#!/bin/bash

# generate some helpfull information
if [ "$1" == "help" ]; then
 echo "usage: call the script with at least 1 parameter:"
 echo "the board you want to dump"
 echo ""
 echo "the second parameter is optional and can be used"
 echo "to download the first 'x' pages (from 0 tot 15) default 0"
 echo "you can also provide all to dump them all"
 echo ""
 echo "examples:"
 echo "4chandump.sh p (downloads the first page of /p/)"
 echo "4chandump.sh p 4 (downloads page 0 through 4)"
 echo "4chandump.sh p all (downloads every page)"
 exit 1
fi
# TODO check for both arguments and default them to something if an argument is not provided

# parse the arguments
if [ "$1" == "" ]; then
 echo error && exit
else
 BOARD="$1"
fi

if [ "$2" == "all" ]; then
 PAGES="15"
elif [ "$2" == "" ]; then
 PAGES="15"
else
 PAGES="$2"
fi

# ikno, dirty but whatevs
if [ ! -d $BOARD ]; then
 mkdir $BOARD
fi

cd $BOARD

# remove old temp files
rm urls.tmp
rm posts.tmp
rm imageurls.tmp
rm 4chan.html

# get the html file
for (( p = 0; p <= $PAGES; p++ ))
do
 echo "http://boards.4chan.org/$BOARD/$p" >> urls.tmp
done

wget -i urls.tmp -O 4chan.html -nv -nc

# parse the html file to get all the threads
cat 4chan.html | sed s/\</\\n\</g | grep "View Thread" | sed 's/.*\(res\/[0-9]*\).*/\1/g' | sed "s/\(.*\)/http:\/\/boards.4chan.org\/$BOARD\/\1/g" | uniq > posts.tmp

# wget all the files and output them to 1 single file
wget -i posts.tmp -O 4chan.html -nv -nc

# parse all the files to get the image urls
cat 4chan.html | sed s/\</\\n\</g | grep ".*\.jpg$" | grep "a href" | sed 's/.*>\([0-9]*\.[jpgs][pniw][fg]\).*/\1/' | sed "s/\(.*\)/http:\/\/images\.4chan\.org\/$BOARD\/src\/\1/" | uniq > imageurls.tmp



# wget ALL DEM IMAGES
wget -i imageurls.tmp -nc

# zip all them shit and send them via email
#zip images.zip *
# remove the images after being zipped, takes up a lot of space so...
#rm images/*

# remove temp files

rm urls.tmp
rm posts.tmp
rm imageurls.tmp
rm 4chan.html

cd ..