L13 Planung: Geocoding der Objekte über OpenStreetMap Nominatim

Provider-Interface (nominatim|none), strukturierte Suche mit User-Agent und Accept-Language, Drosselung 1/s je Prozess, Job geocode-site (Worker: Concurrency 1 + Limiter), Cache am Objekt, manuelle Koordinaten bleiben, Auslöser Anlage/Adressänderung/Import-Bestätigung nur per Queue, Backfill-Skript, Env-Beispiele. Tests mit Fake-Provider, Nominatim nie aufgerufen. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-09-15 10:13:26 +02:00
parent 604f2bfdc5
commit 213bcca3a1
14 changed files with 600 additions and 2 deletions
@@ -0,0 +1,42 @@
+import packageJson from "../../../../package.json";
+
+/**
+ * Geocoding + map configuration (L13 Planung). Defaults follow the OpenStreetMap usage policies:
+ * Nominatim only server side, max. 1 request/s, identifying User-Agent, results cached per site.
+ * For production with more than a handful of addresses a self-hosted / contracted geocoding and
+ * tile service is recommended (see docs/craftvia/lanes/planung.md).
+ */
+
+export const GEOCODING_PROVIDERS = ["nominatim", "none"] as const;
+export type GeocodingProviderName = (typeof GEOCODING_PROVIDERS)[number];
+
+export const DEFAULT_NOMINATIM_URL = "https://nominatim.openstreetmap.org";
+export const DEFAULT_MAP_TILE_URL = "https://tile.openstreetmap.org/{z}/{x}/{y}.png";
+export const DEFAULT_MAP_ATTRIBUTION = "© OpenStreetMap-Mitwirkende";
+
+type Env = Record<string, string | undefined>;
+const val = (v: string | undefined) => (v && v.trim() ? v.trim() : undefined);
+
+export function geocodingProviderName(env: Env = process.env): GeocodingProviderName {
+  const v = val(env.GEOCODING_PROVIDER)?.toLowerCase();
+  return v === "none" ? "none" : "nominatim";
+}
+
+export function geocodingConfig(env: Env = process.env) {
+  const baseUrl = val(env.APP_BASE_URL) ?? val(env.AUTH_URL) ?? "http://localhost:3000";
+  return {
+    provider: geocodingProviderName(env),
+    url: (val(env.GEOCODING_URL) ?? DEFAULT_NOMINATIM_URL).replace(/\/+$/, ""),
+    userAgent: val(env.GEOCODING_USER_AGENT) ?? `Craftvia/${packageJson.version} (+${baseUrl})`,
+    timeoutMs: 8_000,
+    minIntervalMs: 1_000,
+  };
+}
+
+/** Tile URL + attribution for the live map (passed from server pages to the client). */
+export function mapConfig(env: Env = process.env) {
+  return {
+    tileUrl: val(env.MAP_TILE_URL) ?? DEFAULT_MAP_TILE_URL,
+    attribution: val(env.MAP_ATTRIBUTION) ?? DEFAULT_MAP_ATTRIBUTION,
+  };
+}
@@ -0,0 +1,24 @@
+import { enqueueJob } from "@/server/jobs/queues";
+import { geocodingProviderName } from "@/server/services/geo/config";
+
+/**
+ * Ask the worker to geocode a site (after create / address change / import confirmation).
+ * Never runs inline: no external request inside a user request (OSM policy: no bulk geocoding in
+ * requests). If the queue is not reachable yet (lazy Redis connection) one delayed retry is made;
+ * otherwise the site stays without coordinates ("ohne Ortsangabe") until the next change or
+ * `scripts/geocode-backfill.ts`. Never throws.
+ */
+export async function requestSiteGeocoding(tenantId: string, siteId: string): Promise<boolean> {
+  if (geocodingProviderName() === "none") return false;
+  const payload = { tenantId, entityId: siteId };
+  try {
+    if (await enqueueJob("geocode-site", payload)) return true;
+  } catch (err) {
+    console.error("[geo] enqueue geocode-site failed:", (err as Error).message);
+  }
+  const timer = setTimeout(() => {
+    enqueueJob("geocode-site", payload).catch(() => undefined);
+  }, 1_500);
+  timer.unref?.();
+  return false;
+}
@@ -0,0 +1,92 @@
+import { isValidLatLng } from "@/lib/geo/distance";
+import { writeAuditLog } from "@/server/audit";
+import { dbForTenant } from "@/server/db";
+import { geocodeQueryFor } from "@/server/services/geo/normalize";
+import { getGeocodingProvider, type GeocodingProvider } from "@/server/services/geo/provider";
+
+/**
+ * Geocode one site and cache the result on the row (job `geocode-site`, backfill script).
+ *
+ * Rules:
+ * - Manually entered coordinates (coordinates without `geocodedAt`) are never overwritten.
+ * - Unchanged address (`geocodeQuery`) with status ok/not_found → no provider call ("cached").
+ * - No usable address or provider `none` → status `skipped` (retried once enabled/complete).
+ * - Transport/HTTP error → status `failed` (the job rethrows so BullMQ retries).
+ * - Address changed and not found → previous automatic coordinates are removed (they would be wrong).
+ */
+
+export type GeocodeOutcome = "ok" | "not_found" | "failed" | "skipped" | "cached" | "manual" | "missing";
+
+const SITE_SELECT = {
+  id: true,
+  street: true,
+  houseNumber: true,
+  postalCode: true,
+  city: true,
+  country: true,
+  latitude: true,
+  longitude: true,
+  geocodedAt: true,
+  geocodeStatus: true,
+  geocodeQuery: true,
+} as const;
+
+export async function geocodeSite(
+  tenantId: string,
+  siteId: string,
+  opts: { provider?: GeocodingProvider | null; now?: Date } = {},
+): Promise<GeocodeOutcome> {
+  const db = dbForTenant(tenantId);
+  const site = await db.site.findFirst({ where: { id: siteId, deletedAt: null }, select: SITE_SELECT });
+  if (!site) return "missing";
+  const now = opts.now ?? new Date();
+  const query = geocodeQueryFor(site);
+
+  if (isValidLatLng(site) && !site.geocodedAt) {
+    if (site.geocodeStatus !== "skipped" || site.geocodeQuery !== query) {
+      await db.site.update({ where: { id: site.id }, data: { geocodeStatus: "skipped", geocodeQuery: query } });
+    }
+    return "manual";
+  }
+
+  if (!query) {
+    const clearAuto = site.geocodedAt !== null && (site.latitude !== null || site.longitude !== null);
+    await db.site.update({
+      where: { id: site.id },
+      data: { geocodeStatus: "skipped", geocodeQuery: null, ...(clearAuto ? { latitude: null, longitude: null } : {}) },
+    });
+    return "skipped";
+  }
+
+  if (site.geocodeQuery === query && (site.geocodeStatus === "ok" || site.geocodeStatus === "not_found")) return "cached";
+
+  const provider = opts.provider === undefined ? await getGeocodingProvider() : opts.provider;
+  if (!provider) {
+    if (site.geocodeStatus !== "skipped") await db.site.update({ where: { id: site.id }, data: { geocodeStatus: "skipped" } });
+    return "skipped";
+  }
+
+  let result;
+  try {
+    result = await provider.geocode({ street: site.street, houseNumber: site.houseNumber, postalCode: site.postalCode, city: site.city, country: site.country });
+  } catch (err) {
+    console.error(`[geo] geocoding site ${site.id} failed:`, (err as Error).message);
+    await db.site.update({ where: { id: site.id }, data: { geocodeStatus: "failed", geocodeQuery: query, geocodedAt: now } });
+    return "failed";
+  }
+
+  const before = { latitude: site.latitude, longitude: site.longitude, geocodeStatus: site.geocodeStatus };
+  const coords = result.status === "ok" ? { latitude: result.latitude, longitude: result.longitude } : { latitude: null, longitude: null };
+  await db.site.update({ where: { id: site.id }, data: { ...coords, geocodeStatus: result.status, geocodeQuery: query, geocodedAt: now } });
+  if (before.latitude !== coords.latitude || before.longitude !== coords.longitude) {
+    await writeAuditLog({
+      tenantId,
+      action: "update",
+      entity: "site",
+      entityId: site.id,
+      before,
+      after: { ...coords, geocodeStatus: result.status, source: "geocoding", provider: provider.name },
+    });
+  }
+  return result.status;
+}
@@ -0,0 +1,58 @@
+import { GeocodingError, type GeocodeAddress, type GeocodeResult, type GeocodingProvider } from "@/server/services/geo/provider";
+import { nominatimLimiter } from "@/server/services/geo/rate-limit";
+
+/**
+ * Nominatim (OpenStreetMap) provider. Usage policy: server side only, max. 1 request/s process-wide,
+ * identifying User-Agent, results are cached in Site.latitude/longitude by the caller, no bulk
+ * geocoding inside requests (only the geocode-site job and scripts/geocode-backfill.ts call this).
+ */
+
+type Config = { url: string; userAgent: string; timeoutMs: number };
+type Deps = { fetch: typeof fetch; limiter: { acquire(): Promise<number> } };
+
+/** Structured search URL (no free text → fewer ambiguous hits). */
+export function buildNominatimUrl(baseUrl: string, address: GeocodeAddress): URL {
+  const url = new URL(`${baseUrl.replace(/\/+$/, "")}/search`);
+  url.searchParams.set("format", "jsonv2");
+  url.searchParams.set("limit", "1");
+  url.searchParams.set("addressdetails", "0");
+  const street = [address.houseNumber, address.street].map((s) => s?.trim()).filter(Boolean).join(" ");
+  if (street) url.searchParams.set("street", street);
+  if (address.postalCode?.trim()) url.searchParams.set("postalcode", address.postalCode.trim());
+  if (address.city?.trim()) url.searchParams.set("city", address.city.trim());
+  const country = (address.country ?? "DE").trim().toLowerCase();
+  if (/^[a-z]{2}$/.test(country)) url.searchParams.set("countrycodes", country);
+  return url;
+}
+
+export function createNominatimProvider(config: Config, deps: Deps = { fetch: globalThis.fetch, limiter: nominatimLimiter }): GeocodingProvider {
+  return {
+    name: "nominatim",
+    async geocode(address: GeocodeAddress): Promise<GeocodeResult> {
+      await deps.limiter.acquire();
+      let res: Response;
+      try {
+        res = await deps.fetch(buildNominatimUrl(config.url, address), {
+          headers: { "User-Agent": config.userAgent, "Accept-Language": "de", Accept: "application/json" },
+          signal: AbortSignal.timeout(config.timeoutMs),
+        });
+      } catch (err) {
+        throw new GeocodingError(`request failed: ${(err as Error).name}`);
+      }
+      if (!res.ok) throw new GeocodingError(`http ${res.status}`);
+      let body: unknown;
+      try {
+        body = await res.json();
+      } catch {
+        throw new GeocodingError("invalid response");
+      }
+      const hit = Array.isArray(body) ? (body[0] as { lat?: unknown; lon?: unknown } | undefined) : undefined;
+      const latitude = Number(hit?.lat);
+      const longitude = Number(hit?.lon);
+      if (!hit || !Number.isFinite(latitude) || !Number.isFinite(longitude) || Math.abs(latitude) > 90 || Math.abs(longitude) > 180) {
+        return { status: "not_found" };
+      }
+      return { status: "ok", latitude, longitude };
+    },
+  };
+}
@@ -0,0 +1,27 @@
+/**
+ * Normalized address string of a site ("hafenstraße 12|20457|hamburg|de"). Stored in
+ * Site.geocodeQuery next to the coordinates: an unchanged address is never geocoded twice.
+ * Returns null when the address is too incomplete to geocode (neither postal code nor city).
+ */
+export type SiteAddress = {
+  street?: string | null;
+  houseNumber?: string | null;
+  postalCode?: string | null;
+  city?: string | null;
+  country?: string | null;
+};
+
+const clean = (v: string | null | undefined) => (v ?? "").normalize("NFC").trim().replace(/\s+/g, " ").toLowerCase();
+
+export function geocodeQueryFor(site: SiteAddress): string | null {
+  const postal = clean(site.postalCode);
+  const city = clean(site.city);
+  if (!postal && !city) return null;
+  const street = [clean(site.street), clean(site.houseNumber)].filter(Boolean).join(" ");
+  return [street, postal, city, clean(site.country) || "de"].join("|");
+}
+
+/** True when the geocoding-relevant part of the address changed. */
+export function addressChanged(before: SiteAddress, after: SiteAddress): boolean {
+  return geocodeQueryFor(before) !== geocodeQueryFor(after);
+}
@@ -0,0 +1,41 @@
+import { geocodingConfig, geocodingProviderName } from "@/server/services/geo/config";
+
+/** Provider-neutral geocoding contract (L13 Planung) — swap Nominatim for another service via env. */
+
+export type GeocodeAddress = {
+  street: string | null;
+  houseNumber: string | null;
+  postalCode: string | null;
+  city: string | null;
+  country: string | null;
+};
+
+export type GeocodeResult = { status: "ok"; latitude: number; longitude: number } | { status: "not_found" };
+
+export interface GeocodingProvider {
+  readonly name: string;
+  /** Resolves ok/not_found; throws GeocodingError for transport/HTTP failures (retryable). */
+  geocode(address: GeocodeAddress): Promise<GeocodeResult>;
+}
+
+export class GeocodingError extends Error {
+  constructor(message: string) {
+    super(message);
+    this.name = "GeocodingError";
+  }
+}
+
+let override: GeocodingProvider | null | undefined;
+
+/** Tests inject a fake provider (never Nominatim); `undefined` restores the env-based provider. */
+export function setGeocodingProvider(provider: GeocodingProvider | null | undefined): void {
+  override = provider;
+}
+
+/** Configured provider, or null when geocoding is disabled (`GEOCODING_PROVIDER=none`). */
+export async function getGeocodingProvider(): Promise<GeocodingProvider | null> {
+  if (override !== undefined) return override;
+  if (geocodingProviderName() === "none") return null;
+  const { createNominatimProvider } = await import("@/server/services/geo/nominatim");
+  return createNominatimProvider(geocodingConfig());
+}
@@ -0,0 +1,24 @@
+/**
+ * Minimum-interval limiter: callers are released one after another with at least `intervalMs`
+ * between two slots (Nominatim policy: max. 1 request per second). Clock is injectable for tests.
+ */
+export type Clock = { now: () => number; sleep: (ms: number) => Promise<void> };
+
+const realClock: Clock = { now: () => Date.now(), sleep: (ms) => new Promise((resolve) => setTimeout(resolve, ms)) };
+
+export function createIntervalLimiter(intervalMs: number, clock: Clock = realClock) {
+  let next = 0;
+  return {
+    /** Waits for the next free slot and returns its timestamp. */
+    async acquire(): Promise<number> {
+      const now = clock.now();
+      const slot = Math.max(now, next);
+      next = slot + intervalMs;
+      if (slot > now) await clock.sleep(slot - now);
+      return slot;
+    },
+  };
+}
+
+/** Process-wide limiter for all Nominatim requests of this process (worker, backfill script). */
+export const nominatimLimiter = createIntervalLimiter(1_000);