function getimages($url)
{
$url = trim($url);
if (strpos(strtoupper(trim($url)), "HTTP") === 0) {
//do nothing
} else {
$url = "http://" . $url;
}
$html3 = file_get_contents($url);
if (isset($ret)) {
$ret = $this->processhtml($html3, $url, $ret);
} else {
$ret = $this->processhtml($html3, $url, array());
}
if (count($ret) < 3) {
if (!function_exists('curl_init')) {
die('CURL is not installed!');
}
$ch = curl_init();
$thepage = get_web_page($url);
$html1 = $thepage['content'];
$ret = processhtml($html1, $url, $ret);
}
if (count($ret) < 3) {
$timeout = 15;
curl_setopt($ch, CURLOPT_URL, $url);
curl_setopt($ch, CURLOPT_ENCODING, "");
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, false);
curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false);
$html2 = curl_exec($ch);
curl_close($ch);
$ret = processhtml($html2, $url, $ret);
}
if (count($ret) < 3) {
$html4 = url_get_contents($url);
$ret = processhtml($html4, $url, $ret);
}
if (count($ret) < 3) {
$html5 = urlload($url);
$ret = processhtml($html5, $url, $ret);
}
//foreach ($ret as $key=>$value) {
// $img = explode("\n", $value);
// if (($img[1] == 99) && ($img[2] == 99)) {
//
// $thumbnail = imagecreatefromstring(file_get_contents($img[3]));
// $width = imagesx($thumbnail);
// $height = imagesy($thumbnail);
// ImageDestroy($thumbnail);
// if (trim($width) != "") {
// $value = sprintf('%07d', $height + $width) . "\n" . sprintf('%06d', $height) . "\n" . sprintf('%06d', $width) . "\n" . $img[3]
// }
// }
//}
//rsort($ret);
return $ret;
}
function grab_pics($params)
{
$cookie_path = "./cookies.txt";
$path = "allthumbs/";
$num = one(q("select count(*) from gallery\n\t\twhere approved = 0"));
$num > 0 or die("Data is empty.");
$i = min($num, 10000);
print "Found {$num} galleries.\nStart grab pics for next {$i} galleries...\n";
//$galleries=q("select * from gallery
// where approved = 0 order by random() limit $i");
//tmp
//$_counttmp=1;
$galleries = q("select * from gallery\n\t\twhere approved = 0 limit {$i}");
while ($row = fetch($galleries)) {
//tmp
//if($_counttmp <= 0) break;
print "\n{$row['url']}\n";
if (strpos($row['url'], ".flv") !== false) {
q("update gallery set approved=3\n\t\t\t\twhere id = %s", array($row["id"]));
continue;
}
!file_exists($cookie_path) or unlink($cookie_path);
$res = urlload($row["url"], array("cookies" => $cookie_path));
if ($res == "") {
q("update gallery set approved=4\n\t\t\t\twhere id = %s", array($row["id"]));
file_put_contents("errors.log", "{$row['id']}\tError load\t{$row['url']}\n", FILE_APPEND);
continue;
}
/*if(!preg_match_all(
'{<a[^<>]+href=([^<>\s]+)\s*[^<>]*>\s*<img[^<>]src=([^<>\s]+(jpeg|jpg)[^<>\s]*)\s*[^<>]*>\s*</a>}si',
$res, $m, PREG_SET_ORDER)){
file_put_contents("errors.log", "{$row['id']}\tNot found thumbs\t{$row['url']}\n", FILE_APPEND);
continue;}*/
$dom = new DOMDocument();
@$dom->loadHTML($res);
$m = array();
foreach ($dom->getElementsByTagName("a") as $node) {
$img = $node->getElementsByTagName("img")->item(0);
$img and $m[] = array(null, $node->getAttribute("href"), $img->getAttribute("src"));
}
//coollect image links
$urls = array();
while (sizeof($m) > 0 && sizeof($urls) < 5) {
$tmp = array_pop($m);
$tmp[1] = trim($tmp[1]);
$tmp[2] = trim($tmp[2]);
if ($tmp[1] == "" || $tmp[2] == "") {
continue;
}
if (!in_array(strtolower(p(pathinfo(p(parse_url($tmp[1]), "path")), "extension")), array("jpeg", "jpg", "wmv", "mpg", "mpeg", "avi", "mp4", "mov", "flv"))) {
continue;
}
$urls[] = $tmp;
}
if (sizeof($urls) == 0) {
q("update gallery set approved=5\n\t\t\t\twhere id = %s", array($row["id"]));
file_put_contents("errors.log", "{$row['id']}\tNot found valid thumbs\t{$row['url']}\n", FILE_APPEND);
continue;
}
$i = array_rand($urls);
$img_url = trim(unquotes($urls[$i][1]));
$thumb_url = trim(unquotes($urls[$i][2]));
$u = parse_url($img_url);
$p = pathinfo($u["path"]);
$is_pic_galler = strtolower($p["extension"]) == strtolower("jpg") || strtolower($p["extension"]) == strtolower("jpeg");
$link_url = $is_pic_galler ? $img_url : $thumb_url;
$img_url = realurl(baseurl($row['url']), $link_url);
print " {$img_url}\n";
list($res, $data) = urlload($img_url, array("getinfo" => true, "referer" => $row["url"], "cookies" => $cookie_path));
if ($res["http_code"] != "200") {
q("update gallery set approved=2\n\t\t\t\twhere id = %s", array($row["id"]));
file_put_contents("errors.log", "{$row['id']}\tHTTP Error - {$res['http_code']}\t{$row['url']}\n\t{$link_url}\n\t{$img_url}\n", FILE_APPEND);
} else {
if ($res["content_type"] == "image/jpeg") {
$thumb = (int) one(q("select num from thumb_num")) + 1;
$folder = $path . ceil($thumb / 1000);
q("update thumb_num set num={$thumb}");
if (!is_dir($folder)) {
mkdir($folder, 0777);
}
if (file_exists($name = "{$folder}/{$thumb}.jpg")) {
unlink($name);
}
file_put_contents($name, $data);
q("update gallery set thumb=%s, approved=1\n\t\t\t\twhere id = %s", array($thumb, $row["id"]));
file_put_contents("downloaded.log", "{$row['id']}\t{$row['url']}\n\t{$link_url}\n\t{$img_url}\n\t{$thumb}\n", FILE_APPEND);
} else {
q("update gallery set approved=2\n\t\t\t\twhere id = %s", array($row["id"]));
file_put_contents("errors.log", "{$row['id']}\tError image/video type\t{$row['url']}\n\t{$link_url}\n\t{$img_url}\n", FILE_APPEND);
}
}
}
}
