codedump: channeldownloader.py annotate

annotate channeldownloader.py @ 63:bdcdd9c42043

Create win95kggui.c committer: GitHub <noreply@github.com>

author	Paper <37962225+mrpapersonic@users.noreply.github.com>
date	Sat, 23 Apr 2022 05:57:34 -0400
parents	c615532e6572
children	9636d5dee08c

rev	line source
47 00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	1 import argparse
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	2 import internetarchive # pip install internetarchive
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	3 import json
61 c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	4 import glob
47 00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	5 import os
59 a3927b2ec6e6 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 58 diff changeset	6 import re
47 00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	7 import urllib.request
60 4e7a9c7c0cce [channeldownloader] Use yt-dlp in place of youtube-dl Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 59 diff changeset	8 import yt_dlp # pip install yt-dlp
47 00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	9 import itertools
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	10 from urllib.error import HTTPError
61 c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	11 from yt_dlp.utils import sanitize_filename
47 00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	12
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	13 class MyLogger(object):
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	14 def debug(self, msg):
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	15 pass
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	16
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	17 def warning(self, msg):
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	18 pass
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	19
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	20 def error(self, msg):
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	21 pass
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	22
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	23 def matroska_find(filelist):
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	24 for myfile in filelist:
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	25 if os.path.splitext(myfile)[1] == ".mkv" or os.path.splitext(myfile)[1] == ".webm":
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	26 return True
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	27 return False
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	28
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	29 def ytdl_hook(d):
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	30 if d["status"] == "finished":
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	31 print(" downloaded {0}: 100% ".format(os.path.basename(d["filename"])))
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	32 if d["status"] == "downloading":
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	33 print(" downloading {0}: {1}\r".format(os.path.basename(d["filename"]), d["_percent_str"]), end="")
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	34 if d["status"] == "error":
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	35 print(" an error occurred downloading {0}!")
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	36
61 c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	37 def load_split_files(path):
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	38 if os.path.isdir(path):
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	39 result = {"videos": []}
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	40 for f in glob.glob(os.path.join(path, "vids*.json")):
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	41 with open(f, "r", encoding="utf-8") as infile:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	42 for i in json.loads(infile.read())["videos"]:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	43 result["videos"].append(i)
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	44 return result
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	45 else:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	46 return json.loads(open(path, "r", encoding="utf-8"))
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	47
47 00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	48
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	49 parser = argparse.ArgumentParser(description="Downloads (deleted) videos from YTPMV creators")
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	50 parser.add_argument("-c", "--channel", help="channel URL", metavar='<url>', required=True)
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	51 parser.add_argument("-d", "--database", help="json database (https://finnrepo.a2hosted.com/YTPMV_Database)", metavar='<path>', required=True)
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	52 parser.add_argument("-o", "--output", help="output directory, defaults to the channel ID", metavar='<output>')
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	53 args = parser.parse_args()
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	54
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	55 if args.channel[:8] == "https://" or args.channel[:7] == "http://":
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	56 channel = args.channel.split("/")[-1]
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	57 else:
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	58 channel = args.channel
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	59
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	60 if args.output:
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	61 output = args.output
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	62 else:
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	63 output = channel
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	64
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	65 if not os.path.exists(output):
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	66 os.mkdir(output)
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	67
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	68 ytdl_opts = {
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	69 "outtmpl": "{0}/%(title)s-%(id)s.%(ext)s".format(output),
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	70 "retries": 100,
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	71 "nooverwrites": True,
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	72 "call_home": False,
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	73 "quiet": True,
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	74 "writeinfojson": True,
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	75 "writedescription": True,
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	76 "writethumbnail": True,
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	77 "writeannotations": True,
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	78 "writesubtitles": True,
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	79 "allsubtitles": True,
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	80 "ignoreerrors": True,
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	81 "addmetadata": True,
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	82 "continuedl": True,
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	83 "embedthumbnail": True,
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	84 "format": "bestvideo+bestaudio/best",
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	85 "restrictfilenames": True,
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	86 "no_warnings": True,
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	87 "progress_hooks": [ytdl_hook],
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	88 "logger": MyLogger(),
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	89 "ignoreerrors": False,
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	90 }
00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	91
61 c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	92 for i in load_split_files(args.database):
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	93 try:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	94 uploader = i["uploader_id"]
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	95 except Exception:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	96 uploader = "unknown"
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	97 if uploader == channel:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	98 print("{0}:".format(i["id"]))
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	99 isalreadydownloaded = 0
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	100 for file in os.listdir(output):
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	101 if os.path.splitext(file)[1] == ".json":
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	102 if file.find("-" + i["id"] + ".info.json") != -1:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	103 isalreadydownloaded = 1
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	104 if isalreadydownloaded == 1: # not sure how to bypass this without having to go out of the for loop, if anyone could tell me how that would be great!
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	105 print(" video already downloaded!")
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	106 continue
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	107 with yt_dlp.YoutubeDL(ytdl_opts) as ytdl:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	108 try:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	109 result = ytdl.download(["https://youtube.com/watch?v={0}".format(i["id"])]) # TODO: add check for existing downloaded items and don't download them
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	110 continue
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	111 except Exception:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	112 print(" video is not available! attempting to find Internet Archive pages of it...")
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	113 if internetarchive.get_item("youtube-{0}".format(i["id"])).exists: # download from internetarchive if available
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	114 fnames = [f.name for f in internetarchive.get_files("youtube-{0}".format(i["id"]))]
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	115 disallowednames = ["__ia_thumb.jpg", "youtube-{0}_archive.torrent".format(i["id"]), "youtube-{0}_files.xml".format(i["id"]), "youtube-{0}_meta.sqlite".format(i["id"]), "youtube-{0}_meta.xml".format(i["id"])] # list of IA-created files we don't need
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	116 flist = []
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	117 for fname in fnames:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	118 if matroska_find(fnames):
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	119 if fname[-4:] == ".mp4":
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	120 continue
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	121 else:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	122 if fname[-7:] == ".ia.mp4":
47 00403c09455c Add channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: diff changeset	123 continue
61 c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	124 if fname.find("/") == -1:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	125 if fname not in disallowednames and fname[-21:] != "{0}_thumb.jpg".format(i["id"]) and fname[-15:] != "{0}.ogv".format(i["id"]):
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	126 flist.append(fname)
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	127 if len(flist) >= 1:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	128 internetarchive.download("youtube-{0}".format(i["id"]), files=flist, verbose=True, destdir=output, no_directory=True, ignore_existing=True)
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	129 else:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	130 print(" video already downloaded!")
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	131 continue
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	132 if os.path.exists(output + "\\" + i["id"] + ".info.json"): # will always exist no matter which setting was used to download
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	133 for fname in flist:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	134 if os.path.exists(output + "\\" + fname) and not os.path.exists(output + "\\" + sanitize_filename(i["title"], restricted=True) + "-" + fname):
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	135 os.rename(output + "\\" + fname, output + "\\" + sanitize_filename(i["title"], restricted=True) + "-" + fname)
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	136 else:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	137 print("ID file not found!")
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	138 else: # download the vid from waybackmachine (NOTE: only tested with youtube links after polymer, however SHOULD work with links created before then)
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	139 print(" video does not have a Internet Archive page! attempting to download from the Wayback Machine...")
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	140 try:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	141 contenttype = urllib.request.urlopen("https://web.archive.org/web/2oe_/http://wayback-fakeurl.archive.org/yt/{0}".format(i["id"])).getheader("Content-Type")
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	142 if contenttype == "video/webm":
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	143 ext = "webm"
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	144 else:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	145 ext = "mp4"
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	146 urllib.request.urlretrieve("https://web.archive.org/web/2oe_/http://wayback-fakeurl.archive.org/yt/{0}".format(i["id"]), "{3}\\{0}-{1}.{2}".format(sanitize_filename(i["title"], restricted=True), i["id"], ext, output))
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	147 print(" downloaded {0}-{1}.{2}".format(sanitize_filename(i["title"], restricted=True), i["id"], ext))
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	148 except HTTPError:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	149 print(" video not available on the Wayback Machine!")
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	150 except Exception as e:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	151 print(" unknown error downloading video!")
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	152 print(e)
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	153 # metadata
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	154 with open("{2}\\{0}-{1}.info.json".format(sanitize_filename(i["title"], restricted=True), i["id"], output), "w") as jsonfile:
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	155 print(json.dumps(i), end="", file=jsonfile)
c615532e6572 Update channeldownloader.py Paper <37962225+mrpapersonic@users.noreply.github.com> parents: 60 diff changeset	156 print(" saved {0}-{1}.info.json".format(sanitize_filename(i["title"], restricted=True), i["id"], output))

Mercurial > codedump

annotate channeldownloader.py @ 63:bdcdd9c42043