feat(tests): switch vision tests to multi-query extraction (count then per-row/field queries) and add logging/summaries

test/test.invoices.qwen3vl.ts

@@ -1,10 +1,8 @@
 /**
  * Invoice extraction using Qwen3-VL 8B Vision (Direct)
  *
- * Single-step pipeline: PDF → Images → Qwen3-VL → JSON
- * Uses /no_think to disable reasoning mode for fast, direct responses.
- *
- * Qwen3-VL outperforms PaddleOCR-VL on certain invoice formats.
+ * Multi-query approach: 5 parallel simple queries to avoid token exhaustion.
+ * Single pass, no consensus voting.
  */
 import { tap, expect } from '@git.zone/tstest/tapbundle';
 import * as fs from 'fs';
@@ -67,11 +65,10 @@ async function queryField(images: string[], question: string): Promise<string> {
       model: VISION_MODEL,
       messages: [{
         role: 'user',
-        content: `/no_think\n${question} Reply with just the value, nothing else.`,
+        content: `${question} Reply with just the value, nothing else.`,
         images: images,
       }],
       stream: false,
-      think: false,
       options: {
         num_predict: 500,
         temperature: 0.1,
@@ -96,35 +93,80 @@ async function extractInvoiceFromImages(images: string[]): Promise<IInvoice> {
   console.log(`    [Vision] Processing ${images.length} page(s) with Qwen3-VL (multi-query)`);
 
   // Query each field separately to avoid excessive thinking tokens
-  const [invoiceNum, invoiceDate, vendor, currency, amounts] = await Promise.all([
-    queryField(images, 'What is the invoice number on this document?'),
-    queryField(images, 'What is the invoice date? Format as YYYY-MM-DD.'),
-    queryField(images, 'What company issued this invoice?'),
-    queryField(images, 'What currency is used? Answer EUR, USD, or GBP.'),
-    queryField(images, 'What are the net amount, VAT amount, and total amount? Format: net,vat,total'),
+  // Use explicit questions to avoid confusion between similar fields
+  // Log each result as it comes in (not waiting for all to complete)
+  const queryAndLog = async (name: string, question: string): Promise<string> => {
+    const result = await queryField(images, question);
+    console.log(`    [Query] ${name}: "${result}"`);
+    return result;
+  };
+
+  const [invoiceNum, invoiceDate, vendor, currency, totalAmount, netAmount, vatAmount] = await Promise.all([
+    queryAndLog('Invoice Number', 'What is the INVOICE NUMBER (not VAT number, not customer ID)? Look for "Invoice No", "Invoice #", "Rechnung Nr", "Facture". Just the number/code.'),
+    queryAndLog('Invoice Date  ', 'What is the INVOICE DATE (not due date, not delivery date)? The date the invoice was issued. Format: YYYY-MM-DD'),
+    queryAndLog('Vendor        ', 'What company ISSUED this invoice (the seller/vendor, not the buyer)? Look at the letterhead or "From" section.'),
+    queryAndLog('Currency      ', 'What CURRENCY is used? Look for € (EUR), $ (USD), or £ (GBP). Answer with 3-letter code: EUR, USD, or GBP'),
+    queryAndLog('Total Amount  ', 'What is the TOTAL AMOUNT INCLUDING TAX (the final amount to pay, with VAT/tax included)? Just the number, e.g. 24.99'),
+    queryAndLog('Net Amount    ', 'What is the NET AMOUNT (subtotal before VAT/tax)? Just the number, e.g. 20.99'),
+    queryAndLog('VAT Amount    ', 'What is the VAT/TAX AMOUNT? Just the number, e.g. 4.00'),
   ]);
 
-  console.log(`    [Vision] Got: ${invoiceNum} | ${invoiceDate} | ${vendor} | ${currency}`);
-
-  // Parse amounts (format: "net,vat,total" or similar)
-  const amountMatch = amounts.match(/([\d.,]+)/g) || [];
+  // Parse amount from string (handles European format)
   const parseAmount = (s: string): number => {
     if (!s) return 0;
+    // Extract number from the response
+    const match = s.match(/([\d.,]+)/);
+    if (!match) return 0;
+    const numStr = match[1];
     // Handle European format: 1.234,56 → 1234.56
-    const normalized = s.includes(',') && s.indexOf(',') > s.lastIndexOf('.')
-      ? s.replace(/\./g, '').replace(',', '.')
-      : s.replace(/,/g, '');
+    const normalized = numStr.includes(',') && numStr.indexOf(',') > numStr.lastIndexOf('.')
+      ? numStr.replace(/\./g, '').replace(',', '.')
+      : numStr.replace(/,/g, '');
     return parseFloat(normalized) || 0;
   };
 
+  // Extract invoice number from potentially verbose response
+  const extractInvoiceNumber = (s: string): string => {
+    let clean = s.replace(/\*\*/g, '').replace(/`/g, '').trim();
+    // Look for common invoice number patterns
+    const patterns = [
+      /\b([A-Z]{2,3}\d{10,})\b/i,           // IEE2022006460244
+      /\b([A-Z]\d{8,})\b/i,                  // R0014359508
+      /\b(INV[-\s]?\d{4}[-\s]?\d+)\b/i,     // INV-2024-001
+      /\b(\d{7,})\b/,                        // 1579087430
+    ];
+    for (const pattern of patterns) {
+      const match = clean.match(pattern);
+      if (match) return match[1];
+    }
+    return clean.replace(/[^A-Z0-9-]/gi, '').trim() || clean;
+  };
+
+  // Extract date (YYYY-MM-DD) from response
+  const extractDate = (s: string): string => {
+    let clean = s.replace(/\*\*/g, '').replace(/`/g, '').trim();
+    const isoMatch = clean.match(/(\d{4}-\d{2}-\d{2})/);
+    if (isoMatch) return isoMatch[1];
+    return clean.replace(/[^\d-]/g, '').trim();
+  };
+
+  // Extract currency
+  const extractCurrency = (s: string): string => {
+    const upper = s.toUpperCase();
+    if (upper.includes('EUR') || upper.includes('€')) return 'EUR';
+    if (upper.includes('USD') || upper.includes('$')) return 'USD';
+    if (upper.includes('GBP') || upper.includes('£')) return 'GBP';
+    return 'EUR';
+  };
+
   return {
-    invoice_number: invoiceNum || '',
-    invoice_date: invoiceDate || '',
-    vendor_name: vendor || '',
-    currency: (currency || 'EUR').toUpperCase().replace(/[^A-Z]/g, '').slice(0, 3) || 'EUR',
-    net_amount: parseAmount(amountMatch[0] || ''),
-    vat_amount: parseAmount(amountMatch[1] || ''),
-    total_amount: parseAmount(amountMatch[2] || amountMatch[0] || ''),
+    invoice_number: extractInvoiceNumber(invoiceNum),
+    invoice_date: extractDate(invoiceDate),
+    vendor_name: vendor.replace(/\*\*/g, '').replace(/`/g, '').trim() || '',
+    currency: extractCurrency(currency),
+    net_amount: parseAmount(netAmount),
+    vat_amount: parseAmount(vatAmount),
+    total_amount: parseAmount(totalAmount),
   };
 }
 
@@ -296,7 +338,7 @@ tap.test('summary', async () => {
   console.log(`\n======================================================`);
   console.log(`   Invoice Extraction Summary (Qwen3-VL Vision)`);
   console.log(`======================================================`);
-  console.log(`  Method:    Qwen3-VL 8B Direct Vision (/no_think)`);
+  console.log(`  Method:    Multi-query (single pass)`);
   console.log(`  Passed:    ${passedCount}/${total}`);
   console.log(`  Failed:    ${failedCount}/${total}`);
   console.log(`  Accuracy:  ${accuracy.toFixed(1)}%`);