houses: one row per listing when a hand-added link is scraped later

The nightly scraper and worker API now take over an existing row (e.g. source=manual)
whose link is the same listing (normalized URL, or the Pararius key / Funda, Kamernet,
Vesteda id in it) instead of inserting a copy; typed values are kept where the scrape has
none. house-listing update/add give manual rows with such links the site key unless
another row has it, update accepts --source/--source-id, and lookups by link ignore
tracking params. scripts/house-rekey.ts backfills existing manual rows.

src/lib/server/house-tracker.ts

@@ -8,7 +8,7 @@ import {
 	HOUSE_NTFY_TOKEN
 } from '$app/env/private';
 import { error } from '@sveltejs/kit';
-import { and, desc, eq, inArray, isNull, lt, sql } from 'drizzle-orm';
+import { and, desc, eq, ilike, inArray, isNull, lt, ne, or, sql } from 'drizzle-orm';
 import { db } from '#lib/server/db/index.js';
 import {
 	houseAlertRules,
@@ -23,6 +23,13 @@ import {
 	type AlertMessage,
 	type AlertRule
 } from '#lib/house-tracker/alerts.js';
+import {
+	containsPattern,
+	linkHost,
+	pickLinkedListing,
+	withoutEmpty,
+	type LinkCandidate
+} from '#lib/house-tracker/link-key.js';
 import { listingRef } from '#lib/house-tracker/pararius.js';
 import type {
 	DetailsRequest,
@@ -57,6 +64,31 @@ async function recordPrice(
 	await tx.insert(housePriceHistory).values({ listingId, priceEur: next });
 }
 
+/** Rows stored under another source (e.g. a hand-added Pararius link) on the same host. */
+async function linkCandidates(
+	tx: Tx,
+	source: string,
+	urls: (string | null | undefined)[]
+): Promise<LinkCandidate[]> {
+	const hosts = [...new Set(urls.map(linkHost).filter((h) => h != null))];
+	if (!hosts.length) return [];
+	return tx
+		.select({
+			id: houseListings.id,
+			source: houseListings.source,
+			sourceId: houseListings.sourceId,
+			url: houseListings.url,
+			duplicateOf: houseListings.duplicateOf
+		})
+		.from(houseListings)
+		.where(
+			and(
+				ne(houseListings.source, source),
+				or(...hosts.map((host) => ilike(houseListings.url, containsPattern(host))))
+			)
+		);
+}
+
 export async function syncCards(input: SyncRequest): Promise<SyncResponse> {
 	const sourceIds = input.cards.map((card) => card.sourceId);
 	const toAlert: { id: number; message: AlertMessage }[] = [];
@@ -99,9 +131,35 @@ export async function syncCards(input: SyncRequest): Promise<SyncResponse> {
 					)
 			: [];
 		const known = new Map(existing.map((row) => [row.sourceId, row]));
+		// Listings already added by hand under another source: take those rows over
+		// (new source key, scraped values fill the gaps) instead of inserting copies.
+		let candidates = await linkCandidates(
+			tx,
+			input.source,
+			input.cards.filter((card) => !known.has(card.sourceId)).map((card) => card.url)
+		);
+		const adopt = async (card: ListingCard) => {
+			const linked = pickLinkedListing(
+				candidates,
+				{ source: input.source, sourceId: card.sourceId },
+				card.url
+			);
+			if (!linked) return null;
+			candidates = candidates.filter((c) => c.id !== linked.id);
+			const [row] = await tx
+				.update(houseListings)
+				.set({ source: input.source, sourceId: card.sourceId })
+				.where(eq(houseListings.id, linked.id))
+				.returning({
+					id: houseListings.id,
+					sourceId: houseListings.sourceId,
+					priceEur: houseListings.priceEur
+				});
+			return row ?? null;
+		};
 
 		for (const card of input.cards) {
-			const previous = known.get(card.sourceId);
+			const previous = known.get(card.sourceId) ?? (await adopt(card));
 			if (previous) {
 				await tx
 					.update(houseListings)
@@ -217,12 +275,27 @@ export async function saveDetails(input: DetailsRequest) {
 	let saved = 0;
 	await db.transaction(async (tx) => {
 		for (const detail of input.details) {
-			const [previous] = await tx
+			let [previous] = await tx
 				.select({ id: houseListings.id, priceEur: houseListings.priceEur })
 				.from(houseListings)
 				.where(
 					and(eq(houseListings.source, input.source), eq(houseListings.sourceId, detail.sourceId))
 				);
+			let adopted = false;
+			if (!previous) {
+				const linked = pickLinkedListing(
+					await linkCandidates(tx, input.source, [detail.url]),
+					{ source: input.source, sourceId: detail.sourceId },
+					detail.url
+				);
+				if (linked) {
+					[previous] = await tx
+						.select({ id: houseListings.id, priceEur: houseListings.priceEur })
+						.from(houseListings)
+						.where(eq(houseListings.id, linked.id));
+					adopted = Boolean(previous);
+				}
+			}
 			const values = {
 				externalId: detail.externalId,
 				url: detail.url,
@@ -256,7 +329,11 @@ export async function saveDetails(input: DetailsRequest) {
 				lastSeenAt: sql`now()`
 			};
 			if (previous) {
-				await tx.update(houseListings).set(values).where(eq(houseListings.id, previous.id));
+				// A hand-added row keeps what was typed where the page has nothing.
+				const set = adopted
+					? { ...withoutEmpty(values), source: input.source, sourceId: detail.sourceId }
+					: values;
+				await tx.update(houseListings).set(set).where(eq(houseListings.id, previous.id));
 				await recordPrice(tx, previous.id, previous.priceEur, detail.priceEur ?? null);
 			} else {
 				const [row] = await tx