exporter.py

import datetime
import json

import config
from models import Tweet


def get_stat_until(datetime_until) -> tuple[int, int]:
    hour_from = datetime_until - datetime.timedelta(hours=6)
    tweets = Tweet.select().where(
        (Tweet.created_at > hour_from)
        & (Tweet.created_at < datetime_until)
        & (
            (Tweet.is_better == True & Tweet.is_worse == False)
            | (Tweet.is_worse == True & Tweet.is_better == False)
        )
    )

    is_worse, is_better = 0, 0

    # TODO Optimize it into sql query someday
    for tweet in tweets:
        is_retweet = tweet.text.startswith("RT @")

        coeficient = 1
        if is_retweet:
            coeficient = 0.20

        if tweet.is_worse:
            is_worse += coeficient
        if tweet.is_better:
            is_better += coeficient
    return is_worse, is_better


def export():
    current_time = datetime.datetime.now().replace(minute=0, second=0, microsecond=0)

    analytics_by_date = {}

    # Last 80 hours with step 4
    for i in range(0, 81, 4):
        previous_hour = current_time - datetime.timedelta(hours=i)
        is_worse, is_better = get_stat_until(previous_hour)

        count = is_worse + is_better

        if not count:
            # Looks like we don't have tweets yet
            continue

        worse_percentage = round(is_worse * 100 / count)
        better_percentage = round(is_better * 100 / count)

        hour_str = previous_hour.isoformat()
        print(hour_str, better_percentage, worse_percentage)
        analytics_by_date[hour_str] = (worse_percentage, better_percentage)

    with open(config.ANALYTICS_FILE, "w") as analytics_file:
        data = {
            "updated": datetime.datetime.now().isoformat(),
            "data": analytics_by_date,
        }
        analytics_file.write(json.dumps(data))


if __name__ == "__main__":
    export()