4chan Thread Archiver

4cha

#!/usr/bin/env bash
# 4chan Thread Archiver
# requires: wget

# Settings
archivefolder=/srv/http/kittykatt.us/served.kittykatt.us/projects/4chanarchive
archivesite='http://served.kittykatt.us/projects/4chanarchive'

thread="$1"
threadreal=$(echo "$thread" | sed 's/http:\/\///')
threadnum=$(echo "$threadreal" | cut -f 4 -d /)

cd "$archivefolder"
mkdir "$threadnum"
[[ ! -d "$threadnum" ]] && echo 'Error creating directory!' && exit 1

cd "${threadnum}"

echo -n "Getting thread"

wget -e robots=off -E -nd -N -np -r -k -H -Dimages.4chan.org,thumbs.4chan.org "" ${thread} 2>/dev/null &

while ps | grep $! &>/dev/null; do
        echo -n "."
        sleep 2
done

if [[ -f ${threadnum}.html ]]; then
        echo 'done!'
        mv ${threadnum}.html index.html
else
	echo "error retrieving thread!"
	exit 1
fi

echo "Your thread has been saved to ${archivefolder}/${threadnum}/index.html!"
echo "It can be viewed by pointing your browser to ${archivesite}/${threadnum}"