wildlife: schema and migration

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

@@ -19,5 +19,6 @@ export * from './groceries';
 export * from './home';
 export * from './cars';
 export * from './planes';
+export * from './wildlife';
 export * from './music';
 export * from './hardware';

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

@@ -21,3 +21,4 @@ export const carsSchema = pgSchema('cars');
 export const musicSchema = pgSchema('music');
 export const hardwareSchema = pgSchema('hardware');
 export const planesSchema = pgSchema('planes');
+export const wildlifeSchema = pgSchema('wildlife');

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

@@ -0,0 +1,135 @@
+import { sql } from 'drizzle-orm';
+import {
+	boolean,
+	index,
+	integer,
+	real,
+	smallint,
+	text,
+	timestamp,
+	uniqueIndex
+} from 'drizzle-orm/pg-core';
+import { wildlifeSchema } from './schemas';
+import { id, timestamps } from './helpers';
+
+// Public wildlife "pokedex" (/trackers/wildlife). Read-only catalogue built from open data by
+// scripts/wildlife-import.ts (re-runnable; upserts species by iNaturalist taxon id so Dex
+// numbers stay stable) and scripts/wildlife-photos.ts (CC-licensed iNaturalist photos,
+// hotlinked, never downloaded). Sources:
+//  - iNaturalist (API, CC0 taxonomy): wild animal species observed in the Netherlands, in
+//    Europe and the most observed species worldwide; Dutch + English names, taxonomy,
+//    IUCN Red List status, research-grade observation counts and Wikipedia intros.
+//  - Wikidata (CC0): Dutch label, IUCN status fallback, GBIF id, body mass, length,
+//    lifespan and the year/author of the scientific name.
+//  - GBIF: occurrence records in the Netherlands.
+
+export const wildlifeGroups = wildlifeSchema.table(
+	'groups',
+	{
+		id: id(),
+		slug: text('slug').notNull(),
+		name: text('name').notNull(),
+		nameNl: text('name_nl').notNull(),
+		emoji: text('emoji').notNull(),
+		sortOrder: smallint('sort_order').notNull().default(0),
+		speciesCount: integer('species_count').notNull().default(0),
+		nlCount: integer('nl_count').notNull().default(0),
+		/** Lead photo of the group's most seen species in NL. */
+		photo: text('photo'),
+		...timestamps
+	},
+	(t) => [uniqueIndex('groups_slug_key').on(t.slug)]
+);
+
+export const wildlifeSpecies = wildlifeSchema.table(
+	'species',
+	{
+		id: id(),
+		dex: integer('dex').notNull(),
+		groupId: integer('group_id')
+			.notNull()
+			.references(() => wildlifeGroups.id, { onDelete: 'cascade' }),
+		/** Scientific name slug ("vulpes-vulpes"). */
+		slug: text('slug').notNull(),
+		scientificName: text('scientific_name').notNull(),
+		nameEn: text('name_en'),
+		nameNl: text('name_nl'),
+		/** Lowercase names (nl, en, scientific, order, family), for ILIKE / trigram search. */
+		search: text('search').notNull(),
+		inatId: integer('inat_id').notNull(),
+		gbifKey: integer('gbif_key'),
+		wikidata: text('wikidata'),
+		wikipediaEn: text('wikipedia_en'),
+		wikipediaNl: text('wikipedia_nl'),
+		className: text('class_name'),
+		orderName: text('order_name'),
+		familyName: text('family_name'),
+		/** IUCN Red List category: LC, NT, VU, EN, CR, EW, EX, DD (null = not evaluated). */
+		iucn: text('iucn'),
+		/** Dutch Red List / national status from iNaturalist, as published ("Kwetsbaar"). */
+		nlStatus: text('nl_status'),
+		/** native / introduced / endemic in the Netherlands (iNaturalist). */
+		nlEstablishment: text('nl_establishment'),
+		extinct: boolean('extinct').notNull().default(false),
+		/** Research-grade wild observations on iNaturalist in the Netherlands / Europe / worldwide. */
+		nlObs: integer('nl_obs').notNull().default(0),
+		europeObs: integer('europe_obs').notNull().default(0),
+		worldObs: integer('world_obs').notNull().default(0),
+		/** GBIF occurrence records in the Netherlands (all datasets). */
+		gbifNl: integer('gbif_nl'),
+		/** Where it made the list: nl, europe or world. */
+		scope: text('scope').notNull(),
+		summary: text('summary'),
+		summaryNl: text('summary_nl'),
+		facts: text('facts').array().notNull().default([]),
+		massKg: real('mass_kg'),
+		lengthM: real('length_m'),
+		lifespanYears: real('lifespan_years'),
+		namedYear: smallint('named_year'),
+		namedBy: text('named_by'),
+		sources: text('sources').array().notNull().default([]),
+		/** Lead photo (iNaturalist "medium" URL), denormalised from wildlife.photos for the cards. */
+		photo: text('photo'),
+		photoCount: smallint('photo_count').notNull().default(0),
+		photosCheckedAt: timestamp('photos_checked_at', { withTimezone: true }),
+		...timestamps
+	},
+	(t) => [
+		uniqueIndex('species_dex_key').on(t.dex),
+		uniqueIndex('species_inat_key').on(t.inatId),
+		uniqueIndex('species_slug_key').on(t.slug),
+		index('species_group_idx').on(t.groupId),
+		index('species_nl_obs_idx').on(t.nlObs),
+		index('species_search_trgm_idx').using('gin', sql`${t.search} public.gin_trgm_ops`)
+	]
+);
+
+/** CC-licensed photos; only the hotlinked URL and the credit are stored. */
+export const wildlifePhotos = wildlifeSchema.table(
+	'photos',
+	{
+		id: id(),
+		speciesId: integer('species_id')
+			.notNull()
+			.references(() => wildlifeSpecies.id, { onDelete: 'cascade' }),
+		position: smallint('position').notNull(),
+		/** taxon (iNaturalist taxon photo) or observation (research-grade observation photo). */
+		kind: text('kind').notNull(),
+		/** iNaturalist photo id. */
+		photoId: integer('photo_id').notNull(),
+		/** "medium" (500px) URL; small/large/original are derived from it. */
+		url: text('url').notNull(),
+		width: integer('width'),
+		height: integer('height'),
+		author: text('author'),
+		/** cc-by, cc-by-nc, cc0, ... */
+		license: text('license').notNull(),
+		/** Observation or photo page on iNaturalist. */
+		pageUrl: text('page_url'),
+		place: text('place')
+	},
+	(t) => [
+		uniqueIndex('photos_species_position_key').on(t.speciesId, t.position),
+		uniqueIndex('photos_species_photo_key').on(t.speciesId, t.photoId)
+	]
+);