Update app.py

app.py

@@ -48,12 +48,11 @@ st.markdown("""
         background: #F3F6FB !important;
         border-radius: 999px;
     }
-    .css-12w0qpk {padding-top: 0rem;} /* Removes extra padding */
-    .css-1kyxreq {padding-top: 0rem;} /* Removes extra padding */
+    .css-12w0qpk {padding-top: 0rem;}
+    .css-1kyxreq {padding-top: 0rem;}
     </style>
 """, unsafe_allow_html=True)
 
-# --- Model Definitions ---
 MODELS = {
     "OpenAI GPT-4.1": {
         "api_url": "https://api.openai.com/v1/chat/completions",
@@ -140,8 +139,72 @@ def get_extraction_prompt(model_choice, txt):
         "Shipment/invoice-level fields such as CAR NUMBER, SHIPPING POINT, SHIPMENT NUMBER, CURRENCY, etc., must go ONLY into the 'invoice_header', not as line item fields.\n"
         "Use this schema:\n"
         '{\n'
-        '  "invoice_header": {...},\n'
-        '  "line_items": [{...}]\n'
+        '  "invoice_header": {\n'
+        '    "car_number": "string or null",\n'
+        '    "shipment_number": "string or null",\n'
+        '    "shipping_point": "string or null",\n'
+        '    "currency": "string or null",\n'
+        '    "invoice_number": "string or null",\n'
+        '    "invoice_date": "string or null",\n'
+        '    "order_number": "string or null",\n'
+        '    "customer_order_number": "string or null",\n'
+        '    "our_order_number": "string or null",\n'
+        '    "sales_order_number": "string or null",\n'
+        '    "purchase_order_number": "string or null",\n'
+        '    "order_date": "string or null",\n'
+        '    "supplier_name": "string or null",\n'
+        '    "supplier_address": "string or null",\n'
+        '    "supplier_phone": "string or null",\n'
+        '    "supplier_email": "string or null",\n'
+        '    "supplier_tax_id": "string or null",\n'
+        '    "customer_name": "string or null",\n'
+        '    "customer_address": "string or null",\n'
+        '    "customer_phone": "string or null",\n'
+        '    "customer_email": "string or null",\n'
+        '    "customer_tax_id": "string or null",\n'
+        '    "ship_to_name": "string or null",\n'
+        '    "ship_to_address": "string or null",\n'
+        '    "bill_to_name": "string or null",\n'
+        '    "bill_to_address": "string or null",\n'
+        '    "remit_to_name": "string or null",\n'
+        '    "remit_to_address": "string or null",\n'
+        '    "tax_id": "string or null",\n'
+        '    "tax_registration_number": "string or null",\n'
+        '    "vat_number": "string or null",\n'
+        '    "payment_terms": "string or null",\n'
+        '    "payment_method": "string or null",\n'
+        '    "payment_reference": "string or null",\n'
+        '    "bank_account_number": "string or null",\n'
+        '    "iban": "string or null",\n'
+        '    "swift_code": "string or null",\n'
+        '    "total_before_tax": "string or null",\n'
+        '    "tax_amount": "string or null",\n'
+        '    "tax_rate": "string or null",\n'
+        '    "shipping_charges": "string or null",\n'
+        '    "discount": "string or null",\n'
+        '    "total_due": "string or null",\n'
+        '    "amount_paid": "string or null",\n'
+        '    "balance_due": "string or null",\n'
+        '    "due_date": "string or null",\n'
+        '    "invoice_status": "string or null",\n'
+        '    "reference_number": "string or null",\n'
+        '    "project_code": "string or null",\n'
+        '    "department": "string or null",\n'
+        '    "contact_person": "string or null",\n'
+        '    "notes": "string or null",\n'
+        '    "additional_info": "string or null"\n'
+        '  },\n'
+        '  "line_items": [\n'
+        '    {\n'
+        '      "quantity": "string or null",\n'
+        '      "units": "string or null",\n'
+        '      "description": "string or null",\n'
+        '      "footage": "string or null",\n'
+        '      "price": "string or null",\n'
+        '      "amount": "string or null",\n'
+        '      "notes": "string or null"\n'
+        '    }\n'
+        '  ]\n'
         '}'
         "\nIf a field is missing for a line item or header, use null. "
         "Do not invent fields. Do not add any header or shipment data to any line item. Return ONLY the JSON object, no explanation.\n"
@@ -150,9 +213,8 @@ def get_extraction_prompt(model_choice, txt):
     )
 
 def ensure_total_due(invoice_header):
-    # Prefer total_before_tax if total_due mismatches in scoring
     if invoice_header.get("total_due") in [None, ""]:
-        for field in ["total_before_tax", "invoice_total", "invoice_value", "balance_due", "amount_paid"]:
+        for field in ["invoice_total", "invoice_value", "total_before_tax", "balance_due", "amount_paid"]:
             if field in invoice_header and invoice_header[field]:
                 invoice_header["total_due"] = invoice_header[field]
                 break
@@ -210,9 +272,7 @@ def find_best_po_match(inv, po_df, weight_supplier, weight_po_number, weight_cur
     inv_supplier = inv_hdr.get("supplier_name") or ""
     inv_po_number = inv_hdr.get("purchase_order_number") or inv_hdr.get("po_number") or inv_hdr.get("order_number") or ""
     inv_currency = inv_hdr.get("currency") or ""
-    # -- Try both total_due and total_before_tax for matching
     inv_total_due = clean_num(inv_hdr.get("total_due"))
-    inv_total_before_tax = clean_num(inv_hdr.get("total_before_tax"))
     inv_line_items = inv.get("line_items", [])
 
     scores = []
@@ -252,15 +312,10 @@ def find_best_po_match(inv, po_df, weight_supplier, weight_po_number, weight_cur
             "score": s_currency
         })
 
-        # Try total_due, fallback to total_before_tax
-        s_total = 0
-        if inv_total_due is not None and po_total is not None and abs(inv_total_due - po_total) < 2:
-            s_total = 100
-        elif inv_total_before_tax is not None and po_total is not None and abs(inv_total_before_tax - po_total) < 2:
-            s_total = 100
+        s_total = 100 if inv_total_due is not None and po_total is not None and abs(inv_total_due - po_total) < 2 else 0
         field_details.append({
-            "field": "Total Due or Before Tax",
-            "invoice": inv_total_due if s_total else inv_total_before_tax,
+            "field": "Total Due",
+            "invoice": inv_total_due,
             "po": po_total,
             "score": s_total
         })
@@ -316,7 +371,7 @@ def find_best_po_match(inv, po_df, weight_supplier, weight_po_number, weight_cur
             f"Supplier match: {s_supplier}/100 (invoice: '{inv_supplier}' vs PO: '{po_supplier}'), "
             f"PO Number: {s_po_number}/100 ({'found anywhere in JSON' if s_po_number else 'not found'}), "
             f"Currency: {s_currency}/100 (invoice: '{inv_currency}' vs PO: '{po_currency}'), "
-            f"Total: {'match' if s_total else 'no match'} (invoice: {inv_total_due if s_total else inv_total_before_tax} vs PO: {po_total}), "
+            f"Total Due: {'match' if s_total else 'no match'} (invoice: {inv_total_due} vs PO: {po_total}), "
             f"Line item best match: {int(line_item_score)}/100. {line_reason}"
         )
 
@@ -330,8 +385,7 @@ def find_best_po_match(inv, po_df, weight_supplier, weight_po_number, weight_cur
             "best_line_detail": best_line_detail,
             "total_score": total_score,
             "line_reason": line_reason,
-            "inv_total_due": inv_total_due,
-            "inv_total_before_tax": inv_total_before_tax
+            "inv_total_due": inv_total_due
         }
         scores.append((row, total_score, reason, debug))