item_tracked_import.py

from collections import defaultdict

from django.db import transaction

from .models import ProductCategory, Product, ProductMedia


class Importer:
    # keep track of all objects that were created or updated ("seen") in the import
    _item_tracker = defaultdict(set)
    
    def _import_category(self, raw_data):
        category = magic(raw_data)  # do something to create or update a category object
        for product_data in raw_data['products']:
            self._import_product(category, product_data)
        self._item_tracker[ProductCategory].add(category.pk)
        
    def _import_products(self, for_category, raw_data):
        product = magic(for_category, raw_data)  # do something to create or update a product object
        for media_data in raw_data['media']:
            self._import_product_media(product, media_data)
        self._item_tracker[Product].add(product.pk)

    def _import_product_media(self, for_product, raw_data):
        media = magic(for_product, raw_data)
        self._item_tracker[ProductMedia].add(media.pk)

    @transaction.atomic()
    def do_import(self, raw_data):
        for line in raw_data:
            self._import_category(line)

        # auto-cleanup all objects that were not "seen" for all registered models
        for model, seen_pks in self._item_tracker.items():
            model.objects.exclude(pk__in=seen_pks).delete()