restore_watch_history.py

"""
Simulate re-watching all your videos from a given date onwards.

If part of your youtube history has been deeleted but is still visible from 'my activity', then
you can export it using google takeout, and use this script to simulate watching all these videos
in order, in a relatively short time.

Note that the history will not be backdated.

License: CC0 / Public domain
"""


import hashlib
import json
import random
import os
import time
from typing import Any, Dict, List

import yt_dlp

RESUME = "2022-08-17T11:50:00.000Z"


def main():
    # An empty file will be created in the 'done' directory to keep track
    # of videos that were already watched, so that this script can be
    # interrupted and resumed.
    try:
       os.makedirs("done")
    except FileExistsError:
       pass

    # This file should be generated using google takeout
    with open("watch-history.json", encoding="utf8") as f:
        data = json.load(f)

    kept: List[Dict[str, Any]] = []

    for event in data:
        # Disregard youtube music
        if event["header"] != "YouTube":
            continue
        if "details" in event and event["details"][0]["name"] == "From Google Ads":
            continue
        if event["time"] < RESUME:
            continue
        if "titleUrl" not in event:
            continue
        kept.append(event)

    print(f"Found {len(kept)} videos to watch")

    # Deduplicate
    kept = [event for event in {event["titleUrl"]: event for event in kept}.values()]

    print(f"Found {len(kept)} videos to watch after de-duplication")

    # Sort
    kept.sort(key=lambda x: x["time"])

    opts = {
        "mark_watched": True,
        "simulate": True,
        "quiet": True,
        "cookiesfrombrowser": ("firefox",),
    }
    with yt_dlp.YoutubeDL(opts) as ydl:
        for i, event in enumerate(kept):
            t = event["time"]
            url = event["titleUrl"]
            title = event["title"][8:]

            m = hashlib.sha256()
            m.update(url.encode("utf-8"))
            marker = "done/" + m.hexdigest()

            print(
                f"{i}/{len(kept)} \t {t} \t {url} \t {title} ... ", end="", flush=True
            )

            try:
                with open(marker, "r"):
                    pass
                print(" -> Already done")
                continue
            except FileNotFoundError:
                pass

            try:
                ydl.download(url)
                print(" -> Sleeping ... ", end="", flush=True)
                time.sleep(3 + random.random() * 8)
                print(" -> Done")
            except yt_dlp.utils.DownloadError:
                print(" -> DownloadError")
            with open(marker, "w"):
                pass


if __name__ == "__main__":
    main()

an update, the history entries started coming back for a lot of people. It seems mine is mostly fixed.