Backup anima-vet.fr

dl_all.sh

#!/bin/bash

rurl="anima-vet.fr/"

function bingrep {
  cat "$2" | tr -d '\000' | grep "$1"
}

function handle_page {
  url=$1
  if ! echo "$url" | grep "$rurl" > /dev/null; then
    url="$rurl$1"
  fi
  if [ -z "$2" ]; then
    page=`echo "$url" | sed 's/^.*\.fr\///'`
  else
    page=$2
  fi
  if test -e "$page"; then
    return
  fi
  wget --quiet "$url" -O "$page"
  if echo "$page" | grep "\.htm" > /dev/null; then
    echo "Downloading $url to $page..."
    dir=`echo "$page" | sed 's/\.htm.*$/_fichiers/'`
    mkdir -p "$dir"
    bingrep "$dir/" "$page" |
      sed 's/.*\(href\|src\|target\)="//' |
      sed 's/".*$//' |
      while read f; do
        handle_page "$f"
      done
    bingrep 'href=[a-z]' "$page" |
      sed 's/^.*href=//' |
      sed 's/>.*$//' |
      while read f; do
        handle_page "$f"
      done
  
    bingrep "href=\"http://\(www\.\)\?$rurl[a-z]" "$page" | 
      sed 's/^.*href="//' |
      sed 's/".*$//' |
      sed 's/\/www\./\//' |
      sort -u |
      while read f; do
        handle_page "$f"
      done
  else
    echo ". $page"
  fi
}

handle_page "$rurl" "index.html"