langchain/docs/extras/integrations/vectorstores/meilisearch.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Meilisearch\n",
    "\n",
    "> [Meilisearch](https://meilisearch.com) is an open-source, lightning-fast, and hyper relevant search engine. It comes with great defaults to help developers build snappy search experiences. \n",
    ">\n",
    "> You can [self-host Meilisearch](https://www.meilisearch.com/docs/learn/getting_started/installation#local-installation) or run on [Meilisearch Cloud](https://www.meilisearch.com/pricing).\n",
    "\n",
    "Meilisearch v1.3 supports vector search. This page guides you through integrating Meilisearch as a vector store and using it to perform vector search."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Setup\n",
    "\n",
    "### Launching a Meilisearch instance\n",
    "\n",
    "You will need a running Meilisearch instance to use as your vector store. You can run [Meilisearch in local](https://www.meilisearch.com/docs/learn/getting_started/installation#local-installation) or create a [Meilisearch Cloud](https://cloud.meilisearch.com/) account.\n",
    "\n",
    "As of Meilisearch v1.3, vector storage is an experimental feature. After launching your Meilisearch instance, you need to **enable vector storage**. For self-hosted Meilisearch, read the docs on [enabling experimental features](https://www.meilisearch.com/docs/learn/experimental/vector-search). On **Meilisearch Cloud**, enable _Vector Store_ via your project _Settings_ page.\n",
    "\n",
    "You should now have a running Meilisearch instance with vector storage enabled. 🎉\n",
    "\n",
    "### Credentials\n",
    "\n",
    "To interact with your Meilisearch instance, the Meilisearch SDK needs a host (URL of your instance) and an API key.\n",
    "\n",
    "**Host**\n",
    "\n",
    "- In **local**, the default host is `localhost:7700`\n",
    "- On **Meilisearch Cloud**, find the host in your project _Settings_ page\n",
    "\n",
    "**API keys**\n",
    "\n",
    "Meilisearch instance provides you with three API keys out of the box: \n",
    "- A `MASTER KEY` — it should only be used to create your Meilisearch instance\n",
    "- A `ADMIN KEY` — use it only server-side to update your database and its settings\n",
    "- A `SEARCH KEY` — a key that you can safely share in front-end applications\n",
    "\n",
    "You can create [additional API keys](https://www.meilisearch.com/docs/learn/security/master_api_keys) as needed."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Installing dependencies\n",
    "\n",
    "This guide uses the [Meilisearch Python SDK](https://github.com/meilisearch/meilisearch-python). You can install it by running:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "!pip install meilisearch"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For more information, refer to the [Meilisearch Python SDK documentation](https://meilisearch.github.io/meilisearch-python/)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Examples\n",
    "\n",
    "There are multiple ways to initialize the Meilisearch vector store: providing a Meilisearch client or the _URL_ and _API key_ as needed. In our examples, the credentials will be loaded from the environment.\n",
    "\n",
    "You can make environment variables available in your Notebook environment by using `os` and `getpass`. You can use this technique for all the following examples."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import getpass\n",
    "\n",
    "os.environ[\"MEILI_HTTP_ADDR\"] = getpass.getpass(\"Meilisearch HTTP address and port:\")\n",
    "os.environ[\"MEILI_MASTER_KEY\"] = getpass.getpass(\"Meilisearch API Key:\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We want to use OpenAIEmbeddings so we have to get the OpenAI API Key."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "os.environ[\"OPENAI_API_KEY\"] = getpass.getpass(\"OpenAI API Key:\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Adding text and embeddings\n",
    "\n",
    "This example adds text to the Meilisearch vector database without having to initialize a Meilisearch vector store."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from langchain.vectorstores import Meilisearch\n",
    "from langchain.embeddings.openai import OpenAIEmbeddings\n",
    "from langchain.text_splitter import CharacterTextSplitter\n",
    "\n",
    "embeddings = OpenAIEmbeddings()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"../../../state_of_the_union.txt\") as f:\n",
    "    state_of_the_union = f.read()\n",
    "text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)\n",
    "texts = text_splitter.split_text(state_of_the_union)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Use Meilisearch vector store to store texts & associated embeddings as vector\n",
    "vector_store = Meilisearch.from_texts(texts=texts, embedding=embeddings)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Behind the scenes, Meilisearch will convert the text to multiple vectors. This will bring us to the same result as the following example."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Adding documents and embeddings\n",
    "\n",
    "In this example, we'll use Langchain TextSplitter to split the text in multiple documents. Then, we'll store these documents along with their embeddings."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from langchain.document_loaders import TextLoader\n",
    "\n",
    "# Load text\n",
    "loader = TextLoader(\"../../../state_of_the_union.txt\")\n",
    "documents = loader.load()\n",
    "text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)\n",
    "\n",
    "# Create documents\n",
    "docs = text_splitter.split_documents(documents)\n",
    "\n",
    "# Import documents & embeddings in the vector store\n",
    "vector_store = Meilisearch.from_documents(documents=documents, embedding=embeddings)\n",
    "\n",
    "# Search in our vector store\n",
    "query = \"What did the president say about Ketanji Brown Jackson\"\n",
    "docs = vector_store.similarity_search(query)\n",
    "print(docs[0].page_content)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Add documents by creating a Meilisearch Vectorstore"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this approach, we create a vector store object and add documents to it."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from langchain.vectorstores import Meilisearch\n",
    "import meilisearch\n",
    "\n",
    "client = meilisearch.Client(url=\"http://127.0.0.1:7700\", api_key=\"***\")\n",
    "vector_store = Meilisearch(\n",
    "    embedding=embeddings, client=client, index_name=\"langchain_demo\", text_key=\"text\"\n",
    ")\n",
    "vector_store.add_documents(documents)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Similarity Search with score\n",
    "\n",
    "This specific method allows you to return the documents and the distance score of the query to them."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "docs_and_scores = vector_store.similarity_search_with_score(query)\n",
    "docs_and_scores[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Similarity Search by vector"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "embedding_vector = embeddings.embed_query(query)\n",
    "docs_and_scores = vector_store.similarity_search_by_vector(embedding_vector)\n",
    "docs_and_scores[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Additional resources\n",
    "\n",
    "Documentation\n",
    "- [Meilisearch](https://www.meilisearch.com/docs/)\n",
    "- [Meilisearch Python SDK](https://python-sdk.meilisearch.com)\n",
    "\n",
    "Open-source repositories\n",
    "- [Meilisearch repository](https://github.com/meilisearch/meilisearch)\n",
    "- [Meilisearch Python SDK](https://github.com/meilisearch/meilisearch-python)"
   ]
  }
 ],
 "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": "3.11.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}