#!/bin/bash

if [ -z "$1" ]; then
  echo "Please give me the initial URI so I know where to start from."
  exit 1
fi

init=$1

album=$(echo $init | cut -d/ -f6 | cut -d- -f1)
start=$(echo $init | cut -d/ -f6 | cut -d- -f2)
baseurl=$(echo $init | cut -d/ -f3)

mkdir -p $album/pages
wget -nc -P $album/pages $init
sed -i "s/>/>\n/g" $album/pages/$album-$start
total=$(grep -B1 -m1 -P "http://.*org/img/l.png" $album/pages/$album-$start| grep -oP "(?<=href=\").*(?=\")" | cut -d/ -f6 | cut -d- -f2)
nextpage=$(grep -m1 -oP "http://$baseurl/.*$album-$((start + 1))" $album/pages/$album-$start)
wget -nc -P $album $(grep -oP "http://([0-9]{1,3}\.){3}[0-9]{1,3}.*\.(jpg|png|gif)" $album/pages/$album-$start | sed "s/\&amp;/\&/g")

for i in $(seq $start $total)
do
  wget -nc -P $album/pages $nextpage
  until [ $(grep 'You are opening pages too fast, thus placing a heavy load on the server.' $album/pages/$album-$i | wc -l) -eq 0 ]
#  until [ $blocked -eq 0 ]
  do
    echo "Received warning from server. Sleeping for 30 seconds."
    sleep 30
    needrefresh=1
  done
  if [ ! -z $needrefresh ]; then
    rm -vf $album/pages/$album-$i
    wget -P $album/pages $nextpage
    unset needrefresh
  fi
  sed -i "s/>/>\n/g" $album/pages/$album-$i
  nextpage=$(grep -m1 -oP "http://$baseurl/.*$album-$((i + 1))" $album/pages/$album-$i)
  wget -nc -P $album $(grep -oP "http://([0-9]{1,3}\.){3}[0-9]{1,3}.*\.(jpg|png|gif)" $album/pages/$album-$i | sed "s/\&amp;/\&/g")
  sleep 2
#  echo image = $(grep -oP "http://([0-9]{1,3}\.){3}[0-9]{1,3}.*\.jpg" $album/pages/$album-$i)
done

du -h $album
echo $(find $album -maxdepth 1 -name "*.jpg" | wc -l) images exist