import { Dataset } from 'digital-products'

const movieDataset = Dataset({
  id: 'movies',
  name: 'Movie Database',
  description: 'Comprehensive movie information dataset',
  version: '2024.1',
  format: 'parquet',
  source: 's3://datasets/movies.parquet',
})

const movieDataset = Dataset({
  id: 'movies',
  name: 'Movie Database',
  format: 'parquet',
  schema: {
    id: 'Movie ID',
    title: 'Movie title',
    year: 'Release year (number)',
    genres: ['Array of genre names'],
    rating: 'Average rating (number)',
    votes: 'Number of votes (number)',
    runtime: 'Runtime in minutes (number)',
    director: 'Director name',
    cast: ['Array of actor names'],
    plot: 'Plot summary',
  },
})

const movieDataset = Dataset({
  id: 'movies',
  name: 'Movie Database',
  description: 'Comprehensive movie information from 1900 to present',
  version: '2024.1',
  format: 'parquet',
  source: 's3://datasets/movies.parquet',
  size: 1000000,           // Number of records
  license: 'CC-BY-4.0',    // License identifier
  updateFrequency: 'daily', // Update schedule
  schema: { /* ... */ },
})

import { Dataset } from 'digital-products'

const stockDataset = Dataset({
  id: 'stock-prices',
  name: 'Historical Stock Prices',
  description: 'Daily stock prices for S&P 500 companies from 2000 to present',
  version: '2024.12',
  format: 'parquet',
  source: 's3://financial-data/stocks.parquet',

  schema: {
    symbol: 'Stock ticker symbol',
    date: 'Trading date (date)',
    open: 'Opening price (number)',
    high: 'Daily high (number)',
    low: 'Daily low (number)',
    close: 'Closing price (number)',
    adjustedClose: 'Adjusted close (number)',
    volume: 'Trading volume (number)',
    dividends: 'Dividends paid (number)',
    splits: 'Stock splits (number)',
  },

  size: 5000000,
  license: 'CC-BY-4.0',
  updateFrequency: 'daily',
})

import { Dataset } from 'digital-products'

const productsDataset = Dataset({
  id: 'product-catalog',
  name: 'Product Catalog',
  description: 'Complete product catalog with pricing and inventory',
  version: '1.0.0',
  format: 'json',
  source: './data/products.json',

  schema: {
    id: 'Product ID',
    sku: 'Stock keeping unit',
    name: 'Product name',
    description: 'Product description',
    category: 'Category path',
    brand: 'Brand name',
    price: 'Current price (number)',
    salePrice: 'Sale price (number)',
    currency: 'Price currency',
    inStock: 'In stock (boolean)',
    quantity: 'Available quantity (number)',
    images: ['Array of image URLs'],
    attributes: {
      color: 'Product color',
      size: 'Product size',
      weight: 'Weight in kg (number)',
    },
    createdAt: 'Created date (date)',
    updatedAt: 'Updated date (date)',
  },

  size: 50000,
  license: 'proprietary',
  updateFrequency: 'hourly',
})

import { Dataset } from 'digital-products'

const sentimentDataset = Dataset({
  id: 'sentiment-reviews',
  name: 'Product Review Sentiment',
  description: 'Labeled product reviews for sentiment analysis training',
  version: '3.0.0',
  format: 'csv',
  source: 's3://ml-datasets/sentiment-reviews.csv',

  schema: {
    id: 'Review ID',
    text: 'Review text',
    rating: 'Star rating 1-5 (number)',
    sentiment: 'positive | neutral | negative',
    category: 'Product category',
    verified: 'Verified purchase (boolean)',
    helpfulVotes: 'Helpful vote count (number)',
    date: 'Review date (date)',
  },

  size: 2500000,
  license: 'CC-BY-4.0',
  updateFrequency: 'monthly',
})

interface DatasetDefinition {
  id: string
  name: string
  description?: string
  version?: string
  format?: 'json' | 'csv' | 'parquet' | 'arrow' | 'avro'
  schema?: SimpleSchema
  source?: string
  size?: number
  license?: string
  updateFrequency?: 'realtime' | 'hourly' | 'daily' | 'weekly' | 'monthly' | 'quarterly' | 'yearly' | 'static'
  metadata?: Record<string, unknown>
}