In a directory structure full of .info.json files (Generated by youtube-dl), check which youtube videos are no longer online (And write them to a file).

check-youtube-online.py

import os
import sys
import json
import subprocess
import requests
import datetime

def log(msg):
    sys.stdout.buffer.write((msg + "\n").encode('utf8'))

def loadjson(file_path):
    if os.path.getsize(file_path) > 0:
        file = open(file_path, "r")
        jsonx = json.loads(file.read())
        file.close()
        return jsonx
    else:
        return dict()

def writejson(file_path, jsonx):
    file = open(file_path, "w")
    file.write(json.dumps(jsonx, indent=4, sort_keys=True))
    file.close()

youtubedir_path = 'U:/Hoard'
checkdatefile_path = "C:/Users/Flo/Documents/check-youtube-online/checkdate.txt"
notonlinefile_path = "C:/Users/Flo/Documents/check-youtube-online/notonline.txt"

checkdate = loadjson(checkdatefile_path)
notonline = loadjson(notonlinefile_path)

for subdir, dirs, files in os.walk(youtubedir_path):
    for file in files:
        if ".info.json" in file:
            f = open(os.path.join(subdir, file), "r")
            j = json.loads(f.read())

            if j['extractor'] == "youtube" and (j['id'] not in checkdate or (datetime.datetime.today() - datetime.datetime.strptime(checkdate[j['id']], '%Y%m%d')).days > 30):
                log("Checking " + os.path.join(subdir, file))
                r = requests.get("https://www.youtube.com/watch?v=" + j['id'])

                checkdate[j['id']] = datetime.datetime.strftime(datetime.datetime.today(), "%Y%m%d")
                writejson(checkdatefile_path, checkdate)

                if not "<meta name=\"twitter:title" in r.text:
                    log("OFFLINE")
                    notonline[j['id']] = datetime.datetime.strftime(datetime.datetime.today(), "%Y-%m-%d") + ": " + os.path.join(subdir, file).replace(".info.json", "")
                    writejson(notonlinefile_path, notonline)
            else:
                log("Skipping " + os.path.join(subdir, file))