#!/bin/bash
conf() {
 git config user.name "Daniil Gentili"
 git config user.email "daniil.gentili.dg@gmail.com"
 git config push.default simple
}

dir="$PWD"
tmp="$dir/tmp"
rm -r $tmp
mkdir -p $tmp
cd $tmp
{
  extractlog() {
    tar -xzf logs.tgz
    cd log
    for f in ./*;do {
      echo "Working... for $f"
      sed -i '/200\|206\|302\|301/!d;s/\sHTTP.*//g;s/.*\s/http:\/\/profile\.gigaset\.net/g;/http:\/\/profile\.gigaset\.net\"\"/d;s/?.*//g;s/\.net\/device/\.net\/chagall/g;s/^\.$//g' $f
      echo "Remove duplicates for $f"
      awk '!seen[$0]++' $f>>$tmp/tmp
    };done

    md5sum $tmp/logs.tgz >$dir/.md5sum
    rm $tmp/logs.tgz
    rdupes $tmp/tmp
  }
  
  rdupes() {
    echo "Remove all duplicates for $*"
    awk '!seen[$0]++' $* $dir/urllist >$dir/final
    sed -i 's/\&hellip\///g' $dir/final

    mv $dir/final $dir/urllist
    rm $*
    
    for dl in $(cat $dir/urllist); do {
      curl -w "%{url_effective}\n" -L -f -s -I -S "$dl" -o /dev/null >> $dir/final
    };done
    sed -i '/HTTP/d;s/\&hellip\///g' $dir/final
    mv $dir/final $dir/urllist
    awk '!seen[$0]++' $dir/urllist >$dir/final
    mv $dir/final $dir/urllist
  }
  
  extractbin() {
    echo "Extracting urls..."
    cd $tmp
    for f in $(grep "\.bin" $dir/urllist);do {
      [ $(wget -S --spider "$f" 2>&1 | sed '/Length/!d;s/Length\: //g;s/\s.*//g;s/\s//g;s/[^0-9]*//g' | tr -d "\n") -lt 52428800 ] && {
       file=$(wget -qO- "$f" | strings)
       echo "$file" | sed '/http:\/\//!d;/profile.gigaset.net\|update.gigaset.net/!d;s/.*http:\/\//http:\/\//g;s/update\.gigaset/profile\.gigaset/g'>>$tmp/tmp
       baseurl="$(echo "$f" | sed 's/[^/]*$//')"
       url=$(echo "$file" | sed '/http/d;/\.bin/!d')
       for e in $url;do echo "$baseurl$e">>$tmp/tmp;done
      }
    }; done
    rdupes $tmp/tmp
  }
  
  
  treegen() {
    echo "Creating tree..."
    mkdir -p $tmp/test
    cd $tmp/test
    for f in $(sed 's/http:\/\/profile.gigaset.net\///g;s/^\/*//g' $dir/urllist);do {
      mkdir $(dirname "$f")&>/dev/null
      touch "$f"
    };done
    tree -T "profile.gigaset.net" -C -H http://profile.gigaset.net -o $dir/tree.html
  
  }
  
  wget http://profile.gigaset.net/logs.tgz -qO logs.tgz
  [ "$(md5sum $tmp/logs.tgz)" != "$(cat $dir/.md5sum)" ] && {
    extractlog
    extractbin
    sort urllist > final
    mv final urllist
  }
  
  treegen
  
  cd $dir
  rm -rf $tmp/
  mkdir $tmp
  
  git clone https://$GH_TOKEN@github.com/danog/gigaclone.git $tmp/git 
  cd $tmp/git
  config 
  cp $dir/* .
  rm -rf tmp gigaclone.log
  git add -A
  git commit -m "Updated url list"
  git push origin master &>/dev/null
  
  git checkout gh-pages
  cp $dir/tree.html .
  git add -A
  git commit -m "Updated dir tree"
  git push origin gh-pages &>/dev/null
  
  cd $dir
  echo "Clean up."
  rm -rf $tmp
} &> $dir/gigaclone.log