website/src/lib/server/db/schema/cars.ts

import { sql } from 'drizzle-orm';
import {
	boolean,
	index,
	integer,
	jsonb,
	real,
	smallint,
	text,
	timestamp,
	uniqueIndex
} from 'drizzle-orm/pg-core';
import { carsSchema } from './schemas';
import { id, timestamps } from './helpers';

// Public car "pokedex" (/trackers/cars). Read-only catalogue built from open data by
// scripts/cars-import.ts (re-runnable; upserts brands/models by slug so Dex numbers stay
// stable, and replaces variants). Sources:
//  - fueleconomy.gov vehicles.csv (US EPA, 1984-now): one `variants` row per EPA test car.
//  - RDW open data (opendata.rdw.nl, Gekentekende voertuigen + brandstof): per make/model
//    aggregates of passenger cars registered in NL (count, years, list price, mass, body),
//    plus fuel/power/consumption read from one sample registration per model-year.
// Unknown stays null; nothing is estimated.

/** Shape of `models.nl_years`: registrations per first-registration year. */
export type CarNlYear = { y: number; n: number; price: number | null };
/** Shape of `models.nl_names`: RDW trade names (handelsbenaming) folded into this model. */
export type CarNlName = { name: string; n: number };

export const carBrands = carsSchema.table(
	'brands',
	{
		id: id(),
		slug: text('slug').notNull(),
		name: text('name').notNull(),
		country: text('country'),
		modelCount: integer('model_count').notNull().default(0),
		nlCount: integer('nl_count').notNull().default(0),
		firstYear: smallint('first_year'),
		lastYear: smallint('last_year'),
		sources: text('sources').array().notNull().default([]),
		/** Wikidata item of the brand/manufacturer and its enwiki article (scripts/cars-photos.ts). */
		wikidata: text('wikidata'),
		wikipedia: text('wikipedia'),
		/** Logo thumbnail (Wikimedia Commons, 250px) and its credit. */
		logo: text('logo'),
		logoFile: text('logo_file'),
		logoAuthor: text('logo_author'),
		logoLicense: text('logo_license'),
		/** Hero photo: lead photo of the brand's most common model. */
		photo: text('photo'),
		...timestamps
	},
	(t) => [uniqueIndex('brands_slug_key').on(t.slug)]
);

export const carModels = carsSchema.table(
	'models',
	{
		id: id(),
		dex: integer('dex').notNull(),
		brandId: integer('brand_id')
			.notNull()
			.references(() => carBrands.id, { onDelete: 'cascade' }),
		slug: text('slug').notNull(),
		name: text('name').notNull(),
		/** Lowercase brand + name + trade names, for ILIKE / trigram search. */
		search: text('search').notNull(),
		firstYear: smallint('first_year'),
		lastYear: smallint('last_year'),
		/** Consumer body types: hatchback, sedan, wagon, suv, mpv, van, pickup, coupe, convertible, sports. */
		bodies: text('bodies').array().notNull().default([]),
		/** US EPA size classes (e.g. "Compact Cars", "Small SUV"). */
		sizeClasses: text('size_classes').array().notNull().default([]),
		/** petrol, diesel, hybrid, plug-in hybrid, electric, lpg, cng, hydrogen. */
		fuels: text('fuels').array().notNull().default([]),
		/** fwd, rwd, awd, 4wd. */
		drives: text('drives').array().notNull().default([]),
		transmissions: text('transmissions').array().notNull().default([]),
		engineLMin: real('engine_l_min'),
		engineLMax: real('engine_l_max'),
		cylindersMin: smallint('cylinders_min'),
		cylindersMax: smallint('cylinders_max'),
		turbo: boolean('turbo'),
		powerKwMin: real('power_kw_min'),
		powerKwMax: real('power_kw_max'),
		/** Combined consumption, litres per 100 km (RDW type approval, NEDC or WLTP). */
		l100Min: real('l100_min'),
		l100Max: real('l100_max'),
		co2Min: smallint('co2_min'), // g/km (RDW)
		co2Max: smallint('co2_max'),
		mpgMin: real('mpg_min'), // combined (EPA; MPGe for EVs)
		mpgMax: real('mpg_max'),
		rangeKmMax: smallint('range_km_max'), // EV range (EPA)
		seats: smallint('seats'),
		doors: smallint('doors'),
		massKg: smallint('mass_kg'), // NL avg kerb weight (massa rijklaar)
		nlCount: integer('nl_count').notNull().default(0),
		nlRank: integer('nl_rank'),
		nlFirstYear: smallint('nl_first_year'),
		nlLastYear: smallint('nl_last_year'),
		nlPrice: integer('nl_price'), // avg catalogue price (EUR, incl. taxes) over all NL registrations
		nlPriceLatest: integer('nl_price_latest'), // avg catalogue price in the latest year with data
		nlPriceLatestYear: smallint('nl_price_latest_year'),
		nlYears: jsonb('nl_years').$type<CarNlYear[]>().notNull().default([]),
		nlNames: jsonb('nl_names').$type<CarNlName[]>().notNull().default([]),
		usYears: smallint('us_years').array().notNull().default([]),
		variantCount: integer('variant_count').notNull().default(0),
		sources: text('sources').array().notNull().default([]),
		/** Matched Wikidata item / enwiki article (scripts/cars-photos.ts); null = no confident match. */
		wikidata: text('wikidata'),
		wikipedia: text('wikipedia'),
		/** Lead photo thumbnail URL (960px), denormalised from cars.photos for the card grid. */
		photo: text('photo'),
		photoCount: smallint('photo_count').notNull().default(0),
		photosCheckedAt: timestamp('photos_checked_at', { withTimezone: true }),
		...timestamps
	},
	(t) => [
		uniqueIndex('models_dex_key').on(t.dex),
		uniqueIndex('models_brand_slug_key').on(t.brandId, t.slug),
		index('models_nl_count_idx').on(t.nlCount),
		index('models_search_trgm_idx').using('gin', sql`${t.search} public.gin_trgm_ops`)
	]
);

/** One row per fueleconomy.gov test vehicle (model-year x engine x transmission x drive). */
export const carVariants = carsSchema.table(
	'variants',
	{
		id: id(),
		modelId: integer('model_id')
			.notNull()
			.references(() => carModels.id, { onDelete: 'cascade' }),
		source: text('source').notNull().default('fueleconomy.gov'),
		sourceId: integer('source_id'),
		year: smallint('year').notNull(),
		name: text('name').notNull(),
		engineL: real('engine_l'),
		cylinders: smallint('cylinders'),
		transmission: text('transmission'),
		drive: text('drive'),
		fuel: text('fuel'),
		sizeClass: text('size_class'),
		mpgCity: real('mpg_city'),
		mpgHwy: real('mpg_hwy'),
		mpgComb: real('mpg_comb'),
		co2Gpm: smallint('co2_gpm'), // tailpipe g/mile
		rangeMi: smallint('range_mi'),
		evMotor: text('ev_motor'),
		turbo: boolean('turbo').notNull().default(false),
		supercharged: boolean('supercharged').notNull().default(false)
	},
	(t) => [index('variants_model_year_idx').on(t.modelId, t.year)]
);

/** Freely licensed photos from Wikimedia Commons; only thumbnail URLs are stored, nothing is downloaded. */
export const carPhotos = carsSchema.table(
	'photos',
	{
		id: id(),
		modelId: integer('model_id')
			.notNull()
			.references(() => carModels.id, { onDelete: 'cascade' }),
		position: smallint('position').notNull(),
		/** lead (Wikidata image), generation, article (enwiki), commons (Commons category). */
		kind: text('kind').notNull(),
		/** Commons file name without the "File:" prefix. */
		file: text('file').notNull(),
		caption: text('caption'),
		/** 960px thumbnail on upload.wikimedia.org; other standard widths are derived from it. */
		thumb: text('thumb').notNull(),
		width: integer('width').notNull(),
		height: integer('height').notNull(),
		author: text('author'),
		license: text('license'),
		licenseUrl: text('license_url')
	},
	(t) => [uniqueIndex('photos_model_position_key').on(t.modelId, t.position)]
);