Automatically match new prices with their PriceTag prediction counterpart

[raphodn]

Story

When a proof is uploaded, we now automatically extract and create related PriceTags.
But the user might be adding prices with the regular workflows (aka not using the PriceTag predictions).

To avoid the risk of having duplicate prices, we need to hide them, by changing their status + linking them to the created price.

How

• a script to manage the history: feat(Price tags): script to match historical PriceTag with their prices #650
• a check after each new price creation

[raphael0202]

To avoid duplicating the work in the new contribution assistant, I would suggest adding a new unlinked_price_tags_count column to the proof table, that represents for PRICE_TAG proofs the number of prices that are not linked to a price tag.

By doing this, we can filter in the price validator assistant only price tags associated to proofs for which we added prices through the price tag assistant only.

Detecting price duplicates on the backend seems like a waste of time for users to me, as they already checked and validated the price: we just need to link the price tag to an existing price, which may be done in another game.

[raphodn]

I was thinking of an automated way, not a user-game way :)

[raphodn]

Matching script

# for each proof, try to match generated price_tags with existing prices
# - skip proofs without price_tags or without prices
# - skip price_tags that already have a price_id or that have no predictions
# - finally loop on each price and try to match with the price_tag prediction data
for proof in Proof.objects.filter(type="PRICE_TAG").prefetch_related("prices", "price_tags", "price_tags__predictions"):
    if proof.price_tags.count() == 0:
        continue
    elif proof.prices.count() == 0:
        continue
    else:
        for price_tag in proof.price_tags.all():
            if price_tag.price_id is not None:
                continue
            elif price_tag.predictions.count() == 0:
                continue
            else:
                price_tag_prediction_data = price_tag.predictions.first().data
                for price in proof.prices.all():
                    if price.price_tags.count() > 0:
                        continue
                    elif price.type == "PRODUCT" and (price.product_code == price_tag_prediction_data["barcode"]) and (str(price.price) == str(price_tag_prediction_data["price"])):
                        price_tag.price_id = price.id
                        price_tag.status = 1
                        price_tag.save()
                        # print("match! product", proof.id, price_tag.id)
                        break
                    elif price.type == "CATEGORY" and (price.category_tag == price_tag_prediction_data["category"]) and (str(price.price) == str(price_tag_prediction_data["price"])):
                        price_tag.price_id = price.id
                        price_tag.status = 1
                        price_tag.save()
                        # print("match! category", proof.id, price_tag.id)
                        break
                    else:
                        continue

Before running the script

PriceTag.objects.count() 
# 34021
Proof.objects.filter(type="PRICE_TAG").count() 
# 11847

from collections import Counter
Counter(PriceTag.objects.all().values_list("status", flat=True))
# Counter({None: 33108, 2: 515, 1: 235, 3: 164})

PriceTag.objects.filter(price_id__isnull=False).count() 
# 233

After running the script

PriceTag.objects.filter(price_id__isnull=False).count() 
# 4592

PR

added the script as a management command: #650