Update app.py

app.py

@@ -226,9 +226,9 @@ def weighted_fuzzy_score(s1, s2):
 def find_best_po_match(inv, po_df):
     inv_hdr = inv["invoice_header"]
     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_ship_to = inv_hdr.get("ship_to_name") or ""
     inv_bill_to = inv_hdr.get("bill_to_name") or ""
-    inv_payment_terms = inv_hdr.get("payment_terms") or ""
     inv_currency = inv_hdr.get("currency") or ""
     inv_total_due = clean_num(inv_hdr.get("total_due"))
     inv_line_items = inv.get("line_items", [])
@@ -236,9 +236,9 @@ def find_best_po_match(inv, po_df):
     scores = []
     for idx, row in po_df.iterrows():
         po_supplier = row.get("Supplier Name", "")
+        po_po_number = str(row.get("PO Number", ""))  # Make sure string
         po_ship_to = row.get("Ship To", "")
         po_bill_to = row.get("Bill To", "")
-        po_payment_terms = row.get("Payment Terms", "")
         po_currency = row.get("Currency", "")
         po_total = clean_num(row.get("PO Total Value", ""))
         po_desc = row.get("Item Description", "")
@@ -256,6 +256,14 @@ def find_best_po_match(inv, po_df):
             "score": s_supplier
         })
 
+        s_po_number = weighted_fuzzy_score(inv_po_number, po_po_number)
+        field_details.append({
+            "field": "PO Number",
+            "invoice": inv_po_number,
+            "po": po_po_number,
+            "score": s_po_number
+        })
+
         s_ship_to = weighted_fuzzy_score(inv_ship_to, po_ship_to)
         field_details.append({
             "field": "Ship To",
@@ -272,14 +280,6 @@ def find_best_po_match(inv, po_df):
             "score": s_bill_to
         })
 
-        s_terms = weighted_fuzzy_score(inv_payment_terms, po_payment_terms)
-        field_details.append({
-            "field": "Payment Terms",
-            "invoice": inv_payment_terms,
-            "po": po_payment_terms,
-            "score": s_terms
-        })
-
         s_currency = weighted_fuzzy_score(inv_currency, po_currency)
         field_details.append({
             "field": "Currency",
@@ -334,21 +334,22 @@ def find_best_po_match(inv, po_df):
                     f"unit_score={unit_score}, amount_score={amount_score}"
                 )
 
+        # Adjust scoring weights: Supplier 20%, PO Number 15%, Ship To 10%, Bill To 10%, Currency 10%, Total Due 20%, Line Item 15%
         total_score = (
-            s_supplier * 0.25 +
-            s_ship_to * 0.1 +
-            s_bill_to * 0.1 +
-            s_terms * 0.1 +
-            s_currency * 0.05 +
-            s_total * 0.2 +
-            line_item_score * 0.2
+            s_supplier * 0.20 +
+            s_po_number * 0.15 +
+            s_ship_to * 0.10 +
+            s_bill_to * 0.10 +
+            s_currency * 0.10 +
+            s_total * 0.20 +
+            line_item_score * 0.15
         )
 
         reason = (
             f"Supplier match: {s_supplier}/100 (invoice: '{inv_supplier}' vs PO: '{po_supplier}'), "
+            f"PO Number: {s_po_number}/100 (invoice: '{inv_po_number}' vs PO: '{po_po_number}'), "
             f"Ship To: {s_ship_to}/100 (invoice: '{inv_ship_to}' vs PO: '{po_ship_to}'), "
             f"Bill To: {s_bill_to}/100 (invoice: '{inv_bill_to}' vs PO: '{po_bill_to}'), "
-            f"Payment Terms: {s_terms}/100 (invoice: '{inv_payment_terms}' vs PO: '{po_payment_terms}'), "
             f"Currency: {s_currency}/100 (invoice: '{inv_currency}' vs PO: '{po_currency}'), "
             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}"
@@ -357,6 +358,7 @@ def find_best_po_match(inv, po_df):
         debug = {
             "po_idx": idx,
             "po_supplier": po_supplier,
+            "po_po_number": po_po_number,
             "po_ship_to": po_ship_to,
             "po_bill_to": po_bill_to,
             "po_total": po_total,
@@ -445,7 +447,7 @@ def extract_text_from_unstract(uploaded_file):
         if status == "processed":
             status_placeholder.info("EZOFIS AI OCR AGENT STATUS: processed! 🎉")
             break
-        status_placeholder.info(f"Unstract status: {status or 'waiting'}... ({i+1})")
+        status_placeholder.info(f"EZOFIS AI OCR AGENT STATUS: {status or 'waiting'}... ({i+1})")
         time.sleep(2)
     else:
         status_placeholder.error("Unstract: Timeout waiting for OCR to finish.")