dotfiles/.local/bin/tools/ripper

57 lines
2.6 KiB
Plaintext
Raw Normal View History

2020-12-26 15:30:55 +01:00
#!/bin/bash
2020-07-04 14:23:27 +02:00
#needed if run as cronjob
export XDG_VIDEOS_DIR=$HOME/vids
DLARCHIVE=$XDG_VIDEOS_DIR/.downloaded
DLLOC=$XDG_VIDEOS_DIR
CHANNELSFILE=$XDG_VIDEOS_DIR/.channels
2020-12-15 13:42:59 +01:00
BLACKLIST=$XDG_VIDEOS_DIR/.blacklist
2020-07-04 14:23:27 +02:00
# Required to display notifications if run as a cronjob:
2020-10-04 15:44:42 +02:00
export DBUS_SESSION_BUS_ADDRESS=unix:path=/run/user/$(id -u)/bus
2020-07-04 14:23:27 +02:00
export DISPLAY=:0.0
2021-01-01 17:25:13 +01:00
APIKEY="$(pass show Misc/Youtube\ Data\ API\ v3 | head -n1 )"
2020-12-26 14:04:07 +01:00
if [ $(pgrep -c ripper) -gt 1 ]; then
echo "Ripper already running, exiting new instance..."
exit
fi
2020-07-04 14:23:27 +02:00
echo "Scanning for new Videos to download"
2021-01-01 17:25:13 +01:00
echo "Scanning on Youtube:"
IDs="$( grep 'youtube' "$CHANNELSFILE" | grep 'channel' | sed 's/https:\/\/www\.youtube\.com\/channel\///')"
not_correctly_formatted="$(grep 'youtube' "$CHANNELSFILE" | grep -v 'channel')"
if [ "$( echo "$not_correctly_formatted" | wc -l )" -gt 0 ]; then
echo Please fix the following channel urls to be scannable:
echo "$not_correctly_formatted"
echo "They need to be in the 'https://www.youtube.com/channel/...' format"
fi
for channel_id in $IDs; do
echo "ID: $channel_id"
curl -s "https://www.googleapis.com/youtube/v3/search?key=$APIKEY&channelId=$channel_id&part=snippet,id&order=date&maxResults=500" | jq '."items" | .[] | .id."videoId"' | tr -d '"' | sed 's/^/https:\/\/www\.youtube\.com\/watch\?v=/' | grep -vf "$BLACKLIST" >> /tmp/todownload$$
done
2020-12-26 14:04:07 +01:00
grep 'youtube' "$DLARCHIVE" | sed 's/youtube /https:\/\/www\.youtube\.com\/watch?v=/' > /tmp/alreadydownloaded$$
2020-07-04 14:23:27 +02:00
2021-01-01 17:25:13 +01:00
echo "Scanning on Bitchute..."
grep 'bitchute' "$CHANNELSFILE" | xargs -L1 lynx --dump --nonumbers -listonly | grep 'bitchute\.com\/video' | sort -u | grep -vf "$BLACKLIST" >> /tmp/todownload$$
2020-12-26 14:04:07 +01:00
grep 'bitchute' "$DLARCHIVE" | sed 's/bitchute /https:\/\/www\.bitchute\.com\/video\//' >> /tmp/alreadydownloaded$$
2020-07-04 14:23:27 +02:00
2020-12-26 15:30:55 +01:00
grep -vf /tmp/alreadydownloaded$$ /tmp/todownload$$ | sort -u > /tmp/new_videos$$
number=$(wc -l /tmp/new_videos$$ | cut -d ' ' -f 1 )
2020-07-04 14:23:27 +02:00
if [ $number -gt 0 ]; then
2020-12-26 14:04:07 +01:00
[ $number -gt 1 ] && plural="s"
notify-send "Channel Ripper" "$number new video$plural available for download, downloading now."
echo "$number new video$plural for download available, downloading now."
if [ "$number" -lt 10 ];then
2021-01-01 17:25:13 +01:00
youtube-dl --get-filename -o "'%(uploader)s' '%(title)s'" -a /tmp/new_videos$$ | xargs -L1 notify-send
fi
2020-12-26 14:04:07 +01:00
rm -f /tmp/alreadydownloaded$$ /tmp/todownload$$
2021-01-01 17:25:13 +01:00
youtube-dl --hls-prefer-native -i --download-archive $DLARCHIVE -f 'bestvideo[ext=mp4]+bestaudio[ext=m4a]/best[ext=mp4]/best' --add-metadata -o "$DLLOC/%(uploader)s/%(upload_date)s-%(title)s.%(ext)s" -a /tmp/new_videos$$
2020-12-26 14:04:07 +01:00
rm -f /tmp/new_videos$$
2020-07-04 14:23:27 +02:00
notify-send "Channel Ripper" "Finished downloading"
fi
if [ $number -eq 0 ]; then
2021-01-01 17:25:13 +01:00
echo "No new videos"
2020-07-04 14:23:27 +02:00
fi