delete_from_db.py

import os
from pymilvus import connections
from pymilvus import Collection
from datetime import date
import logging


def get_min_age_to_delete():
    BASE_DATE = date(2024, 1, 1)
    today = date.today()
    timedelta = today - BASE_DATE
    return timedelta.days - 30 # 30 days means deleting articles older than a month 

    
def get_secrets():
    uri = os.environ.get("URI")
    token = os.environ.get("TOKEN")
    collection_name = os.environ.get("COLLECTION_NAME")
    return uri, token, collection_name


def get_collection(uri: str, token: str, collection_name: str):
    connections.connect("default", uri=uri, token=token)
    collection = Collection(name=collection_name)
    print("Loaded existing collection")
    return collection        
    
def delete_from_db():
    uri, token, collection_name = get_secrets()
    collection = get_collection(uri, token, collection_name)
    min_age_to_delete = get_min_age_to_delete()
    expr = f"article_age <= {min_age_to_delete}"
    collection.delete(expr)
    logging.warning('Successfully deleted older articles')
    res = collection.query(expr="article_age > -1000", output_fields = ["article_age"])
    # dates = []
    # for x in res:
    #     dates.append(x['article_age'])
    # return dates