<?php
require_once 'functions_backend.php';
echo "<pre>Pool Cleaner(tm)\n\n";
#$meta = get_http_meta('http://news.bbc.co.uk/1/hi/technology/8349923.stm');
#$meta = get_http_meta('http://picasaweb.google.com/lh/photo/xr-C_wqHG8gssUZnzuv1Ow?feat=directlink');
#$meta = get_http_meta('http://bugs.kde.org/show_bug.cgi?id=162326');
#$meta = get_http_meta('http://lala.com/zPSy');
#$meta = get_http_meta('http://tinyurl.com/yjbub48');
echo "\n<hr>\n";
#$meta = get_http_meta('http://ur.ly/3kSj');
#echo "\n"; print_r($meta);
#echo "\n<hr>\n";
$meta = get_http_meta('http://thejesustv.com');
echo "\n";
print_r($meta);
exit;
function get_http_meta($url, $depth = 0, $status_chain = '', $url_chain = '')
{
//echo "get_http_meta('$url', $depth)\n";
// how deep to follow http redirects
if ($depth > 3) {
return;
}
$depth++;
$connect_timeout = 3;
$read_timeout = 6;
ini_set('default_socket_timeout', $connect_timeout);
$http_meta = get_empty_http_meta();
$purl = parse_url($url);
// note: this seems to fail for all addresses
// if a domain name, is it valid?
/*
if (!preg_match("/^[0-9]{3}\.[0-9]{3}\.[0-9]{3}\.[0-9]{3}$/", $purl['host'])) {
$ip = gethostbyname($purl['host']);
if ($ip == $purl['host']) {
echo "(name resolution failed) ";
$http_meta['state'] = 2;
return $http_meta;
}
}
*/
// connect and get http return code
if (array_key_exists('port', $purl)) {
$port = $purl['port'];
} else {
$port = 80;
}
$sock = fsockopen($purl['host'], $port, $errno, $errstr, $connect_timeout);
if (!$sock) {
echo "(connection error - {$errstr} ({$errno})) ";
$http_meta['state'] = 3;
return $http_meta;
}
stream_set_blocking($sock, FALSE);
stream_set_timeout($sock, $read_timeout);
$url_path = '/';
if (array_key_exists('path', $purl)) {
$url_path = $purl['path'];
if (array_key_exists('query', $purl)) {
$url_path .= '?' . $purl['query'];
}
}
// we set 'accept-language' as some sites will have different <title> depending on your geoip
// e.g. flickr
// some sites seem to ignore you if you don't send a user-agent
// e.g. digg (but maybe that's for the better)
fputs($sock, "GET {$url_path} HTTP/1.1\r\n");
fputs($sock, "Host: " . $purl['host'] . "\r\n");
fputs($sock, "Connection: close\r\n");
fputs($sock, "Accept-Language: en-gb, en;q=0.8\r\n");
fputs($sock, "User-Agent: URL catcher\r\n");
fputs($sock, "\r\n");
$status = socket_get_status($sock);
if ($status['timed_out']) {
echo "(status timed_out) ";
$http_meta['state'] = 4;
return $http_meta;
}
if (feof($sock)) {
echo "(socket early EOF) ";
$http_meta['state'] = 5;
return $http_meta;
}
$contents = stream_get_contents($sock, 1024 * 8);
if ($contents === FALSE) {
echo "stream_get_contents returned FALSE ";
return $http_meta;
}
fclose($sock);
$http_status = get_http_status($contents);
//echo "http_status=$http_status\n";
if (!is_numeric($http_status)) {
return $http_meta;
}
$http_meta['http_status'] = $http_status;
$http_meta['location'] = $url;
// handle different status codes
if ($http_status == "404") {
echo "(http 404) ";
$http_meta['state'] = 6;
} elseif (substr($http_status, 0, 1) == 4) {
echo "(http other 40x) ";
$http_meta['state'] = 7;
} elseif (substr($http_status, 0, 1) == 5) {
echo "(http other 50x) ";
$http_meta['state'] = 8;
} elseif ($http_status == "301" or $http_status == "302") {
// 301 - Moved Permanently
// 302 - Found / Moved
$new_url = get_http_location($contents);
if (!$new_url) {
$http_meta['state'] = 0;
return $http_meta;
}
// recurse recurse recurse
// discard our meta in favour of the deepest url
$http_meta = get_http_meta($new_url, $depth);
// set any meta we want to persist here
$http_meta['redirect'] = 1;
} elseif ($http_status == "200") {
$http_meta['state'] = 1;
$http_meta['content_length'] = get_http_content_length($contents);
$http_meta['content_type'] = get_http_content_type($contents);
$check_html = 0;
if ($http_meta['content_type']) {
$check_html = preg_match("/^text\\/html/", $http_meta['content_type']);
} else {
$check_html = !preg_match("/(jpg|jpeg|gif|png|wav|mp3|avi|wmv|mpg)\$/", $url);
}
if ($check_html) {
$http_meta['html_title'] = get_html_title($contents);
if (!$http_meta['html_title']) {
$http_meta['html_title'] = get_html_h1($contents);
}
} else {
//echo "(non text/html type) ";
}
// html meta redirect?
// this could be placed better to avoid doing all the work above
$html_meta_redirect = get_html_meta_redirect($contents);
if ($html_meta_redirect) {
$http_meta = get_http_meta($html_meta_redirect, $depth);
$http_meta['redirect'] = 1;
}
}
return $http_meta;
}