Fix callback, keyword extraction and UPI detection issues

Co-authored-by: Devanshu-BPPC-official <204659429+Devanshu-BPPC-official@users.noreply.github.com>

Author: Copilot

main.py

@@ -182,13 +182,6 @@ async def process_message():
                 )
                 session = SessionManager.get_session(session_id)
 
-            # Check if callback already sent for this session
-            if session and session.get("callback_sent"):
-                return MessageResponse(
-                    status="success",
-                    reply="thank you for the information."
-                )
-            
             # Build conversation history for context
             conversation_texts = [msg.text for msg in request.conversationHistory]
 

modules/detector.py

@@ -113,8 +113,9 @@ def _safe_json_load(s: str) -> Optional[dict]:
 def _match_patterns(text_lower: str, patterns: List[str]) -> List[str]:
     matched = []
     for p in patterns:
-        if re.search(p, text_lower):
-            matched.append(p)
+        match = re.search(p, text_lower)
+        if match:
+            matched.append(match.group(0))
     return matched
 
 def detect_scam(text: str, threshold: float = 0.5) -> Tuple[bool, float, List[str]]:

modules/extractor.py

@@ -80,7 +80,7 @@ class DataExtractor:
 
     # Regex patterns for different data types
     BANK_ACCOUNT_PATTERN = r'\b\d{9,18}\b'  # 9-18 digit bank account numbers
-    UPI_ID_PATTERN = r'\b[\w\.-]+@[\w\.-]+\b'  # UPI ID format: user@bank
+    UPI_ID_PATTERN = r'[\w\.-]+@[\w\.-]+'  # UPI ID format: user@bank
     URL_PATTERN = r'https?://[^\s]+|www\.[^\s]+|bit\.ly/[^\s]+|tinyurl\.com/[^\s]+'
     PHONE_PATTERN = r'\b(?:\+91[\s-]?)?[6-9]\d{9}\b'  # Indian phone numbers
     IFSC_PATTERN = r'\b[A-Z]{4}0[A-Z0-9]{6}\b'  # IFSC code pattern
@@ -123,7 +123,13 @@ def _extract_with_ai(self, text: str) -> ScamData:
                 return ScamData()
 
             if response and response.text:
-                result = json.loads(response.text.strip())
+                # Clean response text - remove markdown code blocks
+                text_clean = response.text.strip()
+                text_clean = re.sub(r'^```(?:json)?\s*', '', text_clean)
+                text_clean = re.sub(r'\s*```$', '', text_clean)
+                
+                # Try to parse JSON
+                result = json.loads(text_clean)
                 ai_data = ScamData()
 
                 for account in result.get("bank_accounts", []):
@@ -137,6 +143,8 @@ def _extract_with_ai(self, text: str) -> ScamData:
 
                 return ai_data
 
+        except json.JSONDecodeError as e:
+            print(f"Error in AI extraction: {e}")
         except Exception as e:
             print(f"Error in AI extraction: {e}")