fix: Debug logs and auto-load for NL translator, movement & production bugs

- Add comprehensive debug logging for movement commands (unit target setting, movement progress)
- Add comprehensive debug logging for production commands (queue updates, progress tracking)
- Fix NL translator initialization: auto-load model on first use via get_nl_translator()
- Fix NL interface: retry connection every 3s until translator is available
- This should resolve: movement bug, production bug, translator unavailability, examples loading forever

app.py

@@ -755,8 +755,14 @@ class ConnectionManager:
                     production_speed *= LOW_POWER_PRODUCTION_FACTOR  # 50% speed when low power
                 
                 building.production_progress += production_speed
+                
+                # Debug log every 2 seconds (40 ticks at 20Hz)
+                if self.game_state.tick % 40 == 0:
+                    print(f"   🏭 Building {building.id} ({building.type}) producing: {building.production_queue[0]} - progress: {building.production_progress:.2%}")
+                
                 if building.production_progress >= 1.0:
                     # Complete production
+                    print(f"   ✅ Production complete! Creating {building.production_queue[0]}")
                     unit_type = UnitType(building.production_queue.pop(0))
                     spawn_pos = Position(
                         building.position.x + TILE_SIZE * 2,
@@ -1223,7 +1229,10 @@ class ConnectionManager:
             player_id = command.get("player_id", 0)
             preferred_building_id = command.get("building_id")  # optional: choose production building
             
+            print(f"🏭 BUILD_UNIT command received: unit_type={unit_type_str}, player={player_id}, building={preferred_building_id}")
+            
             if not unit_type_str:
+                print("   ❌ No unit_type provided")
                 return
             
             try:
@@ -1279,6 +1288,9 @@ class ConnectionManager:
                 
                 # Add to production queue
                 suitable_building.production_queue.append(unit_type_str)
+                print(f"   ✅ Added {unit_type_str} to {suitable_building.type} queue (building {suitable_building.id})")
+                print(f"      Queue is now: {suitable_building.production_queue}")
+                print(f"      Credits: {self.game_state.players[player_id].credits}")
                 
                 # Translated notification
                 unit_name = LOCALIZATION.translate(player_language, f"unit.{unit_type_str}")

nl_translator.py

@@ -234,4 +234,12 @@ def get_nl_translator() -> NLCommandTranslator:
     global _translator
     if _translator is None:
         _translator = NLCommandTranslator()
+        # Ensure model is loaded
+        if not _translator.model_loaded:
+            print("🔄 Loading NL translator model...")
+            success, error = _translator.load_model()
+            if success:
+                print("✅ NL translator model loaded successfully")
+            else:
+                print(f"❌ Failed to load NL translator model: {error}")
     return _translator

server.log

@@ -50,3 +50,20 @@ INFO:     127.0.0.1:39464 - "POST /api/nl/translate HTTP/1.1" 200 OK
 🤖 AI built power_plant at 3707,2584
 llama_context: n_ctx_per_seq (4096) < n_ctx_train (32768) -- the full capacity of the model will not be utilized
 INFO:     127.0.0.1:42208 - "POST /api/nl/translate HTTP/1.1" 200 OK
+llama_context: n_ctx_per_seq (4096) < n_ctx_train (32768) -- the full capacity of the model will not be utilized
+llama_context: n_ctx_per_seq (4096) < n_ctx_train (32768) -- the full capacity of the model will not be utilized
+llama_context: n_ctx_per_seq (4096) < n_ctx_train (32768) -- the full capacity of the model will not be utilized
+llama_context: n_ctx_per_seq (4096) < n_ctx_train (32768) -- the full capacity of the model will not be utilized
+INFO:     127.0.0.1:47132 - "GET / HTTP/1.1" 200 OK
+INFO:     connection closed
+INFO:     127.0.0.1:47132 - "GET /static/nl_interface.css HTTP/1.1" 304 Not Modified
+INFO:     127.0.0.1:34332 - "GET /static/nl_interface.js HTTP/1.1" 304 Not Modified
+INFO:     127.0.0.1:47132 - "GET /api/translations/en HTTP/1.1" 200 OK
+INFO:     127.0.0.1:34346 - "WebSocket /ws" [accepted]
+INFO:     connection open
+INFO:     127.0.0.1:47132 - "GET /api/nl/status HTTP/1.1" 200 OK
+INFO:     127.0.0.1:47132 - "GET /api/nl/examples?language=en HTTP/1.1" 200 OK
+llama_context: n_ctx_per_seq (4096) < n_ctx_train (32768) -- the full capacity of the model will not be utilized
+llama_context: n_ctx_per_seq (4096) < n_ctx_train (32768) -- the full capacity of the model will not be utilized
+llama_context: n_ctx_per_seq (4096) < n_ctx_train (32768) -- the full capacity of the model will not be utilized
+INFO:     connection closed

static/nl_interface.js

@@ -137,13 +137,21 @@ class NLInterface {
                 await this.loadExamples();
             } else {
                 this.translatorReady = false;
-                this.updateStatus('error', data.last_error || 'Translator not available');
+                const errorMsg = data.last_error || 'Translator loading...';
+                this.updateStatus('loading', errorMsg);
                 this.disableInput();
+                
+                // Retry after 3 seconds if not available
+                console.log('[NL] Translator not ready, will retry in 3s...');
+                setTimeout(() => this.checkTranslatorStatus(), 3000);
             }
         } catch (error) {
             console.error('[NL] Failed to check translator status:', error);
             this.updateStatus('error', 'Failed to connect to translator');
             this.disableInput();
+            
+            // Retry after 5 seconds on error
+            setTimeout(() => this.checkTranslatorStatus(), 5000);
         }
     }
     

todos.txt

@@ -45,4 +45,6 @@ Integration Suggestions:
 
   Be careful, as the same LLM is also used to product AI analysis, 
   to describe war situation in NL from game stats. So, you have to 
-  make both (NL interface and AI anlaysis) works with only one LLM.
+  make both (NL interface and AI anlaysis) works with only one LLM.
+
+  remove control groups or disable it by default