SafeBite/update_bakery_dairy.py

import json, re, unicodedata

def slugify(name):
    s = unicodedata.normalize('NFKD', name).encode('ascii', 'ignore').decode('ascii').lower()
    s = re.sub(r'[^a-z0-9\s]', '', s)
    s = s.strip().replace(' ', '_')
    return s

# Items à ajouter
bakery_items = [
    ("Bagel briana", "bagel", "🥯"),
    ("Bagels", "bagel", "🥯"),
    ("Baguette", "pain", "🥖"),
    ("Biscuits breton sans gluten", "biscuit", "🍪"),
    ("Biscuits Feuille D'Érable", "biscuit", "🍁"),
    ("Biscuits sans gluten", "biscuit", "🍪"),
    ("Biscuits swiss", "biscuit", "🍪"),
    ("Bread", "pain", "🍞"),
    ("Buns", "pain", "🍞"),
    ("Crispbread", "biscuit", "🍘"),
    ("Croissant", "viennoiserie", "🥐"),
    ("Croûte à tarte", "patisserie", "🥧"),
    ("Dinner Rolls", "pain", "🍞"),
    ("Donuts", "patisserie", "🍩"),
    ("Fond de tarte", "patisserie", "🥧"),
    ("Galette tortilla", "pain", "🌯"),
    ("Gâteau mille feuilles", "patisserie", "🍰"),
    ("Gaufres", "patisserie", "🧇"),
    ("Muffins", "patisserie", "🧁"),
    ("Muffins Anglais", "pain", "🥯"),
    ("Pain Baguette", "pain", "🥖"),
    ("Pain Blanc", "pain", "🍞"),
    ("Pain Blanc Sans Lactose", "pain", "🍞"),
    ("Pain Bon Matin", "pain", "🍞"),
    ("Pain Briana", "pain", "🍞"),
    ("Pain brun", "pain", "🍞"),
    ("Pain burgers", "pain", "🍔"),
    ("Pain burgers sans gluten", "pain", "🍔"),
    ("Pain Croûte", "pain", "🍞"),
    ("Pain croûté sans gluten", "pain", "🍞"),
    ("Pain hamburger", "pain", "🍔"),
    ("Pain hamburger sans gluten", "pain", "🍔"),
    ("Pain Hot Dog", "pain", "🌭"),
    ("Pain Italien", "pain", "🥖"),
    ("Pain Sans Gluten", "pain", "🍞"),
    ("Pain Sans Lactose", "pain", "🍞"),
    ("Pain Sous Marin", "pain", "🥖"),
    ("Pains Grand Mère", "pain", "🍞"),
    ("Pancakes mix", "patisserie", "🥞"),
    ("Petit Pain Sandwich", "pain", "🥪"),
    ("Petit Pain Sous Marin", "pain", "🥖"),
    ("Pie", "patisserie", "🥧"),
    ("Pizza dough", "patisserie", "🍕"),
    ("Pizza sans gluten", "patisserie", "🍕"),
    ("Puff pastry", "patisserie", "🥐"),
    ("Pumpkin Pie", "patisserie", "🥧"),
    ("Rolls", "pain", "🍞"),
    ("Scones", "patisserie", "🧁"),
    ("Sliced bread", "pain", "🍞"),
    ("Toast", "pain", "🍞"),
    ("Toast melba", "pain", "🍞"),
    ("Tortilla", "pain", "🌯"),
    ("Tortillas", "pain", "🌯"),
    ("Vol au vent", "patisserie", "🥐"),
    ("Waffles", "patisserie", "🧇"),
]

dairy_items = [
    # Beurres, margarines, crèmes
    ("Beurre Sans Lactose", "beurre", "🧈"),
    ("Butter", "beurre", "🧈"),
    ("Margarine", "beurre", "🧈"),
    ("Cream", "creme", "🥛"),
    ("Cream cheese", "fromage", "🧀"),
    ("Crème 10%", "creme", "🥛"),
    ("Crème à café", "creme", "🥛"),
    ("Creme fraiche", "creme", "🥛"),
    ("Crème Glacée", "creme", "🍨"),
    ("Crème glacée Sans Lactose", "creme", "🍨"),
    ("Crème Glacée Sans Lactose Au Chocolat", "creme", "🍨"),
    ("Crème Sans Lactose", "creme", "🥛"),
    ("Crème Sure", "creme", "🥛"),
    ("Sour cream", "creme", "🥛"),
    # Fromages
    ("Blue cheese", "fromage", "🧀"),
    ("Brie", "fromage", "🧀"),
    ("Cheddar", "fromage", "🧀"),
    ("Cheese", "fromage", "🧀"),
    ("Cottage cheese", "fromage", "🧀"),
    ("Crotte De Fromage", "fromage", "🧀"),
    ("Demi Lune", "fromage", "🧀"),
    ("Feta", "fromage", "🧀"),
    ("Fondue Fromage", "fromage", "🧀"),
    ("Frite du petit Québec", "fromage", "🧀"),
    ("Fromage", "fromage", "🧀"),
    ("Fromage à tartiner", "fromage", "🧀"),
    ("Fromage Briana", "fromage", "🧀"),
    ("Fromage chèvre", "fromage", "🧀"),
    ("Fromage Chèvre Confiture", "fromage", "🧀"),
    ("Fromage Cottage", "fromage", "🧀"),
    ("fromage cottage sans lactose", "fromage", "🧀"),
    ("Fromage de chèvre", "fromage", "🧀"),
    ("Fromage Déjà Rappé", "fromage", "🧀"),
    ("Fromage en grains", "fromage", "🧀"),
    ("fromage en tranches", "fromage", "🧀"),
    ("Fromage Mozzarella", "fromage", "🧀"),
    ("Fromage Philadelphia", "fromage", "🧀"),
    ("Fromage Philadelphia sans lactose", "fromage", "🧀"),
    ("Fromage Quick Quick", "fromage", "🧀"),
    ("Fromage Rapé Briana", "fromage", "🧀"),
    ("Fromage rapé sans lactose", "fromage", "🧀"),
    ("Fromage Riviera", "fromage", "🧀"),
    ("Fromage Sans Lactose", "fromage", "🧀"),
    ("Fromage suisse", "fromage", "🧀"),
    ("Fromage tex mex", "fromage", "🧀"),
    ("Fromage Vache Qui Rit", "fromage", "🧀"),
    ("Gorgonzola", "fromage", "🧀"),
    ("Grated cheese", "fromage", "🧀"),
    ("Grilled cheese", "fromage", "🧀"),
    ("Mascarpone", "fromage", "🧀"),
    ("Mozzarella", "fromage", "🧀"),
    ("Parmesan", "fromage", "🧀"),
    ("Parmesan sans lactose", "fromage", "🧀"),
    ("Quark", "fromage", "🧀"),
    ("Ricotta", "fromage", "🧀"),
    # Laits et laits végétaux
    ("lait", "laitier", "🥛"),
    ("Lait avoine", "laitier", "🥛"),
    ("Lait avoine lunch", "laitier", "🥛"),
    ("Lait avoine lunch chocolat", "laitier", "🥛"),
    ("Lait Chocolat", "laitier", "🥛"),
    ("Lait Chocolat Lunch", "laitier", "🥛"),
    ("Lait Chocolat Sans Lactose", "laitier", "🥛"),
    ("Lait condensé coco", "laitier", "🥛"),
    ("Lait D'amande", "laitier", "🥛"),
    ("Lait Sans Lactose", "laitier", "🥛"),
    ("Lait Soya Lunch", "laitier", "🥛"),
    ("Lait Soya Sensationnel Nature", "laitier", "🥛"),
    ("Milk", "laitier", "🥛"),
    ("Soy Milk", "laitier", "🥛"),
    # Yogourts
    ("Soy yogurt", "yogourt", "🥣"),
    ("Yogourt iogo Vanille", "yogourt", "🥣"),
    ("Yogourt logo vanille sans lactose", "yogourt", "🥣"),
    ("Yogourt Sans Lactose", "yogourt", "🥣"),
    ("Yogurt", "yogourt", "🥣"),
]

with open('app/src/main/assets/catalog_seed.json', 'r', encoding='utf-8') as f:
    data = json.load(f)

data['version'] = 3

for domain in data['domains']:
    if domain.get('domainId') != 'grocery':
        continue
    for cat in domain.get('categories', []):
        cat_id = cat.get('categoryId')
        existing_names = {i['name'].lower() for i in cat.get('items', [])}
        existing_ids = {i['itemId'] for i in cat.get('items', [])}
        
        if cat_id == 'bakery':
            new_items = []
            for name, tag, emoji in bakery_items:
                if name.lower() in existing_names:
                    continue
                item_id = slugify(name)
                # éviter doublons d'itemId
                base_id = item_id
                suffix = 1
                while item_id in existing_ids:
                    item_id = f"{base_id}_{suffix}"
                    suffix += 1
                existing_ids.add(item_id)
                new_items.append({
                    "itemId": item_id,
                    "name": name,
                    "emoji": emoji,
                    "tags": tag
                })
            cat['items'] = cat.get('items', []) + new_items
            print(f"Bakery: added {len(new_items)} items, total {len(cat['items'])}")
            
        elif cat_id == 'dairy':
            new_items = []
            for name, tag, emoji in dairy_items:
                if name.lower() in existing_names:
                    continue
                item_id = slugify(name)
                base_id = item_id
                suffix = 1
                while item_id in existing_ids:
                    item_id = f"{base_id}_{suffix}"
                    suffix += 1
                existing_ids.add(item_id)
                new_items.append({
                    "itemId": item_id,
                    "name": name,
                    "emoji": emoji,
                    "tags": tag
                })
            cat['items'] = cat.get('items', []) + new_items
            print(f"Dairy: added {len(new_items)} items, total {len(cat['items'])}")

with open('app/src/main/assets/catalog_seed.json', 'w', encoding='utf-8') as f:
    json.dump(data, f, ensure_ascii=False, indent=0, separators=(',', ':'))
    f.write('\n')

print(f"Done, version={data['version']}")