525 lines
12 KiB
PHP
525 lines
12 KiB
PHP
|
<?php
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
$id = str_pad($_GET['id'], 7, '0', STR_PAD_LEFT);
|
||
|
|
$size = $_GET['size'];
|
||
|
|
if ($size != 'large')
|
||
|
|
$size = 'small';
|
||
|
|
|
||
|
|
$imagetype = array('gif', 'jpeg', 'jpg', 'png');
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
function downloadPosters() {
|
||
|
|
|
||
|
|
global $imagetype;
|
||
|
|
|
||
|
|
set_time_limit(0);
|
||
|
|
|
||
|
|
echo '<tt>';
|
||
|
|
|
||
|
|
// getCriterionPosters();
|
||
|
|
|
||
|
|
$html = file_get_contents('html/0xdb.html');
|
||
|
|
$string = explode('id="iconPoster', $html);
|
||
|
|
for ($i = 1; $i < count($string); $i++) {
|
||
|
|
$substring = explode('"', $string[$i]);
|
||
|
|
$id = $substring[0];
|
||
|
|
if (strlen($id) == 7 || strlen($id) == 34) {
|
||
|
|
|
||
|
|
$file_exists = false;
|
||
|
|
foreach ($imagetype as $value) {
|
||
|
|
if (file_exists('posters/wikipedia.org/' . substr($id, 0, 4) . '/' . $id . '.' . $value)) {
|
||
|
|
$file_exists = true;
|
||
|
|
break;
|
||
|
|
}
|
||
|
|
if (file_exists('posters/wikipedia.org/' . substr($id, 0, 4) . '/_' . $id . '.' . $value)) {
|
||
|
|
$file_exists = true;
|
||
|
|
break;
|
||
|
|
}
|
||
|
|
}
|
||
|
|
if (!$file_exists) {
|
||
|
|
$file = getWikipedia($id);
|
||
|
|
if ($file)
|
||
|
|
echo 'writing ' . $id . '<br />';
|
||
|
|
}
|
||
|
|
else
|
||
|
|
echo 'reading ' . $id . '<br />' . str_repeat(' ', 10000);
|
||
|
|
|
||
|
|
$file_exists = false;
|
||
|
|
foreach ($imagetype as $value) {
|
||
|
|
if (file_exists('posters/imdb.com/' . substr($id, 0, 4) . '/' . $id . '.' . $value)) {
|
||
|
|
$file_exists = true;
|
||
|
|
break;
|
||
|
|
}
|
||
|
|
}
|
||
|
|
if (!$file_exists)
|
||
|
|
getImdb($id);
|
||
|
|
|
||
|
|
if (!file_exists('posters/0xdb.org/' . substr($id, 0, 4) . '/' . $id . '.jpg'))
|
||
|
|
get0xdb($id);
|
||
|
|
|
||
|
|
flush(); ob_flush();
|
||
|
|
|
||
|
|
}
|
||
|
|
}
|
||
|
|
|
||
|
|
// getImpawardsIndex();
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
function returnPoster($id, $size) {
|
||
|
|
|
||
|
|
if ($size == 'small')
|
||
|
|
$source = array('piratecinema', 'karagarga', 'criterion', 'wikipedia', 'impawards', 'imdb', '0xdb');
|
||
|
|
else
|
||
|
|
$source = array('piratecinema', 'karagarga', 'criterion', 'wikipedia', 'impawards');
|
||
|
|
|
||
|
|
foreach ($source as $value) {
|
||
|
|
$function = get . ucfirst($value);
|
||
|
|
$file[$value] = $function($id);
|
||
|
|
// echo $value . time() . ' ';
|
||
|
|
}
|
||
|
|
|
||
|
|
foreach ($source as $value) {
|
||
|
|
if ($file[$value])
|
||
|
|
returnImage($id, $size, $file[$value]);
|
||
|
|
}
|
||
|
|
|
||
|
|
// returnImage($id, $size, 'png/poster' . ucfirst($size) . '.png');
|
||
|
|
if ($size == 'small')
|
||
|
|
returnImage($id, 'small', 'posters/0xdb.org/' . substr($id, 0, 4) . '/' . $id . '.png');
|
||
|
|
else
|
||
|
|
returnImage($id, 'large', 'png/posterLarge.png');
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
function getTitle($id) {
|
||
|
|
|
||
|
|
$html = getUrl('http://www.imdb.com/title/tt' . $id . '/');
|
||
|
|
|
||
|
|
$string = explode('<title>', $html);
|
||
|
|
$string = explode('</title>', $string[1]);
|
||
|
|
$string = explode(' (', $string[0]);
|
||
|
|
$title = html_entity_decode($string[0]);
|
||
|
|
|
||
|
|
return $title;
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
function getImdb($id) {
|
||
|
|
|
||
|
|
global $imagetype;
|
||
|
|
|
||
|
|
foreach ($imagetype as $value) {
|
||
|
|
$file = 'posters/imdb.com/' . substr($id, 0, 4) . '/' . $id . '.' . $value;
|
||
|
|
if (file_exists($file))
|
||
|
|
return $file;
|
||
|
|
}
|
||
|
|
|
||
|
|
$html = getUrl('http://www.imdb.com/title/tt' . $id . '/');
|
||
|
|
|
||
|
|
$string = explode('<a name="poster"', $html);
|
||
|
|
$string = explode('src="', $string[1]);
|
||
|
|
$string = explode('"', $string[1]);
|
||
|
|
$url = $string[0];
|
||
|
|
|
||
|
|
if ($url && $url != 'http://ia.imdb.com/media/imdb/01/I/44/47/29.gif') {
|
||
|
|
$data = getUrl($url);
|
||
|
|
$pathinfo = pathinfo($url);
|
||
|
|
$extension = strtolower($pathinfo['extension']);
|
||
|
|
$file = 'posters/imdb.com/' . substr($id, 0, 4) . '/' . $id . '.' . $extension;
|
||
|
|
// die($url . ' ' . $file);
|
||
|
|
writeFile($file, $data);
|
||
|
|
return $file;
|
||
|
|
}
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
function get0xdb($id) {
|
||
|
|
|
||
|
|
$file = 'posters/0xdb.org/' . substr($id, 0, 4) . '/' . $id . '.png';
|
||
|
|
/*
|
||
|
|
if (file_exists($file))
|
||
|
|
return $file;
|
||
|
|
*/
|
||
|
|
// the following is used to remove broken images from the db
|
||
|
|
if (file_exists($file)) {
|
||
|
|
$image = @imagecreatefrompng($file);
|
||
|
|
if ($image)
|
||
|
|
return $file;
|
||
|
|
else
|
||
|
|
unlink($file);
|
||
|
|
}
|
||
|
|
|
||
|
|
$data = file_get_contents('http://rlx:luxor@services.0xdb.org/posters/0xdbPoster.php?id=' . $id);
|
||
|
|
writeFile($file, $data);
|
||
|
|
|
||
|
|
$image = @imagecreatefrompng($file);
|
||
|
|
if ($image)
|
||
|
|
return $file;
|
||
|
|
else {
|
||
|
|
unlink($file);
|
||
|
|
return 'png/posterSmall.png';
|
||
|
|
}
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
function getWikipedia($id) {
|
||
|
|
|
||
|
|
global $imagetype;
|
||
|
|
|
||
|
|
foreach ($imagetype as $value) {
|
||
|
|
$file = 'posters/wikipedia.org/' . substr($id, 0, 4) . '/' . $id . '.' . $value;
|
||
|
|
if (file_exists($file))
|
||
|
|
return $file;
|
||
|
|
$file = 'posters/wikipedia.org/' . substr($id, 0, 4) . '/_' . $id . '.' . $value;
|
||
|
|
if (file_exists($file))
|
||
|
|
return false;
|
||
|
|
}
|
||
|
|
|
||
|
|
$title = getTitle($id);
|
||
|
|
|
||
|
|
$url = 'http://www.google.com/search?q=' . urlencode('site:en.wikipedia.org "' . $title . '"');
|
||
|
|
$html = getUrl($url);
|
||
|
|
|
||
|
|
$string = explode('" class=l>', $html);
|
||
|
|
for ($i = 0; $i < count($string) - 1; $i++) {
|
||
|
|
$substring = explode('<a href="', $string[$i]);
|
||
|
|
$url = $substring[count($substring) - 1];
|
||
|
|
$html = getUrl($url);
|
||
|
|
if (strstr($html, 'imdb.com/title/tt' . $id)) {
|
||
|
|
// echo "tt ok<br>";
|
||
|
|
if (strstr($html, '"/wiki/Image:')) {
|
||
|
|
// echo "image ok<br>";
|
||
|
|
$string = explode('"/wiki/Image:', $html);
|
||
|
|
$string = explode('"', $string[1]);
|
||
|
|
$url = 'http://en.wikipedia.org/wiki/Image:' . $string[0];
|
||
|
|
$html = getUrl($url);
|
||
|
|
$string = explode('src="http://upload.wikimedia.org/', $html);
|
||
|
|
$string = explode('"', $string[1]);
|
||
|
|
$url = 'http://upload.wikimedia.org/' . $string[0];
|
||
|
|
$pathinfo = pathinfo($url);
|
||
|
|
$extension = strtolower($pathinfo['extension']);
|
||
|
|
$imagesize = getimagesize($url);
|
||
|
|
if ($imagesize[0] > 128 || $imagesize[1] > 128) {
|
||
|
|
$file = 'posters/wikipedia.org/' . substr($id, 0, 4) . '/' . $id . '.' . $extension;
|
||
|
|
$data = getUrl($url);
|
||
|
|
writeFile($file, $data);
|
||
|
|
// echo "$file $title\n";
|
||
|
|
return $file;
|
||
|
|
}
|
||
|
|
}
|
||
|
|
break;
|
||
|
|
}
|
||
|
|
}
|
||
|
|
|
||
|
|
return false;
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
function getImpawards($id) {
|
||
|
|
|
||
|
|
global $imagetype;
|
||
|
|
|
||
|
|
foreach ($imagetype as $value) {
|
||
|
|
$file = 'posters/impawards.com/' . substr($id, 0, 4) . '/' . $id . '.' . $value;
|
||
|
|
if (file_exists($file))
|
||
|
|
return $file;
|
||
|
|
}
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
function getCriterion($id) {
|
||
|
|
|
||
|
|
global $imagetype;
|
||
|
|
|
||
|
|
foreach ($imagetype as $value) {
|
||
|
|
$file = 'posters/criterion.com/' . substr($id, 0, 4) . '/' . $id . '.' . $value;
|
||
|
|
if (file_exists($file))
|
||
|
|
return $file;
|
||
|
|
}
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
function getKaragarga($id) {
|
||
|
|
|
||
|
|
global $imagetype;
|
||
|
|
|
||
|
|
foreach ($imagetype as $value) {
|
||
|
|
$file = 'posters/karagarga.net/' . substr($id, 0, 4) . '/' . $id . '.' . $value;
|
||
|
|
if (file_exists($file))
|
||
|
|
return $file;
|
||
|
|
}
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
function getPiratecinema($id) {
|
||
|
|
|
||
|
|
global $imagetype;
|
||
|
|
|
||
|
|
foreach ($imagetype as $value) {
|
||
|
|
$file = 'posters/piratecinema.org/' . substr($id, 0, 4) . '/' . $id . '.' . $value;
|
||
|
|
if (file_exists($file))
|
||
|
|
return $file;
|
||
|
|
}
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
function returnImage($id, $size, $file) {
|
||
|
|
|
||
|
|
$pathinfo = pathinfo($file);
|
||
|
|
$extension = $pathinfo['extension'];
|
||
|
|
|
||
|
|
if ($extension == 'gif')
|
||
|
|
$source = imagecreatefromgif($file);
|
||
|
|
else if ($extension == 'png')
|
||
|
|
$source = imagecreatefrompng($file);
|
||
|
|
else
|
||
|
|
$source = imagecreatefromjpeg($file);
|
||
|
|
|
||
|
|
if ($size == 'small')
|
||
|
|
$size = 128;
|
||
|
|
else
|
||
|
|
$size = 512;
|
||
|
|
|
||
|
|
$imagesize = getimagesize($file);
|
||
|
|
if ($imagesize[0] > $imagesize[1]) {
|
||
|
|
$width = $size;
|
||
|
|
$height = round($width * $imagesize[1] / $imagesize[0]);
|
||
|
|
}
|
||
|
|
else {
|
||
|
|
$height = $size;
|
||
|
|
$width = round($height * $imagesize[0] / $imagesize[1]);
|
||
|
|
}
|
||
|
|
|
||
|
|
$target = imagecreatetruecolor($width, $height);
|
||
|
|
imagecopyresampled($target, $source, 0, 0, 0, 0, $width, $height, $imagesize[0], $imagesize[1]);
|
||
|
|
|
||
|
|
if (strstr($file, 'png/poster')) {
|
||
|
|
header('Content-Type: image/png');
|
||
|
|
header('Content-Disposition: inline; filename="' . $id . '.png"');
|
||
|
|
imagepng($target);
|
||
|
|
}
|
||
|
|
else {
|
||
|
|
header('Content-Type: image/jpeg');
|
||
|
|
header('Content-Disposition: inline; filename="' . $id . '.jpg"');
|
||
|
|
imagejpeg($target, NULL, 100);
|
||
|
|
}
|
||
|
|
|
||
|
|
exit;
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
|
||
|
|
function getUrl($url) {
|
||
|
|
|
||
|
|
$url = parse_url($url);
|
||
|
|
$host = $url['host'];
|
||
|
|
$file = $url['path'] . str_repeat('?', $url['query'] != '') . $url['query'];
|
||
|
|
// echo $host . ' ' . $file . '<br />';
|
||
|
|
|
||
|
|
$headers = "GET $file HTTP/1.1\r\n";
|
||
|
|
$headers .= "Accept: */*\r\n";
|
||
|
|
$headers .= "Accept-Language: en\r\n";
|
||
|
|
// $headers .= "Accept-Encoding: gzip, deflate\r\n";
|
||
|
|
$headers .= "User-Agent: Mozilla/5.0 (Macintosh; U; PPC Mac OS X; en) AppleWebKit/418 (KHTML, like Gecko) Safari/417.9.3\r\n";
|
||
|
|
$headers .= "Connection: close\r\n";
|
||
|
|
$headers .= "Host: $host\r\n\r\n";
|
||
|
|
|
||
|
|
$f = @fsockopen($host, 80, $errorNumber, $errorString, 60);
|
||
|
|
if ($f) {
|
||
|
|
fwrite($f, $headers);
|
||
|
|
while (!feof($f))
|
||
|
|
$data .= fread($f, 1024);
|
||
|
|
fclose($f);
|
||
|
|
$data = substr($data, strpos($data, "\r\n\r\n") + 4);
|
||
|
|
}
|
||
|
|
|
||
|
|
return $data;
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
function writeFile($file, $data) {
|
||
|
|
|
||
|
|
$pathinfo = pathinfo($file);
|
||
|
|
$dirname = $pathinfo['dirname'];
|
||
|
|
$filename = $dirname . '/' . $pathinfo['basename'];
|
||
|
|
|
||
|
|
$string = explode('/', $dirname);
|
||
|
|
for ($i = 0; $i < count($string); $i++) {
|
||
|
|
if ($i > 0)
|
||
|
|
$subdirname .= '/';
|
||
|
|
$subdirname .= $string[$i];
|
||
|
|
if (!file_exists($subdirname))
|
||
|
|
mkdir($subdirname, 0777);
|
||
|
|
}
|
||
|
|
|
||
|
|
@file_put_contents($filename, $data);
|
||
|
|
@chmod($filename, 0666);
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
function getCriterionPosters() {
|
||
|
|
|
||
|
|
$html = getUrl('http://www.criterion.com/asp/browse.asp?sort=spine');
|
||
|
|
|
||
|
|
$string = explode('height="20" valign="top" class="regtext"><a href="release.asp?id=', $html);
|
||
|
|
for ($i = 1; $i , count($string); $i++) {
|
||
|
|
|
||
|
|
$substring = explode('class="browse">', $string[$i]);
|
||
|
|
$substring = explode('<', $substring[1]);
|
||
|
|
$title = $substring[0];
|
||
|
|
$substring = explode('class="regtext">', $string[$i]);
|
||
|
|
$substring = explode('<', $substring[1]);
|
||
|
|
$director = $substring[0];
|
||
|
|
$substring = explode('class="regtext">', $string[$i]);
|
||
|
|
$substring = explode('<', $substring[4]);
|
||
|
|
$spine = $substring[0];
|
||
|
|
|
||
|
|
$url = 'http://www.google.com/search?q=' . urlencode('site:imdb.com "' . $title . '" ' . $director);
|
||
|
|
$html = getUrl($url);
|
||
|
|
|
||
|
|
$id = 0;
|
||
|
|
$string_ = explode('" class=l>', $html);
|
||
|
|
for ($i_ = 0; $i_ < count($string_) - 1; $i_++) {
|
||
|
|
$substring = explode('<a href="', $string_[$i_]);
|
||
|
|
$url = $substring[count($substring) - 1];
|
||
|
|
if (strstr($url, 'title/tt')) {
|
||
|
|
$substring = explode('title/tt', $url);
|
||
|
|
$id = substr($substring[1], 0, 7);
|
||
|
|
break;
|
||
|
|
}
|
||
|
|
}
|
||
|
|
|
||
|
|
if ($id) {
|
||
|
|
echo $title . ' (' . $director . ') = ' . getTitle($id) . '<br />' . str_repeat(' ', 10000);
|
||
|
|
$file = 'posters/criterion.com/' . substr($id, 0, 4) . '/' . $id . '.jpg';
|
||
|
|
if (!file_exists($file)) {
|
||
|
|
$data = getUrl('http://www.criterion.com/content/images/full_boxshot/' . $spine . '_box_348x490.jpg');
|
||
|
|
writeFile($file, $data);
|
||
|
|
}
|
||
|
|
}
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
function getImpawardsIndex($p = 1) {
|
||
|
|
|
||
|
|
global $url;
|
||
|
|
|
||
|
|
$html = getUrl('http://www.impawards.com/alpha' . $p . '.html');
|
||
|
|
|
||
|
|
// echo $html;
|
||
|
|
|
||
|
|
for ($y = 19; $y <= 20; $y++) {
|
||
|
|
$string = explode('href = ' . $y, $html);
|
||
|
|
for ($i = 1; $i < count($string); $i++) {
|
||
|
|
$substring = explode('>', $string[$i]);
|
||
|
|
if (strstr($substring[0], '_ver1') || !strstr($substring[0], '_ver')) {
|
||
|
|
$key = count($url);
|
||
|
|
$url[$key] = '/' . $y . $substring[0];
|
||
|
|
getImpawardsPage($url[$key]);
|
||
|
|
}
|
||
|
|
}
|
||
|
|
}
|
||
|
|
|
||
|
|
if (strstr($html, 'href = alpha' . ($p + 1)))
|
||
|
|
getImpawardsIndex($p + 1);
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
function getImpawardsPage($url) {
|
||
|
|
|
||
|
|
$html = getUrl('http://www.impawards.com' . $url);
|
||
|
|
|
||
|
|
$string = explode('imdb.com/title/tt', $html);
|
||
|
|
$string = explode(' ', $string[1]);
|
||
|
|
$id = str_pad($string[0], 7, '0', STR_PAD_LEFT);
|
||
|
|
|
||
|
|
if ($id > 0) {
|
||
|
|
|
||
|
|
$dirname = 'posters/impawards.com/' . substr($id, 0, 4);
|
||
|
|
$filename = $dirname . '/' . $id . '.jpg';
|
||
|
|
|
||
|
|
if (strstr($html, '_xlg.html'))
|
||
|
|
$poster = 'http://www.impawards.com' . substr($url, 0, 5) . '/posters' . substr($url, 5, -5) . '_xlg.jpg';
|
||
|
|
else
|
||
|
|
$poster = 'http://www.impawards.com' . substr($url, 0, 5) . '/posters' . substr($url, 5, -5) . '.jpg';
|
||
|
|
|
||
|
|
if (!file_exists($filename)) {
|
||
|
|
|
||
|
|
if (!file_exists($dirname))
|
||
|
|
mkdir($dirname, 0777);
|
||
|
|
|
||
|
|
$jpg = file_get_contents($poster);
|
||
|
|
$f = fopen($filename, 'w');
|
||
|
|
fwrite($f, $jpg);
|
||
|
|
fclose($f);
|
||
|
|
chmod($filename, 0666);
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
echo "$id $poster<br />\n";
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
function crawlPath($path) {
|
||
|
|
|
||
|
|
global $list;
|
||
|
|
|
||
|
|
if (substr($path, -1) == '/')
|
||
|
|
$path = substr($path, 0, -1);
|
||
|
|
|
||
|
|
$dir = dir($path);
|
||
|
|
while ($file = $dir->read()) {
|
||
|
|
if (substr($file, 0, 1) != '.') {
|
||
|
|
$file = $path . '/' . $file;
|
||
|
|
if (is_dir($file))
|
||
|
|
crawlPath($file);
|
||
|
|
else
|
||
|
|
$list[count($list)] = $file;
|
||
|
|
}
|
||
|
|
}
|
||
|
|
|
||
|
|
return $list;
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
?>
|