<?php

chdir ("cache");

//read input file or URL
$buffer = file_get_contents($_SERVER['argv'][1]);

preg_match_all("/(http:\/\/.*?)[\s\"<]/", $buffer, $matches);
$urls = array_unique($matches[1]);
asort($urls);

foreach($urls as $url) {
  if (file_exists(str_replace("http://", "", $url))) {
    echo "$url: already in cache\n";
  } else {
    exec("wget -q --page-requisites --tries=1 --timeout=5 \"$url\"", $dummy, $return_var);
    if (!$return_var) {
      echo "$url: cached\n";
    } else {
      echo "$url: wget gave up\n";
    }
  }
}

print_r ($urls);

?>
