def connect_to_ceph(ceph_bucket_prefix: str, environment: str, bucket: Optional[str] = None) -> CephStore:
"""Connect to Ceph to store SLI metrics for Thoth."""
prefix = _get_sli_metrics_prefix(ceph_bucket_prefix=ceph_bucket_prefix, environment=environment)
ceph = CephStore(prefix=prefix, bucket=bucket)
ceph.connect()
return ceph
def connect_to_ceph(
cls,
ceph_bucket_prefix: str,
processed_data_name: str,
environment: str,
bucket: Optional[str] = None,
) -> CephStore:
"""Connect to Ceph to store processed data."""
prefix = cls._get_processed_data_prefix(
ceph_bucket_prefix=ceph_bucket_prefix,
processed_data_name=processed_data_name,
environment=environment,
)
ceph = CephStore(prefix=prefix, bucket=bucket)
ceph.connect()
return ceph
import os
from typing import List
from thoth.storages import CephStore
store = CephStore(prefix='data/thoth/ash-api/ETM/words')
store.connect()
def retrieve_n_most_similar(token, n) -> List:
if not store.document_exists(token):
return {'error': f'The token: {token} is not in the vocabulary.'}, 404
token_dict = store.retrieve_document(token)
return [t for t in token_dict['most_similar'][:n]]
def get(token):
""" Retrieves the 25 most similar tokens to {token} path param """
return {
'token': token,
'most-similar-tokens': retrieve_n_most_similar(token, n=25),
}, 200