make-mp-data 3.0.3 → 3.0.5

package/dungeons/capstone/generate-product-lookup.js

@@ -0,0 +1,309 @@
+/**
+ * Generate product_id lookup table
+ * This creates a standalone JSONL file with 10,000 products
+ */
+
+import Chance from 'chance';
+import { weighNumRange, integer } from "../brain/utils/utils.js";
+import fs from 'fs';
+import path from 'path';
+import { fileURLToPath } from 'url';
+
+const __filename = fileURLToPath(import.meta.url);
+const __dirname = path.dirname(__filename);
+
+// Initialize chance with seed for reproducibility
+const chance = new Chance('its the capstone baby!');
+
+// Product catalog
+const productCatalog = {
+	electronics: [
+		"Wireless Bluetooth Headphones", "USB-C Charging Cable", "Portable Power Bank 20000mAh",
+		"Smart Watch Series 7", "Wireless Gaming Mouse", "Mechanical Keyboard RGB",
+		"4K Webcam Pro", "Laptop Stand Aluminum", "Phone Case Clear", "Screen Protector Tempered Glass",
+		"Bluetooth Speaker Waterproof", "Ring Light for Streaming", "HDMI Cable 6ft",
+		"Wall Charger 65W Fast Charge", "Wireless Charging Pad", "USB Hub 7-Port",
+		"External SSD 1TB", "Microphone Condenser", "Gaming Headset 7.1 Surround",
+		"Smart LED Light Bulb", "Security Camera 1080p", "Tablet Case Leather",
+		"Stylus Pen for iPad", "Cable Organizer Set", "Laptop Sleeve 15 inch",
+		"Ergonomic Mouse Pad", "Phone Car Mount Magnetic", "Dash Cam Front and Rear",
+		"Fitness Tracker Heart Rate", "Earbuds True Wireless", "Drone with 4K Camera",
+		"Action Camera Waterproof", "VR Headset Meta Quest", "Smart Doorbell Camera",
+		"WiFi Router Dual Band", "Graphics Tablet for Drawing", "Streaming Deck Mini",
+		"Portable Monitor 15.6 inch", "Noise Cancelling Headphones", "Smart Thermostat WiFi"
+	],
+	home_kitchen: [
+		"Air Fryer 6 Quart Digital", "Instant Pot 8 Quart", "Blender High Speed 1500W",
+		"Coffee Maker Programmable 12 Cup", "Toaster 4 Slice Stainless Steel", "Knife Set 15 Piece",
+		"Cutting Board Bamboo Large", "Mixing Bowl Set Stainless", "Measuring Cup and Spoon Set",
+		"Cast Iron Skillet 12 inch", "Non-Stick Cookware Set 10 Piece", "Slow Cooker 6 Quart",
+		"Electric Kettle 1.7L", "Food Processor 12 Cup", "Stand Mixer 5 Quart",
+		"Dish Rack Stainless Steel", "Kitchen Utensil Set 24 Piece", "Spice Rack Organizer",
+		"Storage Containers Glass Set", "Baking Sheet Pan Set", "Vacuum Sealer for Food",
+		"Wine Glasses Set of 6", "Dinnerware Set 16 Piece", "Silverware Set 20 Piece",
+		"Can Opener Electric", "Rice Cooker 10 Cup", "Waffle Maker Belgian",
+		"Panini Press Grill", "Juicer Extractor Centrifugal", "Kitchen Scale Digital",
+		"Trash Can 13 Gallon Step", "Paper Towel Holder Wall Mount", "Soap Dispenser Automatic",
+		"Dish Soap Refill Pack", "Sponges Pack of 24", "Oven Mitts Heat Resistant",
+		"Pot Holders Set of 4", "Apron Adjustable Chef", "Dish Towels Cotton Set"
+	],
+	clothing: [
+		"Men's Cotton T-Shirt Basic", "Women's Yoga Pants High Waist", "Unisex Hoodie Pullover",
+		"Jeans Slim Fit Blue", "Dress Casual Summer Floral", "Button-Down Shirt Long Sleeve",
+		"Sneakers Running Lightweight", "Jacket Windbreaker Waterproof", "Socks Pack of 12",
+		"Baseball Cap Adjustable", "Scarf Winter Knit Warm", "Gloves Touchscreen Compatible",
+		"Belt Leather Reversible", "Tank Top Athletic Performance", "Shorts Cargo Pockets",
+		"Leggings Compression Workout", "Sweater Pullover V-Neck", "Cardigan Open Front",
+		"Blazer Business Casual", "Skirt Midi A-Line", "Polo Shirt Pique Cotton",
+		"Underwear Pack of 6", "Bra Sports High Support", "Pajama Set Soft Cotton",
+		"Slippers Memory Foam Cozy", "Sandals Flip Flops Beach", "Boots Ankle Chelsea",
+		"Tie Silk Classic Stripe", "Sunglasses Polarized UV", "Watch Band Leather Replacement",
+		"Backpack Laptop Compartment", "Wallet Genuine Leather", "Handbag Crossbody Vegan",
+		"Tote Bag Canvas Large", "Duffel Bag Gym Travel", "Beanie Winter Warm Knit",
+		"Bucket Hat Summer UV", "Vest Puffer Down Fill", "Raincoat Packable Lightweight"
+	]
+};
+
+// Generate realistic product names from catalog
+function getProductName() {
+	const allProducts = Object.values(productCatalog).flat();
+	const baseProduct = chance.pickone(allProducts);
+	
+	// Add variation 30% of the time
+	if (chance.bool({ likelihood: 30 })) {
+		const variations = [
+			` - ${chance.pickone(['Black', 'White', 'Blue', 'Red', 'Gray', 'Silver', 'Gold'])}`,
+			` - ${chance.pickone(['Small', 'Medium', 'Large', 'X-Large'])}`,
+			` - ${chance.pickone(['Pro', 'Plus', 'Lite', 'Max', 'Mini'])}`,
+			` (${chance.pickone(['2024 Model', '2025 Edition', 'Gen 2'])})`
+		];
+		return baseProduct + chance.pickone(variations);
+	}
+	
+	return baseProduct;
+}
+
+// Get category for a product
+function getProductCategory() {
+	const categories = [
+		"Electronics", "Home & Kitchen", "Clothing", "Beauty & Personal Care",
+		"Sports & Outdoors", "Books & Media", "Toys & Games", "Pet Supplies",
+		"Office Supplies", "Baby Products", "Automotive", "Garden & Outdoor"
+	];
+	return chance.pickone(categories);
+}
+
+// Generate a single product
+function generateProduct(productId) {
+	const brands = [
+		"TechPro", "HomeEssentials", "FitLife", "StyleCo", "BeautyPlus",
+		"OfficeMax", "EcoGoods", "PremiumChoice", "ValueBrand", "GenericCo",
+		"ProGear", "UltraLife", "SmartHome", "ActiveWear", "LuxeBeauty",
+		"NatureBest", "UrbanLiving", "EliteChoice", "BudgetBuy", "MegaMart",
+		"QualityFirst", "FastShip", "GreenEarth", "ModernLife", "ClassicGoods"
+	];
+	
+	const categories = [
+		"Electronics", "Home & Kitchen", "Clothing", "Beauty & Personal Care",
+		"Sports & Outdoors", "Books & Media", "Toys & Games", "Pet Supplies",
+		"Office Supplies", "Baby Products", "Automotive", "Garden & Outdoor"
+	];
+	
+	const colors = ["Black", "White", "Blue", "Red", "Gray", "Silver", "Gold", "Green", "Pink", "Purple", "Orange", "Brown", "Beige", "Navy", "Teal"];
+	const sizes = ["XS", "S", "M", "L", "XL", "XXL", "One Size", "Adjustable", "Small", "Medium", "Large"];
+	const materials = ["Cotton", "Polyester", "Plastic", "Metal", "Wood", "Glass", "Silicone", "Leather", "Fabric", "Stainless Steel", "Aluminum", "Bamboo"];
+	const conditions = ["New", "New", "New", "New", "Refurbished", "Open Box", "Like New"];
+	const shippingClasses = ["Standard", "Standard", "Express", "Prime", "2-Day", "Overnight", "Economy"];
+	const warehouses = ["US-East-1", "US-West-2", "EU-Central", "Asia-Pacific", "US-Central", "Canada-1"];
+	const suppliers = ["Global Supply Co", "Import Direct", "Wholesale Plus", "Manufacturer Direct", "Trade Partners", "International Goods"];
+	const origins = ["USA", "China", "Germany", "Japan", "Italy", "UK", "France", "South Korea", "Taiwan", "Mexico", "Canada", "Vietnam"];
+	
+	const demand = chance.pickone(["very high", "high", "high", "medium", "medium", "low"]);
+	const supply = chance.pickone(["abundant", "high", "high", "medium", "medium", "low", "limited"]);
+	const isFeatured = chance.pickone([true, false, false, false, false, false, false]);
+	const isOnSale = chance.pickone([true, true, false, false, false, false]);
+	const isNewArrival = chance.pickone([true, false, false, false, false]);
+	const isBestseller = chance.pickone([true, true, false, false, false, false, false]);
+	const isLowStock = chance.pickone([true, false, false, false, false]);
+	
+	const basePrice = parseFloat(chance.pickone(weighNumRange(5, 500, 0.25)).toFixed(2));
+	const costPrice = parseFloat((basePrice * chance.floating({ min: 0.4, max: 0.7 })).toFixed(2));
+	const msrp = parseFloat((basePrice * chance.floating({ min: 1.1, max: 1.5 })).toFixed(2));
+	const discountPercent = isOnSale ? integer(5, 50) : 0;
+	const salePrice = discountPercent > 0 ? parseFloat((basePrice * (1 - discountPercent / 100)).toFixed(2)) : basePrice;
+	
+	const weight = parseFloat(chance.floating({ min: 0.1, max: 50 }).toFixed(2));
+	const length = parseFloat(chance.floating({ min: 1, max: 100 }).toFixed(1));
+	const width = parseFloat(chance.floating({ min: 1, max: 80 }).toFixed(1));
+	const height = parseFloat(chance.floating({ min: 1, max: 60 }).toFixed(1));
+	
+	const totalReviews = chance.pickone(weighNumRange(0, 5000, 1.8));
+	const rating = parseFloat(chance.pickone(weighNumRange(1, 5)).toFixed(1));
+	const stockLevel = chance.pickone(weighNumRange(0, 10000, 0.8));
+	const reorderPoint = Math.floor(stockLevel * 0.2);
+	const reorderQuantity = Math.floor(stockLevel * 0.5);
+	
+	const daysInStock = integer(1, 1000);
+	const lastRestocked = new Date(Date.now() - (daysInStock * 24 * 60 * 60 * 1000)).toISOString().split('T')[0];
+	const dateAdded = new Date(Date.now() - (integer(1, 730) * 24 * 60 * 60 * 1000)).toISOString().split('T')[0];
+	
+	return {
+		product_id: productId,
+		sku: `SKU-${String(productId).padStart(8, '0')}`,
+		upc: chance.string({ length: 12, pool: '0123456789' }),
+		product_name: getProductName(),
+		category: chance.pickone(categories),
+		subcategory: chance.pickone(["Accessories", "Main", "Premium", "Budget", "Specialty"]),
+		brand: chance.pickone(brands),
+		manufacturer: chance.company(),
+		supplier: chance.pickone(suppliers),
+		
+		// Pricing
+		price: basePrice,
+		cost_price: costPrice,
+		msrp: msrp,
+		sale_price: salePrice,
+		discount_percent: discountPercent,
+		profit_margin: parseFloat(((basePrice - costPrice) / basePrice * 100).toFixed(2)),
+		currency: "USD",
+		tax_rate: parseFloat(chance.pickone([0, 5, 7.5, 8.5, 10]).toFixed(2)),
+		
+		// Inventory
+		stock_level: stockLevel,
+		reserved_stock: Math.floor(stockLevel * chance.floating({ min: 0, max: 0.15 })),
+		available_stock: Math.floor(stockLevel * chance.floating({ min: 0.85, max: 1 })),
+		warehouse_location: chance.pickone(warehouses),
+		aisle: `${chance.pickone(['A', 'B', 'C', 'D', 'E'])}-${integer(1, 99)}`,
+		shelf: integer(1, 20),
+		bin: integer(1, 50),
+		reorder_point: reorderPoint,
+		reorder_quantity: reorderQuantity,
+		lead_time_days: integer(1, 60),
+		last_restocked: lastRestocked,
+		
+		// Product Details
+		color: chance.pickone(colors),
+		size: chance.pickone(sizes),
+		material: chance.pickone(materials),
+		weight_lbs: weight,
+		length_in: length,
+		width_in: width,
+		height_in: height,
+		volume_cubic_in: parseFloat((length * width * height).toFixed(2)),
+		condition: chance.pickone(conditions),
+		origin_country: chance.pickone(origins),
+		
+		// Ratings & Reviews
+		rating: rating,
+		total_reviews: totalReviews,
+		five_star: Math.floor(totalReviews * chance.floating({ min: 0.4, max: 0.7 })),
+		four_star: Math.floor(totalReviews * chance.floating({ min: 0.15, max: 0.25 })),
+		three_star: Math.floor(totalReviews * chance.floating({ min: 0.05, max: 0.15 })),
+		two_star: Math.floor(totalReviews * chance.floating({ min: 0.02, max: 0.08 })),
+		one_star: Math.floor(totalReviews * chance.floating({ min: 0.01, max: 0.05 })),
+		
+		// Performance Metrics
+		views_last_30_days: integer(0, 50000),
+		clicks_last_30_days: integer(0, 10000),
+		conversions_last_30_days: integer(0, 1000),
+		revenue_last_30_days: parseFloat((integer(0, 50000)).toFixed(2)),
+		units_sold_total: integer(0, 100000),
+		units_sold_last_30_days: integer(0, 5000),
+		units_returned_total: integer(0, 1000),
+		return_rate: parseFloat(chance.floating({ min: 0, max: 15 }).toFixed(2)),
+		
+		// Marketing & Sales
+		demand: demand,
+		supply: supply,
+		is_featured: isFeatured,
+		is_on_sale: isOnSale,
+		is_new_arrival: isNewArrival,
+		is_bestseller: isBestseller,
+		is_low_stock: isLowStock,
+		is_out_of_stock: stockLevel === 0,
+		is_backordered: chance.bool({ likelihood: 5 }),
+		is_seasonal: chance.bool({ likelihood: 20 }),
+		is_gift_wrap_available: chance.bool({ likelihood: 60 }),
+		is_free_shipping: chance.bool({ likelihood: 40 }),
+		
+		// Shipping
+		shipping_class: chance.pickone(shippingClasses),
+		shipping_cost: parseFloat(chance.pickone([0, 4.99, 7.99, 9.99, 12.99, 19.99]).toFixed(2)),
+		estimated_delivery_days: integer(1, 14),
+		is_hazmat: chance.bool({ likelihood: 2 }),
+		is_fragile: chance.bool({ likelihood: 15 }),
+		is_perishable: chance.bool({ likelihood: 5 }),
+		
+		// Metadata
+		date_added: dateAdded,
+		last_updated: new Date(Date.now() - (integer(0, 30) * 24 * 60 * 60 * 1000)).toISOString().split('T')[0],
+		age_restricted: chance.bool({ likelihood: 8 }),
+		warranty_months: chance.pickone([0, 0, 0, 6, 12, 24, 36]),
+		tags: chance.pickset(["popular", "trending", "eco-friendly", "handmade", "luxury", "budget", "clearance"], integer(0, 3)).join(';'),
+		search_keywords: chance.pickset(["quality", "affordable", "premium", "durable", "bestselling", "top-rated"], integer(1, 4)).join(';')
+	};
+}
+
+// Generate all products
+function generateProductLookup(count = 10000) {
+	console.log(`Generating ${count} products...`);
+	const jsonlFile = path.join(__dirname, 'product_lookup.jsonl');
+	const csvFile = path.join(__dirname, 'product_lookup.csv');
+	
+	// Clear the files if they exist
+	fs.writeFileSync(jsonlFile, '', 'utf8');
+	
+	// Write CSV header - all fields
+	const csvHeaders = [
+		'product_id', 'sku', 'upc', 'product_name', 'category', 'subcategory', 'brand', 
+		'manufacturer', 'supplier', 'price', 'cost_price', 'msrp', 'sale_price', 
+		'discount_percent', 'profit_margin', 'currency', 'tax_rate', 'stock_level', 
+		'reserved_stock', 'available_stock', 'warehouse_location', 'aisle', 'shelf', 
+		'bin', 'reorder_point', 'reorder_quantity', 'lead_time_days', 'last_restocked',
+		'color', 'size', 'material', 'weight_lbs', 'length_in', 'width_in', 'height_in',
+		'volume_cubic_in', 'condition', 'origin_country', 'rating', 'total_reviews',
+		'five_star', 'four_star', 'three_star', 'two_star', 'one_star',
+		'views_last_30_days', 'clicks_last_30_days', 'conversions_last_30_days',
+		'revenue_last_30_days', 'units_sold_total', 'units_sold_last_30_days',
+		'units_returned_total', 'return_rate', 'demand', 'supply', 'is_featured',
+		'is_on_sale', 'is_new_arrival', 'is_bestseller', 'is_low_stock', 
+		'is_out_of_stock', 'is_backordered', 'is_seasonal', 'is_gift_wrap_available',
+		'is_free_shipping', 'shipping_class', 'shipping_cost', 'estimated_delivery_days',
+		'is_hazmat', 'is_fragile', 'is_perishable', 'date_added', 'last_updated',
+		'age_restricted', 'warranty_months', 'tags', 'search_keywords'
+	];
+	fs.writeFileSync(csvFile, csvHeaders.join(',') + '\n', 'utf8');
+	
+	// Generate products in batches to avoid memory issues
+	const batchSize = 1000;
+	for (let i = 1; i <= count; i++) {
+		const product = generateProduct(i);
+		
+		// Write JSONL
+		fs.appendFileSync(jsonlFile, JSON.stringify(product) + '\n', 'utf8');
+		
+		// Write CSV (escape fields with commas or quotes)
+		const escapeCSV = (val) => {
+			if (val === null || val === undefined) return '';
+			const str = String(val);
+			if (str.includes(',') || str.includes('"') || str.includes('\n')) {
+				return `"${str.replace(/"/g, '""')}"`;
+			}
+			return str;
+		};
+		
+		const csvRow = csvHeaders.map(header => escapeCSV(product[header]));
+		fs.appendFileSync(csvFile, csvRow.join(',') + '\n', 'utf8');
+		
+		if (i % batchSize === 0) {
+			console.log(`Generated ${i}/${count} products...`);
+		}
+	}
+	
+	console.log(`✓ Successfully generated ${count} products to:`);
+	console.log(`  - ${jsonlFile} (${csvHeaders.length} fields)`);
+	console.log(`  - ${csvFile} (${csvHeaders.length} columns)`);
+}
+
+// Run the generator - increase this to 100,000 or 1,000,000 for a huge dataset
+generateProductLookup(275_000);