notebooks/1.0-pto-url-chat-sdk-and-api.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import sys\n",
    "sys.path.insert(0, '../src/') \n",
    "from url_chat import URLChat"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "url_chat = URLChat()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# SDK"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Help on URLChat in module url_chat object:\n",
      "\n",
      "class URLChat(builtins.object)\n",
      " |  URLChat(llm_model: str = 'llama3', split_by: str = 'sentence', split_length: int = 15)\n",
      " |  \n",
      " |  Methods defined here:\n",
      " |  \n",
      " |  __init__(self, llm_model: str = 'llama3', split_by: str = 'sentence', split_length: int = 15)\n",
      " |      Model that answers questions based on the content of a given URL\n",
      " |      \n",
      " |      Args:\n",
      " |          llm_model (str, optional): LLM used for text generation. Defaults to \"llama3\".\n",
      " |          split_by (str, optional): Splits the document from the URL into smaller parts before indexing.\n",
      " |              Can split by \"word\", \"sentence\", \"passage\" (paragraph), or \"page\". Defaults to \"sentence\".\n",
      " |          split_length (int, optional): The chunk size, which is the number of words, sentences, or passages. Defaults to 15.\n",
      " |      \n",
      " |      Raises:\n",
      " |          ValueError: User must choose one of the available LLM: 'llama3', 'mixtral', 'gemma', 'gemma2', 'whisper'\n",
      " |  \n",
      " |  ask(self, url: str, question: str) -> str\n",
      " |      Send a url that has been previously indexed using the _index_ method and a question relating to it,\n",
      " |      this will create a chat and you can ask followup questions related to that url.\n",
      " |      \n",
      " |      Args:\n",
      " |          url (str): a url that has been previously indexed using _index_ method. Example: \"https://en.wikipedia.org/wiki/Brazil\"\n",
      " |          question (str): a question relating to the url. Example: \"What is the population of Brazil?\"\n",
      " |      \n",
      " |      Returns:\n",
      " |          str: A string containing the reply generated by the model.\n",
      " |  \n",
      " |  index(self, url: str)\n",
      " |      Send a url so that the model extracts and indexes its content\n",
      " |      \n",
      " |      Args:\n",
      " |          url (str): full URL. Example: \"https://en.wikipedia.org/wiki/Brazil\"\n",
      " |  \n",
      " |  ----------------------------------------------------------------------\n",
      " |  Data descriptors defined here:\n",
      " |  \n",
      " |  __dict__\n",
      " |      dictionary for instance variables (if defined)\n",
      " |  \n",
      " |  __weakref__\n",
      " |      list of weak references to the object (if defined)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "help(url_chat)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You must first index the URL before asking a question about it"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "url_chat.index(\"https://haystack.deepset.ai/\")\n",
    "url_chat.index(\"https://huggingface.co/\")\n",
    "url_chat.index(\"https://en.wikipedia.org/wiki/Brazil\")\n",
    "url_chat.index(\"https://en.wikipedia.org/wiki/New_York\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can ask followup questions and a different chat will be create for each url. Messages will be saved on the chat for each URL and you can return to it at any time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "According to the latest available data, the population of Brazil is approximately 205,375,043 (estimated as of 2024).\n",
      "Your last question was \"what is the population of Brazil?\"\n",
      "You're referring to New York!\n",
      "\n",
      "New York can refer to:\n",
      "\n",
      "1. New York State: a state located in the northeastern United States, bordering Canada, Vermont, Massachusetts, Connecticut, New Jersey, and Pennsylvania.\n",
      "2. New York City (NYC): the most populous city in the United States, located within the state of New York, on the East Coast. NYC is a global hub for finance, culture, and entertainment.\n",
      "\n",
      "If you're looking for New York outside of the United States, there are also:\n",
      "\n",
      "* Nova Iorque, a municipality in the state of Maranhão, Brazil\n",
      "* Nowy Jork, a former name of Łagiewniki, Włocławek County, Poland\n",
      "* New York, Ukraine, a settlement in Donetsk Oblast, Ukraine\n",
      "\n",
      "Let me know if you have any further questions!\n",
      "We've had a conversation about Brazil, and I provided you with information on various topics such as:\n",
      "\n",
      "1. The economy of Brazil\n",
      "2. Politics in Brazil\n",
      "3. Society in Brazil (including race and ethnicity)\n",
      "4. Literature and theatre in Brazil\n",
      "5. Law and justice in Brazil\n",
      "6. Crime and imprisonment in Brazil\n",
      "7. Human rights in Brazil (including LGBTQ+ rights)\n",
      "\n",
      "And, of course, you asked me what the population of Brazil is, which I answered with the latest available data.\n"
     ]
    }
   ],
   "source": [
    "print(url_chat.ask(\"https://en.wikipedia.org/wiki/Brazil\", \"what is the population of Brazil?\"))\n",
    "print(url_chat.ask(\"https://en.wikipedia.org/wiki/Brazil\", \"what was my last question?\"))\n",
    "\n",
    "# since the URL is different this will create a new chat with no memory of previous messages\n",
    "print(url_chat.ask(\"https://en.wikipedia.org/wiki/New_York\", \"where is ny?\"))\n",
    "\n",
    "# but you can continue the old conversation at any time\n",
    "print(url_chat.ask(\"https://en.wikipedia.org/wiki/Brazil\", \"What have we talked about so far?\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# API\n",
    "\n",
    "**Run this from the folder src/ containing api.py**\n",
    "\n",
    "```\n",
    "uvicorn api:app --reload --port 8001\n",
    "curl -X POST \"http://127.0.0.1:8001/index-url/\" -H \"Content-Type: application/json\" -d '{\"url\": \"https://en.wikipedia.org/wiki/Brazil\"}'\n",
    "curl -X POST \"http://127.0.0.1:8001/ask/\" -H \"Content-Type: application/json\" -d '{\"url\": \"https://en.wikipedia.org/wiki/Brazil\", \"question\": \"what is the population of Brazil?\"}'\n",
    "```"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": ".venv",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.18"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}