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.
#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"

scan(){
	if [ $(wc -l "$URLFILE"| cut -d ' ' -f1) -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 $dl_location"
		mkdir -p "$dl_location"
		json_url="$(echo "$url" | sed -E 's/boards\.(4chan|4channel)/a.4cdn/; s/$/.json/')"
		curl -s -L "$json_url" | jq . > /tmp/content$$
		#delete archived threads from watchlist
		if [ -z "$(</tmp/content$$)" ]; then
			echo "Thread $url not found ($dl_location) deleting from cached list of threads to watch"
			continue
		else
			echo "$line" >> "$TMP_URLFILE"
		fi
		files_json="$(jq '.posts[] | if has("filename") then {filename: "\(.no)_\(.filename)\(.ext)", location: "\(.tim)\(.ext)", md5: .md5} else null end ' < /tmp/content$$ | grep -vE '^null$')"
		echo "$files_json" | jq '.filename' | tr -d '"' >> /tmp/filename$$
		echo "$files_json" | jq '.location' | tr -d '"' >> /tmp/location$$
		echo "$files_json" | jq '.md5' | tr -d '"' >> /tmp/md5$$
		paste /tmp/filename$$ /tmp/location$$ >> /tmp/tmp$$
		files="$(paste /tmp/tmp$$ /tmp/md5$$)"
		rm /tmp/tmp$$ /tmp/md5$$ /tmp/location$$ /tmp/filename$$ /tmp/content$$
		echo "$files" | while read -r file_line; do
			filename="$(echo "$file_line" | cut -f1)"
			master_location="$(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"
}
case "$1" in
	"add") if echo "$3" | grep -qE '^/'; then
			dl_location="$3"
		else
			dl_location="$HOME/$3"
		fi
		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 $dl_location_already"
			choice="$(dmenuinput "If you want to change location of thread to download, please insert new directory now:")"
			if echo "$choice" | grep -qE '^/'; then
				new_location="$3"
			else
				new_location="$HOME/$3"
			fi
			[ -z "$new_location" ] && exit
			sed -i "s|$dl_location_already|$new_location|" "$URLFILE"
			mv $dl_location_already/* "$new_location"
			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 $dl_location"
		fi
		scan;;
	"scan") scan;;
	"list") printf "Thread:\t\t\t\t\t\tDownload location:\n"
		sed "s|$HOME|~|" "$URLFILE";;
	"clean")
		echo "Watchlist used up to now:"
		cat "$URLFILE"
		echo "Deleting..."
		rm "$URLFILE"
		touch "$URLFILE";;
esac