Fix: NLP-to-SQL now handles multiple vehicles and multiple locations with proper OR logic

database.py

@@ -363,7 +363,7 @@ class FilterExtractor:
         return None
     
     def extract_location(self, query):
-        """Extract location with variant matching"""
+        """Extract SINGLE location with variant matching (legacy method)"""
         q = query.lower()
         # Sort by length (longest first) to match longer variants first
         for canonical, variants in sorted(
@@ -376,8 +376,34 @@ class FilterExtractor:
                     return canonical
         return None
     
+    def extract_locations(self, query):
+        """Extract MULTIPLE locations from query (e.g., 'adyar and kottupuram')"""
+        q = query.lower()
+        locations = []
+        
+        # Sort by length (longest first) to match longer variants first
+        for canonical, variants in sorted(
+            self.location_variants.items(),
+            key=lambda x: max(len(v) for v in x[1]),
+            reverse=True
+        ):
+            for variant in variants:
+                if variant in q:
+                    locations.append(canonical)
+                    break
+        
+        # Remove duplicates while preserving order
+        seen = set()
+        unique_locations = []
+        for loc in locations:
+            if loc not in seen:
+                seen.add(loc)
+                unique_locations.append(loc)
+        
+        return unique_locations if unique_locations else None
+    
     def extract_vehicle_type(self, query):
-        """Extract vehicle type with synonym resolution"""
+        """Extract SINGLE vehicle type with synonym resolution (legacy method)"""
         q = query.lower()
         # Sort by length (longest first) to match longer synonyms first
         for synonym in sorted(self.vehicle_synonyms.keys(), key=len, reverse=True):
@@ -385,6 +411,27 @@ class FilterExtractor:
                 return self.vehicle_synonyms[synonym]
         return None
     
+    def extract_vehicle_types(self, query):
+        """Extract MULTIPLE vehicle types from query (e.g., 'bike and mini_truck')"""
+        q = query.lower()
+        vehicle_types = []
+        
+        # Sort by length (longest first) to match longer synonyms first
+        for synonym in sorted(self.vehicle_synonyms.keys(), key=len, reverse=True):
+            if re.search(r'\b' + synonym + r'\b', q):
+                normalized = self.vehicle_synonyms[synonym]
+                vehicle_types.append(normalized)
+        
+        # Remove duplicates while preserving order
+        seen = set()
+        unique_types = []
+        for vtype in vehicle_types:
+            if vtype not in seen:
+                seen.add(vtype)
+                unique_types.append(vtype)
+        
+        return unique_types if unique_types else None
+    
     def extract_date_range(self, query):
         """Extract date range (from X to Y, between X and Y)"""
         # Pattern: "from DD-MM-YYYY to DD-MM-YYYY" or "between DD-MM-YYYY and DD-MM-YYYY"
@@ -520,8 +567,8 @@ class FilterExtractor:
         return {
             "plate": self.extract_plate(query),
             "state": self.extract_state(query),
-            "location": self.extract_location(query),
-            "vehicle_type": self.extract_vehicle_type(query),
+            "location": self.extract_locations(query),  # Now returns list or None
+            "vehicle_type": self.extract_vehicle_types(query),  # Now returns list or None
             "date": self.extract_date(query),
             "date_range": self.extract_date_range(query),
             "day": self.extract_day(query),
@@ -549,7 +596,7 @@ class FilterExtractor:
     def build_sql(self, filters, intents):
         """
         Build production-grade SQL from filters and intents.
-        Handles complex aggregations, date ranges, time ranges, and conditions.
+        Handles complex aggregations, date ranges, time ranges, multiple vehicles, and multiple locations.
         """
         
         # =========================================================
@@ -596,7 +643,7 @@ class FilterExtractor:
             """)
         
         # =========================================================
-        # BUILD WHERE CLAUSE FROM FILTERS
+        # BUILD WHERE CLAUSE FROM FILTERS (HANDLE MULTIPLE VALUES)
         # =========================================================
         
         where_conditions = []
@@ -609,13 +656,31 @@ class FilterExtractor:
         if filters["state"]:
             where_conditions.append(f"state = '{filters['state']}'")
         
-        # Location filter
+        # Location filter - HANDLE MULTIPLE LOCATIONS
         if filters["location"]:
-            where_conditions.append(f"LOWER(location) LIKE '%{filters['location'].lower()}%'")
+            if isinstance(filters["location"], list):
+                # Multiple locations with OR logic
+                location_conditions = [
+                    f"LOWER(location) LIKE '%{loc.lower()}%'" 
+                    for loc in filters["location"]
+                ]
+                where_conditions.append(f"({' OR '.join(location_conditions)})")
+            else:
+                # Single location (legacy)
+                where_conditions.append(f"LOWER(location) LIKE '%{filters['location'].lower()}%'")
         
-        # Vehicle type filter
+        # Vehicle type filter - HANDLE MULTIPLE VEHICLE TYPES
         if filters["vehicle_type"]:
-            where_conditions.append(f"LOWER(vehicle_type) LIKE '%{filters['vehicle_type'].lower()}%'")
+            if isinstance(filters["vehicle_type"], list):
+                # Multiple vehicle types with OR logic
+                vehicle_conditions = [
+                    f"LOWER(vehicle_type) LIKE '%{vtype.lower()}%'" 
+                    for vtype in filters["vehicle_type"]
+                ]
+                where_conditions.append(f"({' OR '.join(vehicle_conditions)})")
+            else:
+                # Single vehicle type (legacy)
+                where_conditions.append(f"LOWER(vehicle_type) LIKE '%{filters['vehicle_type'].lower()}%'")
         
         # Date range filter
         if filters["date_range"]:
@@ -752,8 +817,25 @@ def ask_llm(user_query):
         print(f"   Extracted Filters:")
         print(f"     - Plate: {filters['plate']}")
         print(f"     - State: {filters['state']}")
-        print(f"     - Location: {filters['location']}")
-        print(f"     - Vehicle Type: {filters['vehicle_type']}")
+        
+        # Handle multiple locations
+        if filters['location']:
+            if isinstance(filters['location'], list):
+                print(f"     - Locations: {', '.join(filters['location'])}")
+            else:
+                print(f"     - Location: {filters['location']}")
+        else:
+            print(f"     - Location: None")
+        
+        # Handle multiple vehicle types
+        if filters['vehicle_type']:
+            if isinstance(filters['vehicle_type'], list):
+                print(f"     - Vehicle Types: {', '.join(filters['vehicle_type'])}")
+            else:
+                print(f"     - Vehicle Type: {filters['vehicle_type']}")
+        else:
+            print(f"     - Vehicle Type: None")
+        
         print(f"     - Date: {filters['date']}")
         print(f"     - Date Range: {filters['date_range']}")
         print(f"     - Day: {filters['day']}")