notebooks/integrations/gemini/qa-langchain-gemini-elasticsearch.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "2a3143e8-3949-4ecc-905c-8333a43c9c87",
   "metadata": {},
   "source": [
    "# Question Answering using Gemini, Langchain & Elasticsearch\n",
    "\n",
    "This tutorial demonstrates how to use the [Gemini API](https://ai.google.dev/docs) to create [embeddings](https://ai.google.dev/docs/embeddings_guide) and store them in Elasticsearch. We will learn how to connect Gemini to private data stored in Elasticsearch and build question/answer capabilities over it using [LangChian](https://python.langchain.com/docs/get_started/introduction)."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "68c5e34d-28f9-4195-9f9c-2a8aec1effe6",
   "metadata": {},
   "source": [
    "## setup\n",
    "\n",
    "* Elastic Credentials - Create an [Elastic Cloud deployment](https://www.elastic.co/search-labs/tutorials/install-elasticsearch/elastic-cloud) to get all Elastic credentials (`ELASTIC_CLOUD_ID`, `ELASTIC_API_KEY`).\n",
    "\n",
    "* `GOOGLE_API_KEY` - To use the Gemini API, you need to [create an API key in Google AI Studio](https://ai.google.dev/tutorials/setup)."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b8e9a58a-942f-4039-96c0-b276d5b8a97f",
   "metadata": {},
   "source": [
    "## Install packages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5c4781ec-06a5-48dd-963e-fb832b3f7ca2",
   "metadata": {},
   "outputs": [],
   "source": [
    "pip install -q -U google-generativeai langchain-elasticsearch langchain langchain_google_genai"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "851db243-ca7d-4a7c-a93b-d22ab149a1bb",
   "metadata": {},
   "source": [
    "## Import packages and credentials"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "26e7f569-c680-447b-9246-b5140ff47b6b",
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "import os\n",
    "from getpass import getpass\n",
    "from urllib.request import urlopen\n",
    "\n",
    "from langchain_elasticsearch import ElasticsearchStore\n",
    "from langchain.text_splitter import CharacterTextSplitter\n",
    "from langchain_google_genai import GoogleGenerativeAIEmbeddings\n",
    "from langchain_google_genai import ChatGoogleGenerativeAI\n",
    "from langchain.prompts import ChatPromptTemplate\n",
    "from langchain.schema.output_parser import StrOutputParser\n",
    "from langchain.schema.runnable import RunnablePassthrough"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b2f68db5-21ac-47b0-941b-1d816b586e18",
   "metadata": {},
   "source": [
    "## Get Credentials"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "543d27f4-2c53-4726-a324-716900d72338",
   "metadata": {},
   "outputs": [],
   "source": [
    "os.environ[\"GOOGLE_API_KEY\"] = getpass(\"Google API Key :\")\n",
    "ELASTIC_API_KEY = getpass(\"Elastic API Key :\")\n",
    "ELASTIC_CLOUD_ID = getpass(\"Elastic Cloud ID :\")\n",
    "elastic_index_name = \"gemini-qa\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e8bd47b9-b946-46d1-ba02-adbda118415a",
   "metadata": {},
   "source": [
    "## Add documents"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bd04e921-206e-4c8b-937a-277d2c5a02e6",
   "metadata": {},
   "source": [
    "### Let's download the sample dataset and deserialize the document."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "44a5b79d-326e-4317-82a3-7918a11ff7b7",
   "metadata": {},
   "outputs": [],
   "source": [
    "url = \"https://raw.githubusercontent.com/ashishtiwari1993/langchain-elasticsearch-RAG/main/data.json\"\n",
    "\n",
    "response = urlopen(url)\n",
    "\n",
    "workplace_docs = json.loads(response.read())"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8591dce2-3fe6-4c87-b268-4694bb86e803",
   "metadata": {},
   "source": [
    "### Split Documents into Passages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3963b0db-80d5-4908-897c-bec6357adc0a",
   "metadata": {},
   "outputs": [],
   "source": [
    "metadata = []\n",
    "content = []\n",
    "\n",
    "for doc in workplace_docs:\n",
    "    content.append(doc[\"content\"])\n",
    "    metadata.append(\n",
    "        {\n",
    "            \"name\": doc[\"name\"],\n",
    "            \"summary\": doc[\"summary\"],\n",
    "            \"rolePermissions\": doc[\"rolePermissions\"],\n",
    "        }\n",
    "    )\n",
    "\n",
    "text_splitter = CharacterTextSplitter(chunk_size=50, chunk_overlap=0)\n",
    "docs = text_splitter.create_documents(content, metadatas=metadata)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a066d5dc-dbc9-495f-934f-bfe96e0fdeec",
   "metadata": {},
   "source": [
    "## Index Documents into Elasticsearch using Gemini Embeddings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "42ba370a-e4b4-4375-b71a-2aee7c40a330",
   "metadata": {},
   "outputs": [],
   "source": [
    "query_embedding = GoogleGenerativeAIEmbeddings(\n",
    "    model=\"models/embedding-001\", task_type=\"retrieval_document\"\n",
    ")\n",
    "\n",
    "es = ElasticsearchStore.from_documents(\n",
    "    docs,\n",
    "    es_cloud_id=ELASTIC_CLOUD_ID,\n",
    "    es_api_key=ELASTIC_API_KEY,\n",
    "    index_name=elastic_index_name,\n",
    "    embedding=query_embedding,\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dbdb2d55-3349-4e95-8087-68f927f0d864",
   "metadata": {},
   "source": [
    "## Create a retriever using Elasticsearch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "17920c1e-9228-42f5-893d-29b666d6f7b2",
   "metadata": {},
   "outputs": [],
   "source": [
    "query_embedding = GoogleGenerativeAIEmbeddings(\n",
    "    model=\"models/embedding-001\", task_type=\"retrieval_query\"\n",
    ")\n",
    "\n",
    "es = ElasticsearchStore(\n",
    "    es_cloud_id=ELASTIC_CLOUD_ID,\n",
    "    es_api_key=ELASTIC_API_KEY,\n",
    "    embedding=query_embedding,\n",
    "    index_name=elastic_index_name,\n",
    ")\n",
    "\n",
    "retriever = es.as_retriever(search_kwargs={\"k\": 3})"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3647d005-d70e-4c3a-b784-052b21e9f143",
   "metadata": {},
   "source": [
    "## Format Docs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "04ee4d3d-09fb-4a35-bc66-8a2951c402a8",
   "metadata": {},
   "outputs": [],
   "source": [
    "def format_docs(docs):\n",
    "    return \"\\n\\n\".join(doc.page_content for doc in docs)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "864afb6a-a671-434a-bd30-006c79ccda24",
   "metadata": {},
   "source": [
    "## Create a Chain using Prompt Template + `gemini-pro` model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4818aef7-3535-494d-a5d4-16ef6d0581af",
   "metadata": {},
   "outputs": [],
   "source": [
    "template = \"\"\"Answer the question based only on the following context:\\n\n",
    "\n",
    "{context}\n",
    "\n",
    "Question: {question}\n",
    "\"\"\"\n",
    "prompt = ChatPromptTemplate.from_template(template)\n",
    "\n",
    "\n",
    "chain = (\n",
    "    {\"context\": retriever | format_docs, \"question\": RunnablePassthrough()}\n",
    "    | prompt\n",
    "    | ChatGoogleGenerativeAI(model=\"gemini-pro\", temperature=0.7)\n",
    "    | StrOutputParser()\n",
    ")\n",
    "\n",
    "chain.invoke(\"what is our sales goals?\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.11.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}