groceries: add PLUS, Xenos and Amazing Oriental stores, is_food flag, PLUS EAN backfill

drizzle/20261025171621_groceries_more_stores/migration.sql

@@ -0,0 +1,12 @@
+ALTER TABLE "groceries"."groups" ADD COLUMN "is_food" boolean DEFAULT true NOT NULL;--> statement-breakpoint
+ALTER TABLE "groceries"."items" ADD COLUMN "is_food" boolean DEFAULT true NOT NULL;--> statement-breakpoint
+-- New stores (Plus already exists): Xenos and Amazing Oriental (orientalwebshop.nl).
+INSERT INTO "groceries"."stores" ("slug", "name", "website", "color", "sort_order") VALUES
+	('xenos', 'Xenos', 'https://www.xenos.nl', '#e6007e', 50),
+	('amazing-oriental', 'Amazing Oriental', 'https://www.orientalwebshop.nl/nl/', '#2e7d32', 60)
+ON CONFLICT ("slug") DO NOTHING;--> statement-breakpoint
+-- Existing AH/Jumbo items: drugstore and household are not food (the importer keeps this current).
+UPDATE "groceries"."items" SET "is_food" = false WHERE "category" IN ('Drogisterij', 'Huishouden & dieren');--> statement-breakpoint
+UPDATE "groceries"."groups" g SET "is_food" = false
+WHERE NOT EXISTS (SELECT 1 FROM "groceries"."items" i WHERE i.group_id = g.id AND i.available AND i.is_food)
+	AND EXISTS (SELECT 1 FROM "groceries"."items" i WHERE i.group_id = g.id AND i.available);

scripts/groceries-import.ts

@@ -1,19 +1,28 @@
-// Nightly supermarket catalogue import (Albert Heijn + Jumbo) and product regrouping.
+// Supermarket catalogue import and product regrouping.
 //
-//   bun --env-file=$HOME/.config/groceries-import.env scripts/groceries-import.ts [options]
+//   bun scripts/groceries-import.ts [options]
-//     --store ah|jumbo     only this store (repeatable; default: all)
+//     --store SLUG         only this store (repeatable). Default: the nightly stores
+//                          (ah, jumbo, plus, xenos); amazing-oriental runs weekly on its own timer.
 //     --skip-fetch         no store requests; only regroup what is in the DB
-//     --gtin-limit N       AH product details to fetch for EANs this run (default 1500, 0 = off)
+//     --gtin-limit N       AH and PLUS product details to fetch for EANs this run (default 1500, 0 = off)
-//     --delay-ms N         minimum delay between requests (default 2000)
+//     --gtin-only          skip the catalogue crawl; only the EAN backfill (+ regroup)
+//     --gtin-parallel N    parallel PLUS detail requests for the backfill (default 1, max 4)
+//     --delay-ms N         minimum delay between requests (default: per store, else 2000)
 //
-// Politeness: one request at a time, >= 2 s apart, backoff on 429/5xx, identifying User-Agent,
+// Politeness: one request at a time per store (Amazing Oriental: 25 s, its robots.txt
-// and a same-day cache in $GROCERIES_CACHE_DIR (default ~/.cache/groceries, kept 7 days).
+// Crawl-delay), backoff on 429/5xx, and a same-day cache in $GROCERIES_CACHE_DIR
-// Plus is not imported: plus.nl blocks this server's IP range (Incapsula), and we do not evade it.
+// (default ~/.cache/groceries, kept 7 days). The systemd wrapper holds a flock so two imports
+// never run at once.
 import postgres from 'postgres';
 import { homedir } from 'node:os';
 import { join } from 'node:path';
 import { PoliteFetcher, pruneCache } from '../src/lib/server/groceries/http';
-import { enrichAhGtins, importStore, regroup } from '../src/lib/server/groceries/import';
+import {
+	enrichAhGtins,
+	enrichPlusGtins,
+	importStore,
+	regroup
+} from '../src/lib/server/groceries/import';
 import { SOURCES } from '../src/lib/server/groceries/sources';
 
 const args = process.argv.slice(2);
@@ -26,10 +35,14 @@ const num = (name: string, d: number) => {
 	return n;
 };
 if (!process.env.DB_URL) throw new Error('DB_URL is not set');
-const stores = values('--store').length ? values('--store') : Object.keys(SOURCES);
+const stores = values('--store').length
+	? values('--store')
+	: Object.keys(SOURCES).filter((s) => SOURCES[s].nightly !== false);
 for (const s of stores) if (!SOURCES[s]) throw new Error(`unknown store ${s}`);
-const delay = num('--delay-ms', 2000);
+const delayArg = values('--delay-ms').length ? num('--delay-ms', 2000) : null;
 const gtinLimit = num('--gtin-limit', 1500);
+const gtinParallel = Math.min(4, Math.max(1, num('--gtin-parallel', 1)));
+const gtinOnly = args.includes('--gtin-only');
 const cacheDir = process.env.GROCERIES_CACHE_DIR || join(homedir(), '.cache', 'groceries');
 const log = (m: string) => console.log(`[${new Date().toISOString().slice(11, 19)}Z] ${m}`);
 
@@ -39,17 +52,33 @@ try {
 	pruneCache(cacheDir);
 	if (!args.includes('--skip-fetch')) {
 		for (const slug of stores) {
-			const f = new PoliteFetcher({ store: slug, minDelayMs: delay, cacheDir, log });
+			const src = SOURCES[slug];
+			const fetcher = () =>
+				new PoliteFetcher({
+					store: slug,
+					minDelayMs: delayArg ?? src.minDelayMs ?? 2000,
+					cacheDir,
+					log,
+					...src.fetcher
+				});
+			const f = fetcher();
 			try {
-				log(`${slug}: importing`);
+				if (!gtinOnly) {
-				const r = await importStore(sql, SOURCES[slug], f, log);
+					log(`${slug}: importing`);
-				log(
+					const r = await importStore(sql, src, f, log);
-					`${slug}: ${r.fetched} products (${r.inserted} new, ${r.priceChanges} price changes, ${r.delisted} delisted); ${r.requests} requests, ${r.cacheHits} from cache`
+					log(
-				);
+						`${slug}: ${r.fetched} products (${r.inserted} new, ${r.priceChanges} price changes, ${r.delisted} delisted); ${r.requests} requests, ${r.cacheHits} from cache`
+					);
+				}
 				if (slug === 'ah' && gtinLimit > 0) {
 					const g = await enrichAhGtins(sql, f, gtinLimit, log);
 					log(`ah: EAN lookup ${g.checked} checked, ${g.found} found`);
 				}
+				if (slug === 'plus' && gtinLimit > 0) {
+					const fs = [f, ...Array.from({ length: gtinParallel - 1 }, fetcher)];
+					const g = await enrichPlusGtins(sql, fs, gtinLimit, log);
+					log(`plus: EAN lookup ${g.checked} checked, ${g.found} found`);
+				}
 			} catch (e) {
 				failed = true;
 				log(`${slug}: FAILED ${e instanceof Error ? e.message : e}`);

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

@@ -78,6 +78,8 @@ export const groceryGroups = groceriesSchema.table(
 		minPriceCents: integer('min_price_cents'),
 		promoCount: integer('promo_count').notNull().default(0),
 		imageUrl: text('image_url'),
+		/** True when any item is edible (refreshGroupStats); homeware-only groups are false. */
+		isFood: boolean('is_food').notNull().default(true),
 		searchText: text('search_text').notNull().default(''),
 		...timestamps
 	},
@@ -102,7 +104,7 @@ export const groceryItems = groceriesSchema.table(
 		storeId: integer('store_id')
 			.notNull()
 			.references(() => groceryStores.id, { onDelete: 'cascade' }),
-		storeProductId: text('store_product_id').notNull(), // AH webshopId, Jumbo SKU
+		storeProductId: text('store_product_id').notNull(), // AH webshopId, Jumbo/Plus/Xenos/AO SKU
 		name: text('name').notNull(),
 		brand: text('brand'),
 		sizeText: text('size_text'), // as the store prints it, e.g. "6 x 330 ml"
@@ -122,6 +124,8 @@ export const groceryItems = groceriesSchema.table(
 		imageUrl: text('image_url'),
 		url: text('url'),
 		available: boolean('available').notNull().default(true),
+		/** Edible (food/drink). False for drugstore, household and homeware (Xenos décor, woks…). */
+		isFood: boolean('is_food').notNull().default(true),
 		firstSeenAt: timestamptz('first_seen_at').notNull().defaultNow(),
 		lastSeenAt: timestamptz('last_seen_at').notNull().defaultNow(),
 		groupId: integer('group_id').references(() => groceryGroups.id, { onDelete: 'set null' }),

src/lib/server/groceries/http.ts

@@ -12,6 +12,8 @@ export type FetcherOptions = {
 	store: string;
 	minDelayMs?: number;
 	maxRetries?: number;
+	/** First retry backoff; doubles each try (default 5 s). */
+	retryBaseMs?: number;
 	cacheDir?: string | null;
 	log?: (msg: string) => void;
 };
@@ -25,12 +27,14 @@ export class PoliteFetcher {
 	private last = 0;
 	private readonly minDelayMs: number;
 	private readonly maxRetries: number;
+	private readonly retryBaseMs: number;
 	private readonly dir: string | null;
 	private readonly log: (msg: string) => void;
 
 	constructor(opts: FetcherOptions) {
 		this.minDelayMs = opts.minDelayMs ?? 2000;
 		this.maxRetries = opts.maxRetries ?? 5;
+		this.retryBaseMs = opts.retryBaseMs ?? 5000;
 		this.log = opts.log ?? (() => {});
 		this.dir = opts.cacheDir ? join(opts.cacheDir, today(), opts.store) : null;
 		if (this.dir) mkdirSync(this.dir, { recursive: true });
@@ -38,10 +42,25 @@ export class PoliteFetcher {
 
 	/** Fetch JSON. `cacheKey` (a file-safe string) enables the same-day cache. */
 	async json<T>(url: string, init: RequestInit = {}, cacheKey?: string): Promise<T> {
-		const file = this.dir && cacheKey ? join(this.dir, `${cacheKey}.json.gz`) : null;
+		return this.request(url, init, cacheKey, 'json', (text) => JSON.parse(text) as T);
+	}
+
+	/** Fetch text (HTML). Same politeness, retries and same-day cache as json(). */
+	async text(url: string, init: RequestInit = {}, cacheKey?: string): Promise<string> {
+		return this.request(url, init, cacheKey, 'html', (text) => text);
+	}
+
+	private async request<T>(
+		url: string,
+		init: RequestInit,
+		cacheKey: string | undefined,
+		ext: string,
+		parse: (text: string) => T
+	): Promise<T> {
+		const file = this.dir && cacheKey ? join(this.dir, `${cacheKey}.${ext}.gz`) : null;
 		if (file && existsSync(file)) {
 			this.cacheHits++;
-			return JSON.parse(gunzipSync(readFileSync(file)).toString('utf8')) as T;
+			return parse(gunzipSync(readFileSync(file)).toString('utf8'));
 		}
 		for (let attempt = 0; ; attempt++) {
 			const wait = this.last + this.minDelayMs - Date.now();
@@ -54,7 +73,7 @@ export class PoliteFetcher {
 				const res = await fetch(url, { ...init, signal: AbortSignal.timeout(45_000) });
 				if (res.ok) {
 					const text = await res.text();
-					const data = JSON.parse(text) as T;
+					const data = parse(text);
 					if (file) writeFileSync(file, gzipSync(text));
 					return data;
 				}
@@ -67,18 +86,21 @@ export class PoliteFetcher {
 				}
 				retryAfter = Number(res.headers.get('retry-after')) * 1000 || 0;
 			} catch (e) {
-				if (e instanceof HttpError) throw e;
+				if (e instanceof HttpError || e instanceof FatalFetchError) throw e;
 				problem = e instanceof Error ? e.message : String(e);
 			}
 			if (attempt >= this.maxRetries)
 				throw new Error(`${problem} for ${url} after ${attempt + 1} tries`);
-			const backoff = Math.max(retryAfter, 5000 * 2 ** attempt);
+			const backoff = Math.max(retryAfter, this.retryBaseMs * 2 ** attempt);
 			this.log(`  ${problem}; retrying in ${Math.round(backoff / 1000)}s`);
 			await sleep(backoff);
 		}
 	}
 }
 
+/** Thrown from a parse callback to stop without retrying (e.g. a bot wall instead of data). */
+export class FatalFetchError extends Error {}
+
 export class HttpError extends Error {
 	constructor(
 		readonly status: number,

src/lib/server/groceries/import.ts

@@ -7,11 +7,13 @@ import {
 	brandClass,
 	brandLabel,
 	displayName,
+	fold,
 	matchKey,
 	nameTokens,
 	normaliseUnit
 } from './normalize';
-import { AhClient, type RawItem, type Source } from './sources';
+import { PlusClient } from './plus';
+import { AhClient, itemIsFood, type RawItem, type Source } from './sources';
 
 type Log = (m: string) => void;
 
@@ -34,6 +36,10 @@ export async function importStore(sql: Sql, source: Source, f: PoliteFetcher, lo
 		insert into groceries.import_runs (store_id) values (${sid}) returning id, started_at`;
 	try {
 		const raw = await source.fetchAll(f, log);
+		if (BRAND_FROM_NAME.has(source.slug)) {
+			const n = await inferBrands(sql, sid, raw);
+			log(`  ${source.slug}: brand inferred from the name for ${n} products`);
+		}
 		const [prev] = await sql<{ n: number }[]>`
 			select count(*)::int as n from groceries.items where store_id = ${sid} and available`;
 		const res = await upsertItems(sql, sid, raw, run.started_at);
@@ -91,6 +97,7 @@ async function upsertItems(sql: Sql, sid: number, raw: RawItem[], runStart: Date
 				image_url: r.imageUrl,
 				url: r.url,
 				available: r.available,
+				is_food: itemIsFood(r),
 				match_key: keyFor(r.name, r.brand, n.unit).key
 			};
 		});
@@ -104,7 +111,7 @@ async function upsertItems(sql: Sql, sid: number, raw: RawItem[], runStart: Date
 				promo_unit_price_cents = excluded.promo_unit_price_cents, category = excluded.category,
 				category_path = excluded.category_path, gtin = coalesce(excluded.gtin, items.gtin),
 				image_url = excluded.image_url, url = excluded.url, available = excluded.available,
-				match_key = excluded.match_key, last_seen_at = now()
+				is_food = excluded.is_food, match_key = excluded.match_key, last_seen_at = now()
 			returning (xmax = 0) as inserted`;
 		inserted += res.filter((r) => r.inserted).length;
 	}
@@ -132,6 +139,92 @@ async function upsertItems(sql: Sql, sid: number, raw: RawItem[], runStart: Date
 	return { inserted, priceChanges: hist.filter((h) => h.changed).length };
 }
 
+/** Stores whose listings carry no (or little) brand: take it from the start of the name. */
+const BRAND_FROM_NAME = new Set(['xenos', 'amazing-oriental']);
+
+/**
+ * Fill in missing brands when the name starts with a known brand ("Lotus Biscoff …"). Known
+ * brands: those on >= 5 products at other stores, plus the store's own brands in this batch.
+ * House brands are never inferred. Returns how many items got a brand.
+ */
+export async function inferBrands(sql: Sql, sid: number, raw: RawItem[]): Promise<number> {
+	const rows = await sql<{ brand: string }[]>`
+		select brand from groceries.items
+		where brand is not null and store_id <> ${sid} and length(brand) >= 3
+		group by brand having count(*) >= 5`;
+	const known = new Map<string, string>();
+	for (const b of [...rows.map((r) => r.brand), ...raw.map((r) => r.brand)]) {
+		if (!b || b.trim().length < 3 || isHouseBrand(b)) continue;
+		const k = fold(b).replace(/\s+/g, ' ').trim();
+		if (!known.has(k)) known.set(k, b.trim());
+	}
+	const byFirstWord = new Map<string, string[]>();
+	for (const k of known.keys()) {
+		const w = k.split(' ')[0];
+		byFirstWord.set(w, [...(byFirstWord.get(w) ?? []), k]);
+	}
+	for (const list of byFirstWord.values()) list.sort((a, b) => b.length - a.length);
+	let n = 0;
+	for (const r of raw) {
+		if (r.brand) continue;
+		const name = fold(r.name).replace(/\s+/g, ' ').trim();
+		const first = name.split(' ')[0];
+		for (const k of byFirstWord.get(first) ?? []) {
+			// The brand must be followed by more words, so "Mango" alone never becomes a brand.
+			if (name.startsWith(k + ' ') && name.length > k.length + 2) {
+				r.brand = known.get(k)!;
+				n++;
+				break;
+			}
+		}
+	}
+	return n;
+}
+const isHouseBrand = (b: string) => brandClass(b).startsWith('house');
+
+/** Fill in PLUS EANs from the product detail action, `limit` products per run, once per SKU. */
+export async function enrichPlusGtins(
+	sql: Sql,
+	fetchers: PoliteFetcher[],
+	limit: number,
+	log: Log
+) {
+	if (limit <= 0 || !fetchers.length) return { checked: 0, found: 0 };
+	const sid = await storeId(sql, 'plus');
+	const todo = await sql<{ id: number; store_product_id: string; url: string | null }[]>`
+		select id, store_product_id, url from groceries.items
+		where store_id = ${sid} and gtin is null and available
+			and (gtin_checked_at is null or gtin_checked_at < now() - interval '180 days')
+		order by gtin_checked_at asc nulls first, id
+		limit ${limit}`;
+	const clients = fetchers.map((f) => new PlusClient(f));
+	let found = 0;
+	let errors = 0;
+	let next = 0;
+	let done = 0;
+	await Promise.all(
+		clients.map(async (plus) => {
+			while (next < todo.length) {
+				const t = todo[next++];
+				const slug = t.url?.split('/product/')[1] ?? '';
+				let gtin: string | null;
+				try {
+					gtin = await plus.gtin(t.store_product_id, slug);
+				} catch (e) {
+					errors++;
+					log(`  plus gtin ${t.store_product_id}: ${e instanceof Error ? e.message : e}`);
+					if (errors > 25 && found === 0) throw e; // something is wrong; stop hammering
+					continue;
+				}
+				if (gtin) found++;
+				await sql`update groceries.items set gtin = ${gtin}, gtin_checked_at = now() where id = ${t.id}`;
+				if (++done % 500 === 0) log(`  plus gtin: ${done}/${todo.length} checked, ${found} found`);
+			}
+		})
+	);
+	return { checked: done, found };
+}
+
 /** Fill in AH EANs from the product detail endpoint, `limit` products per run (A-brands first). */
 export async function enrichAhGtins(sql: Sql, f: PoliteFetcher, limit: number, log: Log) {
 	if (limit <= 0) return { checked: 0, found: 0 };
@@ -374,7 +467,7 @@ export async function refreshGroupStats(sql: Sql, ids?: number[]) {
 		update groceries.groups g set
 			item_count = s.item_count, store_count = s.store_count, store_ids = s.store_ids,
 			min_unit_price_cents = s.min_unit, min_price_cents = s.min_price, promo_count = s.promo_count,
-			image_url = s.image_url, search_text = s.search_text
+			image_url = s.image_url, search_text = s.search_text, is_food = s.is_food
 		from (
 			select g2.id,
 				count(i.id)::int as item_count,
@@ -384,6 +477,7 @@ export async function refreshGroupStats(sql: Sql, ids?: number[]) {
 				min(coalesce(i.promo_price_cents, i.price_cents)) as min_price,
 				(count(*) filter (where i.promo_price_cents is not null or i.promo_label is not null))::int as promo_count,
 				(array_agg(i.image_url order by i.store_id, i.id) filter (where i.image_url is not null))[1] as image_url,
+				coalesce(bool_or(i.is_food), true) as is_food,
 				translate(lower(concat_ws(' ', g2.name, g2.brand, g2.category,
 					string_agg(distinct concat_ws(' ', i.brand, i.name), ' '))), ${ACCENTS_FROM}, ${ACCENTS_TO}) as search_text
 			from groceries.groups g2
@@ -392,6 +486,6 @@ export async function refreshGroupStats(sql: Sql, ids?: number[]) {
 			group by g2.id
 		) s
 		where g.id = s.id and (g.item_count, g.store_count, g.store_ids, g.min_unit_price_cents, g.min_price_cents,
-			g.promo_count, g.image_url, g.search_text) is distinct from
+			g.promo_count, g.image_url, g.search_text, g.is_food) is distinct from
-			(s.item_count, s.store_count, s.store_ids, s.min_unit, s.min_price, s.promo_count, s.image_url, s.search_text)`;
+			(s.item_count, s.store_count, s.store_ids, s.min_unit, s.min_price, s.promo_count, s.image_url, s.search_text, s.is_food)`;
 }

src/lib/server/groceries/normalize.ts

@@ -299,7 +299,10 @@ export function nameTokens(title: string, brand?: string | null): string[] {
 		.replace(/\bdurum\s*-?\s*(?:tarwe)?\b/g, ' '); // "Penne durum tarwe" is plain penne
 	const drop = new Set<string>();
 	for (const w of [brand ?? '', ...STORE_WORDS])
-		for (const t of fold(w).replace(/['’]/g, '').split(/[^a-z0-9]+/)) if (t) drop.add(t);
+		for (const t of fold(w)
+			.replace(/['’]/g, '')
+			.split(/[^a-z0-9]+/))
+			if (t) drop.add(t);
 	s = s.replace(/['’]/g, '').replace(/[^a-z0-9%+]+/g, ' ');
 	const out: string[] = [];
 	for (const raw of s.split(' ')) {

src/lib/server/groceries/oriental.ts

@@ -0,0 +1,171 @@
+// Amazing Oriental webshop (orientalwebshop.nl, Magento 2) catalogue source. There is no JSON
+// API (graphql/rest are closed), so we parse the category listing HTML. The site's bot wall
+// (Anubis) lets honest non-browser User-Agents through and expects them to follow robots.txt,
+// which asks for Crawl-delay: 25, so this source makes ONE request per 25 s (~45 min a crawl)
+// and runs weekly on its own timer. A challenge page stops the crawl; we never work around it.
+import { FatalFetchError, HttpError, type PoliteFetcher } from './http';
+import type { RawItem, Source } from './sources';
+
+const BASE = 'https://www.orientalwebshop.nl/nl';
+const UA = 'stan-grocery-tracker/0.1 (personal price tracker; +https://stanrunge.dev)';
+export const AO_CRAWL_DELAY_MS = 25_000; // robots.txt Crawl-delay: 25
+const LIMIT = 95; // the largest product_list_limit the shop offers
+
+/** Top category url key → shared category. Crawled in this order; outlet last. */
+const TOPS: [string, string][] = [
+	['verse-producten', 'Groente, fruit & aardappelen'],
+	['diepvriesproducten', 'Diepvries'],
+	['dranken-alcohol', 'Frisdrank & sappen'],
+	['rijst', 'Pasta, rijst & wereldkeuken'],
+	['noedels', 'Pasta, rijst & wereldkeuken'],
+	['sushi-producten', 'Pasta, rijst & wereldkeuken'],
+	['hot-pot-fondue', 'Pasta, rijst & wereldkeuken'],
+	['sauzen', 'Soepen, sauzen & conserven'],
+	['kruiden', 'Soepen, sauzen & conserven'],
+	['geconserveerd', 'Soepen, sauzen & conserven'],
+	['bakproducten', 'Ontbijt & beleg'],
+	['aziatische-snacks', 'Snoep, koek & chips'],
+	['mooncakes', 'Snoep, koek & chips'], // seasonal; 404 out of season
+	['huis-beauty-welzijn', 'Overig'],
+	['outlet', 'Overig'] // short-dated cross-cut; items keep their real category when seen first
+];
+const TOP_MAP = new Map(TOPS);
+/** Level-2 (data-category segment 2, slugified) overrides. */
+const SUB: Record<string, string> = {
+	'zeevruchten-vlees-eieren': 'Vlees, vis & vega',
+	'tofu-tempeh': 'Vlees, vis & vega',
+	'cake-dessert': 'Brood & bakkerij',
+	thee: 'Koffie & thee',
+	'instant-drank': 'Koffie & thee',
+	koffie: 'Koffie & thee',
+	alcohol: 'Bier & wijn',
+	bier: 'Bier & wijn',
+	wijn: 'Bier & wijn'
+};
+
+const decode = (s: string) =>
+	s
+		.replace(/&amp;/g, '&')
+		.replace(/&apos;|&#0?39;/g, "'")
+		.replace(/&quot;/g, '"')
+		.replace(/&lt;/g, '<')
+		.replace(/&gt;/g, '>')
+		.replace(/\s+/g, ' ')
+		.trim();
+const attr = (blk: string, name: string) => {
+	const m = blk.match(new RegExp(`${name}="([^"]*)"`));
+	return m ? decode(m[1]) : null;
+};
+const slug = (s: string) =>
+	s
+		.toLowerCase()
+		.replace(/&/g, '')
+		.replace(/[^a-z0-9]+/g, '-')
+		.replace(/^-|-$/g, '');
+const cents = (v: string | undefined) => {
+	const n = v ? parseFloat(v) : NaN;
+	return Number.isFinite(n) && n > 0 ? Math.round(n * 100) : null;
+};
+
+/** Parse one listing page into items (exported for tests). */
+export function parseOrientalListing(html: string, top: string): RawItem[] {
+	const arts = html.match(/<article class="item product product-item"[\s\S]*?<\/article>/g) ?? [];
+	const out: RawItem[] = [];
+	for (const a of arts) {
+		const id = attr(a, 'data-sku');
+		const link =
+			a.match(/class="product-item-link"[^>]*href="([^"]+)"[^>]*>([\s\S]*?)<\/a>/) ??
+			a.match(/href="([^"]+)"[^>]*class="product-item-link"[^>]*>([\s\S]*?)<\/a>/);
+		const name = decode(link?.[2]?.replace(/<[^>]+>/g, '') ?? attr(a, 'data-name') ?? '');
+		if (!id || !name) continue;
+		const fin = cents(a.match(/data-price-amount="([\d.]+)" data-price-type="finalPrice"/)?.[1]);
+		const old = cents(a.match(/data-price-amount="([\d.]+)" data-price-type="oldPrice"/)?.[1]);
+		const promo = old && fin && fin < old ? fin : null;
+		const labels = [
+			...a.matchAll(/class="product-label (label-[a-z_-]+)"[^>]*><span>([^<]*)</g)
+		].map((m) => [m[1], decode(m[2])] as const);
+		const sale = labels.find(([c]) => c === 'label-sale')?.[1] || null;
+		const exp = labels.find(([c]) => c.startsWith('label-expiration'))?.[1] || null;
+		const catAttr = attr(a, 'data-category'); // "Verse Producten/Tofu & Tempeh" or "OUTLET"
+		const real = catAttr && catAttr.toUpperCase() !== 'OUTLET' ? catAttr : null;
+		const category = (real && SUB[slug(real.split('/')[1] ?? '')]) ?? TOP_MAP.get(top) ?? 'Overig';
+		out.push({
+			storeProductId: id,
+			name,
+			brand: attr(a, 'data-brand') || null,
+			sizeText:
+				name.match(
+					/(\d+\s*[x×]\s*)?\d+(?:[.,]\d+)?\s*(?:kg|g|gr|ml|cl|l|ltr|liter|st|stuks)\b/i
+				)?.[0] ?? null,
+			priceCents: old ?? fin,
+			promoPriceCents: promo,
+			promoLabel: promo
+				? [sale ?? 'Aanbieding', exp ? `THT ${exp.replace(/^THT\s*/i, '')}` : null]
+						.filter(Boolean)
+						.join(' · ')
+				: null,
+			storeUnitPrice: null,
+			storeUnitBasisCents: null,
+			category,
+			categoryPath: real ? real.replace(/\//g, ' > ') : top === 'outlet' ? 'Outlet' : top,
+			gtin: null,
+			imageUrl: a.match(/class="product-image-photo"[^>]*src="([^"]+)"/)?.[1] ?? null,
+			url: link?.[1] ?? null,
+			available: attr(a, 'data-dimension10') !== 'Niet op voorraad',
+			isFood: top !== 'huis-beauty-welzijn'
+		});
+	}
+	return out;
+}
+
+export const orientalSource: Source = {
+	slug: 'amazing-oriental',
+	minDelayMs: AO_CRAWL_DELAY_MS,
+	nightly: false, // weekly, own timer (groceries-import-oriental.timer)
+	// On 429 back off hard (minutes), never hammer.
+	fetcher: { maxRetries: 3, retryBaseMs: 120_000 },
+	async fetchAll(f: PoliteFetcher, log) {
+		const items = new Map<string, RawItem>();
+		const get = (top: string, p: number) =>
+			f.text(
+				`${BASE}/${top}?product_list_limit=${LIMIT}&p=${p}`,
+				{ headers: { 'User-Agent': UA, Accept: 'text/html' } },
+				`${top}-p${p}`
+			);
+		for (const [top] of TOPS) {
+			const before = items.size;
+			let prevFirst = '';
+			for (let p = 1; p <= 200; p++) {
+				let html: string;
+				try {
+					html = await get(top, p);
+				} catch (e) {
+					if (e instanceof HttpError && e.status === 404) {
+						log(`  ao ${top}: 404, skipped (seasonal?)`);
+						break;
+					}
+					throw e;
+				}
+				if (html.includes('anubis_challenge'))
+					throw new FatalFetchError('amazing-oriental: bot challenge page (policy changed)');
+				const page = parseOrientalListing(html, top);
+				if (!page.length || page[0].storeProductId === prevFirst) break; // Magento clamps p
+				prevFirst = page[0].storeProductId;
+				for (const it of page) {
+					const prev = items.get(it.storeProductId);
+					if (!prev) items.set(it.storeProductId, it);
+					else if (top === 'outlet' && it.promoPriceCents && !prev.promoPriceCents)
+						items.set(it.storeProductId, {
+							...prev,
+							promoPriceCents: it.promoPriceCents,
+							promoLabel: it.promoLabel,
+							priceCents: it.priceCents
+						});
+				}
+				if (page.length < LIMIT) break;
+			}
+			log(`  ao ${top}: +${items.size - before} (${items.size} total, ${f.requests} requests)`);
+		}
+		return [...items.values()];
+	}
+};

src/lib/server/groceries/plus.ts

@@ -0,0 +1,317 @@
+// PLUS (plus.nl) catalogue source: the OutSystems "screenservices" data actions the web shop
+// itself calls (product listing + product detail). Anonymous, no login. Prices are the national
+// webshop prices (StoreNumber 0). Listing pages are fixed at 12 products, so a full crawl is
+// ~1,450 requests; the product detail action is only used to backfill EANs once per SKU.
+import type { PoliteFetcher } from './http';
+import { normaliseGtin } from './normalize';
+import type { RawItem, Source } from './sources';
+
+const BASE = 'https://www.plus.nl';
+// plus.nl serves browsers only (Imperva/Incapsula); a desktop UA, one request at a time.
+const UA =
+	'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/130 Safari/537.36';
+const CSRF = 'T6C+9iB49TLra4jEsMeSckDMNhQ='; // OutSystems' fixed anonymous-user CSRF token
+const PLP_ACTION =
+	'screenservices/ECP_Composition_CW/ProductLists/PLP_Content/DataActionGetProductListAndCategoryInfo';
+const PLP_JS = '/ECOP/scripts/ECP_Composition_CW.ProductLists.PLP_Content.mvc.js';
+const PDP_ACTION =
+	'screenservices/ECP_Product_CW/ProductDetails/PDPContent/DataActionGetProductDetailsAndAgeInfo';
+const PDP_JS = '/ECOP/scripts/ECP_Product_CW.ProductDetails.PDPContent.mvc.js';
+
+/** Top category → shared category. */
+const TOP: Record<string, string> = {
+	'Aardappelen, groente, fruit': 'Groente, fruit & aardappelen',
+	'Vlees, kip, vis, vega': 'Vlees, vis & vega',
+	'Kaas, vleeswaren, tapas': 'Vleeswaren, kaas & tapas',
+	'Zuivel, eieren, boter': 'Zuivel & eieren',
+	'Brood, gebak': 'Brood & bakkerij',
+	'Brood, gebak, bakproducten': 'Brood & bakkerij',
+	'Ontbijtgranen, broodbeleg, tussendoor': 'Ontbijt & beleg',
+	'Verse kant-en-klaarmaaltijden': 'Maaltijden & gemak',
+	'Pasta, rijst, internationale keuken': 'Pasta, rijst & wereldkeuken',
+	'Soepen, conserven, sauzen, smaakmakers': 'Soepen, sauzen & conserven',
+	'Snoep, koek, chocolade, chips, noten': 'Snoep, koek & chips',
+	'Snoep, koek, chips, zoutjes, noten': 'Snoep, koek & chips',
+	Diepvries: 'Diepvries',
+	'Frisdrank, sappen, koffie, thee': 'Frisdrank & sappen',
+	'Wijn, bier, sterke drank': 'Bier & wijn',
+	'Baby, drogisterij': 'Drogisterij',
+	Huishouden: 'Huishouden & dieren',
+	Huisdier: 'Huishouden & dieren',
+	'Bewuste voeding': 'Overig', // like AH "Glutenvrij" and Jumbo "Bewuste voeding"
+	'Sinterklaas assortiment': 'Snoep, koek & chips',
+	Kerstassortiment: 'Overig',
+	Paasassortiment: 'Overig',
+	'BBQ assortiment': 'Overig',
+	'Wonen, bloemen, service': 'Overig'
+};
+/** Level-2 overrides. */
+const SUB: Record<string, string> = {
+	'Koffie, koffiemelk, suiker': 'Koffie & thee',
+	Koffie: 'Koffie & thee',
+	Thee: 'Koffie & thee',
+	Babyvoeding: 'Baby & kind',
+	'Luiers, doekjes': 'Baby & kind',
+	'BBQ sauzen, smaakmakers': 'Soepen, sauzen & conserven',
+	'BBQ dranken': 'Frisdrank & sappen',
+	'Kerst wijn, bier': 'Bier & wijn',
+	'Sint bakproducten': 'Ontbijt & beleg'
+};
+const NON_FOOD_TOP = new Set(['Wonen, bloemen, service', 'Huishouden', 'Huisdier']);
+const NON_FOOD_SUB = new Set(['Kerst Adventkalenders']);
+
+/** Delivery offers are not product promotions. */
+const NOT_PROMO = /gratis\s+bezorg|bezorging/i;
+
+type Versions = { moduleVersion: string; apiVersion: string };
+type PlpProduct = {
+	SKU: string | number;
+	Name: string;
+	Brand?: string;
+	Slug?: string;
+	Product_Subtitle?: string;
+	OriginalPrice?: string;
+	NewPrice?: string;
+	PromotionLabel?: string;
+	PromotionBasedLabel?: string;
+	ImageURL?: string;
+	IsAvailable?: boolean;
+	IsOfflineSaleOnly?: boolean;
+	EAN?: string;
+	Categories?: { List?: { Name: string }[] };
+};
+type PlpData = {
+	TotalNumberItems: number;
+	TotalPages: number;
+	ProductList?: { List?: { PLP_Str: PlpProduct }[] };
+	FilterList?: {
+		List?: {
+			Filter_Str: { Name: string; FilterItem: { List: { Name: string; URL: string }[] } };
+		}[];
+	};
+};
+
+const euroCents = (v: unknown): number | null => {
+	if (v === null || v === undefined || v === '') return null;
+	const n = typeof v === 'number' ? v : parseFloat(String(v).replace(',', '.'));
+	return Number.isFinite(n) && n > 0 ? Math.round(n * 100) : null;
+};
+
+/** Effective unit price for common Dutch multibuy labels ("1+1 GRATIS", "3 VOOR 6.00", …). */
+export function promoFromLabel(label: string | null, priceCents: number | null): number | null {
+	if (!label || !priceCents) return null;
+	const l = label.toLowerCase();
+	let m;
+	if ((m = l.match(/(\d+)\s*\+\s*(\d+)\s*gratis/)))
+		return Math.round((priceCents * +m[1]) / (+m[1] + +m[2]));
+	if (/2e\s*halve\s*prijs/.test(l)) return Math.round(priceCents * 0.75);
+	if ((m = l.match(/(\d+)\s*%\s*korting/))) return Math.round(priceCents * (1 - +m[1] / 100));
+	if ((m = l.match(/^(\d+)\s*voor\s*€?\s*(\d+(?:[.,]\d+)?)/))) {
+		const each = (parseFloat(m[2].replace(',', '.')) * 100) / +m[1];
+		return each < priceCents ? Math.round(each) : null;
+	}
+	if ((m = l.match(/^(\d+(?:[.,]\d+)?)\s*korting/))) {
+		const off = Math.round(parseFloat(m[1].replace(',', '.')) * 100);
+		return off < priceCents ? priceCents - off : null;
+	}
+	return null;
+}
+
+export function plusItem(p: PlpProduct): RawItem {
+	const price = euroCents(p.OriginalPrice);
+	const now = euroCents(p.NewPrice);
+	const labels = [p.PromotionLabel, p.PromotionBasedLabel]
+		.map((l) => l?.trim())
+		.filter((l): l is string => !!l && !NOT_PROMO.test(l));
+	const label = labels.length ? [...new Set(labels)].join(' · ') : null;
+	const promo = now && price && now < price ? now : promoFromLabel(label, price);
+	const cats = p.Categories?.List?.map((c) => c.Name) ?? [];
+	const category = SUB[cats[1]] ?? TOP[cats[0]] ?? 'Overig';
+	return {
+		storeProductId: String(p.SKU),
+		name: p.Name.trim(),
+		brand: p.Brand?.trim() || null,
+		sizeText: p.Product_Subtitle?.replace(/^Per\s+/i, '').trim() || null,
+		priceCents: price ?? now,
+		promoPriceCents: promo,
+		promoLabel: label,
+		storeUnitPrice: null,
+		storeUnitBasisCents: null,
+		category,
+		categoryPath: cats.join(' > ') || null,
+		gtin: normaliseGtin(p.EAN),
+		imageUrl: p.ImageURL || null,
+		url: p.Slug ? `${BASE}/product/${p.Slug}` : null,
+		available: !!p.IsAvailable && !p.IsOfflineSaleOnly,
+		isFood: !(
+			NON_FOOD_TOP.has(cats[0]) ||
+			NON_FOOD_SUB.has(cats[1]) ||
+			(cats[0] === 'Baby, drogisterij' && cats[1] !== 'Babyvoeding')
+		)
+	};
+}
+
+export class PlusClient {
+	private cookie = '';
+	private versions: { plp?: Versions; pdp?: Versions } = {};
+	constructor(private f: PoliteFetcher) {}
+
+	private headers(json = false): Record<string, string> {
+		const h: Record<string, string> = { 'User-Agent': UA, Accept: 'application/json' };
+		if (this.cookie) h.Cookie = this.cookie;
+		if (json) {
+			h['Content-Type'] = 'application/json; charset=UTF-8';
+			h['X-CSRFToken'] = CSRF;
+		}
+		return h;
+	}
+
+	/** Incapsula session cookies from the home page (plain fetch: no JSON, never cached). */
+	private async session() {
+		const res = await fetch(BASE + '/', {
+			headers: { 'User-Agent': UA },
+			signal: AbortSignal.timeout(45_000)
+		});
+		const jar = res.headers
+			.getSetCookie()
+			.map((c) => c.split(';')[0])
+			.filter((c) => c.includes('='));
+		this.cookie = jar.join('; ');
+		const html = await res.text();
+		if (!res.ok || /_Incapsula_Resource|Request unsuccessful/i.test(html))
+			throw new Error(`plus.nl blocked this IP (Incapsula, HTTP ${res.status})`);
+	}
+
+	/** moduleVersion + the action's apiVersion; both change on every PLUS deploy. */
+	private async discover(kind: 'plp' | 'pdp'): Promise<Versions> {
+		if (!this.cookie) await this.session();
+		const { versionToken } = await this.f.json<{ versionToken: string }>(
+			`${BASE}/moduleservices/moduleversioninfo`,
+			{ headers: this.headers() }
+		);
+		const info = await this.f.json<{ manifest: { urlVersions: Record<string, string> } }>(
+			`${BASE}/moduleservices/moduleinfo`,
+			{ headers: this.headers() }
+		);
+		const path = kind === 'plp' ? PLP_JS : PDP_JS;
+		const q = info.manifest.urlVersions[path];
+		if (q === undefined) throw new Error(`plus: ${path} not in module manifest`);
+		const js = await this.f.text(`${BASE}${path.replace('/ECOP', '')}${q}`, {
+			headers: { 'User-Agent': UA, Cookie: this.cookie }
+		});
+		const action =
+			kind === 'plp'
+				? 'DataActionGetProductListAndCategoryInfo'
+				: 'DataActionGetProductDetailsAndAgeInfo';
+		const m = js.match(new RegExp(`"${action}",\\s*"[^"]+",\\s*"([^"]+)"`));
+		if (!m) throw new Error(`plus: apiVersion for ${action} not found`);
+		const v = { moduleVersion: versionToken, apiVersion: m[1] };
+		this.versions[kind] = v;
+		return v;
+	}
+
+	private async action<T>(
+		kind: 'plp' | 'pdp',
+		url: string,
+		viewName: string,
+		variables: Record<string, unknown>,
+		cacheKey?: string
+	): Promise<T> {
+		for (let attempt = 0; ; attempt++) {
+			const versionInfo = this.versions[kind] ?? (await this.discover(kind));
+			const r = await this.f.json<{ versionInfo?: { hasApiVersionChanged?: boolean }; data: T }>(
+				url,
+				{
+					method: 'POST',
+					headers: this.headers(true),
+					body: JSON.stringify({ versionInfo, viewName, screenData: { variables } })
+				},
+				attempt === 0 ? cacheKey : undefined
+			);
+			if (!r.versionInfo?.hasApiVersionChanged) return r.data;
+			if (attempt >= 1) throw new Error('plus: apiVersion keeps changing; PLUS is deploying?');
+			delete this.versions[kind];
+		}
+	}
+
+	listing(filter: string, page: number): Promise<PlpData> {
+		const today = new Date().toLocaleDateString('sv-SE', { timeZone: 'Europe/Amsterdam' });
+		return this.action<PlpData>(
+			'plp',
+			`${BASE}/${PLP_ACTION}`,
+			'MainFlow.ProductListPage',
+			{
+				PageNumber: page,
+				CategorySlug: '',
+				SearchKeyword: '',
+				IsSearch: false,
+				IsDesktop: true,
+				IsMobile: false,
+				IsTablet: false,
+				URLPageNumber: 0,
+				FilterQueryURL: filter,
+				SelectedSort: '',
+				StoreNumber: 0,
+				StoreChannel: '',
+				OneWelcomeUserId: '',
+				Monitoring_FlowTypeId: 0,
+				IsCustomerUnderAge: false,
+				AppliedFiltersList: { List: [] },
+				Period: { FromDate: today, ToDate: today }
+			},
+			`plp-${filter.replace(/[^a-z0-9]+/gi, '_').slice(0, 60) || 'all'}-p${page}`
+		);
+	}
+
+	/** EAN from the product detail action (the listing has none). */
+	async gtin(sku: string, slug: string): Promise<string | null> {
+		const today = new Date().toLocaleDateString('sv-SE', { timeZone: 'Europe/Amsterdam' });
+		const d = await this.action<unknown>(
+			'pdp',
+			`${BASE}/${PDP_ACTION}`,
+			'MainFlow.ProductDetailsPage',
+			{ SKU: sku, ProductSlug: slug, IsDesktop: true, Period: { FromDate: today, ToDate: today } }
+		);
+		const m = JSON.stringify(d).match(/"EAN":"(\d{8,14})"/);
+		return normaliseGtin(m?.[1]);
+	}
+}
+
+export const plusSource: Source = {
+	slug: 'plus',
+	minDelayMs: 300,
+	async fetchAll(f, log) {
+		const plus = new PlusClient(f);
+		const items = new Map<string, RawItem>();
+		const add = (d: PlpData) => {
+			for (const row of d.ProductList?.List ?? []) {
+				const p = row.PLP_Str;
+				if (!p?.SKU || !p.Name) continue;
+				const key = String(p.SKU);
+				if (!items.has(key)) items.set(key, plusItem(p));
+			}
+		};
+		// Sweep the unfiltered listing (all SKUs, 12 per page) ...
+		const first = await plus.listing('', 1);
+		add(first);
+		for (let page = 2; page <= first.TotalPages; page++) {
+			add(await plus.listing('', page));
+			if (page % 200 === 0) log(`  plus sweep: page ${page}/${first.TotalPages}, ${items.size}`);
+		}
+		log(`  plus sweep: ${items.size} of ${first.TotalNumberItems}`);
+		// ... then, if ordering shifted during the sweep, fill the gaps per category facet.
+		if (items.size < first.TotalNumberItems * 0.995) {
+			const facet = first.FilterList?.List?.map((f) => f.Filter_Str).find(
+				(f) => f.Name === 'Categorie'
+			);
+			for (const fv of facet?.FilterItem.List ?? []) {
+				if (items.size >= first.TotalNumberItems) break;
+				const p1 = await plus.listing(fv.URL, 1);
+				add(p1);
+				for (let page = 2; page <= p1.TotalPages; page++) add(await plus.listing(fv.URL, page));
+				log(`  plus ${fv.Name}: ${items.size} total`);
+			}
+		}
+		return [...items.values()];
+	}
+};

src/lib/server/groceries/sources.ts

@@ -1,10 +1,13 @@
 // Store catalogue sources. Each returns the full assortment as RawItem[].
 //  - Albert Heijn: the anonymous mobile-app API (api.ah.nl), crawled per top category.
 //  - Jumbo: the public GraphQL search endpoint used by jumbo.com, crawled per top category.
-//  - Plus: not available from the server (plus.nl blocks datacenter IPs via Incapsula).
+//  - Plus, Xenos, Amazing Oriental: see plus.ts, xenos.ts, oriental.ts.
 import type { GroceryUnit } from '../db/schema/groceries';
 import { HttpError, PoliteFetcher, USER_AGENT } from './http';
 import { normaliseGtin, parseAhUnitPrice, storeUnit } from './normalize';
+import { orientalSource } from './oriental';
+import { plusSource } from './plus';
+import { xenosSource } from './xenos';
 
 export type RawItem = {
 	storeProductId: string;
@@ -22,13 +25,26 @@ export type RawItem = {
 	imageUrl: string | null;
 	url: string | null;
 	available: boolean;
+	/** Edible (food/drink). Default: from the category (drugstore/household = false). */
+	isFood?: boolean;
 };
 
 export type Source = {
 	slug: string;
+	/** Minimum delay between requests for this store (default: the importer's --delay-ms). */
+	minDelayMs?: number;
+	/** Part of the default nightly run (default true); false = only with --store. */
+	nightly?: boolean;
+	/** PoliteFetcher overrides (retries/backoff). */
+	fetcher?: { maxRetries?: number; retryBaseMs?: number };
 	fetchAll(f: PoliteFetcher, log: (m: string) => void): Promise<RawItem[]>;
 };
 
+const NON_FOOD_CATEGORIES = new Set(['Drogisterij', 'Huishouden & dieren']);
+/** isFood for an item: the source's flag, else derived from its shared category. */
+export const itemIsFood = (r: Pick<RawItem, 'isFood' | 'category'>) =>
+	r.isFood ?? !NON_FOOD_CATEGORIES.has(r.category);
+
 const cents = (euros: unknown) =>
 	typeof euros === 'number' && Number.isFinite(euros) ? Math.round(euros * 100) : null;
 
@@ -140,7 +156,9 @@ function findKey(o: unknown, key: string, depth = 0): string | null {
 	return null;
 }
 
-function ahItem(p: AhProduct, category: string): RawItem {
+const AH_NON_FOOD_IDS = new Set([44025]); // Bloemenshop
+
+function ahItem(p: AhProduct, category: string, isFood?: boolean): RawItem {
 	const before = cents(p.priceBeforeBonus);
 	const now = cents(p.currentPrice);
 	const promoPrice = p.isBonus && now != null && before != null && now < before ? now : null;
@@ -163,7 +181,8 @@ function ahItem(p: AhProduct, category: string): RawItem {
 		gtin: null,
 		imageUrl: img?.url ?? null,
 		url: `https://www.ah.nl/producten/product/wi${p.webshopId}`,
-		available: p.orderAvailabilityStatus ? p.orderAvailabilityStatus === 'IN_ASSORTMENT' : true
+		available: p.orderAvailabilityStatus ? p.orderAvailabilityStatus === 'IN_ASSORTMENT' : true,
+		isFood
 	};
 }
 
@@ -198,7 +217,8 @@ async function crawlAhCategory(
 		for (const p of r.products ?? []) {
 			if (!p.webshopId || !p.title) continue;
 			const key = String(p.webshopId);
-			if (!items.has(key)) items.set(key, ahItem(p, category));
+			if (!items.has(key))
+				items.set(key, ahItem(p, category, AH_NON_FOOD_IDS.has(id) ? false : undefined));
 		}
 		if (!r.products?.length) break;
 	}
@@ -282,7 +302,7 @@ type JumboProduct = {
 
 const JUMBO_NOT_PROMO = /gratis bezorging|^vanaf\b|bezorg/i;
 
-function jumboItem(p: JumboProduct, category: string): RawItem {
+function jumboItem(p: JumboProduct, category: string, isFood?: boolean): RawItem {
 	const price = p.price?.price ?? null;
 	const promo = p.price?.promoPrice ?? null;
 	const ppu = p.price?.pricePerUnit;
@@ -306,7 +326,8 @@ function jumboItem(p: JumboProduct, category: string): RawItem {
 		gtin: normaliseGtin(p.ean),
 		imageUrl: p.image || null,
 		url: p.link ? new URL(p.link, 'https://www.jumbo.com').toString() : null,
-		available: (p.inAssortment ?? true) && (p.availability?.isAvailable ?? true)
+		available: (p.inAssortment ?? true) && (p.availability?.isAvailable ?? true),
+		isFood
 	};
 }
 
@@ -339,7 +360,11 @@ export const jumboSource: Source = {
 				const sp = r.data?.searchProducts;
 				for (const p of sp?.products ?? []) {
 					if (!p.id || !p.title) continue;
-					if (!items.has(p.id)) items.set(p.id, jumboItem(p, category));
+					if (!items.has(p.id))
+						items.set(
+							p.id,
+							jumboItem(p, category, root === 'nonfood-en-servicebalie' ? false : undefined)
+						);
 				}
 				if (!sp?.products?.length || offSet + JUMBO_PAGE >= sp.count) break;
 			}
@@ -349,4 +374,10 @@ export const jumboSource: Source = {
 	}
 };
 
-export const SOURCES: Record<string, Source> = { ah: ahSource, jumbo: jumboSource };
+export const SOURCES: Record<string, Source> = {
+	ah: ahSource,
+	jumbo: jumboSource,
+	plus: plusSource,
+	xenos: xenosSource,
+	'amazing-oriental': orientalSource
+};

src/lib/server/groceries/stores.test.ts

@@ -0,0 +1,35 @@
+import { describe, expect, it } from 'vitest';
+import { promoFromLabel } from './plus';
+import { xenosCategory } from './xenos';
+import { parseOrientalListing } from './oriental';
+
+describe('plus promoFromLabel', () => {
+	it('handles common multibuy labels', () => {
+		expect(promoFromLabel('1+1 GRATIS', 200)).toBe(100);
+		expect(promoFromLabel('2e halve prijs', 200)).toBe(150);
+		expect(promoFromLabel('25% korting', 400)).toBe(300);
+		expect(promoFromLabel('3 voor 5.00', 200)).toBe(167);
+		expect(promoFromLabel('0.50 KORTING', 200)).toBe(150);
+		expect(promoFromLabel('Gratis bezorging', 200)).toBeNull();
+	});
+});
+
+describe('xenosCategory', () => {
+	it('maps food and non-food', () => {
+		expect(xenosCategory(['Eten en drinken', 'Drinken', 'Koffie'])).toEqual({
+			category: 'Koffie & thee',
+			isFood: true
+		});
+		expect(xenosCategory(['Eten en drinken', 'Bakken', 'Bakvormen'])).toEqual({
+			category: 'Overig',
+			isFood: false
+		});
+		expect(xenosCategory(['Wonen', 'Kaarsen'])).toEqual({ category: 'Overig', isFood: false });
+	});
+});
+
+describe('parseOrientalListing', () => {
+	it('returns nothing for a page without products', () => {
+		expect(parseOrientalListing('<html><body>Geen producten</body></html>', 'rijst')).toEqual([]);
+	});
+});

src/lib/server/groceries/xenos.ts

@@ -0,0 +1,182 @@
+// Xenos (xenos.nl) catalogue source. Top categories come from Xenos' public Tweakwise navigation
+// gateway; products from the storefront's own JSON API (needs the session + XSRF cookies of a
+// plain home page visit). Xenos is mostly homeware: only "Eten en drinken" is food, so every
+// item gets an isFood flag and non-food lands in 'Overig' (hidden by default in the tracker).
+import type { PoliteFetcher } from './http';
+import type { RawItem, Source } from './sources';
+
+const BASE = 'https://www.xenos.nl';
+const TWEAKWISE = 'https://gateway.tweakwisenavigator.net/navigation/829e5ad2'; // public key
+const NL_ROOT = '100012'; // "Xenos NL - Master catalogus"
+const UA = 'Mozilla/5.0 (compatible; stan-grocery-tracker/0.1; +https://stanrunge.dev)';
+
+/** "Eten en drinken" level 2 → shared category. */
+const FOOD_L2: Record<string, string> = {
+	Drinken: 'Frisdrank & sappen',
+	Eten: 'Snoep, koek & chips',
+	Smaakmakers: 'Soepen, sauzen & conserven',
+	Wereldkeukens: 'Pasta, rijst & wereldkeuken'
+};
+/** Level 3 overrides. */
+const FOOD_L3: Record<string, string> = {
+	Koffie: 'Koffie & thee',
+	Thee: 'Koffie & thee',
+	Chocolademelk: 'Koffie & thee',
+	'Pasta, rijst en wraps': 'Pasta, rijst & wereldkeuken',
+	Noodles: 'Pasta, rijst & wereldkeuken',
+	Sushi: 'Pasta, rijst & wereldkeuken',
+	Panettone: 'Brood & bakkerij'
+};
+/** "Bakken" is mostly bakeware; these sub-categories are ingredients. */
+const BAKING_FOOD = new Set([
+	'Bakmixen',
+	'Fondant',
+	'Fondant en marsepein',
+	'Kleurstoffen',
+	'Taartdecoratie'
+]);
+
+type XenosProduct = {
+	sku: string | number;
+	name: string;
+	url?: string;
+	url_key?: string;
+	price_incl_tax?: number | string;
+	final_price_incl_tax?: number | string;
+	discount_percentage?: number;
+	is_salable?: boolean;
+	in_stock?: boolean;
+	default_image?: { url?: string };
+	category_paths?: { label: string }[][];
+};
+type XenosPage = { items?: XenosProduct[]; total_items?: number; last_page?: number };
+
+const euroCents = (v: unknown): number | null => {
+	const n = typeof v === 'number' ? v : parseFloat(String(v ?? ''));
+	return Number.isFinite(n) && n > 0 ? Math.round(n * 100) : null;
+};
+
+export function xenosCategory(labels: string[]): { category: string; isFood: boolean } {
+	if (labels[0] === 'Eten en drinken') {
+		if (labels[1] === 'Bakken')
+			return BAKING_FOOD.has(labels[2])
+				? { category: 'Ontbijt & beleg', isFood: true }
+				: { category: 'Overig', isFood: false };
+		return {
+			category: FOOD_L3[labels[2]] ?? FOOD_L2[labels[1]] ?? 'Overig',
+			isFood: true
+		};
+	}
+	if (labels[0] === 'Huishouden') return { category: 'Huishouden & dieren', isFood: false };
+	return { category: 'Overig', isFood: false };
+}
+
+export function xenosItem(p: XenosProduct): RawItem {
+	const paths = p.category_paths ?? [];
+	const path =
+		paths.find((x) => x[0]?.label === 'Eten en drinken') ??
+		paths.find((x) => x[0]?.label === 'Huishouden') ??
+		paths[0] ??
+		[];
+	const labels = path.map((x) => x.label);
+	const { category, isFood } = xenosCategory(labels);
+	const price = euroCents(p.price_incl_tax);
+	const final = euroCents(p.final_price_incl_tax);
+	const promo = final && price && final < price ? final : null;
+	const name = String(p.name).trim();
+	// Names end in the pack size: "Tamarinde - 360 gram".
+	const size = name.match(
+		/-\s*((?:\d+\s*x\s*)?[\d.,]+\s*(?:g|gr|gram|kg|ml|cl|l|liter|stuks?))\s*$/i
+	)?.[1];
+	return {
+		storeProductId: String(p.sku),
+		name,
+		brand: null, // not in the listing; inferred from the name at import
+		sizeText: size ?? null,
+		priceCents: price ?? final,
+		promoPriceCents: promo,
+		promoLabel: promo
+			? p.discount_percentage
+				? `${Math.round(p.discount_percentage)}% korting`
+				: 'Actie'
+			: null,
+		storeUnitPrice: null,
+		storeUnitBasisCents: null,
+		category,
+		categoryPath: labels.join(' > ') || null,
+		gtin: null,
+		imageUrl: p.default_image?.url?.replace('{size}', '500x500') ?? null,
+		url: p.url ? new URL(p.url, BASE).toString() : p.url_key ? BASE + p.url_key : null,
+		available: !!p.is_salable && !!p.in_stock,
+		isFood
+	};
+}
+
+async function session(): Promise<Record<string, string>> {
+	const res = await fetch(BASE + '/', {
+		headers: { 'User-Agent': UA },
+		signal: AbortSignal.timeout(45_000)
+	});
+	const set = res.headers.getSetCookie();
+	await res.arrayBuffer();
+	if (!res.ok) throw new Error(`xenos home -> HTTP ${res.status}`);
+	const xsrf = set.find((c) => c.startsWith('XSRF-TOKEN='));
+	if (!xsrf) throw new Error('xenos: no XSRF-TOKEN cookie');
+	return {
+		'User-Agent': UA,
+		Accept: 'application/json',
+		'X-Requested-With': 'XMLHttpRequest',
+		'X-XSRF-TOKEN': decodeURIComponent(xsrf.split(';')[0].slice('XSRF-TOKEN='.length)),
+		'X-STORE-ID': '1',
+		Cookie: set.map((c) => c.split(';')[0]).join('; ')
+	};
+}
+
+export const xenosSource: Source = {
+	slug: 'xenos',
+	minDelayMs: 300,
+	async fetchAll(f: PoliteFetcher, log) {
+		const headers = await session();
+		const nav = await f.json<{
+			facets: {
+				facetsettings: { urlkey: string };
+				attributes: { title: string; url: string }[];
+			}[];
+		}>(`${TWEAKWISE}?tn_cid=${NL_ROOT}&tn_ps=1&format=json`, { headers: { 'User-Agent': UA } });
+		const facet = nav.facets.find((x) => x.facetsettings.urlkey === 'categorie');
+		if (!facet) throw new Error('xenos: no categorie facet');
+		// Tweakwise id = "1000" + store id (1) + Magento entity id, e.g. 100011259 → 1259.
+		const cats = facet.attributes.map((a) => ({
+			id: String(a.url.split('-').pop()).slice(5),
+			title: a.title
+		}));
+		// Food first so an item listed in food and a cross-cut (Acties, Cadeaus) keeps food data.
+		cats.sort(
+			(a, b) => Number(b.title === 'Eten en drinken') - Number(a.title === 'Eten en drinken')
+		);
+		const items = new Map<string, RawItem>();
+		for (const c of cats) {
+			const before = items.size;
+			const list = (page: number) =>
+				f.json<XenosPage>(
+					`${BASE}/api/catalog/category/${c.id}/product/list?page=${page}`,
+					{
+						headers
+					},
+					`cat-${c.id}-p${page}`
+				);
+			const first = await list(1);
+			const last = first.last_page ?? 1;
+			for (let page = 1; page <= last; page++) {
+				const r = page === 1 ? first : await list(page);
+				for (const p of r.items ?? []) {
+					if (!p.sku || !p.name) continue;
+					const key = String(p.sku);
+					if (!items.has(key)) items.set(key, xenosItem(p));
+				}
+			}
+			log(`  xenos ${c.title}: +${items.size - before} (${items.size} total)`);
+		}
+		return [...items.values()];
+	}
+};

src/routes/trackers/groceries/+page.server.ts

@@ -73,6 +73,8 @@ export const load: PageServerLoad = async (event) => {
 	const favorites = p.get('fav') === '1';
 	const promo = p.get('promo') === '1';
 	const multi = p.get('multi') === '1';
+	// Homeware (Xenos décor, woks, …) is hidden unless asked for; drugstore/household stay.
+	const nonfood = p.get('nonfood') === '1';
 	const page = Math.max(1, Math.min(10_000, Number(p.get('page')) || 1));
 	const detailId = Number(p.get('g')) || null;
 
@@ -92,6 +94,7 @@ export const load: PageServerLoad = async (event) => {
 	if (favorites) conds.push(s`g.favorite`);
 	if (promo) conds.push(s`g.promo_count > 0`);
 	if (multi) conds.push(s`g.store_count >= 2`);
+	if (!nonfood) conds.push(s`(g.is_food or g.category is distinct from 'Overig')`);
 	const where = conds.reduce((a, c) => s`${a} and ${c}`);
 	const order = q
 		? s`g.favorite desc, (lower(g.name) = ${q}) desc, public.word_similarity(${q}::text, lower(g.name)) desc, g.store_count desc, g.item_count desc, g.name`
@@ -187,7 +190,15 @@ export const load: PageServerLoad = async (event) => {
 		total,
 		page,
 		pageSize: PAGE_SIZE,
-		filters: { q: p.get('q') ?? '', store: store?.slug ?? '', category, favorites, promo, multi },
+		filters: {
+			q: p.get('q') ?? '',
+			store: store?.slug ?? '',
+			category,
+			favorites,
+			promo,
+			multi,
+			nonfood
+		},
 		categories: catList,
 		stats,
 		runs,

src/routes/trackers/groceries/+page.svelte

@@ -272,6 +272,9 @@
 		<label class="check"
 			><input type="checkbox" name="multi" value="1" checked={data.filters.multi} /> In 2+ stores</label
 		>
+		<label class="check" title="Also show homeware and other non-food (mostly Xenos)"
+			><input type="checkbox" name="nonfood" value="1" checked={data.filters.nonfood} /> Non-food</label
+		>
 		<Button type="submit" variant="outline">Search</Button>
 		<div class="segmented ml-auto" role="group" aria-label="Compare by">
 			<button type="button" aria-pressed={mode === 'unit'} onclick={() => (mode = 'unit')}