Incoming Invoice GPT Update

backend/src/modules/cron/prepareIncomingInvoices.ts

@@ -8,9 +8,108 @@ import {
     files,
     filetags,
     incominginvoices,
+    vendors,
 } from "../../../db/schema"
 
-import { eq, and, isNull, not } from "drizzle-orm"
+import { eq, and, isNull, not, desc } from "drizzle-orm"
+
+type InvoiceAccount = {
+    account?: number | null
+    description?: string | null
+    taxType?: string | number | null
+}
+
+const normalizeAccounts = (accounts: unknown): InvoiceAccount[] => {
+    if (!Array.isArray(accounts)) return []
+    return accounts
+        .map((entry: any) => ({
+            account: typeof entry?.account === "number" ? entry.account : null,
+            description: typeof entry?.description === "string" ? entry.description : null,
+            taxType: entry?.taxType ?? null,
+        }))
+        .filter((entry) => entry.account !== null || entry.description || entry.taxType !== null)
+}
+
+const buildLearningContext = (historicalInvoices: any[]) => {
+    if (!historicalInvoices.length) return null
+
+    const vendorProfiles = new Map<number, {
+        vendorName: string
+        paymentTypes: Map<string, number>
+        accountUsage: Map<number, number>
+        sampleDescriptions: string[]
+    }>()
+
+    const recentExamples: any[] = []
+
+    for (const invoice of historicalInvoices) {
+        const accounts = normalizeAccounts(invoice.accounts)
+        const vendorId = typeof invoice.vendorId === "number" ? invoice.vendorId : null
+        const vendorName = typeof invoice.vendorName === "string" ? invoice.vendorName : "Unknown"
+
+        if (vendorId) {
+            if (!vendorProfiles.has(vendorId)) {
+                vendorProfiles.set(vendorId, {
+                    vendorName,
+                    paymentTypes: new Map(),
+                    accountUsage: new Map(),
+                    sampleDescriptions: [],
+                })
+            }
+
+            const profile = vendorProfiles.get(vendorId)!
+            if (invoice.paymentType) {
+                const key = String(invoice.paymentType)
+                profile.paymentTypes.set(key, (profile.paymentTypes.get(key) ?? 0) + 1)
+            }
+            for (const account of accounts) {
+                if (typeof account.account === "number") {
+                    profile.accountUsage.set(account.account, (profile.accountUsage.get(account.account) ?? 0) + 1)
+                }
+            }
+            if (invoice.description && profile.sampleDescriptions.length < 3) {
+                profile.sampleDescriptions.push(String(invoice.description).slice(0, 120))
+            }
+        }
+
+        if (recentExamples.length < 20) {
+            recentExamples.push({
+                vendorId,
+                vendorName,
+                paymentType: invoice.paymentType ?? null,
+                accounts: accounts.map((entry) => ({
+                    account: entry.account,
+                    description: entry.description ?? null,
+                    taxType: entry.taxType ?? null,
+                })),
+            })
+        }
+    }
+
+    const vendorPatterns = Array.from(vendorProfiles.entries())
+        .map(([vendorId, profile]) => {
+            const commonPaymentType = Array.from(profile.paymentTypes.entries())
+                .sort((a, b) => b[1] - a[1])[0]?.[0] ?? null
+            const topAccounts = Array.from(profile.accountUsage.entries())
+                .sort((a, b) => b[1] - a[1])
+                .slice(0, 4)
+                .map(([accountId, count]) => ({ accountId, count }))
+
+            return {
+                vendorId,
+                vendorName: profile.vendorName,
+                commonPaymentType,
+                topAccounts,
+                sampleDescriptions: profile.sampleDescriptions,
+            }
+        })
+        .slice(0, 50)
+
+    return JSON.stringify({
+        vendorPatterns,
+        recentExamples,
+    })
+}
 
 export function prepareIncomingInvoices(server: FastifyInstance) {
     const processInvoices = async (tenantId:number) => {
@@ -72,13 +171,34 @@ export function prepareIncomingInvoices(server: FastifyInstance) {
                 continue
             }
 
+            const historicalInvoices = await server.db
+                .select({
+                    vendorId: incominginvoices.vendor,
+                    vendorName: vendors.name,
+                    paymentType: incominginvoices.paymentType,
+                    description: incominginvoices.description,
+                    accounts: incominginvoices.accounts,
+                })
+                .from(incominginvoices)
+                .leftJoin(vendors, eq(incominginvoices.vendor, vendors.id))
+                .where(
+                    and(
+                        eq(incominginvoices.tenant, tenantId),
+                        eq(incominginvoices.archived, false)
+                    )
+                )
+                .orderBy(desc(incominginvoices.createdAt))
+                .limit(120)
+
+            const learningContext = buildLearningContext(historicalInvoices)
+
             // -------------------------------------------------------------
             // 3️⃣ Jede Datei einzeln durch GPT jagen & IncomingInvoice erzeugen
             // -------------------------------------------------------------
             for (const file of filesRes) {
                 console.log(`Processing file ${file.id} for tenant ${tenantId}`)
 
-                const data = await getInvoiceDataFromGPT(server,file, tenantId)
+                const data = await getInvoiceDataFromGPT(server,file, tenantId, learningContext ?? undefined)
 
                 if (!data) {
                     server.log.warn(`GPT returned no data for file ${file.id}`)