Update app.py

app.py

@@ -131,86 +131,114 @@ def fallback_supplier(text):
     return None
 
 def get_extraction_prompt(model_choice, txt):
-    return (
-        "You are an expert invoice parser. "
-        "Extract data according to the visible table structure and column headers in the invoice. "
-        "For every line item, only extract fields that correspond to the table columns for that row (do not include header/shipment fields in line items). "
-        "Merge all multi-line content within a single cell into that field (especially for the 'description' and 'notes'). "
-        "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'
-        '    "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"
-        "\nInvoice Text:\n"
-        f"{txt}"
-    )
+    return f"""
+You are an expert invoice parser.
+Extract data according to the visible table structure and column headers in the invoice.
+For every line item, only extract fields that correspond to the table columns for that row (do not include header/shipment fields in line items).
+Merge all multi-line content within a single cell into that field (especially for the "description" and "notes").
+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.
+
+**For example, if the source document shows Payment Terms: 1% 10 ADI, ADF, NET 20 DAYS**
+**Do not alter the sequence, combine, or separate phrases in any way.**
+
+Discount Extraction:
+Extract only fields explicitly labeled "Discount", "Special Discount", "Rebate", or "Early Payment Discount".
+Do not extract "deduct" amounts as a discount if they are part of payment terms or incentives.
+Exclude any ambiguous phrases from the discount field.
+
+Use this schema:
+{{
+ // ============================
+ // Invoice Header Details
+ // ============================
+ "invoice_header": {{
+   "id_number": "string or null",
+   "document_type": "string or null",
+   "invoice_number": "string or null",
+   "invoice_date": "string or null",
+   "due_date": "string or null",
+   "invoice_status": "string or null",
+   "mobile_no": "string or null",
+   "landline_no": "string or null",
+   "fax": "string or null",
+   "e_mail_id": "string or null",
+   "taxable_amount": "string or null",
+   "tax_amount": "string or null",
+   "tax_rate": "string or null",
+   "total_before_tax": "string or null",
+   "shipping_charges": "string or null",
+   "discount": "string or null",
+   "round_off": "string or null",
+   "grand_total": "string or null",
+   "order_number": "string or null",
+   "order_date": "string or null",
+   "purchase_order_number": "string or null",
+   "purchase_order_date": "string or null",
+   "customer_order_number": "string or null",
+   "our_order_number": "string or null",
+   "sales_order_number": "string or null",
+   "supplier_name": "string or null",
+   "supplier_address": "string or null",
+   "supplier_phone": "string or null",
+   "supplier_email": "string or null",
+   "supplier_tax_id": "string or null",
+   "customer_name": "string or null",
+   "customer_no": "string or null",
+   "customer_address": "string or null",
+   "customer_phone": "string or null",
+   "customer_email": "string or null",
+   "customer_tax_id": "string or null",
+   "bill_to_name": "string or null",
+   "bill_to_address": "string or null",
+   "ship_to_address": "string or null",
+   "ship_to_name": "string or null",
+   "remit_to": "string or null",
+   "pay_to": "string or null",
+   "shipment_number": "string or null",
+   "shipping_point": "string or null",
+   "shipped_via": "string or null",
+   "transportation_mode": "string or null",
+   "delivery_doc_no": "string or null",
+   "delivery_doc_date": "string or null",
+   "payment_terms": "string or null",
+   "payment_method": "string or null",
+   "payment_reference": "string or null",
+   "bank_account_number": "string or null",
+   "iban": "string or null",
+   "bank_account_name": "string or null",
+   "bank_name": "string or null",
+   "bank_branch": "string or null",
+   "notes_remarks": "string or null",
+   "additional_info": "string or null",
+   "delivery_terms": "string or null",
+   "aca_admin_fee": "string or null"
+ }},
+ // ============================
+ // Line Items Details
+ // ============================
+ "line_items": [
+   {{
+     "line_no": "string or null",
+     "item_no": "string or null",
+     "quantity": "string or null",
+     "uom": "string or null",
+     "units": "string or null",
+     "rate": "string or null",
+     "rate_per": "string or null",
+     "price": "string or null",
+     "line_amount": "string or null",
+     "description": "string or null",
+     "footage": "string or null",
+     "notes": "string or null"
+   }}
+ ]
+}}
+If 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.
+Invoice Text:
+{txt}
+"""
+
 
 def ensure_total_due(invoice_header):
     if invoice_header.get("total_due") in [None, ""]: