{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "# \ud83d\ude80 OpenSource Clipping \u2014 Quick Start Notebook\n",
        "\n",
        "> **\u26a1 Quick Start**: If you just want to get up and running as fast as possible,\n",
        "> simply run all cells from top to bottom. The only thing you need to change is\n",
        "> the **YouTube URL** and your **API Key** in the Configuration section below.\n",
        "\n",
        "---\n",
        "\n",
        "## What is this?\n",
        "\n",
        "This notebook is a **minimal, beginner-friendly** implementation of\n",
        "[OpenSource Clipping](https://github.com/NaufalRizqullah/opensource-clipping) \u2014\n",
        "an AI-powered tool that automatically transforms long-form videos into\n",
        "cinematic short-form highlight clips with:\n",
        "\n",
        "- \ud83c\udfaf AI-curated highlight moments (powered by Google Gemini)\n",
        "- \ud83c\udf99\ufe0f Word-level transcription (powered by Faster-Whisper)\n",
        "- \ud83d\udcf1 Smart face-tracking auto-framing for vertical (9:16) format\n",
        "- \ud83d\udcac Karaoke-style animated subtitles\n",
        "- \ud83c\udfac Cinematic hook teaser intro with glitch transition\n",
        "\n",
        "## How to use this notebook\n",
        "\n",
        "| Step | Section | What to do |\n",
        "|------|---------|------------|\n",
        "| 1 | **Setup** | Run the setup cells to clone the repo & install dependencies |\n",
        "| 2 | **Configuration** | Set your YouTube URL, API key, and clip preferences |\n",
        "| 3 | **Run** | Execute the pipeline \u2014 sit back and wait for your clips! |\n",
        "| 4 | **Download** | Download the generated clips from the `outputs/` folder |\n",
        "\n",
        "### \u26a0\ufe0f Requirements\n",
        "\n",
        "- **Runtime**: Set your Colab runtime to **T4 GPU** (`Runtime > Change runtime type > T4 GPU`)\n",
        "- **API Key**: You need a free [Google Gemini API Key](https://aistudio.google.com/apikey)\n",
        "\n",
        "---"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 1. Setup \u2014 Clone Repository & Install Dependencies\n",
        "\n",
        "This section clones the OpenSource Clipping repository and installs all\n",
        "required Python packages. It also installs **FFmpeg** (needed for video\n",
        "processing) and **Deno** (used internally by some modules).\n",
        "\n",
        "> \ud83d\udca1 **Tip**: This cell only needs to run once per Colab session."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n",
        "# 1a. Clone the repository into the current working directory\n",
        "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n",
        "!rm -rf ./* ./.*\n",
        "!git clone https://github.com/NaufalRizqullah/opensource-clipping.git ."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "%%capture\n",
        "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n",
        "# 1b. Install system dependencies (FFmpeg + Deno) and Python packages\n",
        "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n",
        "import os\n",
        "\n",
        "# FFmpeg \u2014 required for video encoding/decoding\n",
        "!apt-get -qq update\n",
        "!apt-get -qq install -y ffmpeg\n",
        "\n",
        "# Deno \u2014 used by internal subtitle rendering modules\n",
        "!curl -fsSL https://deno.land/install.sh | sh\n",
        "os.environ[\"PATH\"] += \":/root/.deno/bin\"\n",
        "\n",
        "# Python dependencies\n",
        "!pip install -r requirements.txt"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 2. Configuration \u2014 Set Your Preferences\n",
        "\n",
        "This is the **only section you need to edit**. Configure the following:\n",
        "\n",
        "| Parameter | Description | Default |\n",
        "|-----------|-------------|---------|\n",
        "| `YOUTUBE_URL` | The YouTube video URL you want to clip | *(required)* |\n",
        "| `TOTAL_CLIPS` | How many highlight clips to generate | `3` |\n",
        "| `WHISPER_MODEL` | Whisper model size for transcription accuracy | `large-v3` |\n",
        "| `WHISPER_DEVICE` | Device for Whisper inference (`cuda` or `cpu`) | `cuda` |\n",
        "| `WHISPER_COMPUTE_TYPE` | Compute precision (`float16` for Colab T4, `float32` for CPU) | `float16` |\n",
        "\n",
        "### Whisper Model Options\n",
        "\n",
        "| Model | Speed | Accuracy | VRAM |\n",
        "|-------|-------|----------|------|\n",
        "| `tiny` | \u26a1\u26a1\u26a1\u26a1\u26a1 | \u2605\u2606\u2606\u2606\u2606 | ~1 GB |\n",
        "| `base` | \u26a1\u26a1\u26a1\u26a1 | \u2605\u2605\u2606\u2606\u2606 | ~1 GB |\n",
        "| `small` | \u26a1\u26a1\u26a1 | \u2605\u2605\u2605\u2606\u2606 | ~2 GB |\n",
        "| `medium` | \u26a1\u26a1 | \u2605\u2605\u2605\u2605\u2606 | ~5 GB |\n",
        "| `large-v3` | \u26a1 | \u2605\u2605\u2605\u2605\u2605 | ~10 GB |\n",
        "\n",
        "> \ud83d\udca1 **Recommendation**: Use `large-v3` with `float16` on Colab T4 GPU for best\n",
        "> results. If you're on a free Kaggle notebook or CPU-only environment, use\n",
        "> `small` or `medium` with `float32`."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# \u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\n",
        "#  CONFIGURATION \u2014 Edit the values below to match your preferences\n",
        "# \u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\n",
        "\n",
        "# \u2500\u2500 YouTube URL \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n",
        "# Paste the YouTube video URL you want to generate clips from.\n",
        "YOUTUBE_URL = \"https://www.youtube.com/watch?v=YOUR_VIDEO_ID\"  # <-- CHANGE THIS\n",
        "\n",
        "# \u2500\u2500 Total Clips \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n",
        "# Number of highlight clips to generate from the video.\n",
        "# More clips = longer processing time. Start small (1-3) for testing.\n",
        "TOTAL_CLIPS = 3\n",
        "\n",
        "# \u2500\u2500 Whisper Configuration \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n",
        "# These control the speech-to-text transcription step.\n",
        "WHISPER_MODEL = \"large-v3\"        # Model size (see table above)\n",
        "WHISPER_DEVICE = \"cuda\"            # \"cuda\" for GPU, \"cpu\" for CPU-only\n",
        "WHISPER_COMPUTE_TYPE = \"float16\"   # \"float16\" for Colab T4, \"float32\" for CPU/Kaggle\n",
        "\n",
        "# \u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\n",
        "#  OPTIONAL \u2014 Advanced settings (safe to leave as defaults)\n",
        "# \u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\n",
        "\n",
        "ASPECT_RATIO = \"9:16\"              # Output aspect ratio: \"9:16\", \"16:9\", \"1:1\"\n",
        "FONT_STYLE = \"HORMOZI\"             # Subtitle style: \"DEFAULT\", \"HORMOZI\", \"CINEMATIC\", \"STORYTELLER\"\n",
        "HOOK_DURATION = 3                  # Hook teaser duration in seconds\n",
        "WORDS_PER_SUBTITLE = 5             # Max words per subtitle group\n",
        "\n",
        "print(\"\u2705 Configuration loaded!\")\n",
        "print(f\"   URL           : {YOUTUBE_URL}\")\n",
        "print(f\"   Total Clips   : {TOTAL_CLIPS}\")\n",
        "print(f\"   Whisper Model : {WHISPER_MODEL} ({WHISPER_DEVICE}, {WHISPER_COMPUTE_TYPE})\")\n",
        "print(f\"   Aspect Ratio  : {ASPECT_RATIO}\")\n",
        "print(f\"   Font Style    : {FONT_STYLE}\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 3. API Key Setup\n",
        "\n",
        "Your **Google Gemini API Key** is required for the AI analysis step.\n",
        "The recommended way is to store it in **Colab Secrets**:\n",
        "\n",
        "1. Click the \ud83d\udd11 **Key icon** in the left sidebar of Colab\n",
        "2. Add a new secret named `GOOGLE_API_KEY`\n",
        "3. Paste your API key as the value\n",
        "4. Toggle the **\"Notebook access\"** switch ON\n",
        "\n",
        "> \ud83d\udd17 Get a free API key at: https://aistudio.google.com/apikey\n",
        "\n",
        "**Optional keys** (for extra features):\n",
        "- `PEXELS_API_KEY` \u2014 Enables automatic B-roll stock footage ([get one here](https://www.pexels.com/api/))\n",
        "- `HF_TOKEN` \u2014 Enables split-screen / camera-switch mode ([get one here](https://huggingface.co/settings/tokens))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n",
        "# Load API keys from Colab Secrets and create the .env file\n",
        "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n",
        "from google.colab import userdata\n",
        "from pathlib import Path\n",
        "\n",
        "# Required\n",
        "GOOGLE_API_KEY = userdata.get(\"GOOGLE_API_KEY\") or \"\"\n",
        "\n",
        "# Optional \u2014 leave blank if you don't have them\n",
        "PEXELS_API_KEY = userdata.get(\"PEXELS_API_KEY\") or \"\"\n",
        "HF_TOKEN = userdata.get(\"HF_TOKEN\") or \"\"\n",
        "\n",
        "# Write the .env file so the pipeline can read the keys\n",
        "env_text = f\"\"\"# Auto-generated by Quick Start notebook\n",
        "GOOGLE_API_KEY={GOOGLE_API_KEY}\n",
        "PEXELS_API_KEY={PEXELS_API_KEY}\n",
        "HF_TOKEN={HF_TOKEN}\n",
        "\"\"\"\n",
        "Path(\".env\").write_text(env_text, encoding=\"utf-8\")\n",
        "\n",
        "# Status check\n",
        "print(\"\ud83d\udd11 API Key Status:\")\n",
        "print(f\"   GOOGLE_API_KEY : {'\u2705 Set' if GOOGLE_API_KEY else '\u274c Missing (required!)'}\")\n",
        "print(f\"   PEXELS_API_KEY : {'\u2705 Set' if PEXELS_API_KEY else '\u26aa Not set (optional)'}\")\n",
        "print(f\"   HF_TOKEN       : {'\u2705 Set' if HF_TOKEN else '\u26aa Not set (optional)'}\")\n",
        "\n",
        "if not GOOGLE_API_KEY:\n",
        "    print(\"\\n\u26a0\ufe0f  WARNING: GOOGLE_API_KEY is missing!\")\n",
        "    print(\"   Add it to Colab Secrets (\ud83d\udd11 icon in sidebar) before running the pipeline.\")\n",
        "    print(\"   Get your free key at: https://aistudio.google.com/apikey\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 4. Run the Pipeline \ud83c\udfac\n",
        "\n",
        "This cell runs the full AI clipping pipeline:\n",
        "\n",
        "1. **Download** \u2014 Downloads the YouTube video\n",
        "2. **Transcribe** \u2014 Converts speech to text using Faster-Whisper\n",
        "3. **Analyze** \u2014 Google Gemini AI picks the most engaging moments\n",
        "4. **Render** \u2014 Generates clips with face-tracking, subtitles, and hook teasers\n",
        "\n",
        "> \u23f1\ufe0f **Estimated time**: 5\u201315 minutes depending on video length and number of clips.\n",
        "\n",
        "> \u26a0\ufe0f Make sure you have set the `YOUTUBE_URL` in the Configuration section above!"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n",
        "# YouTube Cookies Setup (Optional) - Solves yt-dlp 403 Forbidden errors\n",
        "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n",
        "import os\n",
        "import shutil\n",
        "\n",
        "COOKIES_PATH = os.path.abspath(\"youtube_cookies.txt\")\n",
        "cookie_candidates = [\n",
        "    \"/kaggle/input/datasets/muhammadnaufal/tiktok-secret-cookies/youtube_cookies.txt\",\n",
        "    \"/content/youtube_cookies.txt\",\n",
        "    COOKIES_PATH,\n",
        "]\n",
        "\n",
        "source_cookie = next((p for p in cookie_candidates if os.path.isfile(p)), None)\n",
        "YT_COOKIES_ARG = \"\"\n",
        "\n",
        "if source_cookie:\n",
        "    if os.path.realpath(source_cookie) != os.path.realpath(COOKIES_PATH):\n",
        "        shutil.copy(source_cookie, COOKIES_PATH)\n",
        "    print(f\"\u2705 YouTube cookies aktif: {COOKIES_PATH}\")\n",
        "    YT_COOKIES_ARG = f\"--yt-cookies {COOKIES_PATH}\"\n",
        "else:\n",
        "    print(\"\u2139\ufe0f youtube_cookies.txt tidak ditemukan. Download akan dicoba tanpa cookies.\")\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n",
        "# Execute the OpenSource Clipping pipeline with your configuration\n",
        "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n",
        "\n",
        "!python main.py \\\n",
        "  --url \"{YOUTUBE_URL}\" \\\n",
        "  --clips {TOTAL_CLIPS} \\\n",
        "  --ratio \"{ASPECT_RATIO}\" \\\n",
        "  --font-style \"{FONT_STYLE}\" \\\n",
        "  --hook-duration {HOOK_DURATION} \\\n",
        "  --words-per-sub {WORDS_PER_SUBTITLE} \\\n",
        "  --whisper-model \"{WHISPER_MODEL}\" \\\n",
        "  --whisper-device \"{WHISPER_DEVICE}\" \\\n",
        "  --whisper-compute-type \"{WHISPER_COMPUTE_TYPE}\" \\\n",
        "  {YT_COOKIES_ARG}"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 5. View & Download Results\n",
        "\n",
        "After the pipeline completes, your clips are saved in the `outputs/` folder.\n",
        "Run the cells below to preview and download them."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n",
        "# List all generated output files\n",
        "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n",
        "import os\n",
        "\n",
        "outputs_dir = \"outputs\"\n",
        "if os.path.exists(outputs_dir):\n",
        "    files = os.listdir(outputs_dir)\n",
        "    video_files = [f for f in files if f.endswith(\".mp4\")]\n",
        "    print(f\"\ud83c\udfac Generated {len(video_files)} clip(s):\\n\")\n",
        "    for f in sorted(files):\n",
        "        size_mb = os.path.getsize(os.path.join(outputs_dir, f)) / (1024 * 1024)\n",
        "        icon = \"\ud83c\udfa5\" if f.endswith(\".mp4\") else \"\ud83d\uddbc\ufe0f\" if f.endswith((\".jpg\", \".png\")) else \"\ud83d\udcc4\"\n",
        "        print(f\"   {icon} {f} ({size_mb:.1f} MB)\")\n",
        "else:\n",
        "    print(\"\u274c No outputs found. Make sure the pipeline ran successfully.\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n",
        "# Download all output files as a ZIP archive\n",
        "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n",
        "import shutil\n",
        "from google.colab import files\n",
        "\n",
        "if os.path.exists(outputs_dir) and os.listdir(outputs_dir):\n",
        "    zip_name = \"clipping_outputs\"\n",
        "    shutil.make_archive(zip_name, \"zip\", outputs_dir)\n",
        "    print(f\"\ud83d\udce6 Created {zip_name}.zip \u2014 downloading now...\")\n",
        "    files.download(f\"{zip_name}.zip\")\n",
        "else:\n",
        "    print(\"\u274c No outputs to download.\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "---\n",
        "\n",
        "## \ud83d\udcd6 Troubleshooting\n",
        "\n",
        "| Problem | Solution |\n",
        "|---------|----------|\n",
        "| `GOOGLE_API_KEY not found` | Add your API key to Colab Secrets (\ud83d\udd11 sidebar) |\n",
        "| `CUDA out of memory` | Use a smaller Whisper model (`small` or `medium`) |\n",
        "| `float16 not supported` | Change `WHISPER_COMPUTE_TYPE` to `float32` |\n",
        "| Pipeline takes too long | Reduce `TOTAL_CLIPS` to `1` for testing |\n",
        "| No B-roll footage | Add `PEXELS_API_KEY` to Colab Secrets |\n",
        "\n",
        "## \ud83d\udd17 Links\n",
        "\n",
        "- **GitHub**: [NaufalRizqullah/opensource-clipping](https://github.com/NaufalRizqullah/opensource-clipping)\n",
        "- **Full Documentation**: See the [README](https://github.com/NaufalRizqullah/opensource-clipping/blob/main/README.md) for all CLI options\n",
        "- **Get Gemini API Key**: [aistudio.google.com/apikey](https://aistudio.google.com/apikey)\n",
        "- **Get Pexels API Key**: [pexels.com/api](https://www.pexels.com/api/)"
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "gpuType": "T4",
      "provenance": [],
      "toc_visible": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}