Create brain_fast_lazy.py

brain_fast_lazy.py

@@ -0,0 +1,90 @@
+"""
+brain_fast_lazy.py
+Combines:
+- Lazy loading (bypass Hugging Face timeout)
+- Speed optimizations (fast replies after load)
+"""
+
+import os
+import time
+import threading
+import functools
+from collections import OrderedDict
+import importlib
+
+# ------------------------------
+# Persistent cache setup
+# ------------------------------
+os.environ["TRANSFORMERS_CACHE"] = "/home/user/app/cache"
+os.environ["HF_HOME"] = "/home/user/app/cache"
+os.makedirs("/home/user/app/cache", exist_ok=True)
+
+# ------------------------------
+# Cache for replies
+# ------------------------------
+CACHE_LIMIT = 50
+_response_cache = OrderedDict()
+
+def _cache_result(func):
+    @functools.wraps(func)
+    def wrapper(*args, **kwargs):
+        key = str((args, tuple(sorted(kwargs.items()))))
+        if key in _response_cache:
+            _response_cache.move_to_end(key)
+            return _response_cache[key]
+        result = func(*args, **kwargs)
+        _response_cache[key] = result
+        if len(_response_cache) > CACHE_LIMIT:
+            _response_cache.popitem(last=False)
+        return result
+    return wrapper
+
+# ------------------------------
+# Brain loader + warm-up
+# ------------------------------
+_brain = None
+_is_loading = False
+_is_ready = False
+_lock = threading.Lock()
+
+def _load_brain():
+    global _brain, _is_ready, _is_loading
+    with _lock:
+        if _brain is None:
+            _is_loading = True
+            print("⏳ Loading multimodular brain (fast-lazy mode)...")
+            start_time = time.time()
+            _brain = importlib.import_module("multimodular_modul_v7")
+            _is_ready = True
+            _is_loading = False
+            print(f"✅ Brain loaded in {time.time() - start_time:.2f}s")
+            _warm_up()
+    return _brain
+
+def _warm_up():
+    try:
+        print("🔥 Warming up models...")
+        _brain.process_input("Hello")
+        print("✅ Warm-up complete")
+    except Exception as e:
+        print(f"⚠ Warm-up failed: {e}")
+
+def preload_in_background():
+    threading.Thread(target=_load_brain, daemon=True).start()
+
+# Start background preload at import
+preload_in_background()
+
+# ------------------------------
+# Proxies with caching
+# ------------------------------
+@_cache_result
+def process_input(text): return _load_brain().process_input(text)
+@_cache_result
+def search_kb(query): return _load_brain().search_kb(query)
+def upload_media(path): return _load_brain().upload_media(path)
+def backup_brain(): return _load_brain().backup_brain()
+def restore_brain(): return _load_brain().restore_brain()
+def show_creative_skills(): return _load_brain().show_creative_skills()
+def sync_status(): return _load_brain().sync_status()
+def is_ready(): return _is_ready