def validate_fact(indices: List[str], query: dict, fact: str):
""" Check if given fact exists in the selected indices. """
ag = ElasticAggregator(indices=indices, query=deepcopy(query))
fact_values = ag.get_fact_values_distribution(
fact, fact_name_size=choices.DEFAULT_MAX_FACT_AGGREGATION_SIZE)
if not fact_values:
raise ValidationError(
f"Fact '{fact}' not present in any of the selected indices ({indices})."
)
return True
def _get_max_class_size(self) -> int:
"""Aggregates over values of the selected fact and returns the size of the largest class."""
max_class_size = 0
fact_name = self._get_fact_name()
try:
query = json.loads(self.tagger_object.query)
except:
query = self.tagger_object.query
if fact_name:
es_aggregator = ElasticAggregator(indices=self.indices,
query=query)
facts = es_aggregator.get_fact_values_distribution(
fact_name=fact_name, fact_name_size=10, fact_value_size=10)
logging.getLogger(INFO_LOGGER).info(f"Class frequencies: {facts}")
max_class_size = max(facts.values())
return max_class_size
