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 +1,81 @@
+/**
+ * L13 Planung — geocode sites that have no (or outdated) coordinates, respecting the OSM Nominatim
+ * usage policy: strictly sequential, max. 1 request per second (process-wide limiter in the
+ * provider), identifying User-Agent, results cached on the site (unchanged addresses are skipped).
+ * Manually entered coordinates are never overwritten.
+ *
+ * Usage:
+ *   npx tsx scripts/geocode-backfill.ts                    # all tenants
+ *   npx tsx scripts/geocode-backfill.ts --tenant=demo --limit=50
+ *   npx tsx scripts/geocode-backfill.ts --dry-run          # only list what would be geocoded
+ *   npx tsx scripts/geocode-backfill.ts --retry-not-found  # also retry addresses marked not_found
+ *
+ * For larger data sets use a self-hosted or contracted geocoding service (GEOCODING_URL).
+ */
+import "dotenv/config";
+import { isValidLatLng } from "../src/lib/geo/distance";
+import { dbForTenant, prisma } from "../src/server/db";
+import { geocodingConfig } from "../src/server/services/geo/config";
+import { geocodeSite, type GeocodeOutcome } from "../src/server/services/geo/geocode-site";
+import { geocodeQueryFor } from "../src/server/services/geo/normalize";
+
+const arg = (name: string) => process.argv.find((a) => a.startsWith(`--${name}=`))?.split("=")[1];
+const flag = (name: string) => process.argv.includes(`--${name}`);
+
+async function main() {
+  const cfg = geocodingConfig();
+  const dryRun = flag("dry-run");
+  const retryNotFound = flag("retry-not-found");
+  const limit = Number(arg("limit") ?? Infinity);
+  const slug = arg("tenant");
+  if (cfg.provider === "none" && !dryRun) {
+    console.log("GEOCODING_PROVIDER=none — nichts zu tun.");
+    return;
+  }
+  console.log(`Provider ${cfg.provider} (${cfg.url}), User-Agent "${cfg.userAgent}", ≤ 1 Anfrage/s${dryRun ? " — Probelauf" : ""}`);
+
+  const tenants = await prisma.tenant.findMany({ where: slug ? { slug } : {}, select: { id: true, slug: true }, orderBy: { slug: "asc" } });
+  const totals: Record<string, number> = {};
+  let done = 0;
+  for (const tenant of tenants) {
+    const db = dbForTenant(tenant.id);
+    const sites = await db.site.findMany({
+      where: { deletedAt: null },
+      select: { id: true, name: true, street: true, houseNumber: true, postalCode: true, city: true, country: true, latitude: true, longitude: true, geocodedAt: true, geocodeStatus: true, geocodeQuery: true },
+      orderBy: { createdAt: "asc" },
+    });
+    const pending = sites.filter((s) => {
+      const query = geocodeQueryFor(s);
+      if (!query) return false;
+      if (isValidLatLng(s) && !s.geocodedAt) return false; // manual coordinates
+      if (s.geocodeQuery !== query) return true;
+      if (s.geocodeStatus === "ok") return false;
+      if (s.geocodeStatus === "not_found") return retryNotFound;
+      return true;
+    });
+    if (!pending.length) continue;
+    console.log(`\n${tenant.slug}: ${pending.length} Objekt(e)`);
+    for (const site of pending) {
+      if (done >= limit) break;
+      done++;
+      if (dryRun) {
+        console.log(`  · ${site.name} (${[site.postalCode, site.city].filter(Boolean).join(" ")})`);
+        continue;
+      }
+      if (retryNotFound && site.geocodeStatus === "not_found") await db.site.update({ where: { id: site.id }, data: { geocodeStatus: null } });
+      const outcome: GeocodeOutcome = await geocodeSite(tenant.id, site.id);
+      totals[outcome] = (totals[outcome] ?? 0) + 1;
+      console.log(`  ${outcome.padEnd(9)} ${site.name}`);
+    }
+  }
+  console.log(`\nFertig: ${done} Objekt(e) ${JSON.stringify(totals)}`);
+}
+
+main()
+  .catch((err) => {
+    console.error(err);
+    process.exitCode = 1;
+  })
+  .finally(async () => {
+    await prisma.$disconnect();
+  });