dotfiles/.local/bin/threadwatcher

#!/bin/bash
#A script that interacts with 4chans API to checks for media to download out of threads.
#It uses the file name used by the uploader.
#(and adds post no. to distinguish possible duplicate file names)
#consider using it in a cronjob intermittently with something like
#*/10 * * * * /home/<yourname>/.local/bin/threadwatcher scan
THREADWATCHER_DIR=${XDG_DATA_HOME:-$HOME/.local/share}/4chan_watcher
URLFILE="$THREADWATCHER_DIR/threads"
TMP_URLFILE=/tmp/4chan_thread_watcher_tmp$$

[ -d "$THREADWATCHER_DIR" ] || mkdir -p "$THREADWATCHER_DIR"
[ -f "$URLFILE" ] || touch "$URLFILE"

#Cronjob Notifications
DBUS_SESSION_BUS_ADDRESS=unix:path=/run/user/$(id -u)/bus
export DBUS_SESSION_BUS_ADDRESS
export DISPLAY=:0.0

getfilelist(){
	true
}
scan(){
	ping -q -c 1 1.1.1.1 > /dev/null || ping -q -c 1 1.0.0.1 > /dev/null || ping -q -c 1 example.org || { echo "No internet connection detected."; exit ;}
	if [ $(wc -l < "$URLFILE") -gt 0 ]; then
		echo "scanning threads..."
	else
		echo "no threads to watch over currently"
		exit
	fi
	while read -r line; do
		running_dls=0
		url="$(echo "$line" | cut -f1)"
		echo "scanning $url"
		dl_location="$(echo "$line" | cut -f2)"
		mkdir -p "$dl_location"
		echo "downloading to $dl_location"
		json_url="$(echo "$url" | sed -E 's/boards\.(4chan|4channel)/a.4cdn/; s/$/.json/')"
		curl -s -L "$json_url" | jq . > /tmp/content$$
		if [ -z "$(</tmp/content$$)" ]; then
			echo "Thread $url not found ($dl_location) deleting from cached list of threads to watch"
			notify-send "threadwatcher" "Thread downloading $(echo "$dl_location" | sed "s|$HOME|~|") is complete now."
			continue
		else
			echo "$line" >> "$TMP_URLFILE"
			mkdir -p "$dl_location"
		fi
		files_json="$(jq '.posts[] | if has("filename") then {filename: "\(.no)_\(.filename)\(.ext)", location: "\(.tim)\(.ext)", md5: .md5} else null end ' < /tmp/content$$ | grep -vE '^null$')"
		rm /tmp/content$$
		#pastes together a multiline var using process substitution with layout: filename location md5
		files="$(paste <(paste <(echo "$files_json" | jq '.filename' | tr -d '"') <(echo "$files_json" | jq '.location' | tr -d '"')) <(echo "$files_json" | jq '.md5' | tr -d '"'))"
		echo "$files" | while read -r file_line; do
			filename="$(echo "$file_line" | cut -f1 | tr ' ' '_')"
			master_location="$(echo "$file_line" | cut -f2 | tr -d '"')"
			filelocation="$dl_location/$filename"
			correct_md5="$(echo "$file_line" | cut -f3)"
			[ -f "$filelocation" ] && slave_md5="$(openssl dgst -md5 -binary "$filelocation" | openssl enc -base64)"
			board="$(echo "$url" | cut -d '/' -f4)"
			file_url="https://i.4cdn.org/$board/$master_location"
			if [ -f "$filelocation" ]  && [ "$correct_md5" = "$slave_md5" ]; then
				true
			else
				if [ "$correct_md5" != "$slave_md5" ] && [ -f "$filelocation" ]; then
					rm "$filelocation"
					echo "removed $filename because of incorrect checksum, redownloading."
				fi
				[ -f "$filelocation" ] ||
					touch "$filelocation" #to keep atime order correct?
				#limit concurrent dls
				if [ $running_dls -gt 25 ]; then
					wait
					running_dls=0
				fi
				wget -q -O "$filelocation" "$file_url" &
				echo downloading "$filelocation"
				((running_dls=running_dls+1))
			fi
		done
		wait
	done<"$URLFILE"
	mv "$TMP_URLFILE" "$URLFILE"
}

makepathabsolute(){
	if echo "$1" | grep -qE '^/'; then
		dl_location="$1"
	else
		dl_location="$HOME/$1"
	fi
	echo "$dl_location"
}
case "$1" in
	"add") dl_location="$(makepathabsolute "$3")"
		if grep -qP "^$2\t" "$URLFILE"; then
			dl_location_already="$(grep -P "^$2\t" "$URLFILE" | cut -f2)"
			notify-send "threadwatcher" "Thread already being watched. currently downloads to $(echo "$dl_location_already" | sed "s|$HOME|~|")"
			prompt "Do you want to change download directory to $3?" &&
				new_location="$dl_location" ||
				exit 0
			[ -z "$new_location" ] && exit
			sed -i "s|$dl_location_already|$new_location|" "$URLFILE"

			## Move already downloaded files to new location
			mkdir -p "$new_location"
			url="$2"
			json_url="$(echo "$url" | sed -E 's/boards\.(4chan|4channel)/a.4cdn/; s/$/.json/')"
			curl -s -L "$json_url" | jq . > /tmp/content$$
			files_json="$(jq '.posts[] | if has("filename") then {filename: "\(.no)_\(.filename)\(.ext)", location: "\(.tim)\(.ext)"} else null end ' < /tmp/content$$ | grep -vE '^null$')"
			rm /tmp/content$$
			files="$(paste <(echo "$files_json" | jq '.filename' | tr -d '"') <(echo "$files_json" | jq '.location' | tr -d '"'))"
			echo "$files" | while read -r file_line; do
				filename="$(echo "$file_line" | cut -f1 | tr ' ' '_')"
				mv -v "$dl_location_already/$filename" "$new_location"
			done
			rmdir --ignore-fail-on-non-empty "$dl_location_already"
			notify-send "threadwatcher" "already downloaded files moved to $new_location. New files will also be downloaded there"

		else
			printf "%s\t%s\n" "$2" "$dl_location" | tee -ai "$URLFILE"
			echo "added $2 to threadwatcher list. Downloading to $dl_location"
		fi
		echo "dl_location:$dl_location"
		scan;;
	"scan") scan;;
	"list") printf "Thread:\t\t\t\t\t\tDownload location:\n"
		sed "s|$HOME|~|" "$URLFILE";;
	"clean")
		echo "Watchlist used up to now:"
		cat "$URLFILE"
		echo "Deleting..."
		rm "$URLFILE"
		touch "$URLFILE";;
	"edit") ${EDITOR:-vim} "$URLFILE";;
	*)printf "Incorrect usage.\n\tthreadwatcher [add URL DL_LOCATION] [list] [edit] [clean]\n"; exit 1;;
esac