{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "lIYdn1woOS1n"
      },
      "outputs": [],
      "source": [
        "!git clone https://huggingface.co./datasets/muellerzr/RAG-accelerate"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "%cd RAG-accelerate"
      ],
      "metadata": {
        "id": "N_0YZlykKdVS"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "!pip install langchain"
      ],
      "metadata": {
        "id": "AK8XGa24Kekk"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from pathlib import Path"
      ],
      "metadata": {
        "id": "Qt4Zu6mfLvW2"
      },
      "execution_count": 19,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from langchain.text_splitter import MarkdownHeaderTextSplitter, RecursiveCharacterTextSplitter"
      ],
      "metadata": {
        "id": "a9AHp1eZKgUw"
      },
      "execution_count": 12,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "docs_folder = Path(\"docs/source\")"
      ],
      "metadata": {
        "id": "1gxt3ZcTLxAb"
      },
      "execution_count": 20,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "headers_to_split_on = [\n",
        "    (\"#\", \"Header 1\"),\n",
        "    (\"##\", \"Header 2\"),\n",
        "    (\"###\", \"Header 3\"),\n",
        "    (\"####\", \"Header 4\")\n",
        "]"
      ],
      "metadata": {
        "id": "EFwAEg2mKsNL"
      },
      "execution_count": 29,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "markdown_splitter = MarkdownHeaderTextSplitter(\n",
        "    headers_to_split_on=headers_to_split_on,\n",
        "    strip_headers=False\n",
        ")"
      ],
      "metadata": {
        "id": "HpNxkjoZKs2v"
      },
      "execution_count": 30,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "chunk_size = 1024\n",
        "chunk_overlap = 256\n",
        "text_splitter = RecursiveCharacterTextSplitter(\n",
        "    chunk_size=chunk_size, chunk_overlap=chunk_overlap\n",
        ")"
      ],
      "metadata": {
        "id": "ch_UDrYIM4GT"
      },
      "execution_count": 33,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "categories = [\"package_reference\", \"basic_tutorials\", \"concept_guides\", \"usage_guides\"]"
      ],
      "metadata": {
        "id": "kKRNe4qPMM_6"
      },
      "execution_count": 34,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "data = []"
      ],
      "metadata": {
        "id": "D-rajLfOMynB"
      },
      "execution_count": 44,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "for path in docs_folder.glob(\"**/*.md\"):\n",
        "    content = path.read_text()\n",
        "    md_header_splits = markdown_splitter.split_text(content)\n",
        "    for i,split in enumerate(md_header_splits):\n",
        "        if path.parent.name in categories:\n",
        "            md_header_splits[i].metadata.update({\"category\":path.parent.name})\n",
        "    splits = text_splitter.split_documents(md_header_splits)\n",
        "    data += splits"
      ],
      "metadata": {
        "id": "zAqjYMdMMJRH"
      },
      "execution_count": 46,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "import numpy as np"
      ],
      "metadata": {
        "id": "OAaI9OV_NsTC"
      },
      "execution_count": 50,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "np.save(\"processed_documentation_chunks.npy\", data)"
      ],
      "metadata": {
        "id": "H9FzuYdvOGyM"
      },
      "execution_count": 51,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "BgSCAljkOKLG"
      },
      "execution_count": null,
      "outputs": []
    }
  ],
  "metadata": {
    "colab": {
      "name": "scratchpad",
      "provenance": []
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}