def generate_samples(self,
beginning: str,
samples_num: int = 10) -> List[str]:
es = utils.get_elastic_engine()
return es.get(index_name=self.index_name,
phrase=beginning,
count=samples_num)
def get_indices_stats(cls,
index_name: str = 't9-index-*') -> Dict[str, Any]:
es = utils.get_elastic_engine()
return {
index_name: {
'doc_count': stat['primaries']['docs']['count'],
'size': stat['primaries']['store']['size']
}
for index_name, stat in es.get_indices_stats(
index_name=index_name).items()
}
def delete_index(self):
es = utils.get_elastic_engine()
es.delete_index(index_name=self.index_name)
def update_index(self, train_sentences: Iterable[str]):
es = utils.get_elastic_engine()
es.add_many(index_name=self.index_name, sentences=train_sentences)
def create_index(self):
es = utils.get_elastic_engine()
es.add_index(name=self.index_name,
number_of_shards=Config.ELASTIC_SHARDS_NUMBER,
number_of_replicas=Config.ELASTIC_REPLICAS_NUMBER)