.contexts(misses)
contexts += list(new_contexts.values())
self.__contexts_cache.set_many(new_contexts)
return contexts
def __init__(self, cache_name):
self.__cache_name = cache_name
if ES.connection('es') is None:
Logger.log(__name__, 'could not connect to elasticsearch', type='error')
return
ESService.create_index(self.__cache_name, Config.get('elasticsearch')['indices_settings'][self.__cache_name])
def clear_backedoff_storage():
storage_path = Data.get(
Config.get('CTI')['backed_off_search']['storage_name'])
if os.path.isfile(storage_path):
os.remove(storage_path)
Logger.log(__name__, 'the backedoff search storage is now empty')
else:
Logger.log(__name__, 'the backedoff search storage is already empty')
def handle_cti_request(terms, context):
if len(terms) == 0:
return {}
scored = {}
if (Config.get('CTI')['max_processes_per_job'] == 1) or (len(terms) == 1):
for term in terms:
scored[term] = round(cti.term_informativeness(term, context), 2)
else:
with Pool(Config.get('CTI')['max_processes_per_job']) as p:
for term, score in p.map(
partial(cti_job, cti=cti, context=context), terms):
scored[term] = score
return scored
def __init__(self):
self.__config = Config.get('CTI')['knowledge_base']
self.__on_miss_backoff = self.__config['on_miss_backoff']
self.__search_engine = SearchEngine(
top_n_contexts_per_query=self.__config['max_contexts_per_query'])
self.__query_cache = QueryCache.create(self.__config['query_cache'])
self.__contexts_cache = ContextsCache.create(
self.__config['contexts_cache'])
def bootstrap_knowledge_base(dump_directory):
Logger.log(__name__, 'bootstrapping knowledge base...')
contexts_cache = ContextsCache.create(
Config.get('CTI')['knowledge_base']['contexts_cache'])
for article in WikipediaUtils.get_articles(dump_directory,
attributes=['title',
'context']):
contexts_cache.set(article['title'], article['context'])
Logger.log(__name__,
'bootstrapping got context for: ' + article['title'])
def backed_off_search():
config = Config.get('CTI')['backed_off_search']
contexts_cache = ContextsCache.create(
Config.get('CTI')['knowledge_base']['contexts_cache'])
search_engine = SearchEngine()
Logger.log(__name__, 'backed off search process started')
Logger.log(
__name__, 'backed off search storage has ' +
str(len(SQLiteDict.storage(config['storage_name']))) + ' items')
c = 0
while True:
try:
title = SQLiteDict.storage(config['storage_name']).popitem()[0]
except KeyError:
sleep(config['empty_storage_wait_seconds'])
continue
contexts_cache.set(title, search_engine.context(title))
Logger.log(__name__, 'backed off search got context for: ' + title)
c += 1
if c > 30:
c = 0
Logger.log(
__name__, 'backed off search storage has ' +
str(len(SQLiteDict.storage(config['storage_name']))) +
' items')
sleep(config['seconds_between_searches'])
def __setup(self):
self.__conns = {}
self.__configs = Config.get('elasticsearch')['connections']
def clear_knowledge_base():
ESService.delete_index(
Config.get('CTI')['knowledge_base']['query_cache']['name'])
ESService.delete_index(
Config.get('CTI')['knowledge_base']['contexts_cache']['name'])
Logger.log(__name__, 'the knowledge base is now empty')