{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# AnyCoder Demo Notebook\n",
    "\n",
    "This notebook illustrates how to:\n",
    "\n",
    "1. **Instantiate** the unified `hf_client` with automatic provider routing.\n",
    "2. **Call** a chat completion (Groq → OpenAI → Gemini fall‑back).\n",
    "3. **Trigger** the FastAPI `/predict` endpoint served by *app.py*.\n",
    "4. **Run** a quick sentiment‑analysis pipeline using your preferred provider."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 1. Setup inference client\n",
    "from hf_client import get_inference_client\n",
    "\n",
    "# Choose a model (will route to best provider according to prefix rules)\n",
    "model_id = 'openai/gpt-4'   # try 'gemini/pro' or any HF model path\n",
    "client = get_inference_client(model_id, provider='auto')\n",
    "\n",
    "# Simple chat completion\n",
    "resp = client.chat.completions.create(\n",
    "    model=model_id,\n",
    "    messages=[{'role': 'user', 'content': 'Write a Python function to reverse a string.'}]\n",
    ")\n",
    "print(resp.choices[0].message.content)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 2. Sentiment analysis via HF Inference Providers (OpenAI GPT‑4)\n",
    "from transformers import pipeline\n",
    "\n",
    "sentiment = pipeline(\n",
    "    'sentiment-analysis', \n",
    "    model='openai/gpt-4',      # could be 'gemini/pro' etc.\n",
    "    trust_remote_code=True\n",
    ")\n",
    "sentiment('I love building AI‑powered tools!')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 3. Call the FastAPI /predict endpoint exposed by app.py\n",
    "import json, requests\n",
    "\n",
    "payload = {\n",
    "    'prompt': 'Generate a minimal HTML page.',\n",
    "    'model_id': 'gemini/pro',\n",
    "    'language': 'html',\n",
    "    'web_search': False\n",
    "}\n",
    "\n",
    "r = requests.post('http://localhost:7860/predict', json=payload)\n",
    "print('Status:', r.status_code)\n",
    "print(json.loads(r.text)['code'][:400])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "## Next steps\n",
    "\n",
    "* Switch `model_id` to **`'gemini/pro'`**, **`'fireworks-ai/fireworks-v1'`**, or any HF model (e.g. `Qwen/Qwen3-32B`)—routing will adjust automatically.\n",
    "* Explore **`plugins.py`** for Slack / GitHub integrations.\n",
    "* Use **`auth.py`** helpers to pull private Google Drive docs into the pipeline.\n",
    "* Extend `/predict` with temperature, max‑tokens, or stream support."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "name": "python",
   "version": "3.x"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}