#!/bin/sh 

SAVEHTML=0
THREAD=
LOC=
SPEED=30
FILENAME=

usage() {
	cat << EOF
Usage: `basename $0` [-h] [-t FILENAME] [-s SECONDS] [-l DIRECTORY] <chan thread url>
Downloads images until cancelled or thread 404's

  -h  		This
  -l DIRECTORY	Save elsewhere than the threads ID
  -s SECONDS	Speed between updates (Default: 30)
  -t FILENAME	Save the thread (with thumbnails)
  
Daniel Triendl ( http://pew.cc/ ) wrote the outline of this script.
Gosha ( http://besvikel.se/ ) edited stuff.
EOF
	exit 1
}

# Output the usage if there aren't any arguments 
if [ "$#" = 0 ]; then
	usage
fi

while getopts "t:l:s:h" OPTION
do
	case "$OPTION" in
		t)
			SAVEHTML=1
			FILENAME=$OPTARG
			;;
		l)
			LOC=$OPTARG
			;;
		s)
			SPEED=$OPTARG
			;;
		h|?)
			usage
			;;
	esac
done

                     # bash voodoo magic
shift $(($OPTIND-1)) # Removes all options we've recognized
THREAD="$@"          # The first one left is our thread. ( Hopefully )

if [ "$THREAD" = "" ]; then
	echo >&2 "Error: No thread specified"
	usage	
fi

if [ "$LOC" = "" ]; then
	LOC=$(echo "$THREAD" | egrep -o '/[a-zA-Z0-9.]*$' | sed 's/\.html//g' | sed 's@/@@')
fi

if [ "$LOC" = "" ]; then # This is not good. `cd ` brings us to ~/
	echo >&2 "The location can't be empty"
	exit 1
fi

if [ ! -d $LOC ]; then
	mkdir $LOC
fi


echo "Downloading to $LOC/"
echo ""

cd $LOC

while true
do
	TMP=`mktemp`
	TMP2=`mktemp`
	TMP3=`mktemp`

	wget -O "$TMP" "$THREAD" --no-check-certificate
	if [ "$?" != "0" ]; then
		rm $TMP $TMP2
		exit 0
	fi
	
	if [ "$SAVEHTML" = "1" ]; then
		# Edit (and save) the HTML so that it links relatively to the picture
		cat "$TMP" \
            | sed 's@\(https\?:\)\?//[a-zA-Z0-9.]\+/[a-z0-9]\+/thumb/\([0-9]\+s.\(jpg\|png\|gif\)\)@thumbs/\2@g' \
            | sed 's@\(https\?:\)\?//[a-zA-Z0-9.]\+/[a-z0-9]\+/src/\([0-9]\+.\(jpg\|png\|gif\)\)@\2@g' \
		    > "$FILENAME"
		
		# Download thumbnails
		egrep "(https{0,1}:){0,1}//[a-zA-Z0-9.]+/[a-z0-9]+/thumb/[0-9]*s.(jpg|png|gif)" "$TMP" -o > "$TMP2"
		cat "$TMP2" \
		| sed 's@^//@https://@' > "$TMP3"
		cp $TMP3 tmp3
		cp $TMP2 tmp2
		cp $TMP tmp1
		wget -nc -i $TMP3 -P "thumbs"  --no-check-certificate
	fi

	# Strip out all pictures from the thread and download them
	egrep '(https{0,1}:){0,1}//[a-zA-Z0-9.]+/[a-z0-9]+/src/([0-9]*).(jpg|png|gif)' "$TMP" -o > "$TMP2"
	cat "$TMP2" \
	| sed 's@^//@https://@' > "$TMP3"
	wget -nc -i $TMP3 --no-check-certificate
	
	rm $TMP $TMP2
	
	echo "Waiting $SPEED seconds before next run"
	sleep $SPEED
done