public static function getSrcSeedURL($btih)
{
global $USER_AGENT;
/// 1. 从数据库中查询原始页面链接
$res = get_by_btih($btih);
if (!$res) {
LOGW("BTIH 为 {$btih} 的资源在数据库中不存在");
return FALSE;
}
/// 2. 获取 link 页面内容
LOGI("正在获取动漫花园的资源页面内容: {$res['link']}");
$content = NULL;
$ch = curl_init($res['link']);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, TRUE);
curl_setopt($ch, CURLOPT_ENCODING, '');
curl_setopt($ch, CURLOPT_USERAGENT, $USER_AGENT);
$content = curl_exec($ch);
if (!$content) {
LOGE("无法抓取动漫花园的资源页面: {$res['link']}'");
return FALSE;
}
/// 3. 解析 BT 地址
$matches = [];
$pattern = '/\\/\\/.+[a-f0-9]{40}\\.torrent/';
$ret = preg_match($pattern, $content, $matches);
if ($ret >= 1) {
return 'http:' . $matches[0];
} else {
return FALSE;
}
}
/**
* 下载 RSS 内容
*/
protected function _fetchRss($url)
{
global $USER_AGENT;
LOGI("正在获取 {$url}");
$content = NULL;
$ch = curl_init($url);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, TRUE);
curl_setopt($ch, CURLOPT_ENCODING, '');
curl_setopt($ch, CURLOPT_USERAGENT, $USER_AGENT);
$content = curl_exec($ch);
if (!$content) {
LOGE("无法抓取 RSS：`{$RSS_FEED}'");
return FALSE;
}
/// 2. 归档原始数据
LOGI("正在归档数据");
/// FIXME: 将 archive_raw 做成类成员函数
archive_raw($content);
return $content;
}
$guid = $mysqli->real_escape_string($res['guid']);
$link = $mysqli->real_escape_string($res['link']);
$description = $mysqli->real_escape_string($res['description']);
$pubDate = $res['pubDate'];
$btih = $mysqli->real_escape_string($res['btih']);
$magnet = $mysqli->real_escape_string($res['magnet']);
$src = $mysqli->real_escape_string($src);
$ctime = time();
$mysqli->query('start transaction');
$sql = "SELECT * FROM b_resource WHERE btih='{$btih}' LIMIT 1";
$result = $mysqli->query($sql);
if (!$result) {
LOGE("数据库查询失败: " . $mysqli->error);
continue;
}
if ($result->num_rows > 0) {
LOGI("{$res['title']} 已存在 (因为已存在相同的 btih，btih={$btih}， src={$res['src']})");
$mysqli->query('rollback');
continue;
}
LOGI("保存来自 {$src} 的数据：{$res['title']}");
$sql = "INSERT INTO b_resource(title, guid, link, description, btih, pubDate, src, magnet, ctime)\n VALUES('{$title}', '{$guid}', '{$link}', '{$description}', '{$btih}', {$pubDate}, '{$src}', '{$magnet}', {$ctime})";
$ret = $mysqli->query($sql);
if ($ret === FALSE) {
LOGE("无法保存数据: " . $mysqli->error . "，原 SQL: " . $sql);
}
$mysqli->query('commit');
}
}
LOGI("索引完成");
if (!$result) {
LOGW("数据库查询出错，跳过这个资源：" . $mysqli->error);
continue;
}
$row = $result->fetch_assoc();
if ($row['cnt'] > 0) {
LOGI("数据库中已经存在相同 btih 的资源了，跳过这个资源");
continue;
}
/// 3. 将这个资源添加到数据库中
LOGI("将“{$title}”保存到数据库中");
$guid = $mysqli->real_escape_string($res['guid']);
$link = $mysqli->real_escape_string($res['link']);
$pubDate = (int) $res['pubDate'];
$description = '';
$ctime = time();
$sql = "INSERT INTO b_resource(title, guid, link, description, btih, pubDate, ctime)\n VALUES('{$title}', '{$guid}', '{$link}', '{$description}', '{$btih}', {$pubDate}, {$ctime})";
$ret = $mysqli->query($sql);
if ($ret === FALSE) {
LOGW("数据库查询出错：" . $mysqli->error);
} else {
$cnt_new++;
}
}
}
if ($DRY_RUN) {
LOGI("目前运行在测试模式，将回滚数据库");
$mysqli->rollback();
}
LOGI("资源索引完成，分析得到 {$cnt_count} 个资源，共新添加了 {$cnt_new} 个资源");
}
/// 2.2 开始下载
$content = NULL;
$ch = curl_init($url);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, TRUE);
curl_setopt($ch, CURLOPT_ENCODING, '');
curl_setopt($ch, CURLOPT_USERAGENT, $USER_AGENT);
curl_setopt($ch, CURLOPT_REFERER, '');
curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 1);
curl_setopt($ch, CURLOPT_MAXREDIRS, 5);
$content = curl_exec($ch);
if (!$content || curl_error($ch)) {
LOGD("无法从漫游下载种子（{$url}）:" . curl_error($ch));
header('HTTP/1.1 500 Internal Error');
die('<h1>500 Internal Error</h1>');
}
LOGI("保存种子“{$btih}”，并将用户跳转到下载地址");
$path = archive_torrent($content, $btih);
/// 3. 检查下载的内容是否是种子，漫游在发生错误的时候仍会返回 HTTP 200，所以我们需要通过 MIME 来进行检查
if (!$path || mime_content_type($path) != 'application/x-bittorrent') {
if ($path) {
unlink($path);
}
LOGD("从漫游下载到的种子不是合法的 （{$url}）:" . curl_error($ch));
header('HTTP/1.1 500 Internal Error');
die('<h1>500 Internal Error</h1>');
}
header("Content-Disposition: attachment; filename={$btih}.torrent");
header('Content-Type: application/x-bittorrent');
ob_clean();
echo $content;
/**
* 处理announce_peer请求
* @param array $msg 接收到的announce_peer请求数据
* @param array $address 对端链接信息
* @return void
*/
function on_announce_peer($msg, $address)
{
global $nid;
// 获取infohash
$infohash = $msg['a']['info_hash'];
// 获取token
$token = $msg['a']['token'];
// 获取node id
$id = $msg['a']['id'];
// 验证token是否正确
if (substr($infohash, 0, 2) == $token) {
/*$txt = array(
'action' => 'announce_peer',
'msg' => array(
'ip' => $address[0],
'port1' => $address[1],
'port2' => $msg['a']['port'],
'infohash' => $infohash
)
);
var_dump($txt);*/
$nodeid = bin2hex($id);
LOGI("(node_id={$nodeid}) 获取到info_hash: " . strtoupper(bin2hex($infohash)));
logDHTAnnouncePeer($nodeid, bin2hex($infohash));
}
// 生成回复数据
$msg = array('t' => $msg['t'], 'y' => 'r', 'r' => array('id' => $nid));
// 发送请求回复
send_response($msg, $address);
}
LOGI("已保存磁力链接: {$magnet}");
/// 3. 下载 BT 种子文件
$r = rand() % 20 + 10;
LOGI("等待 {$r} 秒后去下载种子文件");
sleep($r);
$content = NULL;
$ch = curl_init($res['guid']);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, TRUE);
curl_setopt($ch, CURLOPT_ENCODING, '');
curl_setopt($ch, CURLOPT_USERAGENT, $USER_AGENT);
curl_setopt($ch, CURLOPT_REFERER, $res['link']);
curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 1);
curl_setopt($ch, CURLOPT_MAXREDIRS, 5);
$content = curl_exec($ch);
if (!$content) {
LOGE("无法下载种子文件：`{$res['guid']}'");
die('');
}
$btih = $res['btih'];
if ($btih == '') {
$match = array();
preg_match('([0-9a-f]{40})', $res['link'], $match);
$btih = $match[0];
}
if ($btih == '') {
LOGE('无法获得种子文件的 BTIH，无法保存种子文件');
die('');
}
archive_torrent($content, $btih);
LOGI("“{$res['title']}”处理完毕");
/**
* 保存一个 announce_peer 数据，同时更新资源的 7 日下载数
*/
function logDHTAnnouncePeer($node_id, $btih)
{
/// 1. 检验数据合法性
$pattern = '/^[0-9a-f]{40}$/i';
if (!preg_match($pattern, $node_id)) {
LOGW("传入的 node_id({$node_id}) 格式不合法");
return FALSE;
}
if (!preg_match($pattern, $btih)) {
LOGW("传入的 btih({$btih}) 格式不合法");
return FALSE;
}
/// 2. 记录此次 announce_peer
$ctime = time();
$sql = "INSERT INTO b_dht_log (node_id, btih, ctime) VALUES (UNHEX('{$node_id}'), UNHEX('{$btih}'), {$ctime})";
$result = db_query($sql);
if (!$result) {
LOGE("数据库查询失败");
return FALSE;
}
/// 3. 更新资源热度（如果对应的 btih 存在）
$sql = "SELECT * FROM b_resource WHERE btih='{$btih}'";
$result = db_query($sql);
if ($result->num_rows <= 0) {
/// 资源不存在，无需更新
return TRUE;
} else {
$res = $result->fetch_assoc();
if ($res['popularity'] < 0) {
LOGI("初始化资源 {$btih} 的热度");
init_popularity($btih);
}
return update_popularity($res, $ctime);
}
}
<?php
/**
* popgo-gone 版本之后，b_resource 表中的 btih 字段变为 UNIQUE 索引，该脚本用于更新 btih 为空的行，以便修改 btih 索引为 UNIQUE 索引
*/
require_once '../header.php';
$sql = "SELECT * FROM b_resource WHERE src='popgo' AND btih=''";
$res = $mysqli->query($sql) or die($mysqli->error);
if (!$res) {
LOGI("已经没有需要更新的资源了");
exit(0);
}
LOGI(sprintf("需要更新 %d 个资源", $res->num_rows));
while ($row = $res->fetch_assoc()) {
$btih = popgo_get_btih_from_link($row['guid']);
$sql = "UPDATE b_resource SET btih='{$btih}' WHERE resource_id={$row['resource_id']}";
$mysqli->query($sql) or die($mysqli->error);
}