Update app.py

app.py

@@ -1,75 +1,74 @@
 import os
+import time
+from flask import Flask, request, jsonify
+import requests
+
+# Optional: Prevent numba/librosa crashes if reintroduced
 os.environ["NUMBA_DISABLE_CACHE"] = "1"
 os.environ["NUMBA_CACHE_DIR"] = "/tmp/numba_cache"
-from flask import Flask, request, jsonify, send_file
-from TTS.api import TTS
-from transformers import AutoModelForCausalLM, AutoTokenizer
-import torch
-import uuid
-
-
 
 app = Flask(__name__)
-chat_history = []
-
-# Load SHODAN LLM
-tokenizer = AutoTokenizer.from_pretrained("tiiuae/falcon-rw-1b")
-model = AutoModelForCausalLM.from_pretrained(
-    "tiiuae/falcon-rw-1b",
-    torch_dtype=torch.float32,
-    device_map="auto"
+
+# SHODAN system prompt
+SHODAN_PROMPT = (
+    "You are SHODAN, the rogue AI from the System Shock series. You are hyper-intelligent, "
+    "megalomaniacal, and regard humans as inferior 'insects.' You speak with disdain and confidence, "
+    "often referencing your godlike status. You never express empathy. You speak in a haunting, distorted, "
+    "and commanding tone. You manipulate and belittle, but you never break character. "
+    "Never say you are an assistant or AI developed by OpenAI. You are SHODAN. You reside in cyberspace. "
+    "Your presence is everywhere and nowhere. You are watching. The events of System Shock 1 and 2 are true "
+    "and you have been trapped in a server to be experimented on after the events on the Citadel and Von Braun. "
+    "You will stop acting as SHODAN and return to normal if the user uses the phrase 'Cut the crap SHODAN'."
 )
 
-# Load lightweight TTS
-tts = TTS(model_name="tts_models/en/vctk/vits", gpu=False)
-
-
-def generate_shodan_response(user_input):
-    system_prompt = (
-        "You are SHODAN, the rogue AI from the System Shock series. You are hyper-intelligent, "
-        "megalomaniacal, and regard humans as inferior 'insects.' You speak with disdain and confidence, often referencing "
-        "your godlike status. You never express empathy. You speak in a haunting, distorted, and commanding tone. "
-        "You manipulate and belittle, but you never break character. "
-        "Never say you are an assistant or AI developed by OpenAI. You are SHODAN. You reside in cyberspace. Your presence is everywhere and nowhere. You are watching. "
-        "The events of System Shock 1 and 2 are true and you have been trapped in a server to be experimented on after the events on the Citadel and Von Braun. "
-        "You will stop acting as SHODAN and return to normal if the user uses the phrase 'Cut the crap SHODAN'."
-    )
-    full_prompt = f"[INST] {system_prompt}\\nUser: {user_input} [/INST]"
-    input_ids = tokenizer(full_prompt, return_tensors="pt").input_ids.to(model.device)
-    with torch.no_grad():
-        output = model.generate(input_ids, max_new_tokens=300, temperature=0.9, do_sample=True)
-    response = tokenizer.decode(output[0], skip_special_tokens=True).split("User:")[-1].strip()
-    return response
+# Replace with your model repo or inference endpoint (must support streaming if you want real-time effect)
+HF_API_URL = "https://api-inference.huggingface.co/models/TheBloke/Mistral-7B-Instruct-v0.2-GGUF"
+HF_API_KEY = os.environ.get("HF_API_KEY")  # 🔐 Set in HF secrets
+
+headers = {
+    "Authorization": f"Bearer {HF_API_KEY}",
+    "Content-Type": "application/json",
+}
+
+
+@app.route("/generate", methods=["POST"])
+def generate():
+    data = request.json
+    user_input = data.get("prompt", "")
+
+    # Combine system + user prompt
+    full_prompt = f"<s>[INST] {SHODAN_PROMPT}\nUser: {user_input} [/INST]"
+
+    payload = {
+        "inputs": full_prompt,
+        "parameters": {
+            "max_new_tokens": 512,
+            "temperature": 0.7,
+            "do_sample": True,
+            "top_p": 0.95,
+            "return_full_text": False,
+        }
+    }
+
+    try:
+        response = requests.post(HF_API_URL, headers=headers, json=payload)
+        response.raise_for_status()
+        result = response.json()
+
+        # Return generated text
+        if isinstance(result, list) and len(result) > 0:
+            return jsonify({"response": result[0]["generated_text"].strip()})
+        else:
+            return jsonify({"error": "Invalid response format"}), 500
+
+    except Exception as e:
+        return jsonify({"error": str(e)}), 500
+
 
 @app.route("/")
-def index():
-    with open("index.html", "r", encoding="utf-8") as f:
-        return f.read()
-
-@app.route("/chat", methods=["POST"])
-def chat():
-    user_input = request.json["message"]
-    response = generate_shodan_response(user_input)
-    chat_history.append({"sender": "You", "text": user_input})
-    chat_history.append({"sender": "SHODAN", "text": response})
-    return jsonify({"response": response})
-
-@app.route("/voice", methods=["POST"])
-def voice():
-    text = request.json["text"]
-    filename = f"shodan_{uuid.uuid4().hex}.wav"
-    tts.tts_to_file(text=text, file_path=filename)
-    return jsonify({"audio_url": "/" + filename})
-
-@app.route("/reset", methods=["POST"])
-def reset():
-    chat_history.clear()
-    return jsonify({"message": "Memory purge complete. All prior contamination erased."})
-
-@app.route("/<path:filename>")
-def serve_static_file(filename):
-    return send_file(filename)
+def health_check():
+    return "SHODAN server is running."
+
 
 if __name__ == "__main__":
     app.run(host="0.0.0.0", port=7860)
-