review-automation-server/yelp/utils.py

import datetime
import random
from time import sleep
from django.utils import timezone
from django.db.models import Max

from .models import YelpReview, YelpLocation
from .scrapper import scrape_reviews


def date_string2timezone(date):
    month, day, year = map(int,  date.split('/'))
    date = datetime.datetime(day=day, month=month, year=year)
    time_zone = timezone.make_aware(date)
    return time_zone


def get_max_date(yelp_location):
    max_date = yelp_location.yelpreview_set.all()\
        .aggregate(Max('date_posted'))['date_posted__max']
    return max_date if max_date is not None else date_string2timezone('7/2/1995')


def store_into_database(reviews, location):
    for rev in reviews:
        name = rev.get('name')
        profile = rev.get('profile')
        rating = rev.get('rating')
        date = rev.get('date_posted')
        date_posted = date_string2timezone(date)
        comment = rev.get('comment')

        # store into database
        obj, created = YelpReview.objects.update_or_create(
                reviewer_name=name,
                profile=profile,
                rating=rating,
                date_posted=date_posted,
                comment=comment,
                location=location
        )
        if created:
            print(f'A new review object has been created for {location}!')
        else:
            print(f'Review already exist')


def populate_yelp_reviews():
    yelp_locations = YelpLocation.objects.all()
    for yl in yelp_locations:
        reviews = scrape_reviews(location_url=yl.url, n_pages=1)
        if reviews:
            store_into_database(reviews, yl)
        # pause for 30-60 sec to make request more human-like.
        sleep(random.randint(30, 60))