L3 Auftragsimport: Extraktion, Plausibilität, Dubletten, Bestätigung (Services + Tests)

- Claude-Extraktion (PDF nativ/Bild, Structured Output, Konfidenzen, Volltext), FakeProvider - Plausibilitätsprüfung, Mapping Extraktion → Formular, Korrektur-Diff - Services Upload/Verarbeitung/Bestätigung/Verwerfen/Neu verarbeiten, Processor import-extraction - Stubs: storeFile (§4.3), findDuplicateCustomers (L1), createWorkOrder (L2) - Tests test-import-rules/-flow/-live, Beispiel-PDFs + Generator Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-09-14 12:17:13 +02:00
parent bf4456718e
commit 12a764786a
22 changed files with 2592 additions and 1 deletions
@@ -0,0 +1,142 @@
+import type { TenantDb } from "@/server/db";
+import type { DuplicateCandidate, SiteCandidate } from "@/lib/imports/extraction";
+
+/**
+ * STUB (lane L3) for the L1 contract `src/lib/customers/duplicates.ts#findDuplicateCustomers(db, candidate) → Candidate[]`
+ * (ARCHITEKTUR §6, spec §7.3). Same signature and result shape `{ customerId, score, reasons[] }`;
+ * replace the import in services/imports/process.ts after the L1 merge.
+ *
+ * Compares customer number, company/person name, address, e-mail and phone. Never merges —
+ * it only proposes candidates; the backoffice decides (US-003).
+ */
+
+export type CustomerCandidateInput = {
+  customerNumber?: string | null;
+  companyName?: string | null;
+  firstName?: string | null;
+  lastName?: string | null;
+  street?: string | null;
+  houseNumber?: string | null;
+  postalCode?: string | null;
+  city?: string | null;
+  email?: string | null;
+  phone?: string | null;
+};
+
+const LEGAL_FORMS = /\b(gmbh|mbh|ag|kg|ohg|gbr|ug|e\.?\s?k|e\.?\s?v|co|haftungsbeschränkt|und|&)\b/g;
+
+export function normalizeCompany(v: string | null | undefined): string {
+  return (v ?? "")
+    .toLowerCase()
+    .replace(/[.,;:()"'`´+/-]/g, " ")
+    .replace(LEGAL_FORMS, " ")
+    .replace(/\s+/g, " ")
+    .trim();
+}
+
+export function normalizeStreet(v: string | null | undefined): string {
+  return (v ?? "")
+    .toLowerCase()
+    .replace(/ß/g, "ss")
+    .replace(/strasse|str\./g, "str")
+    .replace(/[^a-z0-9äöü]/g, "");
+}
+
+const digits = (v: string | null | undefined) => (v ?? "").replace(/\D/g, "").replace(/^49/, "0").replace(/^00/, "0");
+const lc = (v: string | null | undefined) => (v ?? "").trim().toLowerCase();
+
+export async function findDuplicateCustomers(db: TenantDb, candidate: CustomerCandidateInput): Promise<DuplicateCandidate[]> {
+  const company = normalizeCompany(candidate.companyName);
+  const firstWord = company.split(" ").find((w) => w.length >= 3);
+  const or: object[] = [];
+  if (candidate.customerNumber) or.push({ customerNumber: candidate.customerNumber.trim() });
+  if (candidate.email) or.push({ email: { equals: candidate.email.trim(), mode: "insensitive" } });
+  if (firstWord) or.push({ companyName: { contains: firstWord, mode: "insensitive" } });
+  if (candidate.lastName) or.push({ lastName: { equals: candidate.lastName.trim(), mode: "insensitive" } });
+  if (candidate.postalCode) or.push({ postalCode: candidate.postalCode.trim() });
+  if (candidate.phone) or.push({ phone: { not: null } });
+  if (or.length === 0) return [];
+
+  const rows = await db.customer.findMany({
+    where: { deletedAt: null, status: { not: "merged" }, OR: or },
+    select: { id: true, customerNumber: true, companyName: true, firstName: true, lastName: true, street: true, houseNumber: true, postalCode: true, email: true, phone: true, mobile: true },
+    take: 200,
+  });
+
+  const out: DuplicateCandidate[] = [];
+  for (const c of rows) {
+    let score = 0;
+    const reasons: string[] = [];
+    if (candidate.customerNumber && c.customerNumber && c.customerNumber.trim() === candidate.customerNumber.trim()) {
+      score += 0.6;
+      reasons.push("customer_number");
+    }
+    const cc = normalizeCompany(c.companyName);
+    if (company && cc) {
+      if (cc === company) {
+        score += 0.4;
+        reasons.push("company_name");
+      } else if (cc.includes(company) || company.includes(cc)) {
+        score += 0.25;
+        reasons.push("company_name_similar");
+      }
+    }
+    if (candidate.lastName && lc(c.lastName) === lc(candidate.lastName) && (!candidate.firstName || lc(c.firstName) === lc(candidate.firstName))) {
+      score += 0.3;
+      reasons.push("person_name");
+    }
+    if (candidate.email && lc(c.email) === lc(candidate.email)) {
+      score += 0.3;
+      reasons.push("email");
+    }
+    const phone = digits(candidate.phone);
+    if (phone.length >= 6 && [c.phone, c.mobile].some((p) => digits(p) === phone)) {
+      score += 0.2;
+      reasons.push("phone");
+    }
+    if (
+      candidate.postalCode &&
+      c.postalCode === candidate.postalCode.trim() &&
+      normalizeStreet(c.street) !== "" &&
+      normalizeStreet(c.street) === normalizeStreet(candidate.street)
+    ) {
+      score += 0.25;
+      reasons.push("address");
+    }
+    if (score >= 0.25) out.push({ customerId: c.id, score: Math.min(1, Math.round(score * 100) / 100), reasons });
+  }
+  return out.sort((a, b) => b.score - a.score).slice(0, 5);
+}
+
+/** Sites of the candidate customers at the same address (lane L3, spec §9.3 step 7). */
+export async function findSiteCandidates(
+  db: TenantDb,
+  customerIds: string[],
+  address: { name?: string | null; street?: string | null; houseNumber?: string | null; postalCode?: string | null } | null,
+): Promise<SiteCandidate[]> {
+  if (!customerIds.length || !address || (!address.street && !address.name)) return [];
+  const sites = await db.site.findMany({
+    where: { customerId: { in: customerIds }, deletedAt: null },
+    select: { id: true, customerId: true, name: true, street: true, houseNumber: true, postalCode: true },
+    take: 200,
+  });
+  const out: SiteCandidate[] = [];
+  for (const s of sites) {
+    let score = 0;
+    const reasons: string[] = [];
+    if (address.street && normalizeStreet(s.street) === normalizeStreet(address.street) && (!address.postalCode || s.postalCode === address.postalCode)) {
+      score += 0.6;
+      reasons.push("address");
+      if (address.houseNumber && lc(s.houseNumber) === lc(address.houseNumber)) {
+        score += 0.3;
+        reasons.push("house_number");
+      }
+    }
+    if (address.name && lc(s.name) === lc(address.name)) {
+      score += 0.3;
+      reasons.push("site_name");
+    }
+    if (score >= 0.3) out.push({ siteId: s.id, customerId: s.customerId, name: s.name, score: Math.min(1, Math.round(score * 100) / 100), reasons });
+  }
+  return out.sort((a, b) => b.score - a.score).slice(0, 5);
+}