openai_demo.py

"""
OpenAI API in Python — Project Demo
====================================
Covers:
  - Client setup with python-dotenv
  - Basic chat completion
  - System messages
  - Temperature experimentation
  - Max-token limits
  - Interactive chat loop
"""

import os
from dotenv import load_dotenv
from openai import OpenAI

# ── 1. Load API key from .env ──────────────────────────────────────────────────
load_dotenv()

client = OpenAI()          # reads OPENAI_API_KEY from the environment automatically
MODEL  = "gpt-4o-mini"    # cheap & fast; swap for "gpt-4o" when you need more power


# ── 2. Helper: single completion ───────────────────────────────────────────────
def chat(
    user_message: str,
    system_message: str = "You are a helpful assistant.",
    temperature: float  = 0.7,
    max_tokens: int     = 512,
) -> str:
    """
    Send one user message and return the assistant reply as a string.

    Parameters
    ----------
    user_message   : The text prompt from the user.
    system_message : Sets the model's persona / behaviour.
    temperature    : 0 = deterministic, 2 = very random. Default 0.7.
    max_tokens     : Hard cap on reply length (1 token ≈ 4 chars in English).
    """
    response = client.chat.completions.create(
        model       = MODEL,
        temperature = temperature,
        max_tokens  = max_tokens,
        messages    = [
            {"role": "system",  "content": system_message},
            {"role": "user",    "content": user_message},
        ],
    )
    return response.choices[0].message.content


# ── 3. Demo A – Basic call ──────────────────────────────────────────────────────
def demo_basic():
    print("\n" + "="*60)
    print("DEMO A — Basic API call")
    print("="*60)
    reply = chat("What is a large language model? Answer in two sentences.")
    print(reply)


# ── 4. Demo B – Custom system message ──────────────────────────────────────────
def demo_system_message():
    print("\n" + "="*60)
    print("DEMO B — Custom system message")
    print("="*60)

    personas = {
        "Pirate":    "You are a salty pirate. Answer every question using pirate slang.",
        "Professor": "You are a university professor. Give rigorous, academic answers.",
        "ELI5":      "Explain everything as if talking to a 5-year-old.",
    }

    question = "How does the internet work?"
    for name, system_msg in personas.items():
        print(f"\n[{name}]")
        print(chat(question, system_message=system_msg, max_tokens=120))


# ── 5. Demo C – Temperature comparison ─────────────────────────────────────────
def demo_temperature():
    print("\n" + "="*60)
    print("DEMO C — Temperature comparison (same prompt, 3 runs each)")
    print("="*60)

    prompt = "Give me a creative name for a coffee shop."
    for temp in [0.0, 0.9, 1.8]:
        print(f"\n  temperature = {temp}")
        for _ in range(3):
            reply = chat(prompt, temperature=temp, max_tokens=20)
            print(f"    → {reply.strip()}")


# ── 6. Demo D – Max-tokens limit ───────────────────────────────────────────────
def demo_max_tokens():
    print("\n" + "="*60)
    print("DEMO D — max_tokens comparison")
    print("="*60)

    prompt = "Explain quantum computing."
    for limit in [20, 80, 256]:
        print(f"\n  max_tokens = {limit}")
        reply = chat(prompt, max_tokens=limit)
        tokens_used = len(reply.split())          # rough word count
        print(f"  (~{tokens_used} words) {reply[:200]}{'…' if len(reply)>200 else ''}")


# ── 7. Demo E – Interactive chat loop ──────────────────────────────────────────
def demo_interactive():
    """
    A simple multi-turn chat that keeps conversation history
    so the model remembers what was said earlier in the session.
    """
    print("\n" + "="*60)
    print("DEMO E — Interactive chat  (type 'quit' to exit)")
    print("="*60)

    system_msg = "You are a concise, friendly assistant. Keep replies under 3 sentences."
    history: list[dict] = [{"role": "system", "content": system_msg}]

    while True:
        user_input = input("\nYou: ").strip()
        if user_input.lower() in {"quit", "exit", "q"}:
            print("Goodbye!")
            break
        if not user_input:
            continue

        history.append({"role": "user", "content": user_input})

        response = client.chat.completions.create(
            model       = MODEL,
            temperature = 0.7,
            max_tokens  = 256,
            messages    = history,
        )
        assistant_reply = response.choices[0].message.content
        history.append({"role": "assistant", "content": assistant_reply})

        print(f"\nAssistant: {assistant_reply}")


# ── Main ────────────────────────────────────────────────────────────────────────
if __name__ == "__main__":
    demo_basic()
    demo_system_message()
    demo_temperature()
    demo_max_tokens()

    run_interactive = input("\n\nLaunch interactive chat? [y/N]: ").strip().lower()
    if run_interactive == "y":
        demo_interactive()