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;
}
$title = $mysqli->real_escape_string($res['title']);
$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");
$url = "";
$res = get_by_btih($btih);
if (!$res) {
LOGW("BTIH 为 {$btih} 的资源不存在");
header('HTTP/1.1 404 Not Found');
die('<h1>404 Not Found</h1> <h2>BTIH not exists</h2>');
}
switch ($res['src']) {
case 'popgo':
$url = Indexer_Popgo::getSrcSeedURL($btih);
break;
case 'dmhy':
$url = Indexer_DMHY::getSrcSeedURL($btih);
break;
default:
LOGE("代码不应执行到此处");
$url = FALSE;
break;
}
if (!$url) {
LOGW("无法获得 BTIH 为 {$btih} 的资源原始种子地址");
header('HTTP/1.1 404 Not Found');
die('<h1>404 Not Found</h1> <h2>Could not get source torrent URL</h2>');
}
/// 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, '');
$mysqli->query($sql);
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");
/**
* 进行一次 SQL 查询，如果连接断开，则自动重新尝试
*/
function db_query($sql)
{
global $mysqli;
global $DB_HOST;
global $DB_USER;
global $DB_PASSWORD;
global $DB_DATABASE;
$result = $mysqli->query($sql);
if (!$result) {
LOGN("第一次查询 MySQL 失败，重试后准备继续: " . $mysqli->error . ". SQL: {$sql}");
/// 重连 MySQL
/// FIXME: 连接 MySQL 应该独立成一个函数
$ret = $mysqli->real_connect($DB_HOST, $DB_USER, $DB_PASSWORD, $DB_DATABASE);
if (!$ret) {
LOGE('无法创建 MySQL 连接: ' . $mysqli->error);
return FALSE;
}
$mysqli->query("set NAMES 'utf8'");
$result = $mysqli->query($sql);
if (!$result) {
LOGW("第二次查询 MySQL 失败: " . $mysqli->error . ". SQL: {$sql}");
return FALSE;
}
}
return $result;
}
<?php
$__t1 = microtime(TRUE);
if (!file_exists(__DIR__ . '/config.php')) {
die("缺少 config.php 文件，可以尝试 mv config.sample.php config.php");
} else {
require_once __DIR__ . '/config.php';
}
/// 创建数据库连接
$mysqli = new Mysqli($DB_HOST, $DB_USER, $DB_PASSWORD, $DB_DATABASE);
if (!$mysqli) {
LOGE('无法创建 MySQL 连接: ' . $mysqli->error);
die;
}
$mysqli->query("set NAMES 'utf8'");
$CSP_NONCE = uniqid() . 'X' . rand();
require_once __DIR__ . '/func.php';
