cmsmanage/membershipworks/tasks/scrape.py

import logging
from datetime import datetime, timedelta

from django.conf import settings
from django.db import transaction
from django.db.models import QuerySet

from membershipworks.membershipworks_api import FieldType, MembershipWorks
from membershipworks.models import (
    Event,
    EventCategory,
    EventExt,
    Flag,
    Member,
    Transaction,
)

logger = logging.getLogger(__name__)

MAX_MEETING_TIME = timedelta(hours=6)


def flags_for_member(csv_member, all_flags, folders):
    for flag in all_flags:
        if flag.type == "folder":
            if csv_member["Account ID"] in folders[flag.id]:
                yield flag
        elif csv_member[flag.name] == flag.name:
            yield flag


def update_flags(mw_flags) -> list[Flag]:
    for typ, flags_of_type in mw_flags.items():
        for name, id in flags_of_type.items():
            flag = Flag(id=id, name=name, type=typ[:-1])
            flag.save()
            yield flag


def scrape_members(membershipworks: MembershipWorks):
    logger.info("Updating flags (labels, levels, and addons)")
    flags = list(update_flags(membershipworks._parse_flags()))

    logger.info("Getting folder membership")
    folders = {
        folder_id: membershipworks.get_member_ids([folder_name])
        for folder_name, folder_id in membershipworks._parse_flags()["folders"].items()
    }

    logger.info("Getting/Updating members...")
    members = membershipworks.get_all_members()
    for csv_member in members:
        for field in membershipworks._all_fields().values():
            # convert checkboxes to real booleans
            if (
                field.get("typ") == FieldType.CHECKBOX.value
                and field["lbl"] in csv_member
            ):
                csv_member[field["lbl"]] = csv_member[field["lbl"]] == "Y"

        # create/update member
        member = Member.from_api_dict(csv_member)
        member.clean_fields()
        member.save()
        member.flags.set(flags_for_member(csv_member, flags, folders))


def scrape_transactions(membershipworks: MembershipWorks):
    now = datetime.now()
    start_date = datetime(2010, 1, 1)
    last_transaction = Transaction.objects.order_by("timestamp").last()
    if last_transaction is not None:
        # technically this has the potential to lose
        # transactions, but it should be incredibly unlikely
        start_date = last_transaction.timestamp + timedelta(seconds=1)

    logger.info(f"Getting/Updating transactions since {start_date}...")

    transactions_csv = membershipworks.get_transactions(start_date, now)
    transactions_json = membershipworks.get_transactions(start_date, now, json=True)
    # this is terrible, but as long as the dates are the same, should be fiiiine
    transactions = [{**j, **v} for j, v in zip(transactions_csv, transactions_json)]
    assert all(
        t["Account ID"] == t.get("uid", "") and t["Payment ID"] == t.get("sid", "")
        for t in transactions
    )

    for csv_transaction in transactions:
        Transaction.from_api_dict(csv_transaction).save()


@transaction.atomic
def scrape_membershipworks(*args, **options):
    membershipworks = MembershipWorks()
    membershipworks.login(
        settings.MEMBERSHIPWORKS_USERNAME, settings.MEMBERSHIPWORKS_PASSWORD
    )

    scrape_members(membershipworks)
    scrape_transactions(membershipworks)


def scrape_event_details(queryset: QuerySet[EventExt]):
    membershipworks = MembershipWorks()
    membershipworks.login(
        settings.MEMBERSHIPWORKS_USERNAME, settings.MEMBERSHIPWORKS_PASSWORD
    )

    for event in queryset:
        event.details = membershipworks.get_event_by_eid(event.eid)
        event.save()


def scrape_events():
    membershipworks = MembershipWorks()
    membershipworks.login(
        settings.MEMBERSHIPWORKS_USERNAME, settings.MEMBERSHIPWORKS_PASSWORD
    )

    data = membershipworks.get_events_list(
        datetime.fromtimestamp(0), datetime.now() + timedelta(weeks=52), categories=True
    )
    logger.info(f"{len(data)} events retrieved!")

    for category_id, category_data in enumerate(data["_st"]["evg"]):
        category = EventCategory.from_api_dict(category_id, category_data)
        category.clean_fields()
        category.save()

    events = Event.objects.bulk_create(
        [Event.from_api_dict(event_data) for event_data in data["evt"]],
        update_conflicts=True,
        update_fields=[
            field.attname
            for field in Event._meta.get_fields()
            if not (
                field.auto_created
                or field.many_to_many
                or not field.concrete
                or field.generated
                or field.primary_key
            )
        ],
    )

    for event in events:
        try:
            event_ext = EventExt.objects.get(event_ptr=event)
        except EventExt.DoesNotExist:
            event_ext = EventExt(event_ptr=event)
            # create extension model instance
            event_ext.save_base(raw=True)
            event_ext.refresh_from_db()

        if (
            event_ext.end is not None
            and event_ext.end - event_ext.start < MAX_MEETING_TIME
        ):
            meeting_times_count = event_ext.meeting_times.count()
            if meeting_times_count == 0:
                event_ext.meeting_times.create(start=event_ext.start, end=event_ext.end)
            # if there is exactly one meeting time, it should match the event start/end
            elif meeting_times_count == 1:
                event_ext.meeting_times.update(start=event_ext.start, end=event_ext.end)

        # event has no details, or last retrieval was before the event happened
        if event_ext.details is None or event_ext.details_timestamp < (
            event_ext.end or event_ext.start
        ):
            event_ext.details = membershipworks.get_event_by_eid(event.eid)
            event_ext.save()