langchain/docs/docs/how_to/document_loader_markdown.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "d836a98a-ad14-4bed-af76-e1877f7ef8a4",
   "metadata": {},
   "source": [
    "# How to load Markdown\n",
    "\n",
    "[Markdown](https://en.wikipedia.org/wiki/Markdown) is a lightweight markup language for creating formatted text using a plain-text editor.\n",
    "\n",
    "Here we cover how to load `Markdown` documents into LangChain [Document](https://python.langchain.com/api_reference/core/documents/langchain_core.documents.base.Document.html#langchain_core.documents.base.Document) objects that we can use downstream.\n",
    "\n",
    "We will cover:\n",
    "\n",
    "- Basic usage;\n",
    "- Parsing of Markdown into elements such as titles, list items, and text.\n",
    "\n",
    "LangChain implements an [UnstructuredMarkdownLoader](https://python.langchain.com/api_reference/community/document_loaders/langchain_community.document_loaders.markdown.UnstructuredMarkdownLoader.html) object which requires the [Unstructured](https://unstructured-io.github.io/unstructured/) package. First we install it:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c8b147fb-6877-4f7a-b2ee-ee971c7bc662",
   "metadata": {},
   "outputs": [],
   "source": [
    "%pip install \"unstructured[md]\" nltk"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ea8c41f8-a8dc-48cc-b78d-7b3e2427a34c",
   "metadata": {},
   "source": [
    "Basic usage will ingest a Markdown file to a single document. Here we demonstrate on LangChain's readme:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "80c50cc4-7ce9-4418-81b9-29c52c7b3627",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "🦜️🔗 LangChain\n",
      "\n",
      "⚡ Build context-aware reasoning applications ⚡\n",
      "\n",
      "Looking for the JS/TS library? Check out LangChain.js.\n",
      "\n",
      "To help you ship LangChain apps to production faster, check out LangSmith. \n",
      "LangSmith is a unified developer platform for building,\n"
     ]
    }
   ],
   "source": [
    "from langchain_community.document_loaders import UnstructuredMarkdownLoader\n",
    "from langchain_core.documents import Document\n",
    "\n",
    "markdown_path = \"../../../README.md\"\n",
    "loader = UnstructuredMarkdownLoader(markdown_path)\n",
    "\n",
    "data = loader.load()\n",
    "assert len(data) == 1\n",
    "assert isinstance(data[0], Document)\n",
    "readme_content = data[0].page_content\n",
    "print(readme_content[:250])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b7560a6e-ca5d-47e1-b176-a9c40e763ff3",
   "metadata": {},
   "source": [
    "## Retain Elements\n",
    "\n",
    "Under the hood, Unstructured creates different \"elements\" for different chunks of text. By default we combine those together, but you can easily keep that separation by specifying `mode=\"elements\"`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "a986bbce-7fd3-41d1-bc47-49f9f57c7cd1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of documents: 66\n",
      "\n",
      "page_content='🦜️🔗 LangChain' metadata={'source': '../../../README.md', 'category_depth': 0, 'last_modified': '2024-06-28T15:20:01', 'languages': ['eng'], 'filetype': 'text/markdown', 'file_directory': '../../..', 'filename': 'README.md', 'category': 'Title'}\n",
      "\n",
      "page_content='⚡ Build context-aware reasoning applications ⚡' metadata={'source': '../../../README.md', 'last_modified': '2024-06-28T15:20:01', 'languages': ['eng'], 'parent_id': '200b8a7d0dd03f66e4f13456566d2b3a', 'filetype': 'text/markdown', 'file_directory': '../../..', 'filename': 'README.md', 'category': 'NarrativeText'}\n",
      "\n"
     ]
    }
   ],
   "source": [
    "loader = UnstructuredMarkdownLoader(markdown_path, mode=\"elements\")\n",
    "\n",
    "data = loader.load()\n",
    "print(f\"Number of documents: {len(data)}\\n\")\n",
    "\n",
    "for document in data[:2]:\n",
    "    print(f\"{document}\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "117dc6b0-9baa-44a2-9d1d-fc38ecf7a233",
   "metadata": {},
   "source": [
    "Note that in this case we recover three distinct element types:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "75abc139-3ded-4e8e-9f21-d0c8ec40fdfc",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'ListItem', 'NarrativeText', 'Title'}\n"
     ]
    }
   ],
   "source": [
    "print(set(document.metadata[\"category\"] for document in data))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "223b4c11",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.10.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}