trains: schema and migration

src/lib/server/db/schema/trains.ts

@@ -0,0 +1,187 @@
+import { sql } from 'drizzle-orm';
+import {
+	boolean,
+	index,
+	integer,
+	primaryKey,
+	real,
+	smallint,
+	text,
+	timestamp,
+	uniqueIndex,
+	type AnyPgColumn
+} from 'drizzle-orm/pg-core';
+import { trainsSchema } from './schemas';
+import { id, timestamps } from './helpers';
+
+// Public train "pokedex" (/trackers/trains). Read-only catalogue built from open data by
+// scripts/trains-import.ts (re-runnable; upserts types by Wikidata id so Dex numbers stay
+// stable) and scripts/trains-photos.ts (Commons photos + maker/operator logos). Sources:
+//  - English Wikipedia (CC BY-SA): every article with {{Infobox locomotive}}, {{Infobox train}}
+//    or {{Infobox tram}} (builder, operators, years, number built, speed, power, gauge, ...).
+//  - Wikidata (CC0): manufacturer (P176), operator (P137), country, number built, labels.
+// Only classes/types (individual locomotives and list articles are skipped). Unknown stays null.
+
+export const trainMakers = trainsSchema.table(
+	'makers',
+	{
+		id: id(),
+		slug: text('slug').notNull(),
+		name: text('name').notNull(),
+		fullName: text('full_name'),
+		country: text('country'),
+		wikidata: text('wikidata'),
+		wikipedia: text('wikipedia'),
+		typeCount: integer('type_count').notNull().default(0),
+		firstYear: smallint('first_year'),
+		lastYear: smallint('last_year'),
+		/** Sum of number built over the maker's types (families are not double counted). */
+		builtTotal: integer('built_total').notNull().default(0),
+		sources: text('sources').array().notNull().default([]),
+		/** 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 maker's most built type. */
+		photo: text('photo'),
+		...timestamps
+	},
+	(t) => [uniqueIndex('makers_slug_key').on(t.slug)]
+);
+
+export const trainOperators = trainsSchema.table(
+	'operators',
+	{
+		id: id(),
+		slug: text('slug').notNull(),
+		name: text('name').notNull(),
+		fullName: text('full_name'),
+		country: text('country'),
+		wikidata: text('wikidata'),
+		wikipedia: text('wikipedia'),
+		typeCount: integer('type_count').notNull().default(0),
+		logo: text('logo'),
+		logoFile: text('logo_file'),
+		logoAuthor: text('logo_author'),
+		logoLicense: text('logo_license'),
+		photo: text('photo'),
+		...timestamps
+	},
+	(t) => [uniqueIndex('operators_slug_key').on(t.slug)]
+);
+
+export const trainTypes = trainsSchema.table(
+	'types',
+	{
+		id: id(),
+		dex: integer('dex').notNull(),
+		makerId: integer('maker_id')
+			.notNull()
+			.references(() => trainMakers.id, { onDelete: 'cascade' }),
+		/** Parent class in the dex (e.g. ICE 3 -> Velaro), from Wikidata "subclass of". */
+		familyId: integer('family_id').references((): AnyPgColumn => trainTypes.id, {
+			onDelete: 'set null'
+		}),
+		slug: text('slug').notNull(),
+		/** Name without the maker ("Velaro"); full_name is the article title ("Siemens Velaro"). */
+		name: text('name').notNull(),
+		fullName: text('full_name').notNull(),
+		/** Dutch label from Wikidata, when it differs. */
+		nlName: text('nl_name'),
+		/** Lowercase names, aliases, maker, operators and class numbers for ILIKE / trigram search. */
+		search: text('search').notNull(),
+		wikidata: text('wikidata').notNull(),
+		wikipedia: text('wikipedia'),
+		description: text('description'),
+		/** First sentences of the Wikipedia article. */
+		summary: text('summary'),
+		/** steam, diesel, electric (locomotives), high-speed, multiple unit, metro, tram, carriage, other. */
+		category: text('category').notNull(),
+		/** Infobox stock type / power type ("Electric multiple unit"). */
+		categoryLabel: text('category_label'),
+		/** steam, diesel, electric, battery, hydrogen, gas turbine, bi-mode, none. */
+		traction: text('traction'),
+		country: text('country'),
+		/** Country of origin plus the countries of its operators. */
+		countries: text('countries').array().notNull().default([]),
+		makers: text('makers').array().notNull().default([]),
+		operators: text('operators').array().notNull().default([]),
+		firstBuilt: smallint('first_built'),
+		lastBuilt: smallint('last_built'),
+		introducedYear: smallint('introduced_year'),
+		retiredYear: smallint('retired_year'),
+		/** in service, retired, preserved. */
+		status: text('status'),
+		inService: boolean('in_service'),
+		numberBuilt: integer('number_built'),
+		maxSpeedKmh: smallint('max_speed_kmh'),
+		powerKw: integer('power_kw'),
+		gauge: text('gauge'),
+		gaugeMm: smallint('gauge_mm'),
+		electricSystem: text('electric_system'),
+		/** Whyte (4-6-2) for steam, else UIC (Bo'Bo'). */
+		wheels: text('wheels'),
+		lengthM: real('length_m'),
+		weightT: real('weight_t'),
+		capacity: smallint('capacity'),
+		formation: text('formation'),
+		designer: text('designer'),
+		nicknames: text('nicknames'),
+		sources: text('sources').array().notNull().default([]),
+		/** Lead photo thumbnail URL (960px), denormalised from trains.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('types_dex_key').on(t.dex),
+		uniqueIndex('types_wikidata_key').on(t.wikidata),
+		uniqueIndex('types_maker_slug_key').on(t.makerId, t.slug),
+		index('types_number_built_idx').on(t.numberBuilt),
+		index('types_first_built_idx').on(t.firstBuilt),
+		index('types_category_idx').on(t.category),
+		index('types_search_trgm_idx').using('gin', sql`${t.search} public.gin_trgm_ops`)
+	]
+);
+
+export const trainTypeOperators = trainsSchema.table(
+	'type_operators',
+	{
+		typeId: integer('type_id')
+			.notNull()
+			.references(() => trainTypes.id, { onDelete: 'cascade' }),
+		operatorId: integer('operator_id')
+			.notNull()
+			.references(() => trainOperators.id, { onDelete: 'cascade' }),
+		position: smallint('position').notNull().default(0)
+	},
+	(t) => [
+		primaryKey({ columns: [t.typeId, t.operatorId] }),
+		index('type_operators_operator_idx').on(t.operatorId)
+	]
+);
+
+/** Freely licensed photos from Wikimedia Commons; only thumbnail URLs are stored, nothing is downloaded. */
+export const trainPhotos = trainsSchema.table(
+	'photos',
+	{
+		id: id(),
+		typeId: integer('type_id')
+			.notNull()
+			.references(() => trainTypes.id, { onDelete: 'cascade' }),
+		position: smallint('position').notNull(),
+		/** lead (Wikidata image), infobox (article infobox), commons (Commons category), article (enwiki). */
+		kind: text('kind').notNull(),
+		file: text('file').notNull(),
+		caption: text('caption'),
+		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_type_position_key').on(t.typeId, t.position)]
+);