youtube-summarizer-with-langchain-chatgpt.ipynb

youtube-summarizer-with-langchain-chatgpt.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/gist/Sh1n0g1/d69db6dbc5c13ce887c23c6828658570/youtube-summarizer-with-langchain-chatgpt.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# How to use\n",
        "* Input OpenAI API Key\n",
        "* Input YouTube Video ID\n",
        "  * You can get the Video ID from the URL\n",
        "  * If URL is https://www.youtube.com/watch?v=PlQ4Y8knqvA, then the ID i s `PlQ4Y8knqvA`\n",
        "* Configure the caption language (Default: en, ja)\n"
      ],
      "metadata": {
        "id": "N19SrdFqDsGv"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "#@title 1. Settings\n",
        "openai_api_key = \"\" #@param {type:\"string\"}\n",
        "youtube_video_id = \"\" #@param {type:\"string\"}\n",
        "#@markdown Specify the Language separated by comma (e.g. `en, ja`)\n",
        "acceptable_caption_languages = \"en , ja \" #@param {type:\"string\"}\n",
        "#@markdown If you want to get extra info about summarization, enable this\n",
        "verbose = True #@param {type:\"boolean\"}\n",
        "summary_language = \"Japanese\" #@param {type:\"string\"}\n",
        "if summary_language==\"\":\n",
        "  summary_language=\"English\"\n",
        "\n",
        "languages=[]\n",
        "for l in acceptable_caption_languages.split(\",\"):\n",
        "  languages.append(l.strip())\n",
        "if len(languages)==0:\n",
        "  languages=[\"en\"]\n",
        "\n",
        "  "
      ],
      "metadata": {
        "cellView": "form",
        "id": "rlRFCK440YpK"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "RRYSu48huSUW",
        "cellView": "form"
      },
      "outputs": [],
      "source": [
        "#@title 2. Install Module \n",
        "!pip -q install langchain\n",
        "!pip -q install openai\n",
        "!pip -q install tiktoken\n",
        "!pip -q install youtube-transcript-api\n"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#@title 3. Import Module\n",
        "import os\n",
        "import textwrap\n",
        "\n",
        "from langchain import OpenAI, PromptTemplate, LLMChain\n",
        "from langchain.text_splitter import CharacterTextSplitter\n",
        "from langchain.chains.mapreduce import MapReduceChain\n",
        "from langchain.prompts import PromptTemplate\n",
        "from langchain.docstore.document import Document\n",
        "from langchain.chains.summarize import load_summarize_chain\n",
        "from langchain.callbacks import get_openai_callback\n",
        "import tiktoken\n",
        "\n",
        "from youtube_transcript_api import YouTubeTranscriptApi\n"
      ],
      "metadata": {
        "cellView": "form",
        "id": "JBHj00kg0Gqh"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#@title 4. Get Youtube Caption\n",
        "youtube_caption = YouTubeTranscriptApi.get_transcript(youtube_video_id, languages)\n",
        "caption=\"\"\n",
        "previous_caption_end_time=0\n",
        "for c in youtube_caption:\n",
        "  space=c['start'] - previous_caption_end_time\n",
        "  if space > 3:\n",
        "    caption+=c['text'] + ',\\n'\n",
        "  else:\n",
        "    caption+=c['text'] + ' '\n",
        "  previous_caption_end_time = c['start'] \n",
        "print(f\"{caption}\\n{'='*20}\")\n",
        "print(f\"Length of Text:{len(caption)}\")\n",
        "encoding = tiktoken.get_encoding(\"cl100k_base\")\n",
        "print(f\"Tokens: {len(encoding.encode(caption))}\")\n",
        "\n",
        "text_splitter = CharacterTextSplitter(chunk_size=1000, separator=\"\\n\")\n",
        "texts = text_splitter.split_text(caption)\n",
        "print(f\"{len(texts)} chunks created.\")\n",
        "docs = [Document(page_content=t) for t in texts]"
      ],
      "metadata": {
        "cellView": "form",
        "id": "RzL3WQro0TOg"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "M5uYr_-TPKgR"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#@title 5. Use a Custom Prompt (Optional)\n",
        "map_prompt_template =\"\"\"Write a consise summary of the following YouTube transcription:\n",
        "###\n",
        "{text}\n",
        "\n",
        "SUMMARY:\n",
        "\n",
        "\"\"\"\n",
        "\n",
        "final_prompt_template = \"Write a bullet point summary of the following in \" + summary_language + \"\"\":\n",
        "###\n",
        "{text}\n",
        "\n",
        "SUMMARY:\"\"\"\n",
        "\n",
        "map_prompt = PromptTemplate(template=map_prompt_template, \n",
        "                        input_variables=[\"text\"])\n",
        "final_prompt = PromptTemplate(template=final_prompt_template, \n",
        "                        input_variables=[\"text\"])"
      ],
      "metadata": {
        "id": "YiDNATpDHdsz"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#@title 6. Get Summary\n",
        "os.environ[\"OPENAI_API_KEY\"] = openai_api_key\n",
        "llm = OpenAI(temperature=0)\n",
        "with get_openai_callback() as cb:\n",
        "  if use_custom_prompt:\n",
        "    chain = load_summarize_chain(llm, chain_type=\"map_reduce\", verbose=verbose, map_prompt=map_prompt, combine_prompt=final_prompt)\n",
        "  else:\n",
        "    chain = load_summarize_chain(llm, chain_type=\"map_reduce\", verbose=verbose)\n",
        "  output_summary = chain.run(docs)\n",
        "\n",
        "  \n",
        "  \n",
        "  print(cb)\n"
      ],
      "metadata": {
        "id": "h4nRP8oGO2qf"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#@title 7. Print Summary\n",
        "\n",
        "for t in output_summary.split(\"\\n\"):\n",
        "  print(t)"
      ],
      "metadata": {
        "id": "nbW7Yqp4OAlm"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}