refaktorálás javításai

backend/app/workers/vehicle/vehicle_robot_2_researcher.py

@@ -82,6 +82,45 @@ class VehicleResearcher:
             logger.debug(f"Keresési hiba ({label}): {e}")
             return f"[SOURCE: {label}]\nKERESÉSI HIBA.\n"
 
+    def extract_specs_from_text(self, text: str) -> dict:
+        """ Regex alapú kinyerés a nyers szövegből: ccm, kW, motoradatok. """
+        import re
+        specs = {}
+        
+        # CCM (köbcentiméter) minta: 1998 cc, 2.0 L, 2000 cm³
+        ccm_pattern = r'(\d{3,4})\s*(?:cc|ccm|cm³|cm3|cc\.)'
+        match = re.search(ccm_pattern, text, re.IGNORECASE)
+        if match:
+            specs['ccm'] = int(match.group(1))
+        else:
+            # Alternatív minta: 2.0 liter -> 2000 cc
+            liter_pattern = r'(\d+\.?\d*)\s*(?:L|liter|ℓ)'
+            match = re.search(liter_pattern, text, re.IGNORECASE)
+            if match:
+                liters = float(match.group(1))
+                specs['ccm'] = int(liters * 1000)
+        
+        # KW (kilowatt) minta: 150 kW, 150kW, 150 KW
+        kw_pattern = r'(\d{2,4})\s*(?:kW|kw|KW)'
+        match = re.search(kw_pattern, text, re.IGNORECASE)
+        if match:
+            specs['kw'] = int(match.group(1))
+        else:
+            # Le (lóerő) átváltás: 150 LE -> 110 kW (kb)
+            hp_pattern = r'(\d{2,4})\s*(?:HP|hp|LE|le|Ps)'
+            match = re.search(hp_pattern, text, re.IGNORECASE)
+            if match:
+                hp = int(match.group(1))
+                specs['kw'] = int(hp * 0.7355)  # hozzávetőleges átváltás
+        
+        # Motor kód minta: motor kód: 1.8 TSI, engine code: N47
+        engine_pattern = r'(?:motor\s*kód|engine\s*code|motor\s*code)[:\s]+([A-Z0-9\.\- ]+)'
+        match = re.search(engine_pattern, text, re.IGNORECASE)
+        if match:
+            specs['engine_code'] = match.group(1).strip()
+        
+        return specs
+
     async def research_vehicle(self, db, vehicle_id: int, make: str, model: str, engine: str, year: str, current_attempts: int):
         """ Egy jármű átvilágítása és a strukturált 'Akta' elkészítése a GPU számára. """
         engine_safe = engine or ""
@@ -111,6 +150,9 @@ class VehicleResearcher:
         if len(full_context) > 2500:
             full_context = full_context[:2500] + "\n...[TRUNCATED TO SAVE GPU TOKENS]"
         
+        # Regex alapú specifikáció kinyerés
+        extracted_specs = self.extract_specs_from_text(full_context)
+        
         try:
             if len(full_context.strip()) > 150: # Csökkentettük az elvárást, mert a célzott keresés tömörebb
                 await db.execute(
@@ -118,6 +160,7 @@ class VehicleResearcher:
                     .where(VehicleModelDefinition.id == vehicle_id)
                     .values(
                         raw_search_context=full_context,
+                        research_metadata=extracted_specs,
                         status='awaiting_ai_synthesis', # Kész az Akta, mehet az Alkimistának!
                         last_research_at=func.now(),
                         attempts=current_attempts + 1
@@ -155,14 +198,15 @@ class VehicleResearcher:
                 async with AsyncSessionLocal() as db:
                     # ATOMI ZÁROLÁS
                     query = text("""
-                        UPDATE data.vehicle_model_definitions
+                        UPDATE vehicle.vehicle_model_definitions
                         SET status = 'research_in_progress'
                         WHERE id = (
-                            SELECT id FROM data.vehicle_model_definitions
+                            SELECT id FROM vehicle.vehicle_model_definitions
                             WHERE status IN ('unverified', 'awaiting_research', 'ACTIVE')
                               AND attempts < :max_attempts
-                            ORDER BY 
-                                CASE WHEN make = 'TOYOTA' THEN 1 ELSE 2 END, 
+                              AND is_manual = FALSE
+                            ORDER BY
+                                CASE WHEN make = 'TOYOTA' THEN 1 ELSE 2 END,
                                 attempts ASC
                             FOR UPDATE SKIP LOCKED
                             LIMIT 1