Update app.py

app.py

@@ -1,88 +1,52 @@
 import os
 import struct
 import gradio as gr
-from llama_cpp import Llama
 from cryptography.hazmat.primitives.ciphers.aead import AESGCM
 from huggingface_hub import hf_hub_download, login
 
-# --- CONFIGURATION ---
+# CRITICAL IMPORT: We do this inside a try block to catch the error early
+try:
+    from llama_cpp import Llama
+    print("✅ Llama-CPP Loaded Successfully.")
+except Exception as e:
+    print(f"❌ Llama-CPP Load Failed: {e}")
+
+# --- CONFIG ---
 SOURCE_REPO = "metanthropic/metanthropic-phi3-encrypted" 
 SOURCE_FILENAME = "metanthropic-phi3-v1.mguf"
-TEMP_DECRYPTED = "/tmp/session_model.gguf"
-
-# Secrets
+TEMP_DECRYPTED = "/tmp/model.gguf"
 SECRET_KEY_HEX = os.environ.get("DECRYPTION_KEY")
 HF_TOKEN = os.environ.get("HF_TOKEN")
 
-print("🔄 [SYSTEM] Metanthropic Node Starting...")
-
-def setup_model():
-    if os.path.exists(TEMP_DECRYPTED):
-        print("⚡ [CACHE] Model found locally.")
-        return
-
-    # Authenticate
-    if HF_TOKEN:
-        print("🔐 Authenticating...")
-        login(token=HF_TOKEN)
+def unlock():
+    if os.path.exists(TEMP_DECRYPTED): return
+    print(f"⬇️ Fetching {SOURCE_FILENAME}...")
     
-    # Download
-    print(f"⬇️ [NETWORK] Downloading model from {SOURCE_REPO}...")
-    try:
-        encrypted_path = hf_hub_download(
-            repo_id=SOURCE_REPO,
-            filename=SOURCE_FILENAME,
-            local_dir="."
-        )
-    except Exception as e:
-        print(f"❌ Download Failed: {e}")
-        raise e
-
-    # Decrypt
-    if not SECRET_KEY_HEX:
-        raise ValueError("❌ DECRYPTION_KEY not set in Settings.")
-
-    print("🔓 [SECURITY] Decrypting...")
+    if HF_TOKEN: login(token=HF_TOKEN)
+    
+    path = hf_hub_download(repo_id=SOURCE_REPO, filename=SOURCE_FILENAME)
+    
+    print("🔓 Decrypting...")
     key = bytes.fromhex(SECRET_KEY_HEX)
-    aesgcm = AESGCM(key)
+    aes = AESGCM(key)
     
-    with open(encrypted_path, "rb") as f_in, open(TEMP_DECRYPTED, "wb") as f_out:
+    with open(path, "rb") as f_in, open(TEMP_DECRYPTED, "wb") as f_out:
         nonce = f_in.read(12)
-        header_len = struct.unpack("<I", f_in.read(4))[0]
-        encrypted_header = f_in.read(header_len)
-        
-        # Decrypt Header
-        f_out.write(aesgcm.decrypt(nonce, encrypted_header, None))
-        
-        # Stream Body
-        while True:
-            chunk = f_in.read(1024 * 1024 * 64) # 64MB chunks
-            if not chunk: break
-            f_out.write(chunk)
-            
-    print("✅ [SUCCESS] Model Ready.")
+        h_len = struct.unpack("<I", f_in.read(4))[0]
+        f_out.write(aes.decrypt(nonce, f_in.read(h_len), None))
+        while chunk := f_in.read(64*1024*1024): f_out.write(chunk)
+    print("✅ Ready.")
 
-# --- LOAD ---
 llm = None
 try:
-    setup_model()
-    llm = Llama(
-        model_path=TEMP_DECRYPTED,
-        n_ctx=2048,
-        n_threads=2
-    )
+    unlock()
+    llm = Llama(model_path=TEMP_DECRYPTED, n_ctx=2048, n_threads=2)
 except Exception as e:
-    print(f"❌ CRITICAL BOOT ERROR: {e}")
-
-# --- UI ---
-def chat(message, history):
-    if not llm: return "System Error: Model not loaded."
-    
-    prompt = f"<|user|>\n{message}<|end|>\n<|assistant|>"
-    output = llm(prompt, max_tokens=512, stop=["<|end|>"], echo=False)
-    return output['choices'][0]['text'].strip()
+    print(f"❌ Boot Error: {e}")
 
-demo = gr.ChatInterface(fn=chat, title="Metanthropic Node (Phi-3)")
+def chat(msg, history):
+    if not llm: return "System offline."
+    prompt = f"<|user|>\n{msg}<|end|>\n<|assistant|>"
+    return llm(prompt, max_tokens=512, stop=["<|end|>"])['choices'][0]['text'].strip()
 
-if __name__ == "__main__":
-    demo.launch(server_name="0.0.0.0", server_port=7860)
+gr.ChatInterface(chat).launch(server_name="0.0.0.0", server_port=7860)