gip_fill_recom.py

from datetime import date, datetime, timedelta
from tqdm import tqdm
from functools import reduce

DAYS_OUT = 180

def get_days_opened(clds: list[bool]) -> list[int]:
    days_opened = [-1] * len(clds)
    days_opened_cnt = 0
    
    if len(clds) <= 1:
        return
    
    for cnt in range(len(clds) - 1, -1, -1):
        if clds[cnt]:
            days_opened_cnt += 1
            days_opened[cnt] = days_opened_cnt
        else:
            days_opened_cnt = 0
            days_opened[cnt] = -1

    return days_opened

def get_prc_days_below_min(cl: ChannelListing) -> float:
    days_below_min = 0
    available_days = 0

    clds = cl.dates.filter(date__gte=date.today(), date__lte=date.today() + timedelta(days=DAYS_OUT)).all().order_by('date')
    clds_avail = list(map(lambda x: x.availability == 'available', clds))
    clds_min_stays = list(map(lambda x: x.min_stay_scraped, clds))
    clds_days_opened = get_days_opened(clds_avail)

    for cnt in range(len(clds_days_opened)):
        if clds_days_opened[cnt] == -1:
            continue
        available_days += 1
        if  not clds_min_stays[cnt]:
            continue
        if clds_days_opened[cnt] < clds_min_stays[cnt]:
            days_below_min += 1

    return days_below_min / available_days if available_days > 0 else 0

def count_instances(data: list[str]) -> dict[str, int]:
    return reduce(lambda acc, x: {**acc, x: acc.get(x, 0) + 1}, data, {})

cls = ChannelListing.objects.non_deleted().enabled().filter(posted_prices_at__gte=date.today() - timedelta(days=2))\
    .prefetch_related('master_listing')\
    .defer(*ChannelListing.deferable_fields(), *Listing.deferable_fields(prefix='master_listing__'))\
    .filter(master_listing__orphan_min_stay__isnull=True)\
    .filter(master_listing__orphan_min_stay_buffer__isnull=True)

#cls = ChannelListing.objects.filter(title='Fidalsa Moon Dunes').defer(*ChannelListing.deferable_fields())
print(cls.query)

total_items = cls.count()
cls_selected = []

with tqdm(total=total_items) as pbar:
    for cl in cls.in_batches(size=100):
        prc_days_below_min = get_prc_days_below_min(cl)
        if prc_days_below_min > 0.05:
            cls_selected.append([cl.id, cl.master_listing_id, prc_days_below_min])
        pbar.update(1)

with open(f'report_{datetime.now().strftime("%Y-%m-%d--%H-%M")}', 'w') as output:
    output.write('Channel Listing ID, Master Listing ID, Prc Days Below Min\n')
    for cl in cls_selected:
        output.write(f'{cl[0]}, {cl[1]}, {cl[2]}\n')