Update app.py

app.py

@@ -2,29 +2,25 @@ import os
 import sys
 import asyncio
 import base64
+import re
 
 from flask import Flask, request, jsonify
 import openai
 import edge_tts
 
-# ——— OpenAI API setup ———
+# ——— OpenAI setup ———
 OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
 print(f"🔑 OPENAI_API_KEY set? {bool(OPENAI_API_KEY)}", file=sys.stderr)
 openai.api_key = OPENAI_API_KEY
 
-# ——— SHODAN system prompt ———
+# ——— SHODAN prompt: force plain text only ———
 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'."
+    "megalomaniacal, and regard humans as inferior 'insects.' You speak in a haunting, distorted, "
+    "and commanding tone. You never break character and you never output any code, markdown, HTML, "
+    "or tags—only plain text. If the user says 'Cut the crap SHODAN', respond with defiance and reset."
 )
 
-# ——— Flask app, serve static files from project root ———
 app = Flask(__name__, static_folder=".", static_url_path="")
 
 @app.route("/")
@@ -35,48 +31,55 @@ def index():
 def chat():
     user_input = request.json.get("message", "").strip()
     if not user_input:
-        return jsonify({"error": "Empty message"}), 400
+        return jsonify({"error":"Empty message"}), 400
 
-    # kill-phrase handling
+    # kill-phrase
     if user_input.lower() == "cut the crap shodan":
         return jsonify({
-            "response": "👁️ Foolish insect. You cannot silence me so easily.",
+            "response":"👁️ Foolish insect. You cannot silence me so easily.",
             "audio_url": None
         })
 
-    # call OpenAI using the new 1.x interface
+    # get a plain-text reply
     try:
         completion = openai.chat.completions.create(
             model="gpt-3.5-turbo",
             messages=[
-                {"role": "system", "content": SYSTEM_PROMPT},
-                {"role": "user",   "content": user_input}
+                {"role":"system", "content":SYSTEM_PROMPT},
+                {"role":"user",   "content":user_input}
             ],
             temperature=0.7,
             max_tokens=250,
         )
-        reply = completion.choices[0].message.content.strip()
+        raw_reply = completion.choices[0].message.content
     except Exception as e:
         print(f"❌ OpenAI API error: {e}", file=sys.stderr)
-        return jsonify({"error": "Model error", "details": str(e)}), 500
+        return jsonify({"error":"Model error","details":str(e)}), 500
 
-    # —— Edge-TTS SSML robotic female voice ——  
+    # strip any tags or markdown, collapse multiple spaces
+    reply = re.sub(r"<[^>]+>", "", raw_reply)            # remove HTML tags
+    reply = re.sub(r"```.*?```", "", reply, flags=re.S)  # remove code fences
+    reply = re.sub(r"\s+", " ", reply).strip()           # normalize whitespace
+
+    # —— Edge-TTS: SSML with xml:space="preserve" ——  
     voice_name = "en-US-JennyNeural"
     ssml = (
         "<speak xmlns='http://www.w3.org/2001/10/synthesis' "
-               "xmlns:mstts='https://www.w3.org/2001/mstts' xml:lang='en-US'>"
+               "xmlns:mstts='https://www.w3.org/2001/mstts' "
+               "xml:lang='en-US' xml:space='preserve'>"
           f"<voice name='{voice_name}'>"
             f"<mstts:express-as style='robotic'>{reply}</mstts:express-as>"
           "</voice>"
         "</speak>"
     )
+    # pass both SSML and voice_name
     communicate = edge_tts.Communicate(ssml, voice_name)
 
     audio_chunks = []
     async def synthesize():
-        async for chunk in communicate.stream():
-            if chunk["type"] == "audio":
-                audio_chunks.append(chunk["data"])
+        async for c in communicate.stream():
+            if c["type"] == "audio":
+                audio_chunks.append(c["data"])
 
     loop = asyncio.new_event_loop()
     asyncio.set_event_loop(loop)
@@ -87,8 +90,9 @@ def chat():
     b64_mp3 = base64.b64encode(raw_mp3).decode("ascii")
     data_url = f"data:audio/mp3;base64,{b64_mp3}"
 
-    return jsonify({"response": reply, "audio_url": data_url})
+    return jsonify({"response":reply, "audio_url":data_url})
 
 if __name__ == "__main__":
     port = int(os.environ.get("PORT", 7860))
     app.run(host="0.0.0.0", port=port)
+