import argparse
from pathlib import Path

import polars as pl


DROP_CATEGORIES = {
    "amenity/advice",
    "amenity/atm",
    "amenity/bbq",
    "amenity/bench",
    "amenity/bicycle_parking",
    "amenity/clock",
    "amenity/fixme",
    "amenity/grit_bin",
    "amenity/hunting_stand",
    "amenity/motorcycle_parking",
    "amenity/notice_board",
    "amenity/parking",
    "amenity/parking_entrance",
    "amenity/parking_space",
    "amenity/post_box",
    "amenity/telephone",
    "amenity/toilets",
    "amenity/vacuum_cleaner",
    "amenity/waste_basket",
    "building/air_shaft",
    "building/apartments",
    "building/detached",
    "building/entrance",
    "building/entry",
    "building/garage",
    "building/garages",
    "building/house",
    "building/hut",
    "building/no",
    "building/office",
    "building/public",
    "building/residential",
    "building/roof",
    "building/shed",
    "building/terrace",
    "building/yes",
    "emergency/access_point",
    "emergency/ambulance_station",
    "emergency/assembly_point",
    "emergency/bleed_control_kit",
    "emergency/defibrillator",
    "emergency/designated",
    "emergency/dry_riser_inlet",
    "emergency/emergency_ward_entrance",
    "emergency/fire_alarm_box",
    "emergency/fire_extinguisher",
    "emergency/fire_hydrant",
    "emergency/fire_service_inlet",
    "emergency/first_aid_kit",
    "emergency/life_ring",
    "emergency/lifeguard",
    "emergency/no",
    "emergency/phone",
    "emergency/rescue_equipment",
    "emergency/siren",
    "emergency/throw_bag",
    "emergency/water_rescue",
    "emergency/yes",
    "leisure/firepit",
    "leisure/fishing",
    "leisure/picnic_table",
    "office/company",
    "office/yes",
    "tourism/apartment",
    "tourism/apartments",
    "tourism/camp_pitch",
    "tourism/information",
    "tourism/village_sign",
    "tourism/yes",
}

# (friendly_name, emoji) for every category we keep
CATEGORY_MAP: dict[str, tuple[str, str]] = {
    # amenity
    "amenity/animal_boarding": ("Animal Boarding", "🐾"),
    "amenity/animal_breeding": ("Animal Breeding", "🐣"),
    "amenity/animal_shelter": ("Animal Shelter", "🏠"),
    "amenity/arts_centre": ("Arts Centre", "🎨"),
    "amenity/bank": ("Bank", "🏦"),
    "amenity/bar": ("Bar", "🍸"),
    "amenity/bicycle_rental": ("Bike Rental", "🚲"),
    "amenity/bicycle_repair_station": ("Bike Repair", "🔧"),
    "amenity/binoculars": ("Public Binoculars", "🔭"),
    "amenity/boat_rental": ("Boat Rental", "⛵"),
    "amenity/boat_storage": ("Boat Storage", "🚢"),
    "amenity/boot_scraper": ("Boot Scraper", "🥾"),
    "amenity/bureau_de_change": ("Currency Exchange", "💱"),
    "amenity/bus_station": ("Bus Station", "🚌"),
    "amenity/cafe": ("Café", "☕"),
    "amenity/car_rental": ("Car Rental", "🚗"),
    "amenity/car_sharing": ("Car Sharing", "🚙"),
    "amenity/car_wash": ("Car Wash", "🧽"),
    "amenity/care_home": ("Care Home", "🏥"),
    "amenity/casino": ("Casino", "🎰"),
    "amenity/charging_station": ("EV Charging", "🔌"),
    "amenity/check_in": ("Check-In Point", "✅"),
    "amenity/childcare": ("Childcare", "👶"),
    "amenity/cinema": ("Cinema", "🎬"),
    "amenity/clinic": ("Clinic", "🩺"),
    "amenity/club": ("Club", "🏛️"),
    "amenity/college": ("College", "🎓"),
    "amenity/community_centre": ("Community Centre", "🤝"),
    "amenity/compressed_air": ("Compressed Air", "💨"),
    "amenity/conference_centre": ("Conference Centre", "📋"),
    "amenity/courthouse": ("Courthouse", "⚖️"),
    "amenity/coworking_space": ("Co-working Space", "💻"),
    "amenity/crematorium": ("Crematorium", "🕯️"),
    "amenity/dancing_school": ("Dance School", "💃"),
    "amenity/dentist": ("Dentist", "🦷"),
    "amenity/doctors": ("Doctor", "👨‍⚕️"),
    "amenity/dojo": ("Dojo", "🥋"),
    "amenity/donation_box": ("Donation Box", "📦"),
    "amenity/dressing_room": ("Dressing Room", "👗"),
    "amenity/drinking_water": ("Drinking Water", "🚰"),
    "amenity/driving_school": ("Driving School", "🚦"),
    "amenity/escooter_rental": ("E-Scooter Rental", "🛴"),
    "amenity/events_venue": ("Events Venue", "🎪"),
    "amenity/fast_food": ("Fast Food", "🍔"),
    "amenity/feeding_place": ("Feeding Place", "🍽️"),
    "amenity/ferry_terminal": ("Ferry Terminal", "⛴️"),
    "amenity/fire_station": ("Fire Station", "🚒"),
    "amenity/food_court": ("Food Court", "🍴"),
    "amenity/fountain": ("Fountain", "⛲"),
    "amenity/fuel": ("Fuel Station", "⛽"),
    "amenity/gambling": ("Gambling", "🎲"),
    "amenity/grave_yard": ("Graveyard", "🪦"),
    "amenity/hall": ("Hall", "🏛️"),
    "amenity/hookah_lounge": ("Hookah Lounge", "💨"),
    "amenity/hospital": ("Hospital", "🏥"),
    "amenity/ice_cream": ("Ice Cream", "🍦"),
    "amenity/internet_cafe": ("Internet Café", "🌐"),
    "amenity/kick-scooter_rental": ("Kick Scooter Rental", "🛴"),
    "amenity/kindergarten": ("Kindergarten", "💒"),
    "amenity/language_school": ("Language School", "🗣️"),
    "amenity/letter_box": ("Letter Box", "📮"),
    "amenity/library": ("Library", "📚"),
    "amenity/loading_dock": ("Loading Dock", "📥"),
    "amenity/lounge": ("Lounge", "🛋️"),
    "amenity/lounger": ("Public Lounger", "🪑"),
    "amenity/marketplace": ("Market", "🛒"),
    "amenity/money_transfer": ("Money Transfer", "💸"),
    "amenity/mounting_block": ("Mounting Block", "🐴"),
    "amenity/music_school": ("Music School", "🎵"),
    "amenity/music_venue": ("Music Venue", "🎶"),
    "amenity/nightclub": ("Nightclub", "🪩"),
    "amenity/nursing_home": ("Nursing Home", "🏠"),
    "amenity/parcel_locker": ("Parcel Locker", "📦"),
    "amenity/payment_terminal": ("Payment Terminal", "💳"),
    "amenity/pharmacy": ("Pharmacy", "💊"),
    "amenity/photo_booth": ("Photo Booth", "📸"),
    "amenity/piano": ("Public Piano", "🎹"),
    "amenity/place_of_worship": ("Place of Worship", "⛪"),
    "amenity/police": ("Police Station", "🚔"),
    "amenity/post_depot": ("Post Depot", "📬"),
    "amenity/post_office": ("Post Office", "🏤"),
    "amenity/prep_school": ("Prep School", "📖"),
    "amenity/pub": ("Pub", "🍺"),
    "amenity/public_bookcase": ("Public Bookcase", "📕"),
    "amenity/public_building": ("Public Building", "🏢"),
    "amenity/reception_desk": ("Reception Desk", "🛎️"),
    "amenity/recycling": ("Recycling", "♻️"),
    "amenity/restaurant": ("Restaurant", "🍽️"),
    "amenity/sanitary_dump_station": ("Sanitary Dump Station", "🚿"),
    "amenity/school": ("School", "🏫"),
    "amenity/scout_hut": ("Scout Hut", "⚜️"),
    "amenity/shelter": ("Shelter", "🛖"),
    "amenity/shower": ("Public Shower", "🚿"),
    "amenity/smoking_area": ("Smoking Area", "🚬"),
    "amenity/social_centre": ("Social Centre", "🏘️"),
    "amenity/social_club": ("Social Club", "🤝"),
    "amenity/social_facility": ("Social Facility", "🫂"),
    "amenity/stripclub": ("Strip Club", "🔞"),
    "amenity/studio": ("Studio", "🎙️"),
    "amenity/table": ("Public Table", "🪑"),
    "amenity/taxi": ("Taxi Stand", "🚕"),
    "amenity/telescope": ("Public Telescope", "🔭"),
    "amenity/theatre": ("Theatre", "🎭"),
    "amenity/ticket_validator": ("Ticket Validator", "🎫"),
    "amenity/townhall": ("Town Hall", "🏛️"),
    "amenity/training": ("Training Centre", "📝"),
    "amenity/trolley_bay": ("Trolley Bay", "🛒"),
    "amenity/university": ("University", "🏫"),
    "amenity/vehicle_inspection": ("Vehicle Inspection", "🔍"),
    "amenity/vending_machine": ("Vending Machine", "🏧"),
    "amenity/veterinary": ("Vet", "🐕"),
    "amenity/washing_machine": ("Washing Machine", "🧺"),
    "amenity/washingline": ("Washing Line", "👕"),
    "amenity/waste_disposal": ("Waste Disposal", "🗑️"),
    "amenity/waste_transfer_station": ("Waste Transfer Station", "🚛"),
    "amenity/water_point": ("Water Point", "💧"),
    "amenity/watering_place": ("Watering Place", "🚰"),
    "amenity/weighbridge": ("Weighbridge", "⚖️"),
    # building
    "building/barn": ("Barn", "🏚️"),
    "building/bunker": ("Bunker", "🏗️"),
    "building/chapel": ("Chapel", "⛪"),
    "building/church": ("Church", "⛪"),
    "building/commercial": ("Commercial Building", "🏬"),
    "building/construction": ("Construction Site", "🚧"),
    "building/farm": ("Farmhouse", "🌾"),
    "building/greenhouse": ("Greenhouse", "🌿"),
    "building/industrial": ("Industrial Building", "🏭"),
    "building/kiosk": ("Kiosk", "🏪"),
    "building/retail": ("Retail Building", "🏬"),
    "building/ruins": ("Ruins", "🏚️"),
    "building/school": ("School Building", "🏫"),
    "building/semidetached_house": ("Semi-Detached House", "🏠"),
    "building/service": ("Service Building", "🔧"),
    "building/university": ("University Building", "🎓"),
    "building/warehouse": ("Warehouse", "🏭"),
    # craft
    "craft/agricultural_engines": ("Agricultural Engines", "🚜"),
    "craft/atelier": ("Atelier", "🎨"),
    "craft/blacksmith": ("Blacksmith", "🔨"),
    "craft/bookbinder": ("Bookbinder", "📖"),
    "craft/brewery": ("Brewery", "🍺"),
    "craft/builder": ("Builder", "🧱"),
    "craft/carpenter": ("Carpenter", "🪚"),
    "craft/caterer": ("Caterer", "🍱"),
    "craft/cleaning": ("Cleaning Service", "🧹"),
    "craft/confectionery": ("Confectioner", "🍬"),
    "craft/distillery": ("Distillery", "🥃"),
    "craft/dressmaker": ("Dressmaker", "👗"),
    "craft/electrician": ("Electrician", "⚡"),
    "craft/electronics_repair": ("Electronics Repair", "🔌"),
    "craft/floorer": ("Flooring Specialist", "🪵"),
    "craft/gardener": ("Gardener", "🌱"),
    "craft/glaziery": ("Glazier", "🪟"),
    "craft/handicraft": ("Handicraft", "✂️"),
    "craft/hvac": ("HVAC", "❄️"),
    "craft/jeweller": ("Jeweller", "💎"),
    "craft/joiner": ("Joiner", "🪚"),
    "craft/key_cutter": ("Key Cutter", "🔑"),
    "craft/locksmith": ("Locksmith", "🔐"),
    "craft/metal_construction": ("Metal Fabrication", "🔩"),
    "craft/painter": ("Painter & Decorator", "🖌️"),
    "craft/photographer": ("Photographer", "📷"),
    "craft/photographic_laboratory": ("Photo Lab", "🖼️"),
    "craft/plumber": ("Plumber", "🔧"),
    "craft/pottery": ("Pottery", "🏺"),
    "craft/printer": ("Printer", "🖨️"),
    "craft/roofer": ("Roofer", "🏠"),
    "craft/sawmill": ("Sawmill", "🪵"),
    "craft/scaffolder": ("Scaffolder", "🏗️"),
    "craft/sculptor": ("Sculptor", "🗿"),
    "craft/shoemaker": ("Shoemaker", "👞"),
    "craft/signmaker": ("Sign Maker", "🪧"),
    "craft/stonemason": ("Stonemason", "🪨"),
    "craft/tailor": ("Tailor", "🧵"),
    "craft/upholsterer": ("Upholsterer", "🛋️"),
    "craft/watchmaker": ("Watchmaker", "⌚"),
    "craft/window_construction": ("Window Fitter", "🪟"),
    "craft/winery": ("Winery", "🍷"),
    "craft/yes": ("Craft Workshop", "🛠️"),
    # healthcare
    "healthcare/alternative": ("Alternative Medicine", "🌿"),
    "healthcare/audiologist": ("Audiologist", "👂"),
    "healthcare/centre": ("Health Centre", "🏥"),
    "healthcare/clinic": ("Health Clinic", "🩺"),
    "healthcare/counselling": ("Counselling", "🧠"),
    "healthcare/dentist": ("Dental Practice", "🦷"),
    "healthcare/doctor": ("GP Surgery", "👨‍⚕️"),
    "healthcare/hospital": ("Hospital", "🏥"),
    "healthcare/laboratory": ("Medical Lab", "🔬"),
    "healthcare/optometrist": ("Optometrist", "👁️"),
    "healthcare/pharmacy": ("Pharmacy", "💊"),
    "healthcare/physiotherapist": ("Physiotherapist", "🏃"),
    "healthcare/podiatrist": ("Podiatrist", "🦶"),
    "healthcare/psychotherapist": ("Psychotherapist", "🧠"),
    "healthcare/rehabilitation": ("Rehabilitation Centre", "♿"),
    "healthcare/vaccination_centre": ("Vaccination Centre", "💉"),
    "healthcare/yes": ("Healthcare Facility", "🏥"),
    # leisure
    "leisure/adult_gaming_centre": ("Adult Gaming Centre", "🎮"),
    "leisure/amusement_arcade": ("Amusement Arcade", "🕹️"),
    "leisure/bandstand": ("Bandstand", "🎺"),
    "leisure/bathing_place": ("Bathing Spot", "🏖️"),
    "leisure/bird_hide": ("Bird Hide", "🐦"),
    "leisure/bowling_alley": ("Bowling Alley", "🎳"),
    "leisure/common": ("Common Land", "🌳"),
    "leisure/dance": ("Dance Venue", "💃"),
    "leisure/dog_park": ("Dog Park", "🐕"),
    "leisure/escape_game": ("Escape Room", "🔓"),
    "leisure/fitness_centre": ("Gym", "🏋️"),
    "leisure/fitness_station": ("Outdoor Gym", "💪"),
    "leisure/garden": ("Garden", "🌷"),
    "leisure/golf_course": ("Golf Course", "⛳"),
    "leisure/hackerspace": ("Hackerspace", "💻"),
    "leisure/horse_riding": ("Horse Riding", "🐎"),
    "leisure/indoor_play": ("Indoor Play Area", "🧒"),
    "leisure/marina": ("Marina", "⚓"),
    "leisure/miniature_golf": ("Mini Golf", "⛳"),
    "leisure/nature_reserve": ("Nature Reserve", "🦔"),
    "leisure/outdoor_seating": ("Outdoor Seating", "🪑"),
    "leisure/park": ("Park", "🌳"),
    "leisure/pitch": ("Sports Pitch", "⚽"),
    "leisure/playground": ("Playground", "🛝"),
    "leisure/sauna": ("Sauna", "🧖"),
    "leisure/slipway": ("Slipway", "🚤"),
    "leisure/social_club": ("Social Club", "🍻"),
    "leisure/sports_centre": ("Sports Centre", "🏟️"),
    "leisure/sports_hall": ("Sports Hall", "🏀"),
    "leisure/swimming_pool": ("Swimming Pool", "🏊"),
    "leisure/tanning_salon": ("Tanning Salon", "☀️"),
    "leisure/track": ("Running Track", "🏃"),
    "leisure/trampoline_park": ("Trampoline Park", "🤸"),
    "leisure/water_park": ("Water Park", "🌊"),
    "leisure/wildlife_hide": ("Wildlife Hide", "🦌"),
    "leisure/yes": ("Leisure Facility", "🎉"),
    # office
    "office/accountant": ("Accountant", "🧮"),
    "office/advertising_agency": ("Advertising Agency", "📢"),
    "office/architect": ("Architect", "📐"),
    "office/association": ("Association", "🏛️"),
    "office/charity": ("Charity", "❤️"),
    "office/construction_company": ("Construction Company", "🏗️"),
    "office/consulting": ("Consulting Firm", "📊"),
    "office/courier": ("Courier Service", "📦"),
    "office/coworking": ("Co-working Space", "💻"),
    "office/design": ("Design Studio", "🎨"),
    "office/diplomatic": ("Diplomatic Office", "🏛️"),
    "office/educational_institution": ("Education Office", "🎓"),
    "office/employment_agency": ("Employment Agency", "💼"),
    "office/energy_supplier": ("Energy Supplier", "⚡"),
    "office/engineer": ("Engineering Firm", "⚙️"),
    "office/estate_agent": ("Estate Agent", "🏠"),
    "office/financial": ("Financial Services", "💰"),
    "office/financial_advisor": ("Financial Advisor", "📈"),
    "office/foundation": ("Foundation", "🏛️"),
    "office/government": ("Government Office", "🏛️"),
    "office/graphic_design": ("Graphic Design", "🖌️"),
    "office/healthcare": ("Healthcare Office", "🏥"),
    "office/home_care": ("Home Care Service", "🏠"),
    "office/insurance": ("Insurance", "🛡️"),
    "office/interior_design": ("Interior Design", "🛋️"),
    "office/it": ("IT Company", "💻"),
    "office/lawyer": ("Lawyer", "⚖️"),
    "office/logistics": ("Logistics", "🚚"),
    "office/marketing": ("Marketing Agency", "📣"),
    "office/mortgage": ("Mortgage Broker", "🏦"),
    "office/moving_company": ("Moving Company", "📦"),
    "office/newspaper": ("Newspaper Office", "📰"),
    "office/ngo": ("NGO", "🌍"),
    "office/notary": ("Notary", "📜"),
    "office/political_party": ("Political Party", "🗳️"),
    "office/politician": ("Politician Office", "🏛️"),
    "office/property_management": ("Property Management", "🏘️"),
    "office/recruitment": ("Recruitment Agency", "👥"),
    "office/religion": ("Religious Office", "✝️"),
    "office/research": ("Research Office", "🔬"),
    "office/security": ("Security Company", "🔒"),
    "office/solicitor": ("Solicitor", "⚖️"),
    "office/surveyor": ("Surveyor", "📏"),
    "office/tax_advisor": ("Tax Advisor", "🧾"),
    "office/taxi": ("Taxi Office", "🚕"),
    "office/telecommunication": ("Telecoms Office", "📡"),
    "office/therapist": ("Therapist", "🧠"),
    "office/travel_agent": ("Travel Agent", "✈️"),
    "office/union": ("Trade Union", "✊"),
    "office/university": ("University Office", "🎓"),
    "office/vacant": ("Vacant Office", "🏚️"),
    "office/web_design": ("Web Design", "🌐"),
    # public_transport
    "public_transport/entrance": ("Transport Entrance", "🚪"),
    "public_transport/platform": ("Platform", "🚉"),
    "public_transport/station": ("Station", "🚉"),
    "public_transport/stop_position": ("Stop", "🚏"),
    # shop
    "shop/accessories": ("Accessories Shop", "👜"),
    "shop/agrarian": ("Farm Supply Shop", "🌾"),
    "shop/alcohol": ("Off-Licence", "🍷"),
    "shop/antiques": ("Antiques Shop", "🏺"),
    "shop/appliance": ("Appliance Shop", "🔌"),
    "shop/art": ("Art Shop", "🎨"),
    "shop/baby_goods": ("Baby Shop", "🍼"),
    "shop/bag": ("Bag Shop", "👜"),
    "shop/bakery": ("Bakery", "🥐"),
    "shop/bathroom": ("Bathroom Shop", "🛁"),
    "shop/bathroom_furnishing": ("Bathroom Furnishings", "🚿"),
    "shop/beauty": ("Beauty Shop", "💄"),
    "shop/bed": ("Bed Shop", "🛏️"),
    "shop/beverages": ("Drinks Shop", "🥤"),
    "shop/bicycle": ("Bike Shop", "🚲"),
    "shop/boat": ("Boat Shop", "⛵"),
    "shop/bookmaker": ("Bookmaker", "🏇"),
    "shop/books": ("Bookshop", "📚"),
    "shop/boutique": ("Boutique", "👗"),
    "shop/building_materials": ("Building Materials", "🧱"),
    "shop/butcher": ("Butcher", "🥩"),
    "shop/camera": ("Camera Shop", "📷"),
    "shop/candles": ("Candle Shop", "🕯️"),
    "shop/car": ("Car Dealership", "🚗"),
    "shop/car;car_repair": ("Car Sales & Repair", "🚗"),
    "shop/car_parts": ("Car Parts", "🔩"),
    "shop/car_repair": ("Car Repair", "🔧"),
    "shop/caravan": ("Caravan Dealer", "🚐"),
    "shop/carpet": ("Carpet Shop", "🧶"),
    "shop/catalogue": ("Catalogue Shop", "📋"),
    "shop/charity": ("Charity Shop", "❤️"),
    "shop/cheese": ("Cheese Shop", "🧀"),
    "shop/chemist": ("Chemist", "🧪"),
    "shop/chocolate": ("Chocolate Shop", "🍫"),
    "shop/clothes": ("Clothes Shop", "👕"),
    "shop/coffee": ("Coffee Shop", "☕"),
    "shop/collector": ("Collector Shop", "🏆"),
    "shop/computer": ("Computer Shop", "🖥️"),
    "shop/confectionery": ("Sweet Shop", "🍬"),
    "shop/convenience": ("Convenience Store", "🏪"),
    "shop/copyshop": ("Copy Shop", "🖨️"),
    "shop/cosmetics": ("Cosmetics Shop", "💅"),
    "shop/country_store": ("Country Store", "🏡"),
    "shop/craft": ("Craft Shop", "✂️"),
    "shop/curtain": ("Curtain Shop", "🪟"),
    "shop/dairy": ("Dairy Shop", "🥛"),
    "shop/deli": ("Delicatessen", "🧆"),
    "shop/department_store": ("Department Store", "🏬"),
    "shop/discount": ("Discount Store", "💲"),
    "shop/doityourself": ("DIY Store", "🔨"),
    "shop/doors": ("Door Shop", "🚪"),
    "shop/dry_cleaning": ("Dry Cleaner", "👔"),
    "shop/e-cigarette": ("Vape Shop", "💨"),
    "shop/electrical": ("Electrical Shop", "⚡"),
    "shop/electronics": ("Electronics Shop", "📱"),
    "shop/erotic": ("Adult Shop", "🔞"),
    "shop/esoteric": ("Esoteric Shop", "🔮"),
    "shop/estate_agent": ("Estate Agent", "🏠"),
    "shop/fabric": ("Fabric Shop", "🧵"),
    "shop/fan": ("Fan Shop", "🏅"),
    "shop/farm": ("Farm Shop", "🥕"),
    "shop/fashion_accessories": ("Fashion Accessories", "👒"),
    "shop/fireplace": ("Fireplace Shop", "🔥"),
    "shop/fishing": ("Fishing Shop", "🎣"),
    "shop/flooring": ("Flooring Shop", "🪵"),
    "shop/florist": ("Florist", "💐"),
    "shop/food": ("Food Shop", "🍞"),
    "shop/frame": ("Framing Shop", "🖼️"),
    "shop/frozen_food": ("Frozen Food Shop", "🧊"),
    "shop/fuel": ("Fuel Shop", "⛽"),
    "shop/funeral_directors": ("Funeral Director", "⚰️"),
    "shop/furniture": ("Furniture Shop", "🪑"),
    "shop/games": ("Games Shop", "🎮"),
    "shop/garden_centre": ("Garden Centre", "🌻"),
    "shop/gas": ("Gas Shop", "🔥"),
    "shop/general": ("General Store", "🏪"),
    "shop/gift": ("Gift Shop", "🎁"),
    "shop/glaziery": ("Glazier", "🪟"),
    "shop/greengrocer": ("Greengrocer", "🥬"),
    "shop/grocery": ("Grocery Shop", "🛒"),
    "shop/haberdashery": ("Haberdashery", "🧵"),
    "shop/hairdresser": ("Hairdresser", "💇"),
    "shop/hairdresser_supply": ("Hairdresser Supply", "💇"),
    "shop/hardware": ("Hardware Shop", "🔩"),
    "shop/health": ("Health Shop", "🌿"),
    "shop/health_food": ("Health Food Shop", "🥗"),
    "shop/hearing_aids": ("Hearing Aid Shop", "👂"),
    "shop/herbalist": ("Herbalist", "🌿"),
    "shop/hifi": ("Hi-Fi Shop", "🔊"),
    "shop/household": ("Household Shop", "🏠"),
    "shop/household_linen": ("Linen Shop", "🛏️"),
    "shop/houseware": ("Houseware Shop", "🍳"),
    "shop/ice_cream": ("Ice Cream Shop", "🍦"),
    "shop/interior_decoration": ("Interior Decoration", "🖼️"),
    "shop/jewelry": ("Jewellery Shop", "💍"),
    "shop/kiosk": ("Kiosk", "🏪"),
    "shop/kitchen": ("Kitchen Shop", "🍳"),
    "shop/laundry": ("Laundry", "🧺"),
    "shop/leather": ("Leather Shop", "🧳"),
    "shop/lighting": ("Lighting Shop", "💡"),
    "shop/locksmith": ("Locksmith", "🔐"),
    "shop/mall": ("Shopping Centre", "🏬"),
    "shop/massage": ("Massage Parlour", "💆"),
    "shop/medical_supply": ("Medical Supply", "🩺"),
    "shop/military_surplus": ("Military Surplus", "🎖️"),
    "shop/mobile_phone": ("Mobile Phone Shop", "📱"),
    "shop/mobile_phone_accessories": ("Phone Accessories", "📱"),
    "shop/mobility": ("Mobility Shop", "♿"),
    "shop/mobility_scooter": ("Mobility Scooter Shop", "🦽"),
    "shop/model": ("Model Shop", "✈️"),
    "shop/money_lender": ("Money Lender", "💰"),
    "shop/motorcycle": ("Motorcycle Shop", "🏍️"),
    "shop/motorcycle_repair": ("Motorcycle Repair", "🔧"),
    "shop/music": ("Music Shop", "🎵"),
    "shop/musical_instrument": ("Musical Instrument Shop", "🎸"),
    "shop/newsagent": ("Newsagent", "📰"),
    "shop/nutrition_supplements": ("Nutrition Shop", "💪"),
    "shop/optician": ("Optician", "👓"),
    "shop/outdoor": ("Outdoor Shop", "🏕️"),
    "shop/outpost": ("Outpost", "📦"),
    "shop/paint": ("Paint Shop", "🎨"),
    "shop/party": ("Party Shop", "🎈"),
    "shop/pastry": ("Pastry Shop", "🥐"),
    "shop/pawnbroker": ("Pawnbroker", "💰"),
    "shop/perfumery": ("Perfumery", "🌸"),
    "shop/pet": ("Pet Shop", "🐾"),
    "shop/pet_grooming": ("Pet Grooming", "🐩"),
    "shop/photo": ("Photo Shop", "📸"),
    "shop/piercing": ("Piercing Studio", "💎"),
    "shop/plant_hire": ("Plant Hire", "🚜"),
    "shop/pottery": ("Pottery Shop", "🏺"),
    "shop/printer_ink": ("Ink & Toner Shop", "🖨️"),
    "shop/printing": ("Print Shop", "🖨️"),
    "shop/psychic": ("Psychic", "🔮"),
    "shop/pyrotechnics": ("Fireworks Shop", "🎆"),
    "shop/religion": ("Religious Shop", "✝️"),
    "shop/rental": ("Rental Shop", "🔑"),
    "shop/repair": ("Repair Shop", "🔧"),
    "shop/scuba_diving": ("Scuba Diving Shop", "🤿"),
    "shop/seafood": ("Fishmonger", "🐟"),
    "shop/second_hand": ("Second-Hand Shop", "♻️"),
    "shop/security": ("Security Shop", "🔒"),
    "shop/sewing": ("Sewing Shop", "🪡"),
    "shop/shoe_repair": ("Shoe Repair", "👞"),
    "shop/shoes": ("Shoe Shop", "👟"),
    "shop/sports": ("Sports Shop", "⚽"),
    "shop/stationery": ("Stationery Shop", "✏️"),
    "shop/storage_rental": ("Self Storage", "📦"),
    "shop/supermarket": ("Supermarket", "🛒"),
    "shop/swimming_pool": ("Pool Supplies", "🏊"),
    "shop/tailor": ("Tailor", "🧵"),
    "shop/tattoo": ("Tattoo Studio", "🖋️"),
    "shop/taxi": ("Taxi Booking", "🚕"),
    "shop/tea": ("Tea Shop", "🫖"),
    "shop/telecommunication": ("Telecoms Shop", "📡"),
    "shop/ticket": ("Ticket Office", "🎫"),
    "shop/tiles": ("Tile Shop", "🔲"),
    "shop/tobacco": ("Tobacconist", "🚬"),
    "shop/tool_hire": ("Tool Hire", "🧰"),
    "shop/toys": ("Toy Shop", "🧸"),
    "shop/trade": ("Trade Supplier", "🏭"),
    "shop/travel_agency": ("Travel Agency", "✈️"),
    "shop/trophy": ("Trophy Shop", "🏆"),
    "shop/tyres": ("Tyre Shop", "🛞"),
    "shop/vacant": ("Vacant Shop", "🏚️"),
    "shop/variety_store": ("Variety Store", "🏪"),
    "shop/video": ("Video Shop", "📀"),
    "shop/video_games": ("Video Game Shop", "🎮"),
    "shop/watches": ("Watch Shop", "⌚"),
    "shop/water_sports": ("Water Sports Shop", "🏄"),
    "shop/weapons": ("Weapons Shop", "🗡️"),
    "shop/wedding": ("Wedding Shop", "💒"),
    "shop/wholesale": ("Wholesaler", "📦"),
    "shop/wigs": ("Wig Shop", "💇"),
    "shop/window_blind": ("Blinds Shop", "🪟"),
    "shop/windows": ("Window Shop", "🪟"),
    "shop/wine": ("Wine Shop", "🍷"),
    "shop/wool": ("Wool Shop", "🧶"),
    "shop/yes": ("Shop", "🛍️"),
    # tourism
    "tourism/artwork": ("Public Artwork", "🎨"),
    "tourism/attraction": ("Tourist Attraction", "📸"),
    "tourism/camp_site": ("Campsite", "⛺"),
    "tourism/caravan_site": ("Caravan Site", "🚐"),
    "tourism/chalet": ("Chalet", "🏔️"),
    "tourism/gallery": ("Gallery", "🖼️"),
    "tourism/guest_house": ("Guest House", "🏡"),
    "tourism/hostel": ("Hostel", "🛏️"),
    "tourism/hotel": ("Hotel", "🏨"),
    "tourism/motel": ("Motel", "🏨"),
    "tourism/museum": ("Museum", "🏛️"),
    "tourism/picnic_site": ("Picnic Site", "🧺"),
    "tourism/preserved_railway": ("Heritage Railway", "🚂"),
    "tourism/theme_park": ("Theme Park", "🎢"),
    "tourism/viewpoint": ("Viewpoint", "🔭"),
    "tourism/zoo": ("Zoo", "🦁"),
}


def transform(input_path: Path) -> pl.LazyFrame:
    lf = pl.scan_parquet(input_path)

    # Get all unique categories present in the data
    all_categories = lf.select("category").unique().collect().to_series().to_list()

    # Verify every non-dropped category has a mapping
    unmapped = []
    for cat in all_categories:
        if cat not in DROP_CATEGORIES and cat not in CATEGORY_MAP:
            unmapped.append(cat)
    if unmapped:
        raise ValueError(f"Categories missing from CATEGORY_MAP: {sorted(unmapped)}")

    # Verify every CATEGORY_MAP key actually exists in the data (catch typos)
    mapped_but_absent = []
    all_set = set(all_categories)
    for cat in CATEGORY_MAP:
        if cat not in all_set:
            mapped_but_absent.append(cat)
    if mapped_but_absent:
        raise ValueError(
            f"CATEGORY_MAP contains categories not in data: {sorted(mapped_but_absent)}"
        )

    # Drop unwanted categories
    lf = lf.filter(~pl.col("category").is_in(list(DROP_CATEGORIES)))

    # Build name and emoji lookup expressions
    name_mapping = {k: v[0] for k, v in CATEGORY_MAP.items()}
    emoji_mapping = {k: v[1] for k, v in CATEGORY_MAP.items()}

    # Check no friendly names are missing (defensive)
    missing_names = [k for k, v in CATEGORY_MAP.items() if not v[0]]
    if missing_names:
        raise ValueError(f"Empty friendly names for: {missing_names}")
    missing_emojis = [k for k, v in CATEGORY_MAP.items() if not v[1]]
    if missing_emojis:
        raise ValueError(f"Empty emojis for: {missing_emojis}")

    lf = lf.with_columns(
        pl.col("category").replace_strict(name_mapping).alias("category"),
        pl.col("category").replace_strict(emoji_mapping).alias("emoji"),
    )

    return lf


def main():
    parser = argparse.ArgumentParser(
        description="Transform raw POIs to filtered version with friendly names"
    )
    parser.add_argument(
        "--input", type=Path, required=True, help="Raw POIs parquet file"
    )
    parser.add_argument(
        "--output", type=Path, required=True, help="Output filtered POIs parquet file"
    )
    args = parser.parse_args()

    df = transform(args.input).collect()

    df.write_parquet(args.output)

    size_mb = args.output.stat().st_size / (1024 * 1024)
    print(f"Wrote {args.output} ({size_mb:.1f} MB, {len(df):,} POIs)")
    print(f"\nCategories ({df['category'].n_unique()}):")
    counts = df.group_by("category", "emoji").len().sort("len", descending=True)
    for row in counts.iter_rows(named=True):
        print(f"  {row['emoji']} {row['category']}: {row['len']:,}")


if __name__ == "__main__":
    main()