Upload folder using huggingface_hub

app.py

@@ -1,3 +1,4 @@
+
 import os
 
 # ✅ Always use /tmp for Hugging Face cache in Spaces
@@ -91,23 +92,31 @@ def extract_entities(text):
     entities = {}
 
     # PAN format: AAAAA9999A
-    pan_match = re.search(r"\b[A-Z]{5}\d{4}[A-Z]\b", text, re.IGNORECASE)
+    # pan_match = re.search(r"\b[A-Z]{5}\d{4}[A-Z]\b", text, re.IGNORECASE)
+    pan_match = re.search(r"PAN (\w{5}\d{4}\w)", text)
 
     # Account Number
-    acc_match = re.search(r"account\s*number\s*[:\-]?\s*([A-Za-z0-9]+)", text, re.IGNORECASE)
+    # acc_match = re.search(r"account\s*number\s*[:\-]?\s*([A-Za-z0-9]+)", text, re.IGNORECASE)
+    acc_match = re.search(r"Account Number (\w+)", text, re.IGNORECASE)
 
     # Penalty (accepts 'penalty' or 'penalties')
-    penalty_match = re.search(r"\bpenalt(?:y|ies)\s*[:\-]?\s*([\d,]+)", text, re.IGNORECASE)
+    # penalty_match = re.search(r"\bpenalt(?:y|ies)\s*[:\-]?\s*([\d,]+)", text, re.IGNORECASE)
+    penalty_match = re.search(r"INR ([\d,]+)", text, re.IGNORECASE)
 
     # Deactivation keywords
     deactivate_match = re.search(r"\bdeactivat(?:e|ed|ion)\b", text, re.IGNORECASE)
 
     if pan_match:
-        entities["PAN"] = pan_match.group(0).upper()
+        #entities["PAN"] = pan_match.group(0).upper()
+        entities["PAN"] = pan_match.group(1) if pan_match else None
+
     if acc_match:
-        entities["Account_Number"] = acc_match.group(1)
+        entities["Account_Number"] = acc_match.group(1) if acc_match else None
+
     if penalty_match:
-        entities["Penalty"] = penalty_match.group(1).replace(",", "")
+        #entities["Penalty"] = penalty_match.group(1).replace(",", "")
+        entities["Penalty"] = penalty_match.group(1) if penalty_match else None
+
     if deactivate_match:
         entities["Deactivate"] = deactivate_match.group(0).lower()