Create noRag.py

noRag.py

@@ -0,0 +1,120 @@
+# noRag.py
+
+import asyncio
+import os
+from groq import Groq
+from pymongo import MongoClient
+from config import (
+    CONNECTION_STRING,
+    CHATGROQ_API_KEY,
+    CUSTOM_PROMPT
+)
+
+# --- Setup Groq client and MongoDB ---
+client = Groq(api_key=CHATGROQ_API_KEY)
+mongo  = MongoClient(CONNECTION_STRING)
+db     = mongo["edulearnai"]
+chats  = db["chats"]
+
+SYSTEM_PROMPT = "You are a helpful assistant which helps people in their tasks."
+
+# --- Session management in MongoDB ---
+def get_session(session_id: str) -> dict:
+    """Fetch or create a session doc with keys: session_id, history (list), summary (str)."""
+    doc = chats.find_one({"session_id": session_id})
+    if not doc:
+        doc = {"session_id": session_id, "history": [], "summary": ""}
+        chats.insert_one(doc)
+    return doc
+
+def save_session(doc: dict):
+    """Overwrite the session document in Mongo."""
+    chats.replace_one({"session_id": doc["session_id"]}, doc)
+
+# --- History summarization ---
+async def summarize_history(prev_summary: str, history_msgs: list[str]) -> str:
+    """Ask the LLM to produce a short summary of the combined previous summary + new messages."""
+    combined = prev_summary + "\n" + "\n".join(history_msgs)
+    prompt = (
+        "Summarize the following chat history in one or two short sentences:\n\n"
+        f"{combined}\n\nSummary:"
+    )
+    resp = client.chat.completions.create(
+        model="meta-llama/llama-4-scout-17b-16e-instruct",
+        messages=[{"role": "user", "content": prompt}],
+        temperature=0.3,
+        max_completion_tokens=150,
+        top_p=1,
+        stream=False,
+    )
+    # the first (and only) completion
+    return resp.choices[0].message.content.strip()
+
+# --- Core chat logic ---
+async def chat(session_id: str, question: str):
+    session = get_session(session_id)
+    history = session["history"]
+    summary = session["summary"]
+
+    # If history is too long, summarize it and clear
+    if len(history) >= 10:
+        msgs_to_summarize = [f"{m['role']}: {m['content']}" for m in history]
+        new_summary = await summarize_history(summary, msgs_to_summarize)
+        session["summary"] = new_summary
+        session["history"] = []
+        history = []
+
+    # Build the prompt
+    chat_history_text = "\n".join([f"{m['role']}: {m['content']}" for m in history])
+    full_prompt = CUSTOM_PROMPT.format(
+        context=SYSTEM_PROMPT,
+        chat_history=chat_history_text or "(no prior messages)",
+        question=question
+    )
+
+    # Call the model, streaming
+    completion = client.chat.completions.create(
+        model="meta-llama/llama-4-scout-17b-16e-instruct",
+        messages=[{"role": "user", "content": full_prompt}],
+        temperature=1,
+        max_completion_tokens=1024,
+        top_p=1,
+        stream=True,
+    )
+
+    # Print & accumulate the assistant’s reply
+    assistant_response = ""
+    print("Assistant:", end=" ", flush=True)
+    for chunk in completion:
+        delta = chunk.choices[0].delta.content or ""
+        print(delta, end="", flush=True)
+        assistant_response += delta
+    print()  # newline after done
+
+    # Persist the new exchange
+    session["history"].append({"role": "user",      "content": question})
+    session["history"].append({"role": "assistant", "content": assistant_response})
+    save_session(session)
+
+# --- CLI loop ---
+if __name__ == "__main__":
+    import argparse
+
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--session",
+        "-s",
+        default="default",
+        help="Session ID (used to key chat history in MongoDB)"
+    )
+    args = parser.parse_args()
+
+    print(f"Starting noRag chat (session={args.session}).  Type Ctrl+C to quit.\n")
+    try:
+        while True:
+            user_q = input("You: ")
+            if not user_q.strip():
+                continue
+            asyncio.run(chat(args.session, user_q))
+    except KeyboardInterrupt:
+        print("\nGoodbye!")