Gemini_02_Prompting_long_video_with_gemini_github.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "256859c1-4ea8-45c1-be06-0e2e4d017b7b",
   "metadata": {},
   "source": [
    "# Prompting long 👀 Videos with Gemini 1.5 🔥🔥🔥"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5fa2a677-0ea1-4eb9-bf4e-f70226c1e456",
   "metadata": {},
   "source": [
    "Une démo sur l'utilisation de plusieurs API de Google pour la génération automatique d'un transcript timé d'une vidéo.\n",
    "\n",
    "\n"
   ]
  },
  {
   "attachments": {
    "9ac9ca54-eb53-46ac-ab53-4ae94dabc6e4.png": {
     "image/png": "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"
    }
   },
   "cell_type": "markdown",
   "id": "626a2413-90ab-4b9c-b4af-f857fbeac41c",
   "metadata": {},
   "source": [
    "![image.png](attachment:9ac9ca54-eb53-46ac-ab53-4ae94dabc6e4.png)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "de45f081-1b4f-417a-b4df-d24af1387e3c",
   "metadata": {},
   "source": [
    "# -4) Seting up a venv in Vertex AI Workbench\n",
    "\n",
    "Look at : https://cloud.google.com/vertex-ai/docs/workbench/instances/add-environment to create a python env"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "9bb57413-1aa8-4109-ab88-577cbedf5aa9",
   "metadata": {},
   "outputs": [],
   "source": [
    "requirements_txt = \"\"\"\n",
    "google-cloud-aiplatform\n",
    "varname\n",
    "unidecode\n",
    "pytube\n",
    "google-cloud-videointelligence\n",
    "moviepy\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "fed12a26-3d71-46d4-bfac-8ca8aa904779",
   "metadata": {},
   "outputs": [],
   "source": [
    "shell_code = \"\"\"\n",
    "\n",
    "    myenv=\"gemini1.5\" \n",
    "    \n",
    "    conda create -n $myenv\n",
    "    conda activate $myenv\n",
    "    \n",
    "    conda install pip\n",
    "    pip install -r requirements.txt\n",
    "    \n",
    "# installting & creating the env : \n",
    "    DL_ANACONDA_ENV_HOME=\"${DL_ANACONDA_HOME}/envs/myenv\"\n",
    "    python -m ipykernel install --prefix \"${DL_ANACONDA_ENV_HOME}\" --name $myenv --display-name $myenv\n",
    "\n",
    "#simlinking to vertex's default folder\n",
    "    ln -s /opt/conda/envs/myenv/share/jupyter/kernels/$myenv ls /opt/conda/share/jupyter/kernels\n",
    "\n",
    "\"\"\"# to execute as a shell command"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "df7e3d6c-1415-495e-b868-99cd34bca3e5",
   "metadata": {},
   "source": [
    "# -3) Install python libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "06c9918d-9380-4f57-be6b-f5bd4136c2b0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[33mWARNING: Ignoring invalid distribution -oogle-auth (/opt/conda/lib/python3.10/site-packages)\u001b[0m\u001b[33m\n",
      "\u001b[0mRequirement already satisfied: moviepy in /opt/conda/lib/python3.10/site-packages (1.0.3)\n",
      "Requirement already satisfied: decorator<5.0,>=4.0.2 in /opt/conda/lib/python3.10/site-packages (from moviepy) (4.4.2)\n",
      "Requirement already satisfied: tqdm<5.0,>=4.11.2 in /opt/conda/lib/python3.10/site-packages (from moviepy) (4.65.0)\n",
      "Requirement already satisfied: requests<3.0,>=2.8.1 in /opt/conda/lib/python3.10/site-packages (from moviepy) (2.31.0)\n",
      "Requirement already satisfied: proglog<=1.0.0 in /opt/conda/lib/python3.10/site-packages (from moviepy) (0.1.10)\n",
      "Requirement already satisfied: numpy>=1.17.3 in /opt/conda/lib/python3.10/site-packages (from moviepy) (1.23.5)\n",
      "Requirement already satisfied: imageio<3.0,>=2.5 in /opt/conda/lib/python3.10/site-packages (from moviepy) (2.31.1)\n",
      "Requirement already satisfied: imageio-ffmpeg>=0.2.0 in /opt/conda/lib/python3.10/site-packages (from moviepy) (0.4.9)\n",
      "Requirement already satisfied: pillow>=8.3.2 in /opt/conda/lib/python3.10/site-packages (from imageio<3.0,>=2.5->moviepy) (9.5.0)\n",
      "Requirement already satisfied: setuptools in /opt/conda/lib/python3.10/site-packages (from imageio-ffmpeg>=0.2.0->moviepy) (67.7.2)\n",
      "Requirement already satisfied: charset-normalizer<4,>=2 in /opt/conda/lib/python3.10/site-packages (from requests<3.0,>=2.8.1->moviepy) (3.1.0)\n",
      "Requirement already satisfied: idna<4,>=2.5 in /opt/conda/lib/python3.10/site-packages (from requests<3.0,>=2.8.1->moviepy) (3.4)\n",
      "Requirement already satisfied: urllib3<3,>=1.21.1 in /opt/conda/lib/python3.10/site-packages (from requests<3.0,>=2.8.1->moviepy) (2.0.3)\n",
      "Requirement already satisfied: certifi>=2017.4.17 in /opt/conda/lib/python3.10/site-packages (from requests<3.0,>=2.8.1->moviepy) (2024.2.2)\n",
      "\u001b[33mWARNING: Error parsing requirements for executing: [Errno 2] No such file or directory: '/opt/conda/lib/python3.10/site-packages/executing-1.2.0.dist-info/METADATA'\u001b[0m\u001b[33m\n",
      "\u001b[0m\u001b[33mWARNING: Ignoring invalid distribution -oogle-auth (/opt/conda/lib/python3.10/site-packages)\u001b[0m\u001b[33m\n",
      "\u001b[0m\u001b[33mWARNING: Ignoring invalid distribution -oogle-auth (/opt/conda/lib/python3.10/site-packages)\u001b[0m\u001b[33m\n",
      "\u001b[0mRequirement already satisfied: varname in /opt/conda/lib/python3.10/site-packages (0.13.0)\n",
      "Collecting executing<3.0,>=2.0 (from varname)\n",
      "  Using cached executing-2.0.1-py2.py3-none-any.whl.metadata (9.0 kB)\n",
      "Using cached executing-2.0.1-py2.py3-none-any.whl (24 kB)\n",
      "\u001b[33mWARNING: Error parsing requirements for executing: [Errno 2] No such file or directory: '/opt/conda/lib/python3.10/site-packages/executing-1.2.0.dist-info/METADATA'\u001b[0m\u001b[33m\n",
      "\u001b[0m\u001b[33mWARNING: Ignoring invalid distribution -oogle-auth (/opt/conda/lib/python3.10/site-packages)\u001b[0m\u001b[33m\n",
      "\u001b[0mInstalling collected packages: executing\n",
      "  Attempting uninstall: executing\n",
      "\u001b[33m    WARNING: No metadata found in /opt/conda/lib/python3.10/site-packages\u001b[0m\u001b[33m\n",
      "\u001b[0m    Found existing installation: executing 1.2.0\n",
      "\u001b[31mERROR: Cannot uninstall executing 1.2.0, RECORD file not found. You might be able to recover from this via: 'pip install --force-reinstall --no-deps executing==1.2.0'.\u001b[0m\u001b[31m\n",
      "\u001b[0m\u001b[33mWARNING: Ignoring invalid distribution -oogle-auth (/opt/conda/lib/python3.10/site-packages)\u001b[0m\u001b[33m\n",
      "\u001b[0mRequirement already satisfied: unidecode in /opt/conda/lib/python3.10/site-packages (1.3.8)\n",
      "\u001b[33mWARNING: Error parsing requirements for executing: [Errno 2] No such file or directory: '/opt/conda/lib/python3.10/site-packages/executing-1.2.0.dist-info/METADATA'\u001b[0m\u001b[33m\n",
      "\u001b[0m\u001b[33mWARNING: Ignoring invalid distribution -oogle-auth (/opt/conda/lib/python3.10/site-packages)\u001b[0m\u001b[33m\n",
      "\u001b[0m\u001b[33mWARNING: Ignoring invalid distribution -oogle-auth (/opt/conda/lib/python3.10/site-packages)\u001b[0m\u001b[33m\n",
      "\u001b[0mRequirement already satisfied: pytube in /opt/conda/lib/python3.10/site-packages (15.0.0)\n",
      "\u001b[33mWARNING: Error parsing requirements for executing: [Errno 2] No such file or directory: '/opt/conda/lib/python3.10/site-packages/executing-1.2.0.dist-info/METADATA'\u001b[0m\u001b[33m\n",
      "\u001b[0m\u001b[33mWARNING: Ignoring invalid distribution -oogle-auth (/opt/conda/lib/python3.10/site-packages)\u001b[0m\u001b[33m\n",
      "\u001b[0m\u001b[33mWARNING: Ignoring invalid distribution -oogle-auth (/opt/conda/lib/python3.10/site-packages)\u001b[0m\u001b[33m\n",
      "\u001b[0mRequirement already satisfied: google-cloud-videointelligence in /opt/conda/lib/python3.10/site-packages (2.11.4)\n",
      "Requirement already satisfied: google-api-core!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0 in /opt/conda/lib/python3.10/site-packages (from google-api-core[grpc]!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-cloud-videointelligence) (2.18.0)\n",
      "Requirement already satisfied: proto-plus<2.0.0dev,>=1.22.0 in /opt/conda/lib/python3.10/site-packages (from google-cloud-videointelligence) (1.23.0)\n",
      "Requirement already satisfied: protobuf!=3.20.0,!=3.20.1,!=4.21.0,!=4.21.1,!=4.21.2,!=4.21.3,!=4.21.4,!=4.21.5,<5.0.0dev,>=3.19.5 in /opt/conda/lib/python3.10/site-packages (from google-cloud-videointelligence) (3.20.3)\n",
      "Requirement already satisfied: googleapis-common-protos<2.0.dev0,>=1.56.2 in /opt/conda/lib/python3.10/site-packages (from google-api-core!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-api-core[grpc]!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-cloud-videointelligence) (1.59.1)\n",
      "Requirement already satisfied: google-auth<3.0.dev0,>=2.14.1 in /opt/conda/lib/python3.10/site-packages (from google-api-core!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-api-core[grpc]!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-cloud-videointelligence) (2.27.0)\n",
      "Requirement already satisfied: requests<3.0.0.dev0,>=2.18.0 in /opt/conda/lib/python3.10/site-packages (from google-api-core!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-api-core[grpc]!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-cloud-videointelligence) (2.31.0)\n",
      "Requirement already satisfied: grpcio<2.0dev,>=1.33.2 in /opt/conda/lib/python3.10/site-packages (from google-api-core[grpc]!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-cloud-videointelligence) (1.51.3)\n",
      "Requirement already satisfied: grpcio-status<2.0.dev0,>=1.33.2 in /opt/conda/lib/python3.10/site-packages (from google-api-core[grpc]!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-cloud-videointelligence) (1.48.2)\n",
      "Requirement already satisfied: cachetools<6.0,>=2.0.0 in /opt/conda/lib/python3.10/site-packages (from google-auth<3.0.dev0,>=2.14.1->google-api-core!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-api-core[grpc]!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-cloud-videointelligence) (5.3.1)\n",
      "Requirement already satisfied: pyasn1-modules>=0.2.1 in /opt/conda/lib/python3.10/site-packages (from google-auth<3.0.dev0,>=2.14.1->google-api-core!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-api-core[grpc]!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-cloud-videointelligence) (0.3.0)\n",
      "Requirement already satisfied: rsa<5,>=3.1.4 in /opt/conda/lib/python3.10/site-packages (from google-auth<3.0.dev0,>=2.14.1->google-api-core!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-api-core[grpc]!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-cloud-videointelligence) (4.9)\n",
      "Requirement already satisfied: charset-normalizer<4,>=2 in /opt/conda/lib/python3.10/site-packages (from requests<3.0.0.dev0,>=2.18.0->google-api-core!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-api-core[grpc]!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-cloud-videointelligence) (3.1.0)\n",
      "Requirement already satisfied: idna<4,>=2.5 in /opt/conda/lib/python3.10/site-packages (from requests<3.0.0.dev0,>=2.18.0->google-api-core!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-api-core[grpc]!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-cloud-videointelligence) (3.4)\n",
      "Requirement already satisfied: urllib3<3,>=1.21.1 in /opt/conda/lib/python3.10/site-packages (from requests<3.0.0.dev0,>=2.18.0->google-api-core!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-api-core[grpc]!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-cloud-videointelligence) (2.0.3)\n",
      "Requirement already satisfied: certifi>=2017.4.17 in /opt/conda/lib/python3.10/site-packages (from requests<3.0.0.dev0,>=2.18.0->google-api-core!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-api-core[grpc]!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-cloud-videointelligence) (2024.2.2)\n",
      "Requirement already satisfied: pyasn1<0.6.0,>=0.4.6 in /opt/conda/lib/python3.10/site-packages (from pyasn1-modules>=0.2.1->google-auth<3.0.dev0,>=2.14.1->google-api-core!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-api-core[grpc]!=2.0.*,!=2.1.*,!=2.10.*,!=2.2.*,!=2.3.*,!=2.4.*,!=2.5.*,!=2.6.*,!=2.7.*,!=2.8.*,!=2.9.*,<3.0.0dev,>=1.34.0->google-cloud-videointelligence) (0.5.0)\n",
      "\u001b[33mWARNING: Error parsing requirements for executing: [Errno 2] No such file or directory: '/opt/conda/lib/python3.10/site-packages/executing-1.2.0.dist-info/METADATA'\u001b[0m\u001b[33m\n",
      "\u001b[0m\u001b[33mWARNING: Ignoring invalid distribution -oogle-auth (/opt/conda/lib/python3.10/site-packages)\u001b[0m\u001b[33m\n",
      "\u001b[0m"
     ]
    }
   ],
   "source": [
    "!pip install moviepy\n",
    "!pip install varname\n",
    "!pip install unidecode\n",
    "!pip install pytube\n",
    "!pip install google-cloud-videointelligence"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "285d16c4-9d60-40d1-9384-93280a3b582a",
   "metadata": {},
   "source": [
    "# -2) import des librairies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "8f8ff16c-7dc1-45b9-a207-76017f689aea",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\"\\nDans l'import from vertexai.generative_models import part, l'objet part représente une composante d'un modèle génératif Vertex AI. Il s'agit d'une unité de base qui peut être assemblée avec d'autres parties pour créer un modèle plus complexe.\\n\\nChaque partie possède ses propres fonctionnalités et caractéristiques, et peut être utilisée pour générer différents types de contenu. Voici quelques exemples de parties :\\n\\nTexte: génère du texte, comme des phrases, des paragraphes ou des articles entiers.\\nCode: génère du code, comme des scripts Python ou des fonctions Java.\\nImage: génère des images, comme des photos ou des illustrations.\\nAudio: génère du son, comme de la musique ou des voix.\\nEn plus de ces types de base, il existe également des parties plus spécialisées qui peuvent être utilisées pour des tâches spécifiques, comme la traduction, la synthèse de discours ou la génération de questions/réponses.\\n\""
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\"\"\"Transcribe speech from a video stored on GCS.\"\"\"\n",
    "\n",
    "import vertexai\n",
    "import json\n",
    "import unidecode\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "import base64\n",
    "\n",
    "from google.cloud import storage\n",
    "from google.cloud import videointelligence\n",
    "from pytube import YouTube\n",
    "from varname import nameof\n",
    "from time import sleep\n",
    "from moviepy.editor import VideoFileClip\n",
    "from math import ceil\n",
    "\n",
    "from IPython.display import Markdown, display\n",
    "from sklearn.metrics.pairwise import cosine_similarity\n",
    "from vertexai.generative_models import GenerativeModel\n",
    "\n",
    "from vertexai.generative_models import Part, FinishReason\n",
    "\n",
    "import vertexai.preview.generative_models as generative_models\n",
    "from vertexai.preview.language_models import (ChatModel, InputOutputTextPair,\n",
    "                                              TextEmbeddingModel,\n",
    "                                              TextGenerationModel)\n",
    "\n",
    "\n",
    "\n",
    "\"\"\"\n",
    "Dans l'import from vertexai.generative_models import part, l'objet part représente une composante d'un modèle génératif Vertex AI. Il s'agit d'une unité de base qui peut être assemblée avec d'autres parties pour créer un modèle plus complexe.\n",
    "\n",
    "Chaque partie possède ses propres fonctionnalités et caractéristiques, et peut être utilisée pour générer différents types de contenu. Voici quelques exemples de parties :\n",
    "\n",
    "Texte: génère du texte, comme des phrases, des paragraphes ou des articles entiers.\n",
    "Code: génère du code, comme des scripts Python ou des fonctions Java.\n",
    "Image: génère des images, comme des photos ou des illustrations.\n",
    "Audio: génère du son, comme de la musique ou des voix.\n",
    "En plus de ces types de base, il existe également des parties plus spécialisées qui peuvent être utilisées pour des tâches spécifiques, comme la traduction, la synthèse de discours ou la génération de questions/réponses.\n",
    "\"\"\"\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f076da06-a4fc-4918-ae56-72e316cfb0e6",
   "metadata": {
    "tags": []
   },
   "source": [
    "# -1) Helper functions for bucket management"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a800ad3a-74b5-400f-be1e-2f1faaa91657",
   "metadata": {
    "tags": []
   },
   "source": [
    "## -1.1) Managing buckets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "1bd00d79-e617-40c7-a292-f78fb7f0ce4e",
   "metadata": {},
   "outputs": [],
   "source": [
    "def list_buckets_names():\n",
    "    storage_client = storage.Client()\n",
    "    buckets        = storage_client.list_buckets()\n",
    "    return [bucket.name for bucket in buckets]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "9b63c1da-892f-44a9-9a70-23e9da1b27a9",
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_bucket_class_location(bucket_name, storage_class = \"STANDARD\", location=\"us\"):\n",
    "    storage_client       = storage.Client()\n",
    "    bucket               = storage_client.bucket(bucket_name)\n",
    "    bucket.storage_class = storage_class\n",
    "    new_bucket           = storage_client.create_bucket(bucket, location=location)\n",
    "    print(f\"Created bucket {new_bucket.name} in {new_bucket.location} with storage class {new_bucket.storage_class}\")\n",
    "    return new_bucket"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "cf523e6a-842d-4449-814e-4a8ccaa83dca",
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_bucket_if_doesnt_exist(bucket_name):\n",
    "    if bucket_name not in list_buckets_names():\n",
    "        print(f\"Going to create bucket :{bucket_name}\")\n",
    "        create_bucket_class_location(bucket_name)\n",
    "        \n",
    "    assert(bucket_name in list_buckets_names())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "198ff24d-375f-47d0-9029-ff5bf014e271",
   "metadata": {},
   "outputs": [],
   "source": [
    "def upload_file_to_gcs(bucket_name, source_file_path, destination_blob_name):\n",
    "    client = storage.Client()\n",
    "    bucket = client.bucket(bucket_name)\n",
    "    blob = bucket.blob(destination_blob_name)\n",
    "    blob.upload_from_filename(source_file_path)\n",
    "    return blob"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "31038d14-bde4-4af1-8a62-ece2ba17e2cd",
   "metadata": {},
   "outputs": [],
   "source": [
    "def upload_file_to_bucket(bucket_name, source_file_path): \n",
    "    source_file_path = f\"./{source_file_path}\"\n",
    "    destination_blob_name = source_file_path.replace(' ', '_')\n",
    "    destination_blob_name = unidecode.unidecode(destination_blob_name)[2:]\n",
    "    uploaded_blob         = upload_file_to_gcs(bucket_name, source_file_path, destination_blob_name)\n",
    "    print(f\"File [{source_file_path}] \\nUploaded to [{uploaded_blob.name}] \\nIn bucket [{bucket_name}]\")\n",
    "    return destination_blob_name"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "93c66310-3d58-43ed-b0b7-81c2bc4b1543",
   "metadata": {
    "tags": []
   },
   "source": [
    "## -1.2) Fetching from youtube "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "1711581b-80f3-4068-bbbb-6a21ac000487",
   "metadata": {},
   "outputs": [],
   "source": [
    "def download_video(video_url):\n",
    "    # Create a YouTube object\n",
    "    yt = YouTube(video_url)\n",
    "    \n",
    "    # Filter to get the highest resolution stream that includes both video and audio\n",
    "    stream = yt.streams.filter(progressive=True, file_extension='mp4').order_by('resolution').desc().first()\n",
    "    \n",
    "    # Download the video\n",
    "    if stream:\n",
    "        print(f\"Going to download {video_url}\")\n",
    "        stream.download()\n",
    "        print(f\"Downloaded into {stream.default_filename}\")\n",
    "    else:\n",
    "        print(\"No suitable stream found\")\n",
    "    return stream.default_filename"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "538d1707-8173-4659-b36d-7680b3266b32",
   "metadata": {},
   "outputs": [],
   "source": [
    "def upload_local_video_sur_bucket(local_video_path, bucket_name):\n",
    "    uploaded_video       = upload_file_to_bucket(bucket_name, local_video_path)\n",
    "    \n",
    "    uri_video_sur_bucket = f\"gs://{bucket_name}/{uploaded_video}\"\n",
    "\n",
    "    return uri_video_sur_bucket"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "c2763199-df3f-4184-ba5b-20d24e2fac43",
   "metadata": {},
   "outputs": [],
   "source": [
    "def upload_video_sur_bucket(video_url):\n",
    "    downloaded_video     = download_video(video_url)\n",
    "    uploaded_video       = upload_file_to_bucket(bucket_name, downloaded_video)\n",
    "    \n",
    "    uri_video_sur_bucket = f\"gs://{bucket_name}/{uploaded_video}\"\n",
    "\n",
    "    return uri_video_sur_bucket"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9fbc024b-f7cb-4b2e-adfa-ba5b7fb1be1c",
   "metadata": {
    "tags": []
   },
   "source": [
    "## -1.3) Divide videos in sub videos 🔥"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "41bfc65e-87ce-4c7d-9e0c-bcf259ea12a0",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_sub_timestamps(video_path, subclip_nb_of_sec):\n",
    "    full_video    = VideoFileClip(video_path)\n",
    "    duration      = full_video.duration\n",
    "    nb_sub_video  = ceil(duration/subclip_nb_of_sec)\n",
    "    start, end    = 0,0\n",
    "    extracts      = []\n",
    "    while end < full_video.duration :\n",
    "        end     = min(start+subclip_nb_of_sec, full_video.duration )\n",
    "        extract = (start, end)\n",
    "        extracts.append(extract)\n",
    "        start   = start+subclip_nb_of_sec\n",
    "    return extracts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "52b7b84b-7412-4646-98bb-f970ac04b98a",
   "metadata": {},
   "outputs": [],
   "source": [
    "def write_subclip(video_path, start,end):\n",
    "    output_path   = f'{video_path.replace(\".mp4\", \"\")}_{start}_{end}.mp4'\n",
    "    full_video    = VideoFileClip(video_path)\n",
    "    sub_video     = full_video.subclip(start, end)\n",
    "    sub_video.write_videofile(output_path, verbose=False, threads=4)\n",
    "    sub_video.close()\n",
    "    return output_path"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "88c657c5-0b42-4b73-b17f-bdfdfbf84f00",
   "metadata": {},
   "outputs": [],
   "source": [
    "def chunk_video(video_path, subclip_nb_of_sec = 1800):\n",
    "    sub_timestamps = get_sub_timestamps(video_path, subclip_nb_of_sec)\n",
    "    output_paths = []\n",
    "    \n",
    "    if len(sub_timestamps)==1:\n",
    "        output_paths.append(video_path)\n",
    "    else:    \n",
    "        for (start, end) in sub_timestamps:\n",
    "            output_path = write_subclip(video_path, start, end)\n",
    "            output_paths.append(output_path)\n",
    "    return output_paths"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9c2993d8-f692-4f79-b9cc-cbc6a8726abb",
   "metadata": {
    "tags": []
   },
   "source": [
    "## - 1.3) Calling Gemini 🔥"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "bd33028e-8f7a-4975-99aa-2c19a2c3e4af",
   "metadata": {},
   "outputs": [],
   "source": [
    "def generate(input_array, max_output_tokens=8192, temperature=1, top_p=0.95):\n",
    "    model = GenerativeModel(\"gemini-1.5-pro-preview-0409\")\n",
    "    params = {}\n",
    "    params[\"safety_settings\"] = {\n",
    "        generative_models.HarmCategory.HARM_CATEGORY_HATE_SPEECH: generative_models.HarmBlockThreshold.BLOCK_MEDIUM_AND_ABOVE,\n",
    "        generative_models.HarmCategory.HARM_CATEGORY_DANGEROUS_CONTENT: generative_models.HarmBlockThreshold.BLOCK_MEDIUM_AND_ABOVE,\n",
    "        generative_models.HarmCategory.HARM_CATEGORY_SEXUALLY_EXPLICIT: generative_models.HarmBlockThreshold.BLOCK_MEDIUM_AND_ABOVE,\n",
    "        generative_models.HarmCategory.HARM_CATEGORY_HARASSMENT: generative_models.HarmBlockThreshold.BLOCK_MEDIUM_AND_ABOVE,\n",
    "    }\n",
    "    params[\"generation_config\"] = {\"max_output_tokens\": max_output_tokens,\"temperature\": temperature,\"top_p\": top_p}\n",
    "    params[\"stream\"] = True\n",
    "    responses = model.generate_content(input_array,**params)\n",
    "    return responses\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "fa4e97bc-582f-426f-993a-dca5bd14ef17",
   "metadata": {},
   "outputs": [],
   "source": [
    "def prompt_video(prompt, uri_video_sur_bucket, verbose=False):\n",
    "    mime_type            = \"video/mp4\"\n",
    "    parameters           = {\"mime_type\"  : mime_type, \n",
    "                            \"uri\"        : uri_video_sur_bucket }\n",
    "    video                = Part.from_uri(**parameters)    \n",
    "    if verbose: print(f\"going to generate : \\n{prompt}\\n {parameters}\")\n",
    "    responses            = generate([prompt, video])\n",
    "    return responses\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "af5052fa-1212-41e8-8717-b20e5c3fad26",
   "metadata": {},
   "outputs": [],
   "source": [
    "def ask_list_of_videos(prompt, list_of_video_uri_sur_bucket):\n",
    "    result     = {}\n",
    "    for uri in list_of_video_uri_sur_bucket:\n",
    "        print(f\"\\nGoing to ask the prompt on {uri}\")\n",
    "        responses  = prompt_video(prompt, uri)\n",
    "        result[uri] = []\n",
    "        for response in responses :\n",
    "            result[uri].append(response.text)\n",
    "            print(\".\",end=\"\")\n",
    "    return result"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c5ae5118-8cf5-441c-913d-6c5f09edd8b3",
   "metadata": {},
   "source": [
    "# 0) Parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "24e78c06-b0d7-4ad7-b10e-705d3dcc10f0",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define the video URL\n",
    "video_url = \"https://www.youtube.com/watch?v=MFp4avqwJME\"\n",
    "video_url = \"https://youtu.be/REeUEBmvJqY\"\n",
    "video_url = \"https://www.youtube.com/watch?v=mEhoD5cGDOs\"\n",
    "video_url = \"https://www.youtube.com/watch?v=TowOva8ll3Q\"\n",
    "video_url = \"https://www.youtube.com/watch?v=bNN3PMkMSfY\"\n",
    "\n",
    "# Set with your own parameters\n",
    "#PROJECT_ID = \"YOUR_PROJECT_ID\"\n",
    "#bucket_name = \"NAME_YOUR_BUCKET\"\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "24c0ba65-56e6-4c5b-a941-c2f45319fed8",
   "metadata": {},
   "source": [
    "## 0.1) verify Vertex AI & bucket accessibility"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "611947aa-5e41-416c-94d8-b38bef176997",
   "metadata": {},
   "outputs": [],
   "source": [
    "vertexai.init(project=PROJECT_ID, location=\"us-central1\")\n",
    "create_bucket_if_doesnt_exist(bucket_name)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0ddb0ec8-d380-4ae4-8871-8d72f4574108",
   "metadata": {
    "tags": []
   },
   "source": [
    "# 1) Prompting long videos"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8e24671b-faa1-409a-a89f-6a8e3574395e",
   "metadata": {},
   "source": [
    "## 1.1) subcliping"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "6562c5b4-d967-4969-99f4-8fdf6b36c2ee",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Going to download https://www.youtube.com/watch?v=bNN3PMkMSfY\n",
      "Downloaded into LA GOUVERNEMENTALITÉ ALGORITHMIQUE OU LART DE NE PAS CHANGER LE MONDE.mp4\n"
     ]
    }
   ],
   "source": [
    "video_path = download_video(video_url)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "683b885d-eba6-4d14-a1e5-42dd2e55d45b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Moviepy - Building video LA GOUVERNEMENTALITÉ ALGORITHMIQUE OU LART DE NE PAS CHANGER LE MONDE_0_1800.mp4.\n",
      "MoviePy - Writing audio in LA GOUVERNEMENTALITÉ ALGORITHMIQUE OU LART DE NE PAS CHANGER LE MONDE_0_1800TEMP_MPY_wvf_snd.mp3\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                                                        \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MoviePy - Done.\n",
      "Moviepy - Writing video LA GOUVERNEMENTALITÉ ALGORITHMIQUE OU LART DE NE PAS CHANGER LE MONDE_0_1800.mp4\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                                                  \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Moviepy - Done !\n",
      "Moviepy - video ready LA GOUVERNEMENTALITÉ ALGORITHMIQUE OU LART DE NE PAS CHANGER LE MONDE_0_1800.mp4\n",
      "Moviepy - Building video LA GOUVERNEMENTALITÉ ALGORITHMIQUE OU LART DE NE PAS CHANGER LE MONDE_1800_3600.mp4.\n",
      "MoviePy - Writing audio in LA GOUVERNEMENTALITÉ ALGORITHMIQUE OU LART DE NE PAS CHANGER LE MONDE_1800_3600TEMP_MPY_wvf_snd.mp3\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                                                        \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MoviePy - Done.\n",
      "Moviepy - Writing video LA GOUVERNEMENTALITÉ ALGORITHMIQUE OU LART DE NE PAS CHANGER LE MONDE_1800_3600.mp4\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                                                  \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Moviepy - Done !\n",
      "Moviepy - video ready LA GOUVERNEMENTALITÉ ALGORITHMIQUE OU LART DE NE PAS CHANGER LE MONDE_1800_3600.mp4\n",
      "Moviepy - Building video LA GOUVERNEMENTALITÉ ALGORITHMIQUE OU LART DE NE PAS CHANGER LE MONDE_3600_5400.mp4.\n",
      "MoviePy - Writing audio in LA GOUVERNEMENTALITÉ ALGORITHMIQUE OU LART DE NE PAS CHANGER LE MONDE_3600_5400TEMP_MPY_wvf_snd.mp3\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                                                        \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MoviePy - Done.\n",
      "Moviepy - Writing video LA GOUVERNEMENTALITÉ ALGORITHMIQUE OU LART DE NE PAS CHANGER LE MONDE_3600_5400.mp4\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "t:  52%|█████▏    | 23229/45000 [09:16<08:43, 41.58it/s, now=None]"
     ]
    }
   ],
   "source": [
    "uri_videos = []\n",
    "subvideo_nb_of_seconds = 60*30 \n",
    "for output_path in chunk_video(video_path, subvideo_nb_of_seconds):\n",
    "    print(output_path)\n",
    "    uri = upload_local_video_sur_bucket(output_path, bucket_name)\n",
    "    uri_videos.append(uri)\n",
    "uri_videos"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4063a7a6-b60e-4569-bbb8-118d05059d3d",
   "metadata": {},
   "source": [
    "## 1.2) asking a question to all subclips"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "15d9541e-ef44-4ff3-a785-de3575358ce5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['gs://input-video-files-jr/LA_GOUVERNEMENTALITE_ALGORITHMIQUE_OU_LART_DE_NE_PAS_CHANGER_LE_MONDE_0_1800.mp4',\n",
       " 'gs://input-video-files-jr/LA_GOUVERNEMENTALITE_ALGORITHMIQUE_OU_LART_DE_NE_PAS_CHANGER_LE_MONDE_1800_3600.mp4',\n",
       " 'gs://input-video-files-jr/LA_GOUVERNEMENTALITE_ALGORITHMIQUE_OU_LART_DE_NE_PAS_CHANGER_LE_MONDE_3600_5400.mp4',\n",
       " 'gs://input-video-files-jr/LA_GOUVERNEMENTALITE_ALGORITHMIQUE_OU_LART_DE_NE_PAS_CHANGER_LE_MONDE_5400_5825.0.mp4']"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "uri_videos"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "86d36885-87da-47d8-af1d-b60100ea12a1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Going to ask the prompt on gs://input-video-files-jr/LA_GOUVERNEMENTALITE_ALGORITHMIQUE_OU_LART_DE_NE_PAS_CHANGER_LE_MONDE_0_1800.mp4\n",
      "...........\n",
      "Going to ask the prompt on gs://input-video-files-jr/LA_GOUVERNEMENTALITE_ALGORITHMIQUE_OU_LART_DE_NE_PAS_CHANGER_LE_MONDE_1800_3600.mp4\n",
      ".........\n",
      "Going to ask the prompt on gs://input-video-files-jr/LA_GOUVERNEMENTALITE_ALGORITHMIQUE_OU_LART_DE_NE_PAS_CHANGER_LE_MONDE_3600_5400.mp4\n",
      "....\n",
      "Going to ask the prompt on gs://input-video-files-jr/LA_GOUVERNEMENTALITE_ALGORITHMIQUE_OU_LART_DE_NE_PAS_CHANGER_LE_MONDE_5400_5825.0.mp4\n",
      "........"
     ]
    }
   ],
   "source": [
    "prompt = \"\"\"\n",
    "Fais moi un chapitrage de la vidéo : \n",
    "pour chaque personne qui parle, donne moi : \n",
    "- son nom\n",
    "- un résumé de ce qu'il dit \n",
    "- le timestamp de début et de fin de son intervention\n",
    "\"\"\"\n",
    "results = ask_list_of_videos(prompt, uri_videos)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "657ff5e7-b61f-4c43-9204-cd21347e256d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Fais moi un chapitrage de la vidéo : \n",
      "pour chaque personne qui parle, donne moi : \n",
      "- son nom\n",
      "- un résumé de ce qu'il dit \n",
      "- le timestamp de début et de fin de son intervention\n",
      "\n",
      "gs://input-video-files-jr/LA_GOUVERNEMENTALITE_ALGORITHMIQUE_OU_LART_DE_NE_PAS_CHANGER_LE_MONDE_0_1800.mp4\n",
      "## Chapitrage de la vidéo \"La gouvernementalité algorithmique ou l'art de ne pas changer le monde\" - Antoinette Rouvroy\n",
      "\n",
      "**Antoinette Rouvroy** (00:00 - 29:59)\n",
      "\n",
      "*   **Introduction et remerciements** (00:00 - 00:35) : Antoinette Rouvroy se présente et exprime sa gratitude pour l'invitation à intervenir.\n",
      "*   **Positionnement par rapport à Foucault** (00:35 - 01:59) : Elle précise qu'elle ne se considère pas comme une exégète de Foucault, mais qu'elle emprunte son concept de \"gouvernementalité\" pour analyser les nouvelles formes de production du savoir, d'exercice du pouvoir et de subjectivation.\n",
      "*   **Nouveaux modes de production du savoir** (01:59 - 03:59) : Elle introduit l'idée d'une \"mise en nombre\" généralisée de la vie et de l'individu, soulignant l'importance des données dans la société contemporaine. \n",
      "*   **Dépassement du prisme institutionnel et du sujet** (03:59 - 04:59) : Elle affirme que le pouvoir ne se situe plus uniquement au sein des institutions étatiques, mais s'exerce de manière diffuse et implique la participation des individus. Elle met en avant le concept de \"servitude volontaire\" et de \"récalcitrance\" pour comprendre ces nouvelles dynamiques de pouvoir.\n",
      "*   **Neutralisation de l'incertitude** (04:59 - 08:59) : Elle présente la gouvernementalité algorithmique comme un mode de gouvernement qui n'a pas de projet politique particulier, mais vise à \"immuniser l'actuel\" contre toute potentialité de changement. Elle explique que les \"big data\" sont constituées de données brutes (\"soft data\") et de données administratives structurées (\"hard data\"), auxquels s'ajoutent les métadonnées. Ces données, bien qu'initialement insignifiantes, permettent de produire du sens et d'exercer une influence sur les comportements.\n",
      "*   **Analyse des données et action sur les comportements** (08:59 - 13:59) : Elle souligne les différences entre l'analyse des données massives et les usages classiques de la statistique. Elle explique que les algorithmes agissent comme des sismographes qui enregistrent et analysent les \"signaux\" émanant de nos comportements, permettant ainsi de modéliser et d'influencer le social.\n",
      "*   **Le projet de la gouvernementalité algorithmique** (13:59 - 22:59) : Elle critique le discours \"publicitaire\" qui présente les \"big data\" comme un moyen d'augmenter notre rationalité et de maîtriser l'avenir. Elle affirme que la gouvernementalité algorithmique vise plutôt à neutraliser l'incertitude et à maintenir l'immobilité de l'ordre établi. Elle met en avant la notion de \"crise de la représentation\" et le glissement vers un \"langage sans langage\" pour expliquer cette tendance.\n",
      "*   **Conclusion** (22:59 - 29:59) : Elle souligne que l'objectif de la gouvernementalité algorithmique est de \"ne pas changer le monde\", mais de le figer dans son état actuel. Elle invite à la réflexion sur les formes de vie et d'altération qui échappent à la quantification et à la modélisation. \n",
      "\n",
      "**Remarque** : Le chapitrage est basé sur la transcription du début de la vidéo. Il est possible que d'autres sujets soient abordés dans la suite de la vidéo.\n",
      "\n",
      "gs://input-video-files-jr/LA_GOUVERNEMENTALITE_ALGORITHMIQUE_OU_LART_DE_NE_PAS_CHANGER_LE_MONDE_1800_3600.mp4\n",
      "## Chapitrage de la vidéo :\n",
      "\n",
      "**Intervenant :** Antoinette Rouvroy (00:00 - 29:59)\n",
      "\n",
      "**Résumé :**\n",
      "\n",
      "Antoinette Rouvroy, chercheuse en droit et spécialiste des enjeux liés aux technologies de l'information, analyse les différences fondamentales entre les statistiques classiques et les méthodes d'analyse de données massives (\"big data\") utilisées dans la gouvernance algorithmique.\n",
      "\n",
      "**Points clés :**\n",
      "\n",
      "* **Finalité :**\n",
      "    * **Statistiques classiques :** Objectiver et quantifier le réel pour confirmer ou infirmer des hypothèses préétablies.\n",
      "    * **Big data :** Faire surgir des hypothèses sur le réel à partir des données elles-mêmes, sans a priori.\n",
      "* **Sélectivité :**\n",
      "    * **Statistiques classiques :** Sélectionnent des données pertinentes et représentatives pour éviter le \"bruit\".\n",
      "    * **Big data :** Exhaustives, prennent en compte toutes les données disponibles, y compris le \"bruit\", pour une meilleure prédiction.\n",
      "* **Critique :**\n",
      "    * **Statistiques classiques :** Possibilité de critique basée sur la sélectivité des données.\n",
      "    * **Big data :** La prétention à l'exhaustivité rend la critique difficile, voire impossible.\n",
      "* **Machine learning :** Les algorithmes apprennent et s'affinent par eux-mêmes, sans intervention humaine, se rapprochant ainsi d'un système \"aveugle\" aux événements singuliers.\n",
      "* **Effets sur la gouvernance :**\n",
      "    * **Statistiques classiques :** Permettent des débats et une remise en question des catégories statistiques.\n",
      "    * **Big data :** La prétention à l'objectivité et à l'exhaustivité rend difficile la contestation des résultats algorithmiques.\n",
      "* **Exemple :**\n",
      "    * **Statistiques classiques :** Comparaison des parcours académiques de différentes disciplines par quantification des données.\n",
      "    * **Big data :** Profilage des individus à des fins de marketing ou de sécurité, anticipant leurs comportements et les privant de leur capacité à s'écarter des prédictions.\n",
      "* **Comparaison avec la gouvernance par la loi :**\n",
      "    * **Droit :** Fonctionne par la menace de sanctions et présuppose les capacités d'entendement et de volonté des individus.\n",
      "    * **Gouvernance algorithmique :** Exploite les faiblesses de la volonté et anticipe les comportements en contournant la rationalité.\n",
      "\n",
      "**Conclusion :** La gouvernance algorithmique, malgré ses prétentions à l'objectivité et à l'efficacité, pose des questions éthiques et politiques quant à la liberté individuelle et aux possibilités de contestation. \n",
      "\n",
      "gs://input-video-files-jr/LA_GOUVERNEMENTALITE_ALGORITHMIQUE_OU_LART_DE_NE_PAS_CHANGER_LE_MONDE_3600_5400.mp4\n",
      "## Chapitrage de la vidéo :\n",
      "\n",
      "Malheureusement, je ne peux pas traiter votre demande car elle nécessite de voir et d'analyser la vidéo pour identifier les personnes et résumer leurs propos. Mes capacités se limitent à l'analyse de texte et je ne peux pas accéder à des contenus externes comme des vidéos. \n",
      "\n",
      "gs://input-video-files-jr/LA_GOUVERNEMENTALITE_ALGORITHMIQUE_OU_LART_DE_NE_PAS_CHANGER_LE_MONDE_5400_5825.0.mp4\n",
      "## Chapitrage de la vidéo :\n",
      "\n",
      "**Intervenante 1 : Nom inconnu (probablement une philosophe)**\n",
      "\n",
      "* **Résumé :** Elle parle des systèmes algorithmiques et de leur impact sur nos vies. Elle souligne que ces systèmes sont réflexifs, c'est-à-dire qu'ils s'adaptent et nous influencent en retour. Elle s'interroge sur la nécessité de sortir de la dualité sujet-objet pour mieux comprendre ces systèmes et propose une approche trivalente : sujet, objet, et système réflexif. Elle met en garde contre l'animisme informatique qui attribue aux algorithmes une conscience et une intentionnalité qu'ils n'ont pas. Elle conclut en soulignant que les algorithmes sont principalement \"oublieux\" et \"amnésiques\", ne gardant de traces que de ce qui a été matérialisé et réalisé.\n",
      "* **Timestamp :** 00:00 - 06:53 \n",
      "\n",
      "**Intervenant 2 : Nom inconnu**\n",
      "\n",
      "* **Résumé :** Il pose des questions à l'intervenante 1 sur la notion de \"visibilité\" utilisée pour décrire les algorithmes et s'interroge sur la nécessité pour les algorithmes d'une fonction de sensorialité similaire à la perception humaine. Il suggère que les algorithmes opèrent à différents niveaux d'intégration des données. \n",
      "* **Timestamp :** 01:22 - 02:59, 03:01 - 03:05, 05:04 - 05:05, 05:43 - 05:50, 06:47 - 06:49\n",
      "\n",
      "**Remarque :**  Il n'est pas possible de connaître les noms des intervenants à partir de la vidéo.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(prompt)\n",
    "for k in results.keys():\n",
    "    print(k)\n",
    "    print(\"\".join(results[k]))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7ee3fa3c-c6e4-4b65-b88d-6fc7dc9a951f",
   "metadata": {},
   "source": [
    "# End"
   ]
  }
 ],
 "metadata": {
  "environment": {
   "kernel": "gemini1.5",
   "name": "common-cpu.m109",
   "type": "gcloud",
   "uri": "gcr.io/deeplearning-platform-release/base-cpu:m109"
  },
  "kernelspec": {
   "display_name": "gemini1.5",
   "language": "python",
   "name": "gemini1.5"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.11"
  },
  "toc-autonumbering": false,
  "toc-showmarkdowntxt": false,
  "toc-showtags": false
 },
 "nbformat": 4,
 "nbformat_minor": 5
}