How to query from a csv repeatedly throughout the day in a simulation code

I'm trying to write a simulation code in python. This simulation code relies on inputs for a large csv file, and there is a separate csv file for each day in the simulation. I need to make numerous queries (the queries are based on time, which are columns in the csv file) each simulation day.

I'm thinking of using pandas.read_csv to read this in as a dataframe, and store the result and then query from this dataframe. One coding requirement is I don't want the dataframe stored at the query site.

I think the easiest way to do this is with a class, e.g.,

import pandas as pd
class DailyCSVLoader:
  def __init__(filepath):
    self.df = pd.read_csv(filepath)
  def query(time):
    # return the rows corresponding to time

with usage:

import datetime

path = "/path/to/csv/file/filename.csv"
time = datetime.datetime(year=2020, month=1, day=1, hour=12, minute=2, second=0)
loader = DailyCSVLoader(path)
loader.query()

However, for my particular codebase, it might be slightly easier to do this outside of a class and with just a function and perhaps a static variable that holds the dataframe, e.g.,

import pandas as pd

# because I don't want the calling site to store df, I decided to keep it as a static variable here
def daily_csv_loader(filepath):
  daily_csv_loader.df = pd.read_csv(filepath)


def query(time, df):
  # return rows from df corresponding to time

with usage

import datetime

path = "/path/to/csv/file/filename.csv"
time = datetime.datetime(year=2020, month=1, day=1, hour=12, minute=2, second=0)
daily_csv_loader(filepath)
query(time, daily_csv_loader.)

Are there any other approaches here, preferably a functional approach (I would prefer not to use OOP here as alluded to previously)? Is there a functional approach that can be done with a single function, perhaps with nested functions?

An example that is functional-esque, with nested functions, doesn't expose the underlying csv/dataframe object at the callsite, ultimately invoked via a single function:

import csv
import datetime as dt
import io
from typing import Any, Dict

def load_queryable_csv(csv_str):
    # a quick & easy setup, could have used pandas instead and/or generalized parsing
    rows = csv.DictReader(io.StringIO(csv_str))
    parsed_rows = [
        {'a': row['a'], 'b': int(row['b']), 'time': dt.datetime.fromisoformat(row['time'])}
        for row in rows
    ]

    # "querying"
    def query_csv(**column_matchers):
        def matcher(row: Dict[str, Any]):
            return all(row[col] == val for col, val in column_matchers.items())

        return list(filter(matcher, parsed_rows))

    return query_csv

query_csv = load_queryable_csv("""\
a,b,time
x,2,2020-01-01 12:02:00
y,4,2020-01-01 12:02:01
""")

time = dt.datetime(year=2020, month=1, day=1, hour=12, minute=2, second=0)

query_csv(time=time)
# => [{'a': 'x', 'b': 2, 'time': datetime.datetime(2020, 1, 1, 12, 2)}]

query_csv(a='y')
# => [{'a': 'y', 'b': 4, 'time': datetime.datetime(2020, 1, 1, 12, 2, 1)}]