5080 lines
205 KiB
PHP
5080 lines
205 KiB
PHP
<?php
|
||
if (!defined('__TYPECHO_ROOT_DIR__')) exit;
|
||
|
||
/**
|
||
* 网址导航、RSS信息阅读
|
||
*
|
||
* @package UrlNav
|
||
* @author 石头厝
|
||
* @version 1.5.0
|
||
* @link https://www.shitoucuo.com
|
||
*/
|
||
class UrlNav_Plugin implements Typecho_Plugin_Interface
|
||
{
|
||
/**
|
||
* 数据库文件路径
|
||
*/
|
||
private static $dbPath;
|
||
|
||
/**
|
||
* RSS刷新管理器实例
|
||
*/
|
||
private static $rssManager = null;
|
||
|
||
/**
|
||
* 激活插件方法
|
||
*/
|
||
public static function activate()
|
||
{
|
||
// 初始化数据库
|
||
self::initDbPath();
|
||
self::initDatabase();
|
||
self::migrateDatabase();
|
||
|
||
// 添加后台管理菜单
|
||
Helper::addPanel(3, 'UrlNav/Manage.php', '网址导航', '网址导航管理', 'administrator');
|
||
Helper::addPanel(3, 'UrlNav/Rss.php', 'RSS信息', 'RSS信息阅读', 'administrator');
|
||
|
||
// 添加动作处理
|
||
Helper::addAction('urlnav', 'UrlNav_Action');
|
||
|
||
// 注册路由 - 使用与Collection相同的格式
|
||
Helper::addRoute('urlnav_action', '/action/urlnav', 'UrlNav_Action', 'action');
|
||
|
||
// 注册独立的定时任务路由（网站状态检查）
|
||
Helper::addRoute('urlnav_status_cron', '/urlnav-status-cron', 'UrlNav_Action', 'statusCron');
|
||
|
||
// 注册独立的定时任务路由（RSS刷新）
|
||
Helper::addRoute('urlnav_rss_cron', '/urlnav-rss-cron', 'UrlNav_Action', 'rssCron');
|
||
|
||
return _t('网址导航插件已激活');
|
||
}
|
||
|
||
/**
|
||
* 禁用插件方法
|
||
*/
|
||
public static function deactivate()
|
||
{
|
||
// 移除管理菜单
|
||
Helper::removePanel(3, 'UrlNav/Manage.php');
|
||
Helper::removePanel(3, 'UrlNav/Rss.php');
|
||
|
||
Helper::removeAction('urlnav');
|
||
Helper::removeRoute('urlnav_action');
|
||
Helper::removeRoute('urlnav_status_cron');
|
||
Helper::removeRoute('urlnav_rss_cron');
|
||
|
||
return _t('网址导航插件已禁用');
|
||
}
|
||
|
||
/**
|
||
* 获取插件配置面板
|
||
*/
|
||
public static function config(Typecho_Widget_Helper_Form $form)
|
||
{
|
||
// ================== 网址管理配置 ==================
|
||
$form->addItem(new Typecho_Widget_Helper_Layout('div', array('class' => 'typecho-page-title')), '<h2>网址管理配置</h2>');
|
||
|
||
// 每页显示数量
|
||
$pageSize = new Typecho_Widget_Helper_Form_Element_Text('pageSize', null, '20',
|
||
_t('每页显示数量'), _t('后台管理中每页显示的网址数量'));
|
||
$pageSize->input->setAttribute('class', 'mini');
|
||
$form->addInput($pageSize);
|
||
|
||
// 是否开启网址验证
|
||
$validateUrl = new Typecho_Widget_Helper_Form_Element_Radio('validateUrl', array(
|
||
'1' => _t('开启'),
|
||
'0' => _t('关闭')
|
||
), '1', _t('网址验证'), _t('新增网址时是否验证网址有效性'));
|
||
$form->addInput($validateUrl);
|
||
|
||
// ================== 全文抓取配置 ==================
|
||
$form->addItem(new Typecho_Widget_Helper_Layout('div', array('class' => 'typecho-page-title')), '<h2>全文抓取配置</h2>');
|
||
|
||
// 是否开启全文抓取
|
||
$enableFullText = new Typecho_Widget_Helper_Form_Element_Radio('enableFullText', array(
|
||
'1' => _t('开启'),
|
||
'0' => _t('关闭')
|
||
), '0', _t('开启全文抓取'), _t('开启后会对白名单中的网站自动抓取全文'));
|
||
$form->addInput($enableFullText);
|
||
|
||
// 白名单配置（多行文本）
|
||
$fullTextWhitelist = new Typecho_Widget_Helper_Form_Element_Textarea('fullTextWhitelist', null,
|
||
"https://wiki.eryajf.net/learning-weekly.xml|.markdown-body\nhttps://example.com/rss|#content",
|
||
_t('全文抓取白名单'),
|
||
_t('每行一个，格式：RSS地址|内容选择器（CSS选择器）<br>示例：https://wiki.eryajf.net/learning-weekly.xml|.post-content'));
|
||
$form->addInput($fullTextWhitelist);
|
||
|
||
// 每个站点抓取全文的篇数
|
||
$fullTextPerSite = new Typecho_Widget_Helper_Form_Element_Text('fullTextPerSite', null, '3',
|
||
_t('每站抓取全文篇数'), _t('每个RSS源最多抓取几篇的全文，建议1-5'));
|
||
$fullTextPerSite->input->setAttribute('class', 'mini');
|
||
$form->addInput($fullTextPerSite);
|
||
|
||
// 页面抓取超时时间（单篇文章）
|
||
$pageFetchTimeout = new Typecho_Widget_Helper_Form_Element_Text('pageFetchTimeout', null, '8',
|
||
_t('页面抓取超时时间（秒）'), _t('抓取单篇文章页面时的超时时间，建议8-15秒'));
|
||
$pageFetchTimeout->input->setAttribute('class', 'mini');
|
||
$form->addInput($pageFetchTimeout);
|
||
|
||
// ================== RSS配置 ==================
|
||
$form->addItem(new Typecho_Widget_Helper_Layout('div', array('class' => 'typecho-page-title')), '<h2>RSS配置</h2>');
|
||
|
||
// RSS页面每页显示数量
|
||
$rssPageSize = new Typecho_Widget_Helper_Form_Element_Text('rssPageSize', null, '30',
|
||
_t('RSS页面每页显示数量'), _t('RSS信息页面每页显示的文章数量'));
|
||
$rssPageSize->input->setAttribute('class', 'mini');
|
||
$form->addInput($rssPageSize);
|
||
|
||
// RSS刷新间隔
|
||
$rssRefresh = new Typecho_Widget_Helper_Form_Element_Text('rssRefresh', null, '3600',
|
||
_t('RSS刷新间隔（秒）'), _t('建议的RSS刷新间隔时间，实际执行时间由宝塔计划任务决定'));
|
||
$rssRefresh->input->setAttribute('class', 'mini');
|
||
$form->addInput($rssRefresh);
|
||
|
||
// 【新增】每次自动刷新网址数量
|
||
$rssRefreshLimit = new Typecho_Widget_Helper_Form_Element_Text('rssRefreshLimit', null, '20',
|
||
_t('每次自动刷新网址数量'), _t('每次定时任务最多刷新的RSS网址数量，建议10-50，根据服务器性能调整'));
|
||
$rssRefreshLimit->input->setAttribute('class', 'mini');
|
||
$form->addInput($rssRefreshLimit);
|
||
|
||
// 每个站点最大文章数
|
||
$maxFeedsPerSite = new Typecho_Widget_Helper_Form_Element_Text('maxFeedsPerSite', null, '5',
|
||
_t('每个站点最大文章数'), _t('每个RSS源最多显示的文章数量'));
|
||
$maxFeedsPerSite->input->setAttribute('class', 'mini');
|
||
$form->addInput($maxFeedsPerSite);
|
||
|
||
// RSS文章保留时间（改为下拉框）
|
||
$rssKeepTime = new Typecho_Widget_Helper_Form_Element_Select('rssKeepTime',
|
||
array(
|
||
'0' => _t('不自动清理（默认）'), // ← 将"默认"标识放在这里
|
||
'86400' => _t('一天之前（24小时前）'),
|
||
'259200' => _t('三天之前（72小时前）'),
|
||
'604800' => _t('一周之前（7天前）'),
|
||
'1296000' => _t('半个月之前（15天前）'),
|
||
'2592000' => _t('一个月之前（30天前）'),
|
||
'7776000' => _t('三个月之前（90天前）'),
|
||
'15552000' => _t('半年之前（180天前）')
|
||
),
|
||
'259200', // ← 这里改为 0，默认不清理
|
||
_t('RSS文章保留时间'),
|
||
_t('自动清理超过此时间的RSS文章，按照文章发布时间判断，默认不自动清理'));
|
||
$form->addInput($rssKeepTime);
|
||
|
||
// RSS最大缓存条数
|
||
$maxCachePerSite = new Typecho_Widget_Helper_Form_Element_Text('maxCachePerSite', null, '5',
|
||
_t('每个站点最大缓存条数'), _t('每个RSS源最多缓存的文章数量，0表示不限制'));
|
||
$maxCachePerSite->input->setAttribute('class', 'mini');
|
||
$form->addInput($maxCachePerSite);
|
||
|
||
// 连接超时时间
|
||
$fetchTimeout = new Typecho_Widget_Helper_Form_Element_Text('fetchTimeout', null, '5',
|
||
_t('RSS抓取超时时间（秒）'), _t('抓取RSS源时的超时时间'));
|
||
$fetchTimeout->input->setAttribute('class', 'mini');
|
||
$form->addInput($fetchTimeout);
|
||
|
||
// 失败重试次数
|
||
$retryTimes = new Typecho_Widget_Helper_Form_Element_Text('retryTimes', null, '2',
|
||
_t('失败重试次数'), _t('RSS抓取失败时的重试次数'));
|
||
$retryTimes->input->setAttribute('class', 'mini');
|
||
$form->addInput($retryTimes);
|
||
|
||
// ================== 网站状态检查配置 ==================
|
||
$form->addItem(new Typecho_Widget_Helper_Layout('div', array('class' => 'typecho-page-title')), '<h2>网站状态检查配置</h2>');
|
||
|
||
// 状态检查超时时间
|
||
$statusCheckTimeout = new Typecho_Widget_Helper_Form_Element_Text('statusCheckTimeout', null, '8',
|
||
_t('状态检查超时时间（秒）'), _t('检查网站状态时的超时时间'));
|
||
$statusCheckTimeout->input->setAttribute('class', 'mini');
|
||
$form->addInput($statusCheckTimeout);
|
||
|
||
// 每次检查的最大数量
|
||
$statusCheckMax = new Typecho_Widget_Helper_Form_Element_Text('statusCheckMax', null, '80',
|
||
_t('每次检查最大数量'), _t('每次自动检查时最多检查的网址数量'));
|
||
$statusCheckMax->input->setAttribute('class', 'mini');
|
||
$form->addInput($statusCheckMax);
|
||
|
||
// ================== 定时任务配置 ==================
|
||
$form->addItem(new Typecho_Widget_Helper_Layout('div', array('class' => 'typecho-page-title')), '<h2>定时任务配置</h2>');
|
||
|
||
// RSS定时任务访问密钥
|
||
$rssCronSecret = new Typecho_Widget_Helper_Form_Element_Text('rssCronSecret', null, self::generateSecret(),
|
||
_t('RSS定时任务密钥'), _t('用于RSS定时任务访问的密钥，请妥善保管'));
|
||
$form->addInput($rssCronSecret);
|
||
|
||
// 状态检查定时任务访问密钥
|
||
$statusCronSecret = new Typecho_Widget_Helper_Form_Element_Text('statusCronSecret', null, self::generateSecret(),
|
||
_t('状态检查定时任务密钥'), _t('用于状态检查定时任务访问的密钥，请妥善保管'));
|
||
$form->addInput($statusCronSecret);
|
||
}
|
||
|
||
/**
|
||
* 个人用户的配置面板
|
||
*/
|
||
public static function personalConfig(Typecho_Widget_Helper_Form $form) {}
|
||
|
||
/**
|
||
* 初始化数据库路径
|
||
*/
|
||
private static function initDbPath()
|
||
{
|
||
$dbDir = __DIR__ . '/db';
|
||
|
||
// 确保目录存在
|
||
if (!is_dir($dbDir)) {
|
||
@mkdir($dbDir, 0755, true);
|
||
}
|
||
|
||
$dbFiles = glob($dbDir . '/urlnav_*.db');
|
||
if (!empty($dbFiles)) {
|
||
self::$dbPath = $dbFiles[0];
|
||
} else {
|
||
$randomStr = substr(md5(uniqid(rand(), true)), 0, 10);
|
||
self::$dbPath = $dbDir . '/urlnav_' . $randomStr . '.db';
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 生成随机密钥
|
||
*/
|
||
private static function generateSecret()
|
||
{
|
||
return substr(md5(uniqid(rand(), true) . time()), 0, 16);
|
||
}
|
||
|
||
public static function getCategoryStats($categoryId) {
|
||
$db = self::getDbConnection();
|
||
|
||
// 获取网址总数
|
||
$stmt = $db->prepare("SELECT COUNT(*) as url_count FROM urlnav_urls WHERE category_id = ? AND is_active = 1");
|
||
$stmt->execute([$categoryId]);
|
||
$urlCount = $stmt->fetchColumn();
|
||
|
||
// 获取有RSS的网址数
|
||
$stmt = $db->prepare("SELECT COUNT(*) as rss_count FROM urlnav_urls WHERE category_id = ? AND rss_url IS NOT NULL AND rss_url != '' AND is_active = 1");
|
||
$stmt->execute([$categoryId]);
|
||
$rssCount = $stmt->fetchColumn();
|
||
|
||
return [
|
||
'url_count' => (int)$urlCount,
|
||
'rss_count' => (int)$rssCount
|
||
];
|
||
}
|
||
|
||
/**
|
||
* 初始化数据库
|
||
*/
|
||
private static function initDatabase()
|
||
{
|
||
if (empty(self::$dbPath)) {
|
||
self::initDbPath();
|
||
}
|
||
|
||
try {
|
||
$db = new PDO('sqlite:' . self::$dbPath);
|
||
$db->setAttribute(PDO::ATTR_ERRMODE, PDO::ERRMODE_EXCEPTION);
|
||
|
||
// 检查分类表是否存在
|
||
$tableCheck = $db->query("SELECT name FROM sqlite_master WHERE type='table' AND name='urlnav_categories'");
|
||
if (!$tableCheck->fetch()) {
|
||
// 创建分类表
|
||
$db->exec("CREATE TABLE urlnav_categories (
|
||
id INTEGER PRIMARY KEY AUTOINCREMENT,
|
||
name TEXT NOT NULL,
|
||
description TEXT,
|
||
sort_order INTEGER DEFAULT 0,
|
||
is_active INTEGER DEFAULT 1,
|
||
created_at DATETIME DEFAULT CURRENT_TIMESTAMP,
|
||
updated_at DATETIME DEFAULT CURRENT_TIMESTAMP
|
||
)");
|
||
|
||
// 插入默认分类
|
||
$db->exec("INSERT INTO urlnav_categories (name, description, sort_order) VALUES
|
||
('常用工具', '日常使用的在线工具', 1),
|
||
('设计资源', '设计相关的素材和资源', 2),
|
||
('开发资源', '程序开发相关资源', 3),
|
||
('技术社区', '技术交流和学习社区', 4)");
|
||
}
|
||
|
||
// 检查网址表是否存在
|
||
$tableCheck2 = $db->query("SELECT name FROM sqlite_master WHERE type='table' AND name='urlnav_urls'");
|
||
if (!$tableCheck2->fetch()) {
|
||
// 创建网址表
|
||
$db->exec("CREATE TABLE urlnav_urls (
|
||
id INTEGER PRIMARY KEY AUTOINCREMENT,
|
||
title TEXT NOT NULL,
|
||
url TEXT NOT NULL,
|
||
description TEXT,
|
||
rss_url TEXT,
|
||
category_id INTEGER,
|
||
star_rating INTEGER DEFAULT 0, -- 新增：星级评分，0-3表示0-3颗星
|
||
sort_order INTEGER DEFAULT 0,
|
||
is_active INTEGER DEFAULT 1,
|
||
is_online INTEGER DEFAULT 1,
|
||
last_status_check DATETIME,
|
||
status_check_count INTEGER DEFAULT 0,
|
||
last_status_code INTEGER,
|
||
created_at DATETIME DEFAULT CURRENT_TIMESTAMP,
|
||
updated_at DATETIME DEFAULT CURRENT_TIMESTAMP,
|
||
last_refresh DATETIME,
|
||
refresh_count INTEGER DEFAULT 0,
|
||
success_count INTEGER DEFAULT 0,
|
||
failure_count INTEGER DEFAULT 0,
|
||
last_error TEXT,
|
||
FOREIGN KEY (category_id) REFERENCES urlnav_categories(id) ON DELETE SET NULL
|
||
)");
|
||
|
||
// 创建索引
|
||
$db->exec("CREATE INDEX idx_category_id ON urlnav_urls(category_id)");
|
||
$db->exec("CREATE INDEX idx_is_active ON urlnav_urls(is_active)");
|
||
$db->exec("CREATE INDEX idx_rss_url ON urlnav_urls(rss_url)");
|
||
$db->exec("CREATE INDEX idx_last_refresh ON urlnav_urls(last_refresh)");
|
||
$db->exec("CREATE INDEX idx_is_online ON urlnav_urls(is_online)");
|
||
$db->exec("CREATE INDEX idx_last_status_check ON urlnav_urls(last_status_check)");
|
||
}
|
||
|
||
// 创建RSS缓存表 - 修改：添加full_content字段
|
||
$tableCheck3 = $db->query("SELECT name FROM sqlite_master WHERE type='table' AND name='urlnav_rss_cache'");
|
||
if (!$tableCheck3->fetch()) {
|
||
$db->exec("CREATE TABLE urlnav_rss_cache (
|
||
id INTEGER PRIMARY KEY AUTOINCREMENT,
|
||
url_id INTEGER NOT NULL,
|
||
feed_title TEXT NOT NULL,
|
||
feed_link TEXT NOT NULL,
|
||
feed_description TEXT,
|
||
full_content TEXT, -- 新增：完整内容字段
|
||
pub_date DATETIME NOT NULL,
|
||
guid TEXT NOT NULL,
|
||
cached_at DATETIME DEFAULT CURRENT_TIMESTAMP,
|
||
is_fresh INTEGER DEFAULT 1,
|
||
FOREIGN KEY (url_id) REFERENCES urlnav_urls(id) ON DELETE CASCADE,
|
||
UNIQUE(url_id, guid)
|
||
)");
|
||
|
||
$db->exec("CREATE INDEX idx_url_id ON urlnav_rss_cache(url_id)");
|
||
$db->exec("CREATE INDEX idx_pub_date ON urlnav_rss_cache(pub_date)");
|
||
$db->exec("CREATE INDEX idx_cached_at ON urlnav_rss_cache(cached_at)");
|
||
$db->exec("CREATE INDEX idx_is_fresh ON urlnav_rss_cache(is_fresh)");
|
||
$db->exec("CREATE UNIQUE INDEX idx_url_guid ON urlnav_rss_cache(url_id, guid)");
|
||
}
|
||
|
||
// 创建收藏表 - 新增
|
||
$tableCheck8 = $db->query("SELECT name FROM sqlite_master WHERE type='table' AND name='urlnav_favorites'");
|
||
if (!$tableCheck8->fetch()) {
|
||
$db->exec("CREATE TABLE urlnav_favorites (
|
||
id INTEGER PRIMARY KEY AUTOINCREMENT,
|
||
user_id INTEGER DEFAULT 0,
|
||
feed_id INTEGER NOT NULL,
|
||
feed_title TEXT NOT NULL,
|
||
feed_link TEXT NOT NULL,
|
||
feed_description TEXT,
|
||
full_content TEXT, -- 新增：完整内容字段
|
||
pub_date DATETIME NOT NULL,
|
||
site_title TEXT,
|
||
site_url TEXT,
|
||
category_name TEXT,
|
||
favorited_at DATETIME DEFAULT CURRENT_TIMESTAMP,
|
||
UNIQUE(user_id, feed_id)
|
||
)");
|
||
|
||
$db->exec("CREATE INDEX idx_favorite_user_id ON urlnav_favorites(user_id)");
|
||
$db->exec("CREATE INDEX idx_favorite_feed_id ON urlnav_favorites(feed_id)");
|
||
$db->exec("CREATE INDEX idx_favorite_created_at ON urlnav_favorites(favorited_at)");
|
||
}
|
||
|
||
// 创建RSS刷新记录表
|
||
$tableCheck4 = $db->query("SELECT name FROM sqlite_master WHERE type='table' AND name='urlnav_refresh_log'");
|
||
if (!$tableCheck4->fetch()) {
|
||
$db->exec("CREATE TABLE urlnav_refresh_log (
|
||
id INTEGER PRIMARY KEY AUTOINCREMENT,
|
||
refresh_type TEXT NOT NULL,
|
||
success_count INTEGER DEFAULT 0,
|
||
total_feeds INTEGER DEFAULT 0,
|
||
url_count INTEGER DEFAULT 0,
|
||
new_articles INTEGER DEFAULT 0,
|
||
error_message TEXT,
|
||
refresh_time DATETIME DEFAULT CURRENT_TIMESTAMP,
|
||
duration INTEGER DEFAULT 0,
|
||
cron_type TEXT DEFAULT 'rss' -- 新增：区分RSS和状态检查
|
||
)");
|
||
|
||
$db->exec("CREATE INDEX idx_refresh_time ON urlnav_refresh_log(refresh_time)");
|
||
$db->exec("CREATE INDEX idx_refresh_type ON urlnav_refresh_log(refresh_type)");
|
||
$db->exec("CREATE INDEX idx_cron_type ON urlnav_refresh_log(cron_type)");
|
||
}
|
||
|
||
// 创建定时任务记录表
|
||
$tableCheck5 = $db->query("SELECT name FROM sqlite_master WHERE type='table' AND name='urlnav_cron_log'");
|
||
if (!$tableCheck5->fetch()) {
|
||
$db->exec("CREATE TABLE urlnav_cron_log (
|
||
id INTEGER PRIMARY KEY AUTOINCREMENT,
|
||
cron_type TEXT NOT NULL,
|
||
executed_time DATETIME DEFAULT CURRENT_TIMESTAMP,
|
||
result TEXT,
|
||
error_message TEXT
|
||
)");
|
||
|
||
$db->exec("CREATE INDEX idx_executed_time ON urlnav_cron_log(executed_time)");
|
||
$db->exec("CREATE INDEX idx_cron_type ON urlnav_cron_log(cron_type)");
|
||
}
|
||
|
||
// 创建状态检查记录表
|
||
$tableCheck6 = $db->query("SELECT name FROM sqlite_master WHERE type='table' AND name='urlnav_status_log'");
|
||
if (!$tableCheck6->fetch()) {
|
||
$db->exec("CREATE TABLE urlnav_status_log (
|
||
id INTEGER PRIMARY KEY AUTOINCREMENT,
|
||
url_id INTEGER NOT NULL,
|
||
is_online INTEGER DEFAULT 0,
|
||
status_code INTEGER,
|
||
response_time INTEGER,
|
||
check_time DATETIME DEFAULT CURRENT_TIMESTAMP,
|
||
error_message TEXT,
|
||
FOREIGN KEY (url_id) REFERENCES urlnav_urls(id) ON DELETE CASCADE
|
||
)");
|
||
|
||
$db->exec("CREATE INDEX idx_url_id_status ON urlnav_status_log(url_id)");
|
||
$db->exec("CREATE INDEX idx_check_time ON urlnav_status_log(check_time)");
|
||
$db->exec("CREATE INDEX idx_is_online_status ON urlnav_status_log(is_online)");
|
||
}
|
||
|
||
// 创建状态检查统计表
|
||
$tableCheck7 = $db->query("SELECT name FROM sqlite_master WHERE type='table' AND name='urlnav_status_stats'");
|
||
if (!$tableCheck7->fetch()) {
|
||
$db->exec("CREATE TABLE urlnav_status_stats (
|
||
id INTEGER PRIMARY KEY AUTOINCREMENT,
|
||
total_checks INTEGER DEFAULT 0,
|
||
success_checks INTEGER DEFAULT 0,
|
||
failed_checks INTEGER DEFAULT 0,
|
||
avg_response_time REAL DEFAULT 0,
|
||
last_check_time DATETIME,
|
||
created_at DATETIME DEFAULT CURRENT_TIMESTAMP,
|
||
updated_at DATETIME DEFAULT CURRENT_TIMESTAMP
|
||
)");
|
||
|
||
// 初始化一条记录
|
||
$db->exec("INSERT INTO urlnav_status_stats (total_checks, success_checks, failed_checks, avg_response_time) VALUES (0, 0, 0, 0)");
|
||
}
|
||
|
||
// 创建更新时间触发器
|
||
$db->exec("CREATE TRIGGER IF NOT EXISTS update_category_time
|
||
AFTER UPDATE ON urlnav_categories
|
||
BEGIN
|
||
UPDATE urlnav_categories SET updated_at = CURRENT_TIMESTAMP WHERE id = NEW.id;
|
||
END");
|
||
|
||
$db->exec("CREATE TRIGGER IF NOT EXISTS update_url_time
|
||
AFTER UPDATE ON urlnav_urls
|
||
BEGIN
|
||
UPDATE urlnav_urls SET updated_at = CURRENT_TIMESTAMP WHERE id = NEW.id;
|
||
END");
|
||
|
||
$db = null;
|
||
} catch (PDOException $e) {
|
||
error_log('UrlNav: 数据库初始化失败: ' . $e->getMessage());
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 数据库迁移
|
||
*/
|
||
private static function migrateDatabase()
|
||
{
|
||
try {
|
||
$db = new PDO('sqlite:' . self::$dbPath);
|
||
$db->setAttribute(PDO::ATTR_ERRMODE, PDO::ERRMODE_EXCEPTION);
|
||
|
||
// 检查是否需要添加字段
|
||
$tableInfo = $db->query("PRAGMA table_info(urlnav_urls)");
|
||
$columns = $tableInfo->fetchAll(PDO::FETCH_ASSOC);
|
||
|
||
$newColumns = array(
|
||
'rss_url' => "ALTER TABLE urlnav_urls ADD COLUMN rss_url TEXT",
|
||
'last_refresh' => "ALTER TABLE urlnav_urls ADD COLUMN last_refresh DATETIME",
|
||
'refresh_count' => "ALTER TABLE urlnav_urls ADD COLUMN refresh_count INTEGER DEFAULT 0",
|
||
'success_count' => "ALTER TABLE urlnav_urls ADD COLUMN success_count INTEGER DEFAULT 0",
|
||
'failure_count' => "ALTER TABLE urlnav_urls ADD COLUMN failure_count INTEGER DEFAULT 0",
|
||
'last_error' => "ALTER TABLE urlnav_urls ADD COLUMN last_error TEXT",
|
||
'is_online' => "ALTER TABLE urlnav_urls ADD COLUMN is_online INTEGER DEFAULT 1",
|
||
'last_status_check' => "ALTER TABLE urlnav_urls ADD COLUMN last_status_check DATETIME",
|
||
'status_check_count' => "ALTER TABLE urlnav_urls ADD COLUMN status_check_count INTEGER DEFAULT 0",
|
||
'last_status_code' => "ALTER TABLE urlnav_urls ADD COLUMN last_status_code INTEGER"
|
||
);
|
||
|
||
foreach ($newColumns as $columnName => $sql) {
|
||
$hasColumn = false;
|
||
foreach ($columns as $column) {
|
||
if ($column['name'] === $columnName) {
|
||
$hasColumn = true;
|
||
break;
|
||
}
|
||
}
|
||
|
||
if (!$hasColumn) {
|
||
$db->exec($sql);
|
||
}
|
||
}
|
||
|
||
// 检查缓存表是否需要添加is_fresh字段
|
||
$cacheTableInfo = $db->query("PRAGMA table_info(urlnav_rss_cache)");
|
||
$cacheColumns = $cacheTableInfo->fetchAll(PDO::FETCH_ASSOC);
|
||
|
||
$hasIsFresh = false;
|
||
foreach ($cacheColumns as $column) {
|
||
if ($column['name'] === 'is_fresh') {
|
||
$hasIsFresh = true;
|
||
break;
|
||
}
|
||
}
|
||
|
||
if (!$hasIsFresh) {
|
||
$db->exec("ALTER TABLE urlnav_rss_cache ADD COLUMN is_fresh INTEGER DEFAULT 1");
|
||
$db->exec("CREATE INDEX IF NOT EXISTS idx_is_fresh ON urlnav_rss_cache(is_fresh)");
|
||
}
|
||
|
||
// 检查是否需要添加star_rating字段
|
||
$hasStarRating = false;
|
||
foreach ($columns as $column) {
|
||
if ($column['name'] === 'star_rating') {
|
||
$hasStarRating = true;
|
||
break;
|
||
}
|
||
}
|
||
|
||
if (!$hasStarRating) {
|
||
$db->exec("ALTER TABLE urlnav_urls ADD COLUMN star_rating INTEGER DEFAULT 0");
|
||
error_log("UrlNav: 已添加star_rating字段到urlnav_urls表");
|
||
}
|
||
|
||
// 检查缓存表是否需要添加full_content字段
|
||
$hasFullContent = false;
|
||
foreach ($cacheColumns as $column) {
|
||
if ($column['name'] === 'full_content') {
|
||
$hasFullContent = true;
|
||
break;
|
||
}
|
||
}
|
||
|
||
if (!$hasFullContent) {
|
||
$db->exec("ALTER TABLE urlnav_rss_cache ADD COLUMN full_content TEXT");
|
||
$db->exec("ALTER TABLE urlnav_favorites ADD COLUMN full_content TEXT");
|
||
}
|
||
|
||
// 检查refresh_log表是否需要添加cron_type字段
|
||
$refreshLogTableInfo = $db->query("PRAGMA table_info(urlnav_refresh_log)");
|
||
$refreshLogColumns = $refreshLogTableInfo->fetchAll(PDO::FETCH_ASSOC);
|
||
|
||
$hasCronType = false;
|
||
foreach ($refreshLogColumns as $column) {
|
||
if ($column['name'] === 'cron_type') {
|
||
$hasCronType = true;
|
||
break;
|
||
}
|
||
}
|
||
|
||
if (!$hasCronType) {
|
||
$db->exec("ALTER TABLE urlnav_refresh_log ADD COLUMN cron_type TEXT DEFAULT 'rss'");
|
||
$db->exec("CREATE INDEX IF NOT EXISTS idx_cron_type ON urlnav_refresh_log(cron_type)");
|
||
}
|
||
|
||
// ===== 修复关键：添加缺失的new_articles字段 =====
|
||
$hasNewArticles = false;
|
||
foreach ($refreshLogColumns as $column) {
|
||
if ($column['name'] === 'new_articles') {
|
||
$hasNewArticles = true;
|
||
break;
|
||
}
|
||
}
|
||
|
||
if (!$hasNewArticles) {
|
||
$db->exec("ALTER TABLE urlnav_refresh_log ADD COLUMN new_articles INTEGER DEFAULT 0");
|
||
error_log("UrlNav: 已添加new_articles字段到urlnav_refresh_log表");
|
||
}
|
||
|
||
// 🔴 新增：检查是否需要添加message字段
|
||
$hasMessage = false;
|
||
foreach ($refreshLogColumns as $column) {
|
||
if ($column['name'] === 'message') {
|
||
$hasMessage = true;
|
||
break;
|
||
}
|
||
}
|
||
|
||
if (!$hasMessage) {
|
||
$db->exec("ALTER TABLE urlnav_refresh_log ADD COLUMN message TEXT");
|
||
error_log("UrlNav: 已添加message字段到urlnav_refresh_log表");
|
||
}
|
||
|
||
// 🔴 新增：检查是否需要添加details字段
|
||
$hasDetails = false;
|
||
foreach ($refreshLogColumns as $column) {
|
||
if ($column['name'] === 'details') {
|
||
$hasDetails = true;
|
||
break;
|
||
}
|
||
}
|
||
|
||
if (!$hasDetails) {
|
||
$db->exec("ALTER TABLE urlnav_refresh_log ADD COLUMN details TEXT");
|
||
error_log("UrlNav: 已添加details字段到urlnav_refresh_log表");
|
||
}
|
||
// ===== 修复结束 =====
|
||
|
||
$db = null;
|
||
} catch (PDOException $e) {
|
||
error_log('UrlNav数据库迁移失败: ' . $e->getMessage());
|
||
}
|
||
}
|
||
|
||
|
||
/**
|
||
* 获取数据库连接 - 优化版，解决数据库锁问题
|
||
*/
|
||
public static function getDbConnection()
|
||
{
|
||
if (empty(self::$dbPath)) {
|
||
self::initDbPath();
|
||
}
|
||
|
||
if (!file_exists(self::$dbPath)) {
|
||
self::initDatabase();
|
||
}
|
||
|
||
$maxRetries = 3;
|
||
$retryDelay = 1; // 秒
|
||
|
||
for ($retry = 0; $retry < $maxRetries; $retry++) {
|
||
if ($retry > 0) {
|
||
error_log("UrlNav: 数据库连接重试 {$retry}，等待 {$retryDelay} 秒...");
|
||
sleep($retryDelay);
|
||
}
|
||
|
||
try {
|
||
$db = new PDO('sqlite:' . self::$dbPath);
|
||
$db->setAttribute(PDO::ATTR_ERRMODE, PDO::ERRMODE_EXCEPTION);
|
||
$db->exec('PRAGMA foreign_keys = ON');
|
||
$db->exec('PRAGMA busy_timeout = 3000'); // 设置3秒超时
|
||
$db->exec('PRAGMA journal_mode = WAL'); // 使用WAL模式提高并发性能
|
||
|
||
return $db;
|
||
} catch (PDOException $e) {
|
||
if (strpos($e->getMessage(), 'database is locked') !== false && $retry < $maxRetries - 1) {
|
||
continue;
|
||
}
|
||
throw new Exception('数据库连接失败: ' . $e->getMessage());
|
||
}
|
||
}
|
||
|
||
throw new Exception('数据库连接失败：重试' . $maxRetries . '次后仍被锁定');
|
||
}
|
||
|
||
/**
|
||
* 获取插件配置
|
||
*/
|
||
public static function getConfig()
|
||
{
|
||
static $config = null;
|
||
if ($config === null) {
|
||
$options = Typecho_Widget::widget('Widget_Options');
|
||
$config = $options->plugin('UrlNav');
|
||
}
|
||
return $config;
|
||
}
|
||
|
||
/**
|
||
* 获取RSS管理器
|
||
*/
|
||
private static function getRssManager()
|
||
{
|
||
if (self::$rssManager === null) {
|
||
self::$rssManager = new UrlNav_RssManager();
|
||
}
|
||
return self::$rssManager;
|
||
}
|
||
|
||
public static function executeRssCronTask()
|
||
{
|
||
// 立即设置响应头，防止502
|
||
if (!headers_sent()) {
|
||
header('Content-Type: application/json; charset=utf-8');
|
||
header('Cache-Control: no-cache, no-store, must-revalidate');
|
||
header('Pragma: no-cache');
|
||
header('Expires: 0');
|
||
|
||
// 立即输出，让Nginx知道脚本在运行
|
||
echo json_encode(['status' => 'starting', 'timestamp' => time()]);
|
||
flush();
|
||
ob_flush();
|
||
}
|
||
|
||
$startTime = microtime(true);
|
||
|
||
try {
|
||
error_log("UrlNav RSS定时任务: 开始执行 " . date('Y-m-d H:i:s'));
|
||
|
||
// 设置更长的执行时间
|
||
@set_time_limit(300); // 5分钟
|
||
@ini_set('max_execution_time', 300);
|
||
|
||
// 添加简单的锁检查，防止多个进程同时执行
|
||
$lockFile = __DIR__ . '/db/rss_cron_running.lock';
|
||
$lockTimeout = 1800; // 30分钟超时
|
||
|
||
if (file_exists($lockFile)) {
|
||
$lockTime = @filemtime($lockFile);
|
||
if ($lockTime && (time() - $lockTime) < $lockTimeout) {
|
||
error_log("UrlNav RSS定时任务: 跳过执行，已在运行中");
|
||
return array(
|
||
'success' => false,
|
||
'message' => '定时任务已在运行中，跳过本次执行',
|
||
'timestamp' => time()
|
||
);
|
||
}
|
||
// 锁已超时，删除它
|
||
@unlink($lockFile);
|
||
}
|
||
|
||
// 创建锁文件
|
||
@touch($lockFile);
|
||
@file_put_contents($lockFile, "Started at: " . date('Y-m-d H:i:s'));
|
||
|
||
register_shutdown_function(function() use ($lockFile) {
|
||
if (file_exists($lockFile)) {
|
||
@unlink($lockFile);
|
||
error_log("UrlNav: shutdown函数删除RSS锁文件");
|
||
}
|
||
});
|
||
|
||
// 执行刷新任务
|
||
$refreshResult = self::refreshAllRssFeeds(true);
|
||
|
||
$duration = round(microtime(true) - $startTime, 2);
|
||
|
||
// 删除锁文件
|
||
if (file_exists($lockFile)) {
|
||
@unlink($lockFile);
|
||
}
|
||
|
||
// 🆕 修改：确保result包含RSS地址信息
|
||
$result = array(
|
||
'success' => $refreshResult['success'],
|
||
'refreshed' => true,
|
||
'refresh_result' => $refreshResult,
|
||
'timestamp' => time(),
|
||
'duration' => $duration,
|
||
'message' => $refreshResult['message'],
|
||
// 🆕 关键：直接包含RSS地址信息
|
||
'successRssUrls' => $refreshResult['successRssUrls'] ?? array(),
|
||
'failedRssUrls' => $refreshResult['failedRssUrls'] ?? array()
|
||
);
|
||
|
||
// 记录日志（会自动将上面的result转为JSON存入数据库）
|
||
self::logCron('rss_auto_refresh', $result);
|
||
|
||
error_log("UrlNav RSS定时任务: 执行完成，耗时 {$duration} 秒");
|
||
|
||
return $result;
|
||
|
||
} catch (Exception $e) {
|
||
error_log("UrlNav RSS定时任务异常: " . $e->getMessage());
|
||
|
||
// 确保锁文件被删除
|
||
$lockFile = __DIR__ . '/db/rss_cron_running.lock';
|
||
if (file_exists($lockFile)) {
|
||
@unlink($lockFile);
|
||
}
|
||
|
||
return array(
|
||
'success' => false,
|
||
'error' => $e->getMessage(),
|
||
'timestamp' => time(),
|
||
'message' => 'RSS定时任务执行异常',
|
||
'successRssUrls' => array(),
|
||
'failedRssUrls' => array()
|
||
);
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 执行状态检查定时任务 - 完全移除锁机制
|
||
*/
|
||
public static function executeStatusCronTask()
|
||
{
|
||
try {
|
||
$startTime = microtime(true);
|
||
error_log("UrlNav 状态检查定时任务: 开始执行 " . date('Y-m-d H:i:s'));
|
||
|
||
// 修改这里：调用正确的自动检查方式
|
||
$statusResult = self::manualCheckStatus(null, false); // $urlIds=null, $isBatchCheck=false
|
||
|
||
$endTime = microtime(true);
|
||
$duration = round($endTime - $startTime, 2);
|
||
|
||
// 更新状态检查统计
|
||
self::updateStatusStats($statusResult);
|
||
|
||
// 记录状态检查专用的定时任务日志
|
||
self::logCron('status_auto_check', json_encode(array_merge($statusResult, array(
|
||
'duration' => $duration,
|
||
'timestamp' => time()
|
||
))));
|
||
|
||
if ($statusResult['success']) {
|
||
error_log("UrlNav 状态检查定时任务: 执行成功，耗时 {$duration} 秒");
|
||
return array(
|
||
'success' => true,
|
||
'status_checked' => $statusResult['total'] > 0,
|
||
'status_result' => $statusResult,
|
||
'timestamp' => time(),
|
||
'duration' => $duration,
|
||
'message' => '状态检查定时任务执行成功'
|
||
);
|
||
} else {
|
||
error_log("UrlNav 状态检查定时任务: 执行失败: " . $statusResult['message']);
|
||
return array(
|
||
'success' => false,
|
||
'status_checked' => false,
|
||
'status_result' => $statusResult,
|
||
'timestamp' => time(),
|
||
'duration' => $duration,
|
||
'message' => '状态检查定时任务执行失败'
|
||
);
|
||
}
|
||
|
||
} catch (Exception $e) {
|
||
error_log("UrlNav 状态检查定时任务异常: " . $e->getMessage());
|
||
return array(
|
||
'success' => false,
|
||
'error' => $e->getMessage(),
|
||
'timestamp' => time(),
|
||
'message' => '状态检查定时任务执行异常'
|
||
);
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 通用的锁定任务执行器 - 优化版，减少锁竞争
|
||
*/
|
||
private static function executeLockedTask($lockFile, $taskType, $callback)
|
||
{
|
||
$lockTimeout = 3600; // 延长到1小时超时
|
||
|
||
// 简化的锁检查：如果锁文件存在且未超时，直接跳过
|
||
if (file_exists($lockFile)) {
|
||
$lockTime = @filemtime($lockFile);
|
||
if ($lockTime && (time() - $lockTime) < $lockTimeout) {
|
||
$lockDuration = time() - $lockTime;
|
||
error_log("UrlNav {$taskType}: 跳过执行，锁文件存在 {$lockDuration} 秒");
|
||
return array(
|
||
'success' => false,
|
||
'message' => "{$taskType}定时任务正在运行，跳过本次执行",
|
||
'timestamp' => time(),
|
||
'lock_time' => $lockTime,
|
||
'lock_duration' => $lockDuration
|
||
);
|
||
}
|
||
// 锁已超时，删除它
|
||
@unlink($lockFile);
|
||
error_log("UrlNav {$taskType}: 删除超时的锁文件（已存在超过 {$lockTimeout} 秒）");
|
||
}
|
||
|
||
// 创建锁文件
|
||
if (!@touch($lockFile)) {
|
||
error_log("UrlNav {$taskType}: 无法创建锁文件");
|
||
return array(
|
||
'success' => false,
|
||
'message' => '无法创建锁文件',
|
||
'timestamp' => time()
|
||
);
|
||
}
|
||
|
||
// 在锁文件中记录开始时间
|
||
file_put_contents($lockFile, "Started at: " . date('Y-m-d H:i:s') . "\nTask type: {$taskType}");
|
||
|
||
error_log("UrlNav {$taskType}: 开始执行定时任务 " . date('Y-m-d H:i:s'));
|
||
|
||
try {
|
||
// 确保锁文件会被删除（即使脚本意外终止）
|
||
register_shutdown_function(function() use ($lockFile, $taskType) {
|
||
if (file_exists($lockFile)) {
|
||
$lockDuration = time() - filemtime($lockFile);
|
||
@unlink($lockFile);
|
||
error_log("UrlNav {$taskType}: shutdown函数删除锁文件，锁持续了 {$lockDuration} 秒");
|
||
}
|
||
});
|
||
|
||
// 执行回调函数
|
||
$result = $callback();
|
||
|
||
// 删除锁文件
|
||
if (file_exists($lockFile)) {
|
||
$lockDuration = time() - filemtime($lockFile);
|
||
@unlink($lockFile);
|
||
error_log("UrlNav {$taskType}: 任务完成，删除锁文件，任务耗时 {$lockDuration} 秒");
|
||
}
|
||
|
||
return $result;
|
||
|
||
} catch (Exception $e) {
|
||
// 确保锁文件被删除
|
||
if (file_exists($lockFile)) {
|
||
$lockDuration = time() - filemtime($lockFile);
|
||
@unlink($lockFile);
|
||
error_log("UrlNav {$taskType}: 异常时删除锁文件，锁持续了 {$lockDuration} 秒");
|
||
}
|
||
|
||
error_log("UrlNav {$taskType}定时任务异常: " . $e->getMessage());
|
||
self::logCron('error', $e->getMessage());
|
||
|
||
return array(
|
||
'success' => false,
|
||
'error' => $e->getMessage(),
|
||
'timestamp' => time(),
|
||
'message' => "{$taskType}定时任务执行异常"
|
||
);
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 手动解锁定时任务（供调试使用）
|
||
*/
|
||
public static function unlockCron($cronType = 'rss')
|
||
{
|
||
if ($cronType === 'rss') {
|
||
$lockFile = __DIR__ . '/db/rss_cron.lock';
|
||
} elseif ($cronType === 'status') {
|
||
$lockFile = __DIR__ . '/db/status_cron.lock';
|
||
} else {
|
||
$lockFile = __DIR__ . '/db/cron.lock';
|
||
}
|
||
|
||
if (file_exists($lockFile)) {
|
||
if (@unlink($lockFile)) {
|
||
error_log("UrlNav: 手动解锁{$cronType}成功");
|
||
return array(
|
||
'success' => true,
|
||
'message' => "{$cronType}定时任务锁已解除",
|
||
'timestamp' => time()
|
||
);
|
||
} else {
|
||
error_log("UrlNav: 手动解锁{$cronType}失败");
|
||
return array(
|
||
'success' => false,
|
||
'message' => '无法删除锁文件',
|
||
'timestamp' => time()
|
||
);
|
||
}
|
||
} else {
|
||
return array(
|
||
'success' => true,
|
||
'message' => "{$cronType}没有锁文件存在",
|
||
'timestamp' => time()
|
||
);
|
||
}
|
||
}
|
||
|
||
public static function refreshAllRssFeeds($isCron = false)
|
||
{
|
||
$startTime = microtime(true);
|
||
|
||
// 🆕 新增：在定时任务中自动清理过期缓存
|
||
if ($isCron) {
|
||
self::cleanExpiredCache();
|
||
}
|
||
|
||
// 立即设置响应头，避免502
|
||
if ($isCron && !headers_sent()) {
|
||
header('Content-Type: application/json; charset=utf-8');
|
||
header('Cache-Control: no-cache, no-store, must-revalidate');
|
||
header('Pragma: no-cache');
|
||
header('Expires: 0');
|
||
|
||
// 立即输出一些内容，让Nginx知道脚本还在运行
|
||
echo '{"status":"starting","message":"RSS刷新任务开始...","timestamp":' . time() . '}';
|
||
flush();
|
||
ob_flush();
|
||
}
|
||
|
||
// 设置更长的执行时间
|
||
if ($isCron) {
|
||
@set_time_limit(300); // 5分钟
|
||
@ini_set('max_execution_time', 300);
|
||
}
|
||
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
// 使用后台配置的数量
|
||
$config = self::getConfig();
|
||
$limit = intval($config->rssRefreshLimit ?? 10);
|
||
$limit = max(1, min($limit, 30)); // 限制在1-30之间
|
||
|
||
error_log("===== UrlNav RSS刷新开始，时间: " . date('Y-m-d H:i:s') . " =====");
|
||
error_log("配置数量: {$limit}");
|
||
|
||
// 优化查询：优先处理从未刷新或很久没刷新的
|
||
// 关键修复：添加时间条件，避免重复刷新刚刷过的
|
||
$sql = "
|
||
SELECT id, rss_url, url, title, last_refresh, failure_count, success_count, created_at
|
||
FROM urlnav_urls
|
||
WHERE is_active = 1
|
||
AND rss_url IS NOT NULL
|
||
AND TRIM(rss_url) != ''
|
||
AND (
|
||
-- 从未刷新过的
|
||
last_refresh IS NULL
|
||
-- 或者超过1小时没刷新的
|
||
OR last_refresh < datetime('now', '-1 hour')
|
||
-- 或者失败次数多且超过30分钟没重试
|
||
OR (failure_count > success_count AND last_refresh < datetime('now', '-30 minutes'))
|
||
)
|
||
ORDER BY
|
||
CASE
|
||
-- 最高优先级：从未刷新过的
|
||
WHEN last_refresh IS NULL THEN 0
|
||
-- 次高优先级：失败次数多于成功次数的
|
||
WHEN failure_count > success_count THEN 1
|
||
-- 中等优先级：新添加的网址（最近3天内）
|
||
WHEN created_at > datetime('now', '-3 days') THEN 2
|
||
-- 低优先级：正常的
|
||
ELSE 3
|
||
END,
|
||
-- 按刷新时间从早到晚排序
|
||
CASE
|
||
WHEN last_refresh IS NULL THEN created_at
|
||
ELSE last_refresh
|
||
END ASC
|
||
LIMIT ?
|
||
";
|
||
|
||
$stmt = $db->prepare($sql);
|
||
$stmt->execute(array($limit));
|
||
$urls = $stmt->fetchAll(PDO::FETCH_ASSOC);
|
||
|
||
// 如果没有符合条件的，放宽条件选择一些
|
||
if (empty($urls)) {
|
||
error_log("UrlNav: 没有需要立即刷新的RSS源，选择一些较久没刷新的");
|
||
|
||
$sql = "
|
||
SELECT id, rss_url, url, title, last_refresh, failure_count, success_count
|
||
FROM urlnav_urls
|
||
WHERE is_active = 1
|
||
AND rss_url IS NOT NULL
|
||
AND TRIM(rss_url) != ''
|
||
ORDER BY last_refresh ASC NULLS FIRST
|
||
LIMIT ?
|
||
";
|
||
|
||
$stmt = $db->prepare($sql);
|
||
$stmt->execute(array(min($limit, 5))); // 少选几个
|
||
$urls = $stmt->fetchAll(PDO::FETCH_ASSOC);
|
||
}
|
||
|
||
if (empty($urls)) {
|
||
error_log("UrlNav: 没有需要刷新的RSS网址");
|
||
return array(
|
||
'success' => true,
|
||
'message' => '没有需要刷新的RSS网址',
|
||
'successCount' => 0,
|
||
'failureCount' => 0,
|
||
'newArticles' => 0,
|
||
'totalFeeds' => 0,
|
||
'urlCount' => 0,
|
||
'successRssUrls' => array(),
|
||
'failedRssUrls' => array()
|
||
);
|
||
}
|
||
|
||
error_log("UrlNav: 获取到 " . count($urls) . " 个需要刷新的RSS源");
|
||
|
||
// 记录获取到的URL信息
|
||
foreach ($urls as $url) {
|
||
$refreshStatus = $url['last_refresh'] ?
|
||
"最后刷新: " . $url['last_refresh'] :
|
||
"从未刷新";
|
||
error_log("UrlNav: 选中 - ID: {$url['id']}, {$refreshStatus}, URL: {$url['rss_url']}");
|
||
}
|
||
|
||
$successCount = 0;
|
||
$failureCount = 0;
|
||
$totalFeeds = 0;
|
||
$newArticles = 0;
|
||
|
||
// 🆕 修改：记录成功和失败的RSS地址
|
||
$successRssUrls = array();
|
||
$failedRssUrls = array();
|
||
|
||
// 配置参数
|
||
$timeout = intval($config->fetchTimeout ?? 15); // 默认15秒
|
||
$retryTimes = intval($config->retryTimes ?? 2); // 重试2次
|
||
$maxFeeds = intval($config->maxFeedsPerSite ?? 20); // 每个站点20条
|
||
|
||
// 关键：定期输出内容，保持连接活跃
|
||
$lastOutputTime = $startTime;
|
||
|
||
foreach ($urls as $index => $url) {
|
||
$currentTime = microtime(true);
|
||
$elapsedTime = $currentTime - $startTime;
|
||
|
||
// 检查总执行时间（4分钟限制）
|
||
if ($isCron && $elapsedTime > 240) {
|
||
error_log('UrlNav: 接近总超时（4分钟），停止处理');
|
||
break;
|
||
}
|
||
|
||
// 每3秒输出一次，保持连接活跃（防502关键）
|
||
if ($isCron && ($currentTime - $lastOutputTime) > 3) {
|
||
if (!headers_sent()) {
|
||
echo '{"status":"processing","progress":"' . ($index+1) . '/' . count($urls) . '","timestamp":' . time() . '}';
|
||
flush();
|
||
ob_flush();
|
||
}
|
||
$lastOutputTime = $currentTime;
|
||
}
|
||
|
||
try {
|
||
error_log("UrlNav: [开始] 处理RSS #" . ($index+1) . " - ID: " . $url['id'] . ", URL: " . $url['rss_url']);
|
||
error_log("UrlNav: 最后刷新时间: " . ($url['last_refresh'] ?: '从未刷新'));
|
||
|
||
$urlResult = self::refreshSingleRssUrl($url, $timeout, $retryTimes, $maxFeeds);
|
||
|
||
if ($urlResult['success']) {
|
||
$successCount++;
|
||
$newArticles += $urlResult['new_articles'];
|
||
$totalFeeds += $urlResult['total_feeds'];
|
||
|
||
// 🆕 记录成功的RSS地址
|
||
$successRssUrls[] = $url['rss_url'];
|
||
|
||
error_log("UrlNav: [成功] ID: " . $url['id'] . ", 新增文章: " . $urlResult['new_articles'] . ", RSS: " . $url['rss_url']);
|
||
} else {
|
||
$failureCount++;
|
||
|
||
// 🆕 记录失败的RSS地址
|
||
$failedRssUrls[] = $url['rss_url'];
|
||
|
||
error_log("UrlNav: [失败] ID: " . $url['id'] . ", 错误: " . ($urlResult['error'] ?? '未知错误') . ", RSS: " . $url['rss_url']);
|
||
}
|
||
|
||
// 短暂休息，避免对目标服务器压力过大
|
||
if ($index < count($urls) - 1) { // 不是最后一个时休息
|
||
usleep(800000); // 0.8秒休息
|
||
}
|
||
|
||
} catch (Exception $e) {
|
||
$failureCount++;
|
||
|
||
// 🆕 记录异常的RSS地址
|
||
$failedRssUrls[] = $url['rss_url'] . " [异常]";
|
||
|
||
error_log('UrlNav: [异常] ID: ' . $url['id'] . ', RSS: ' . $url['rss_url'] . ', 异常: ' . $e->getMessage());
|
||
}
|
||
}
|
||
|
||
$duration = round(microtime(true) - $startTime, 2);
|
||
|
||
// 记录日志 - 现在传递成功和失败的RSS地址
|
||
self::logRefresh($isCron ? 'cron' : 'manual', $successCount, $totalFeeds,
|
||
count($urls), $newArticles, null, $duration, 'rss',
|
||
$successRssUrls, $failedRssUrls);
|
||
|
||
$message = "刷新完成：成功 {$successCount} 个，失败 {$failureCount} 个";
|
||
$result = array(
|
||
'success' => $successCount > 0 || count($urls) == 0,
|
||
'successCount' => $successCount,
|
||
'failureCount' => $failureCount,
|
||
'newArticles' => $newArticles,
|
||
'totalFeeds' => $totalFeeds,
|
||
'urlCount' => count($urls),
|
||
'duration' => $duration,
|
||
'message' => $message,
|
||
// 🆕 修改：返回成功和失败的RSS地址
|
||
'successRssUrls' => $successRssUrls,
|
||
'failedRssUrls' => $failedRssUrls
|
||
);
|
||
|
||
error_log("UrlNav: [完成] RSS刷新完成，耗时 {$duration} 秒，{$message}");
|
||
error_log("===== UrlNav RSS刷新结束 =====");
|
||
|
||
return $result;
|
||
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: [全局异常] 刷新失败: ' . $e->getMessage());
|
||
error_log("===== UrlNav RSS刷新异常结束 =====");
|
||
return array(
|
||
'success' => false,
|
||
'message' => '刷新失败: ' . $e->getMessage(),
|
||
'successRssUrls' => array(),
|
||
'failedRssUrls' => array()
|
||
);
|
||
}
|
||
}
|
||
/**
|
||
* 获取RSS刷新状态统计
|
||
*/
|
||
public static function getRssRefreshStatus()
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
// 获取统计信息
|
||
$stats = array();
|
||
|
||
// 总RSS源数量
|
||
$stmt = $db->query("SELECT COUNT(*) as total FROM urlnav_urls WHERE is_active = 1 AND rss_url IS NOT NULL AND TRIM(rss_url) != ''");
|
||
$stats['total_rss_sources'] = $stmt->fetchColumn();
|
||
|
||
// 从未刷新的数量
|
||
$stmt = $db->query("SELECT COUNT(*) as never_refreshed FROM urlnav_urls WHERE is_active = 1 AND rss_url IS NOT NULL AND TRIM(rss_url) != '' AND last_refresh IS NULL");
|
||
$stats['never_refreshed'] = $stmt->fetchColumn();
|
||
|
||
// 今天刷新的数量
|
||
$stmt = $db->query("SELECT COUNT(*) as today_refreshed FROM urlnav_urls WHERE is_active = 1 AND rss_url IS NOT NULL AND TRIM(rss_url) != '' AND date(last_refresh) = date('now')");
|
||
$stats['today_refreshed'] = $stmt->fetchColumn();
|
||
|
||
// 最近7天刷新的数量
|
||
$stmt = $db->query("SELECT COUNT(*) as week_refreshed FROM urlnav_urls WHERE is_active = 1 AND rss_url IS NOT NULL AND TRIM(rss_url) != '' AND last_refresh >= datetime('now', '-7 days')");
|
||
$stats['week_refreshed'] = $stmt->fetchColumn();
|
||
|
||
// 最久未刷新的时间
|
||
$stmt = $db->query("SELECT MIN(last_refresh) as oldest_refresh FROM urlnav_urls WHERE is_active = 1 AND rss_url IS NOT NULL AND TRIM(rss_url) != '' AND last_refresh IS NOT NULL");
|
||
$oldest = $stmt->fetchColumn();
|
||
$stats['oldest_refresh'] = $oldest;
|
||
if ($oldest) {
|
||
$stats['oldest_days'] = round((time() - strtotime($oldest)) / 86400, 1);
|
||
}
|
||
|
||
// 需要刷新的数量（超过1天没刷新的）
|
||
$stmt = $db->query("SELECT COUNT(*) as need_refresh FROM urlnav_urls WHERE is_active = 1 AND rss_url IS NOT NULL AND TRIM(rss_url) != '' AND (last_refresh IS NULL OR last_refresh < datetime('now', '-1 day'))");
|
||
$stats['need_refresh'] = $stmt->fetchColumn();
|
||
|
||
// 成功率统计
|
||
$stmt = $db->query("SELECT
|
||
SUM(success_count) as total_success,
|
||
SUM(failure_count) as total_failure,
|
||
SUM(refresh_count) as total_refreshes
|
||
FROM urlnav_urls WHERE is_active = 1 AND rss_url IS NOT NULL AND TRIM(rss_url) != ''");
|
||
$countStats = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
|
||
$stats['total_success'] = $countStats['total_success'] ?? 0;
|
||
$stats['total_failure'] = $countStats['total_failure'] ?? 0;
|
||
$stats['total_refreshes'] = $countStats['total_refreshes'] ?? 0;
|
||
$stats['success_rate'] = $stats['total_refreshes'] > 0 ?
|
||
round(($stats['total_success'] / $stats['total_refreshes']) * 100, 1) : 0;
|
||
|
||
return $stats;
|
||
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 获取刷新状态失败: ' . $e->getMessage());
|
||
return array();
|
||
}
|
||
}
|
||
|
||
|
||
private static function refreshSingleRssUrl($url, $timeout = 8, $retryTimes = 1, $maxFeeds = 10)
|
||
{
|
||
$urlId = $url['id'];
|
||
$rssUrl = trim($url['rss_url']);
|
||
|
||
error_log("UrlNav: === 开始处理RSS ID: {$urlId} ===");
|
||
error_log("UrlNav: RSS URL: {$rssUrl}");
|
||
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
// 更新刷新统计
|
||
$stmt = $db->prepare("UPDATE urlnav_urls SET refresh_count = refresh_count + 1 WHERE id = ?");
|
||
$stmt->execute(array($urlId));
|
||
error_log("UrlNav: 更新刷新统计成功");
|
||
|
||
// 解析RSS内容
|
||
error_log("UrlNav: 开始解析RSS内容...");
|
||
$feeds = self::parseRssFeedWithRetry($rssUrl, $retryTimes, $timeout);
|
||
error_log("UrlNav: RSS解析完成，获取到 " . count($feeds) . " 篇文章");
|
||
|
||
if (empty($feeds)) {
|
||
error_log("UrlNav: 没有获取到文章数据");
|
||
|
||
$stmt = $db->prepare("
|
||
UPDATE urlnav_urls SET
|
||
last_refresh = CURRENT_TIMESTAMP,
|
||
last_error = '无可用数据'
|
||
WHERE id = ?
|
||
");
|
||
$stmt->execute(array($urlId));
|
||
|
||
error_log("UrlNav: === 处理完成（无数据）===");
|
||
return array(
|
||
'success' => true,
|
||
'new_articles' => 0,
|
||
'total_feeds' => 0,
|
||
'error' => null
|
||
);
|
||
}
|
||
|
||
// 限制每个站点最大文章数
|
||
$feeds = array_slice($feeds, 0, $maxFeeds);
|
||
error_log("UrlNav: 限制后文章数: " . count($feeds));
|
||
|
||
$addedCount = 0;
|
||
foreach ($feeds as $feedIndex => $feed) {
|
||
try {
|
||
// 确保所有必要字段都有值
|
||
$title = !empty($feed['title']) ? substr($feed['title'], 0, 255) : '无标题';
|
||
$link = !empty($feed['link']) ? substr($feed['link'], 0, 500) : $url['url'];
|
||
$description = !empty($feed['description']) ? substr($feed['description'], 0, 1000) : '';
|
||
$fullContent = !empty($feed['full_content']) ? substr($feed['full_content'], 0, 5000) : $description; // 使用完整内容，如果不存在则使用描述
|
||
$pubDate = !empty($feed['pubDate']) ? $feed['pubDate'] : date('Y-m-d H:i:s');
|
||
$guid = !empty($feed['guid']) ? substr($feed['guid'], 0, 255) : md5($link . $pubDate);
|
||
|
||
error_log("UrlNav: 处理文章 #" . ($feedIndex+1) . ": {$title}");
|
||
|
||
// 使用INSERT OR IGNORE避免冲突
|
||
$stmt = $db->prepare("
|
||
INSERT OR IGNORE INTO urlnav_rss_cache
|
||
(url_id, feed_title, feed_link, feed_description, full_content, pub_date, guid, cached_at, is_fresh)
|
||
VALUES (?, ?, ?, ?, ?, ?, ?, CURRENT_TIMESTAMP, 1)
|
||
");
|
||
|
||
$stmt->execute(array(
|
||
$urlId,
|
||
$title,
|
||
$link,
|
||
$description,
|
||
$fullContent,
|
||
$pubDate,
|
||
$guid
|
||
));
|
||
|
||
if ($stmt->rowCount() > 0) {
|
||
$addedCount++;
|
||
error_log("UrlNav: 文章 #" . ($feedIndex+1) . " 插入成功");
|
||
} else {
|
||
error_log("UrlNav: 文章 #" . ($feedIndex+1) . " 已存在，跳过");
|
||
}
|
||
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 文章处理异常: ' . $e->getMessage());
|
||
// 继续处理下一篇文章
|
||
continue;
|
||
}
|
||
}
|
||
|
||
// 更新URL统计信息
|
||
$stmt = $db->prepare("
|
||
UPDATE urlnav_urls SET
|
||
success_count = success_count + 1,
|
||
last_refresh = CURRENT_TIMESTAMP,
|
||
last_error = NULL
|
||
WHERE id = ?
|
||
");
|
||
$stmt->execute(array($urlId));
|
||
|
||
error_log("UrlNav: 成功解析RSS - ID: {$urlId}, 获取到 " . count($feeds) . " 篇文章, 新增 {$addedCount} 篇");
|
||
error_log("UrlNav: === 处理完成（成功）===");
|
||
|
||
return array(
|
||
'success' => true,
|
||
'new_articles' => $addedCount,
|
||
'total_feeds' => count($feeds),
|
||
'error' => null
|
||
);
|
||
|
||
} catch (Exception $e) {
|
||
// 记录错误信息
|
||
$errorMessage = substr($e->getMessage(), 0, 500);
|
||
error_log("UrlNav: RSS解析失败 - 错误: {$errorMessage}");
|
||
|
||
$stmt = $db->prepare("
|
||
UPDATE urlnav_urls SET
|
||
failure_count = failure_count + 1,
|
||
last_refresh = CURRENT_TIMESTAMP,
|
||
last_error = ?
|
||
WHERE id = ?
|
||
");
|
||
$stmt->execute(array($errorMessage, $urlId));
|
||
|
||
error_log("UrlNav: === 处理完成（失败）===");
|
||
|
||
return array(
|
||
'success' => false,
|
||
'new_articles' => 0,
|
||
'total_feeds' => 0,
|
||
'error' => $errorMessage
|
||
);
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 获取需要刷新的网址数量 - 新增方法
|
||
*/
|
||
public static function getUrlsNeedingRefresh()
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
$sql = "
|
||
SELECT COUNT(*) as count FROM urlnav_urls
|
||
WHERE is_active = 1
|
||
AND rss_url IS NOT NULL
|
||
AND TRIM(rss_url) != ''
|
||
AND (
|
||
last_refresh IS NULL
|
||
OR last_refresh <= datetime('now', '-1 hour')
|
||
)
|
||
";
|
||
|
||
$stmt = $db->query($sql);
|
||
$result = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
return $result['count'] ?? 0;
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 获取需要刷新的网址数量失败: ' . $e->getMessage());
|
||
return 0;
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 带重试机制的RSS解析 - 优化版
|
||
*/
|
||
private static function parseRssFeedWithRetry($rssUrl, $retryTimes = 1, $timeout = 8)
|
||
{
|
||
$lastError = null;
|
||
|
||
for ($i = 0; $i <= $retryTimes; $i++) {
|
||
try {
|
||
if ($i > 0) {
|
||
// 重试前等待一段时间
|
||
sleep($i * 2);
|
||
error_log("UrlNav: RSS重试第{$i}次: {$rssUrl}");
|
||
}
|
||
|
||
$feeds = self::parseRssFeed($rssUrl, $timeout);
|
||
return $feeds;
|
||
|
||
} catch (Exception $e) {
|
||
$lastError = $e;
|
||
$errorMsg = $e->getMessage();
|
||
|
||
// 如果是DNS错误，尝试使用IP直接访问（针对特定域名）
|
||
if (strpos($errorMsg, 'getaddrinfo failed') !== false && strpos($rssUrl, 'windful.cn') !== false) {
|
||
// 尝试使用IP访问（需要你知道windful.cn的IP）
|
||
// $rssUrl = str_replace('https://windful.cn/', 'https://[IP地址]/', $rssUrl);
|
||
error_log("UrlNav: DNS解析失败，建议检查windful.cn域名是否正常");
|
||
}
|
||
|
||
if ($i < $retryTimes) {
|
||
error_log("UrlNav: RSS解析失败，第" . ($i+1) . "次重试: " . $errorMsg);
|
||
}
|
||
}
|
||
}
|
||
|
||
// 所有重试都失败
|
||
throw new Exception("RSS解析失败: " . $lastError->getMessage());
|
||
}
|
||
|
||
|
||
/**
|
||
* 解析RSS源 - 完整功能增强版（修改全文字段逻辑）
|
||
*/
|
||
private static function parseRssFeed($rssUrl, $timeout = 8)
|
||
{
|
||
error_log("UrlNav: >>> 开始解析RSS: {$rssUrl}");
|
||
|
||
try {
|
||
// 设置超时时间（保持原样）
|
||
$context = stream_context_create(array(
|
||
'http' => array(
|
||
'timeout' => $timeout,
|
||
'ignore_errors' => true,
|
||
'header' => "User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36\r\n" .
|
||
"Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8\r\n" .
|
||
"Accept-Language: zh-CN,zh;q=0.9,en;q=0.8\r\n"
|
||
),
|
||
'ssl' => array(
|
||
'verify_peer' => false,
|
||
'verify_peer_name' => false,
|
||
'allow_self_signed' => true
|
||
)
|
||
));
|
||
|
||
error_log("UrlNav: 尝试获取RSS内容...");
|
||
$content = @file_get_contents($rssUrl, false, $context);
|
||
|
||
if ($content === false) {
|
||
$error = error_get_last();
|
||
$errorMsg = $error['message'] ?? '未知错误';
|
||
error_log("UrlNav: file_get_contents失败: {$errorMsg}");
|
||
|
||
if (isset($http_response_header)) {
|
||
error_log("UrlNav: HTTP响应头: " . implode(" | ", $http_response_header));
|
||
}
|
||
|
||
throw new Exception('无法获取RSS内容: ' . $errorMsg);
|
||
}
|
||
|
||
error_log("UrlNav: 获取内容成功，长度: " . strlen($content) . " 字节");
|
||
|
||
// 检查HTTP状态码（保持原样）
|
||
if (isset($http_response_header[0])) {
|
||
error_log("UrlNav: HTTP状态: {$http_response_header[0]}");
|
||
if (strpos($http_response_header[0], '404') !== false) {
|
||
throw new Exception('RSS源不存在 (404)');
|
||
}
|
||
if (strpos($http_response_header[0], '403') !== false) {
|
||
throw new Exception('拒绝访问 (403)');
|
||
}
|
||
if (strpos($http_response_header[0], '500') !== false) {
|
||
throw new Exception('服务器内部错误 (500)');
|
||
}
|
||
}
|
||
|
||
if (empty($content) || trim($content) === '') {
|
||
error_log("UrlNav: RSS内容为空");
|
||
throw new Exception('RSS内容为空');
|
||
}
|
||
|
||
// 处理可能存在的BOM头（保持原样）
|
||
if (substr($content, 0, 3) == "\xEF\xBB\xBF") {
|
||
$content = substr($content, 3);
|
||
error_log("UrlNav: 已移除BOM头");
|
||
}
|
||
|
||
// 简单的XML修复（保持原样）
|
||
$content = preg_replace('/[\x00-\x08\x0B\x0C\x0E-\x1F\x7F]/', '', $content);
|
||
$content = preg_replace('/&(?!(amp|lt|gt|quot|apos|#\d+);)/', '&', $content);
|
||
|
||
// 🆕 增强：尝试多种XML解析方式，确保兼容性
|
||
libxml_use_internal_errors(true);
|
||
libxml_clear_errors();
|
||
|
||
$xml = null;
|
||
|
||
// 方式1：先尝试DOMDocument（最兼容WordPress/Typecho）
|
||
try {
|
||
$dom = new DOMDocument();
|
||
$dom->recover = true;
|
||
$dom->strictErrorChecking = false;
|
||
|
||
if (@$dom->loadXML($content)) {
|
||
error_log("UrlNav: 使用DOMDocument解析成功");
|
||
$xml = simplexml_import_dom($dom);
|
||
}
|
||
} catch (Exception $e) {
|
||
error_log("UrlNav: DOMDocument解析失败: " . $e->getMessage());
|
||
}
|
||
|
||
// 方式2：如果DOMDocument失败，使用SimpleXML
|
||
if ($xml === null) {
|
||
error_log("UrlNav: 尝试SimpleXML解析...");
|
||
$xml = simplexml_load_string($content, 'SimpleXMLElement', LIBXML_NOCDATA);
|
||
}
|
||
|
||
if ($xml === false) {
|
||
$errorMsg = 'XML解析失败';
|
||
$xmlErrors = libxml_get_errors();
|
||
if (!empty($xmlErrors)) {
|
||
$errorMsg .= ': ' . $xmlErrors[0]->message;
|
||
error_log("UrlNav: XML错误: " . $xmlErrors[0]->message);
|
||
}
|
||
libxml_clear_errors();
|
||
|
||
if (strpos($content, '<html') !== false) {
|
||
$errorMsg .= ' (可能是HTML页面而非RSS)';
|
||
}
|
||
|
||
throw new Exception($errorMsg);
|
||
}
|
||
|
||
libxml_clear_errors();
|
||
|
||
$feeds = array();
|
||
|
||
// 获取配置（保持原样）
|
||
$config = self::getConfig();
|
||
$fullTextPerSite = intval($config->fullTextPerSite ?? 3);
|
||
$pageFetchTimeout = intval($config->pageFetchTimeout ?? 10);
|
||
$fullTextCount = 0; // 计数器
|
||
|
||
// 检查是否在白名单中（保持原样）
|
||
$selector = self::isInFullTextWhitelist($rssUrl);
|
||
$isInWhitelist = ($selector !== false);
|
||
|
||
error_log("UrlNav: 白名单检查 - 是否在白名单: " . ($isInWhitelist ? '是' : '否') .
|
||
($isInWhitelist ? "，选择器: {$selector}" : ""));
|
||
|
||
// ========== RSS格式解析 ==========
|
||
if (isset($xml->channel) && isset($xml->channel->item)) {
|
||
error_log("UrlNav: 检测到RSS格式 (channel->item)");
|
||
|
||
foreach ($xml->channel->item as $itemIndex => $item) {
|
||
// 🆕 增强：安全处理每个item，防止一个item失败影响全部
|
||
try {
|
||
// 基础内容获取（保持原样）
|
||
$fullContent = '';
|
||
$description = isset($item->description) ? (string)$item->description : '';
|
||
$articleTitle = isset($item->title) ? (string)$item->title : '无标题文章';
|
||
$articleLink = isset($item->link) ? (string)$item->link : '';
|
||
|
||
// 确保标题不为空（保持原样）
|
||
if (empty($articleTitle)) {
|
||
$articleTitle = '未命名文章 ' . date('Y-m-d H:i:s');
|
||
}
|
||
|
||
// 如果链接为空，尝试使用guid（保持原样）
|
||
if (empty($articleLink) && isset($item->guid)) {
|
||
$articleLink = (string)$item->guid;
|
||
}
|
||
|
||
error_log("UrlNav: 处理文章: {$articleTitle}");
|
||
|
||
// 🆕 增强：更好的content:encoded提取（处理WordPress/Typecho）
|
||
$namespaces = $item->getNamespaces(true);
|
||
|
||
// 1. 优先获取content:encoded（WordPress完整内容）
|
||
$encodedContent = '';
|
||
if (isset($namespaces['content'])) {
|
||
$contentNs = $item->children($namespaces['content']);
|
||
if (isset($contentNs->encoded)) {
|
||
$encodedContent = (string)$contentNs->encoded;
|
||
if (!empty($encodedContent) && trim($encodedContent) !== '') {
|
||
$fullContent = $encodedContent;
|
||
error_log("UrlNav: ✓ 找到content:encoded完整内容，长度: " . strlen($fullContent));
|
||
}
|
||
}
|
||
}
|
||
|
||
// 2. 如果没有content:encoded，使用description
|
||
if (empty($fullContent) && !empty($description)) {
|
||
$fullContent = $description;
|
||
error_log("UrlNav: 使用description作为内容，长度: " . strlen($description));
|
||
}
|
||
|
||
// 3. 尝试dc:description命名空间
|
||
if (empty($fullContent) && isset($namespaces['dc'])) {
|
||
$dcNs = $item->children($namespaces['dc']);
|
||
if (isset($dcNs->description) && !empty((string)$dcNs->description)) {
|
||
$fullContent = (string)$dcNs->description;
|
||
error_log("UrlNav: 找到dc:description内容");
|
||
}
|
||
}
|
||
|
||
// 4. 尝试item的直接子元素（保持原样）
|
||
if (empty($fullContent)) {
|
||
foreach ($item->children() as $child) {
|
||
$childName = $child->getName();
|
||
$childContent = (string)$child;
|
||
|
||
// 跳过已知的短字段
|
||
if (in_array($childName, ['title', 'link', 'guid', 'pubDate', 'author', 'category'])) {
|
||
continue;
|
||
}
|
||
|
||
// 如果子元素内容较长，可能是文章内容
|
||
if (strlen($childContent) > 100) {
|
||
$fullContent = $childContent;
|
||
error_log("UrlNav: 从子元素 {$childName} 提取内容");
|
||
break;
|
||
}
|
||
}
|
||
}
|
||
|
||
// ===== 页面抓取判断逻辑（完全保持不变） =====
|
||
$pageContent = null;
|
||
$rssContentLength = strlen($fullContent);
|
||
|
||
// 判断逻辑：只有在白名单中且未超过限制才抓取
|
||
if ($isInWhitelist && $fullTextCount < $fullTextPerSite) {
|
||
$needPageFetch = true;
|
||
$fullTextCount++;
|
||
error_log("UrlNav: 白名单抓取全文 #{$fullTextCount}/{$fullTextPerSite} - {$articleTitle}");
|
||
} else {
|
||
$needPageFetch = false;
|
||
if ($isInWhitelist && $fullTextCount >= $fullTextPerSite) {
|
||
error_log("UrlNav: 已达白名单抓取限制（{$fullTextCount}/{$fullTextPerSite}），跳过");
|
||
} elseif (!$isInWhitelist) {
|
||
error_log("UrlNav: 非白名单网站，使用RSS摘要（{$rssContentLength}字符），不抓取全文");
|
||
}
|
||
}
|
||
|
||
// 执行页面抓取（仅白名单）
|
||
if ($needPageFetch && !empty($articleLink)) {
|
||
// 短暂延迟，避免对服务器压力过大
|
||
if ($itemIndex > 0) {
|
||
usleep(rand(300000, 800000)); // 300-800ms延迟
|
||
}
|
||
|
||
// 使用选择器抓取
|
||
$pageContent = self::fetchFullContentWithSelector($articleLink, $selector, $pageFetchTimeout);
|
||
|
||
if (!empty($pageContent)) {
|
||
$pageLength = strlen($pageContent);
|
||
|
||
if ($pageLength > $rssContentLength + 300) {
|
||
$fullContent = $pageContent;
|
||
error_log("UrlNav: ✓ 页面抓取成功，获得 {$pageLength} 字符内容");
|
||
} elseif ($pageLength > 0) {
|
||
// 合并内容
|
||
$fullContent = $fullContent . "\n\n[页面补充内容]\n" . $pageContent;
|
||
error_log("UrlNav: ✓ 合并页面内容，总长度: " . strlen($fullContent));
|
||
} else {
|
||
error_log("UrlNav: ✗ 页面抓取未获得内容");
|
||
}
|
||
} else {
|
||
error_log("UrlNav: ✗ 页面抓取失败");
|
||
}
|
||
}
|
||
// ===== 页面抓取逻辑结束 =====
|
||
|
||
// 🔴 修改：非白名单网站全文字段处理逻辑
|
||
if (!$isInWhitelist) {
|
||
// 非白名单网站，判断 description 或 content:encoded 是否大于500字
|
||
$descriptionLength = strlen($description);
|
||
$encodedContentLength = strlen($encodedContent);
|
||
|
||
// 只要 description 或 content:encoded 任意一个大于500字就存入全文
|
||
if ($descriptionLength >= 500 || $encodedContentLength >= 500) {
|
||
// 有足够长的内容，存入全文字段
|
||
// 内容清理和截断
|
||
$fullContent = preg_replace('/\s+/', ' ', $fullContent);
|
||
if (strlen($fullContent) > 10000) {
|
||
$fullContent = substr($fullContent, 0, 10000) . '... [内容已截断]';
|
||
error_log("UrlNav: 内容过长，已截断至10000字符");
|
||
}
|
||
error_log("UrlNav: 非白名单网站，description({$descriptionLength})或content:encoded({$encodedContentLength})长度≥500字，存入全文字段");
|
||
} else {
|
||
// 内容太短，留空不存储
|
||
$fullContent = '';
|
||
error_log("UrlNav: 非白名单网站，description({$descriptionLength})和content:encoded({$encodedContentLength})都小于500字，全文字段留空");
|
||
}
|
||
} else {
|
||
// 白名单网站保持原有逻辑
|
||
if (!empty($fullContent)) {
|
||
// 移除过多的空白字符
|
||
$fullContent = preg_replace('/\s+/', ' ', $fullContent);
|
||
|
||
// 截断到合理长度
|
||
if (strlen($fullContent) > 10000) {
|
||
$fullContent = substr($fullContent, 0, 10000) . '... [内容已截断]';
|
||
error_log("UrlNav: 内容过长，已截断至10000字符");
|
||
}
|
||
} else {
|
||
error_log("UrlNav: 警告：未找到任何内容");
|
||
$fullContent = $description;
|
||
}
|
||
}
|
||
|
||
// 获取发布时间（保持原样）
|
||
$pubDate = date('Y-m-d H:i:s', strtotime((string)$item->pubDate));
|
||
|
||
// 获取GUID（保持原样）
|
||
$guid = (string)$item->guid;
|
||
|
||
$feeds[] = array(
|
||
'title' => $articleTitle,
|
||
'link' => $articleLink,
|
||
'description' => $description,
|
||
'full_content' => $fullContent, // 🔴 现在非白名单网站可能为空
|
||
'pubDate' => $pubDate,
|
||
'guid' => $guid
|
||
);
|
||
|
||
error_log("UrlNav: ✓ 文章解析完成: {$articleTitle}");
|
||
|
||
} catch (Exception $e) {
|
||
// 🆕 增强：单个item失败不影响其他item
|
||
error_log("UrlNav: 文章处理失败，跳过: " . $e->getMessage());
|
||
continue;
|
||
}
|
||
}
|
||
}
|
||
// ========== Atom格式解析（保持原样但应用相同逻辑修改） ==========
|
||
elseif (isset($xml->entry) || ($xml->getName() == 'feed' && isset($xml->children('http://www.w3.org/2005/Atom')->entry))) {
|
||
error_log("UrlNav: 检测到Atom格式");
|
||
|
||
// 获取所有entry元素（保持原样）
|
||
$entries = isset($xml->entry) ? $xml->entry : $xml->children('http://www.w3.org/2005/Atom')->entry;
|
||
|
||
foreach ($entries as $entryIndex => $entry) {
|
||
$link = '';
|
||
$title = '';
|
||
$description = '';
|
||
$fullContent = '';
|
||
$pubDate = '';
|
||
$guid = '';
|
||
|
||
// 获取链接（完全保持不变）
|
||
if (isset($entry->link)) {
|
||
foreach ($entry->link as $linkElem) {
|
||
$attributes = $linkElem->attributes();
|
||
if ((string)$attributes['rel'] == 'alternate' || empty((string)$attributes['rel'])) {
|
||
$link = (string)$attributes['href'];
|
||
break;
|
||
}
|
||
}
|
||
}
|
||
|
||
// 如果没有找到链接，使用id作为链接（完全保持不变）
|
||
if (empty($link) && isset($entry->id)) {
|
||
$link = (string)$entry->id;
|
||
}
|
||
|
||
// 获取标题（完全保持不变）
|
||
if (isset($entry->title)) {
|
||
$title = (string)$entry->title;
|
||
}
|
||
|
||
// 获取描述（summary）（完全保持不变）
|
||
if (isset($entry->summary)) {
|
||
$description = (string)$entry->summary;
|
||
}
|
||
|
||
// ===== Atom全文抓取 =====
|
||
// 1. 优先获取content元素（完全保持不变）
|
||
$atomContent = '';
|
||
if (isset($entry->content)) {
|
||
$contentElem = $entry->content;
|
||
$attributes = $contentElem->attributes();
|
||
|
||
// 检查type属性
|
||
$type = (string)($attributes['type'] ?? '');
|
||
|
||
if ($type === 'html' || $type === 'xhtml' || empty($type)) {
|
||
$atomContent = (string)$contentElem;
|
||
$fullContent = $atomContent;
|
||
error_log("UrlNav: 找到Atom content完整内容，类型: {$type}，长度: " . strlen($fullContent));
|
||
} elseif ($type === 'text') {
|
||
$atomContent = htmlspecialchars((string)$contentElem);
|
||
$fullContent = $atomContent;
|
||
error_log("UrlNav: 找到Atom text内容，长度: " . strlen($fullContent));
|
||
}
|
||
}
|
||
|
||
// 2. 如果没有content，尝试summary（完全保持不变）
|
||
if (empty($fullContent) && isset($entry->summary)) {
|
||
$fullContent = $description;
|
||
error_log("UrlNav: 使用Atom summary作为内容，长度: " . strlen($fullContent));
|
||
}
|
||
|
||
// 3. 检查是否有CDATA包裹（完全保持不变）
|
||
if (!empty($fullContent) && strpos($fullContent, '<![CDATA[') !== false) {
|
||
if (preg_match('/<!\[CDATA\[(.*?)\]\]>/s', $fullContent, $matches)) {
|
||
$fullContent = $matches[1];
|
||
error_log("UrlNav: 从CDATA提取Atom内容");
|
||
}
|
||
}
|
||
|
||
// ===== Atom格式的页面抓取判断（完全保持不变） =====
|
||
$pageContent = null;
|
||
$atomContentLength = strlen($fullContent);
|
||
|
||
// 判断逻辑：只有在白名单中且未超过限制才抓取
|
||
if ($isInWhitelist && $fullTextCount < $fullTextPerSite) {
|
||
$needPageFetch = true;
|
||
$fullTextCount++;
|
||
error_log("UrlNav Atom: 白名单抓取全文 #{$fullTextCount}/{$fullTextPerSite} - {$title}");
|
||
} else {
|
||
$needPageFetch = false;
|
||
error_log("UrlNav Atom: " . ($isInWhitelist ? "已达限制" : "非白名单") . "，使用Atom内容（{$atomContentLength}字符）");
|
||
}
|
||
|
||
// 执行Atom页面抓取（仅白名单）
|
||
if ($needPageFetch && !empty($link)) {
|
||
if ($entryIndex > 0) {
|
||
usleep(rand(300000, 800000));
|
||
}
|
||
|
||
$pageContent = self::fetchFullContentWithSelector($link, $selector, $pageFetchTimeout);
|
||
|
||
if (!empty($pageContent) && strlen($pageContent) > $atomContentLength + 300) {
|
||
$fullContent = $pageContent;
|
||
error_log("UrlNav: ✓ Atom页面抓取成功");
|
||
}
|
||
}
|
||
// ===== Atom页面抓取结束 =====
|
||
|
||
// 🔴 修改：Atom格式的非白名单网站全文字段处理
|
||
if (!$isInWhitelist) {
|
||
// 非白名单网站，判断 summary 或 content 是否大于500字
|
||
$descriptionLength = strlen($description);
|
||
$atomContentLength = strlen($atomContent);
|
||
|
||
// 只要 summary 或 content 任意一个大于500字就存入全文
|
||
if ($descriptionLength >= 500 || $atomContentLength >= 500) {
|
||
// 有足够长的内容，存入全文字段
|
||
// 内容截断
|
||
if (strlen($fullContent) > 10000) {
|
||
$fullContent = substr($fullContent, 0, 10000) . '... [内容已截断]';
|
||
error_log("UrlNav: Atom内容过长，已截断");
|
||
}
|
||
error_log("UrlNav Atom: 非白名单网站，summary({$descriptionLength})或content({$atomContentLength})长度≥500字，存入全文字段");
|
||
} else {
|
||
// 内容太短，留空不存储
|
||
$fullContent = '';
|
||
error_log("UrlNav Atom: 非白名单网站，summary({$descriptionLength})和content({$atomContentLength})都小于500字，全文字段留空");
|
||
}
|
||
} else {
|
||
// 白名单网站保持原有逻辑
|
||
// 4. 内容截断（完全保持不变）
|
||
if (!empty($fullContent) && strlen($fullContent) > 10000) {
|
||
$fullContent = substr($fullContent, 0, 10000) . '... [内容已截断]';
|
||
error_log("UrlNav: Atom内容过长，已截断");
|
||
}
|
||
}
|
||
// ===== Atom全文抓取结束 =====
|
||
|
||
// 获取发布时间（updated或published）（完全保持不变）
|
||
if (isset($entry->updated)) {
|
||
$pubDate = date('Y-m-d H:i:s', strtotime((string)$entry->updated));
|
||
} elseif (isset($entry->published)) {
|
||
$pubDate = date('Y-m-d H:i:s', strtotime((string)$entry->published));
|
||
} else {
|
||
$pubDate = date('Y-m-d H:i:s');
|
||
}
|
||
|
||
// 获取guid（id）（完全保持不变）
|
||
if (isset($entry->id)) {
|
||
$guid = (string)$entry->id;
|
||
} else {
|
||
$guid = md5($link . $pubDate);
|
||
}
|
||
|
||
$feeds[] = array(
|
||
'title' => $title,
|
||
'link' => $link,
|
||
'description' => $description,
|
||
'full_content' => $fullContent, // 🔴 现在非白名单网站可能为空
|
||
'pubDate' => $pubDate,
|
||
'guid' => $guid
|
||
);
|
||
}
|
||
}
|
||
// ========== 其他RSS格式解析（保持原样但应用相同逻辑修改） ==========
|
||
elseif (isset($xml->item)) {
|
||
error_log("UrlNav: 检测到RSS格式 (直接item)");
|
||
foreach ($xml->item as $itemIndex => $item) {
|
||
// 优先获取完整内容
|
||
$fullContent = '';
|
||
$description = isset($item->description) ? (string)$item->description : '';
|
||
$articleTitle = (string)$item->title;
|
||
$articleLink = (string)$item->link;
|
||
|
||
// 尝试获取content:encoded（完整内容）
|
||
$encodedContent = '';
|
||
$namespaces = $item->getNamespaces(true);
|
||
if (isset($namespaces['content'])) {
|
||
$contentNs = $item->children($namespaces['content']);
|
||
if (isset($contentNs->encoded)) {
|
||
$encodedContent = (string)$contentNs->encoded;
|
||
$fullContent = $encodedContent;
|
||
error_log("UrlNav: 找到content:encoded完整内容");
|
||
}
|
||
}
|
||
|
||
// 如果没找到content:encoded，使用description
|
||
if (empty($fullContent) && !empty($description)) {
|
||
$fullContent = $description;
|
||
error_log("UrlNav: 使用description作为内容");
|
||
}
|
||
|
||
// ===== 其他格式的页面抓取判断 =====
|
||
$pageContent = null;
|
||
$rssContentLength = strlen($fullContent);
|
||
|
||
// 判断逻辑：只有在白名单中且未超过限制才抓取
|
||
if ($isInWhitelist && $fullTextCount < $fullTextPerSite) {
|
||
$needPageFetch = true;
|
||
$fullTextCount++;
|
||
error_log("UrlNav Other: 白名单抓取全文 #{$fullTextCount}/{$fullTextPerSite} - {$articleTitle}");
|
||
} else {
|
||
$needPageFetch = false;
|
||
}
|
||
|
||
// 页面抓取（仅白名单）
|
||
if ($needPageFetch && !empty($articleLink)) {
|
||
if ($itemIndex > 0) {
|
||
usleep(rand(300000, 800000));
|
||
}
|
||
|
||
$pageContent = self::fetchFullContentWithSelector($articleLink, $selector, $pageFetchTimeout);
|
||
|
||
if (!empty($pageContent) && strlen($pageContent) > strlen($fullContent) + 300) {
|
||
$fullContent = $pageContent;
|
||
}
|
||
}
|
||
// ===== 其他格式页面抓取结束 =====
|
||
|
||
// 🔴 修改：其他格式的非白名单网站全文字段处理
|
||
if (!$isInWhitelist) {
|
||
// 非白名单网站，判断 description 或 content:encoded 是否大于500字
|
||
$descriptionLength = strlen($description);
|
||
$encodedContentLength = strlen($encodedContent);
|
||
|
||
// 只要 description 或 content:encoded 任意一个大于500字就存入全文
|
||
if ($descriptionLength >= 500 || $encodedContentLength >= 500) {
|
||
// 有足够长的内容，存入全文字段
|
||
// 内容截断
|
||
if (!empty($fullContent) && strlen($fullContent) > 10000) {
|
||
$fullContent = substr($fullContent, 0, 10000) . '... [内容已截断]';
|
||
}
|
||
error_log("UrlNav Other: 非白名单网站，description({$descriptionLength})或content:encoded({$encodedContentLength})长度≥500字，存入全文字段");
|
||
} else {
|
||
// 内容太短，留空不存储
|
||
$fullContent = '';
|
||
error_log("UrlNav Other: 非白名单网站，description({$descriptionLength})和content:encoded({$encodedContentLength})都小于500字，全文字段留空");
|
||
}
|
||
} else {
|
||
// 白名单网站保持原有逻辑
|
||
// 内容截断
|
||
if (!empty($fullContent) && strlen($fullContent) > 10000) {
|
||
$fullContent = substr($fullContent, 0, 10000) . '... [内容已截断]';
|
||
}
|
||
}
|
||
|
||
$feeds[] = array(
|
||
'title' => $articleTitle,
|
||
'link' => $articleLink,
|
||
'description' => $description,
|
||
'full_content' => $fullContent, // 🔴 现在非白名单网站可能为空
|
||
'pubDate' => date('Y-m-d H:i:s', strtotime((string)$item->pubDate)),
|
||
'guid' => (string)$item->guid
|
||
);
|
||
}
|
||
}
|
||
// ========== 尝试检测命名空间（保持原样但应用相同逻辑修改） ==========
|
||
else {
|
||
// 检查是否有Atom命名空间
|
||
$namespaces = $xml->getNamespaces(true);
|
||
foreach ($namespaces as $ns) {
|
||
if (strpos($ns, 'www.w3.org/2005/Atom') !== false) {
|
||
$atom = $xml->children($ns);
|
||
if (isset($atom->entry)) {
|
||
error_log("UrlNav: 检测到Atom命名空间格式");
|
||
foreach ($atom->entry as $entryIndex => $entry) {
|
||
$entry = $entry->children($ns);
|
||
|
||
// 获取完整内容
|
||
$fullContent = '';
|
||
$atomContent = '';
|
||
$entryDescription = '';
|
||
|
||
if (isset($entry->content)) {
|
||
$atomContent = (string)$entry->content;
|
||
$fullContent = $atomContent;
|
||
}
|
||
|
||
if (isset($entry->summary)) {
|
||
$entryDescription = (string)$entry->summary;
|
||
if (empty($fullContent)) {
|
||
$fullContent = $entryDescription;
|
||
}
|
||
}
|
||
|
||
// ===== 命名空间格式的页面抓取判断 =====
|
||
$needPageFetch = false;
|
||
$entryLink = isset($entry->link) ? (string)$entry->link : '';
|
||
$rssContentLength = strlen($fullContent);
|
||
|
||
// 判断逻辑：只有在白名单中且未超过限制才抓取
|
||
if ($isInWhitelist && $fullTextCount < $fullTextPerSite) {
|
||
$needPageFetch = true;
|
||
$fullTextCount++;
|
||
}
|
||
|
||
// 页面抓取（仅白名单）
|
||
if ($needPageFetch && !empty($entryLink)) {
|
||
if ($entryIndex > 0) {
|
||
usleep(rand(300000, 800000));
|
||
}
|
||
|
||
$pageContent = self::fetchFullContentWithSelector($entryLink, $selector, $pageFetchTimeout);
|
||
|
||
if (!empty($pageContent)) {
|
||
$fullContent = $pageContent;
|
||
}
|
||
}
|
||
// ===== 命名空间格式页面抓取结束 =====
|
||
|
||
// 🔴 修改：命名空间格式的非白名单网站全文字段处理
|
||
if (!$isInWhitelist) {
|
||
// 非白名单网站，判断 summary 或 content 是否大于500字
|
||
$descriptionLength = strlen($entryDescription);
|
||
$contentLength = strlen($atomContent);
|
||
|
||
// 只要 summary 或 content 任意一个大于500字就存入全文
|
||
if ($descriptionLength >= 500 || $contentLength >= 500) {
|
||
// 有足够长的内容，存入全文字段
|
||
// 内容截断
|
||
if (!empty($fullContent) && strlen($fullContent) > 10000) {
|
||
$fullContent = substr($fullContent, 0, 10000) . '... [内容已截断]';
|
||
}
|
||
error_log("UrlNav Namespace: 非白名单网站，summary({$descriptionLength})或content({$contentLength})长度≥500字，存入全文字段");
|
||
} else {
|
||
// 内容太短，留空不存储
|
||
$fullContent = '';
|
||
error_log("UrlNav Namespace: 非白名单网站，summary({$descriptionLength})和content({$contentLength})都小于500字，全文字段留空");
|
||
}
|
||
} else {
|
||
// 白名单网站保持原有逻辑
|
||
// 内容截断
|
||
if (!empty($fullContent) && strlen($fullContent) > 10000) {
|
||
$fullContent = substr($fullContent, 0, 10000) . '... [内容已截断]';
|
||
}
|
||
}
|
||
|
||
$feeds[] = array(
|
||
'title' => isset($entry->title) ? (string)$entry->title : '',
|
||
'link' => $entryLink,
|
||
'description' => $entryDescription,
|
||
'full_content' => $fullContent, // 🔴 现在非白名单网站可能为空
|
||
'pubDate' => isset($entry->updated) ? date('Y-m-d H:i:s', strtotime((string)$entry->updated)) : date('Y-m-d H:i:s'),
|
||
'guid' => isset($entry->id) ? (string)$entry->id : ''
|
||
);
|
||
}
|
||
break;
|
||
}
|
||
}
|
||
}
|
||
|
||
if (empty($feeds)) {
|
||
error_log("UrlNav: 无法识别的RSS格式");
|
||
throw new Exception('无法识别的RSS格式');
|
||
}
|
||
}
|
||
|
||
if (empty($feeds)) {
|
||
error_log("UrlNav: RSS中没有找到文章内容");
|
||
throw new Exception('RSS中没有找到文章内容');
|
||
}
|
||
|
||
error_log("UrlNav: 找到 " . count($feeds) . " 篇文章");
|
||
error_log("UrlNav: <<< RSS解析成功");
|
||
|
||
return $feeds;
|
||
|
||
} catch (Exception $e) {
|
||
error_log("UrlNav: <<< RSS解析失败: " . $e->getMessage());
|
||
throw new Exception("解析RSS失败 [{$rssUrl}]: " . $e->getMessage());
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 提取CDATA内容（处理多层或不规范CDATA）
|
||
* @param string $content 原始内容
|
||
* @param string $source 来源标识（用于日志）
|
||
* @return string 处理后的内容
|
||
*/
|
||
private static function extractCdataContent($content, $source = '')
|
||
{
|
||
if (empty($content)) {
|
||
return $content;
|
||
}
|
||
|
||
// 如果内容包含CDATA标记
|
||
if (strpos($content, '<![CDATA[') !== false) {
|
||
$cdataCount = 0;
|
||
$originalLength = strlen($content);
|
||
|
||
// 尝试最多5次提取CDATA内容
|
||
while (strpos($content, '<![CDATA[') !== false && $cdataCount < 5) {
|
||
$cdataCount++;
|
||
|
||
// 方法1：正则提取最内层CDATA
|
||
if (preg_match('/<!\[CDATA\[(.*?)\]\]>/s', $content, $matches)) {
|
||
$extracted = $matches[1];
|
||
|
||
// 如果提取的内容明显比原来短，说明CDATA格式正确
|
||
if (strlen($extracted) < strlen($content) * 0.9 && strlen($extracted) > 50) {
|
||
$content = $extracted;
|
||
error_log("UrlNav: 从CDATA提取 {$source} 内容 (第{$cdataCount}次)");
|
||
} else {
|
||
// CDATA可能嵌套或不规范，尝试移除CDATA标记
|
||
$content = str_replace('<![CDATA[', '', $content);
|
||
$content = str_replace(']]>', '', $content);
|
||
error_log("UrlNav: 清理不规范的CDATA标记");
|
||
break;
|
||
}
|
||
} else {
|
||
// CDATA格式不正确，直接移除标记
|
||
$content = str_replace('<![CDATA[', '', $content);
|
||
$content = str_replace(']]>', '', $content);
|
||
error_log("UrlNav: 清理不规范的CDATA标记");
|
||
break;
|
||
}
|
||
}
|
||
|
||
$finalLength = strlen($content);
|
||
if ($originalLength != $finalLength) {
|
||
error_log("UrlNav: CDATA处理完成 {$source}，从 {$originalLength} 到 {$finalLength} 字符");
|
||
}
|
||
}
|
||
|
||
return $content;
|
||
}
|
||
|
||
|
||
/**
|
||
* 从文章页面抓取完整内容
|
||
* @param string $articleUrl 文章链接
|
||
* @param string $title 文章标题（用于日志）
|
||
* @param int $timeout 超时时间（秒）
|
||
* @return string|null 抓取到的内容，失败返回null
|
||
*/
|
||
private static function fetchFullContentFromPage($articleUrl, $title = '', $timeout = 10)
|
||
{
|
||
error_log("UrlNav: 尝试从页面抓取完整内容: {$articleUrl}");
|
||
|
||
try {
|
||
// 设置请求头，模拟浏览器
|
||
$context = stream_context_create([
|
||
'http' => [
|
||
'timeout' => $timeout,
|
||
'ignore_errors' => true,
|
||
'header' => "User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36\r\n" .
|
||
"Accept: text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8\r\n" .
|
||
"Accept-Language: zh-CN,zh;q=0.9,en;q=0.8\r\n" .
|
||
"Accept-Encoding: gzip\r\n" .
|
||
"Connection: close\r\n" .
|
||
"Upgrade-Insecure-Requests: 1",
|
||
'method' => 'GET'
|
||
],
|
||
'ssl' => [
|
||
'verify_peer' => false,
|
||
'verify_peer_name' => false,
|
||
'allow_self_signed' => true
|
||
]
|
||
]);
|
||
|
||
$html = @file_get_contents($articleUrl, false, $context);
|
||
|
||
if ($html === false) {
|
||
$error = error_get_last();
|
||
error_log("UrlNav: 无法访问文章页面: " . ($error['message'] ?? '未知错误'));
|
||
return null;
|
||
}
|
||
|
||
if (empty($html)) {
|
||
error_log("UrlNav: 文章页面内容为空");
|
||
return null;
|
||
}
|
||
|
||
$htmlLength = strlen($html);
|
||
error_log("UrlNav: 获取页面成功，长度: {$htmlLength} 字节");
|
||
|
||
// 转换编码为UTF-8（如果检测到其他编码）
|
||
$encoding = 'UTF-8';
|
||
if (preg_match('/<meta[^>]*charset=["\']?([a-zA-Z0-9\-_]+)["\']?/i', $html, $matches)) {
|
||
$encoding = strtoupper($matches[1]);
|
||
if ($encoding !== 'UTF-8') {
|
||
$html = mb_convert_encoding($html, 'UTF-8', $encoding);
|
||
error_log("UrlNav: 检测到编码 {$encoding}，已转换为UTF-8");
|
||
}
|
||
}
|
||
|
||
// 提取内容
|
||
$fullContent = '';
|
||
|
||
// 方法1：尝试提取Open Graph描述
|
||
if (preg_match('/<meta\s+property="og:description"\s+content="([^"]+)"/i', $html, $matches)) {
|
||
$fullContent = html_entity_decode($matches[1], ENT_QUOTES | ENT_HTML5, 'UTF-8');
|
||
error_log("UrlNav: 从og:description提取内容，长度: " . strlen($fullContent));
|
||
}
|
||
|
||
// 方法2：尝试提取meta description
|
||
if (empty($fullContent) && preg_match('/<meta\s+name="description"\s+content="([^"]+)"/i', $html, $matches)) {
|
||
$fullContent = html_entity_decode($matches[1], ENT_QUOTES | ENT_HTML5, 'UTF-8');
|
||
error_log("UrlNav: 从meta description提取内容，长度: " . strlen($fullContent));
|
||
}
|
||
|
||
// 方法3：尝试提取文章主要内容区域
|
||
if (empty($fullContent) || strlen($fullContent) < 500) {
|
||
// 常见的文章内容CSS类/ID模式
|
||
$contentPatterns = [
|
||
// WordPress主题
|
||
'/<article[^>]*>(.*?)<\/article>/is',
|
||
'/<div\s+class="[^"]*post-content[^"]*"[^>]*>(.*?)<\/div>/is',
|
||
'/<div\s+class="[^"]*entry-content[^"]*"[^>]*>(.*?)<\/div>/is',
|
||
// 通用内容区域
|
||
'/<div\s+class="[^"]*content[^"]*"[^>]*>(.*?)<\/div>/is',
|
||
'/<div\s+id="content"[^>]*>(.*?)<\/div>/is',
|
||
// Typecho主题
|
||
'/<div\s+class="[^"]*post[^"]*"[^>]*>(.*?)<\/div>/is',
|
||
'/<div\s+class="[^"]*post-body[^"]*"[^>]*>(.*?)<\/div>/is',
|
||
// 其他常见模式
|
||
'/<div\s+class="[^"]*article-content[^"]*"[^>]*>(.*?)<\/div>/is',
|
||
'/<div\s+class="[^"]*blog-content[^"]*"[^>]*>(.*?)<\/div>/is',
|
||
'/<div\s+class="[^"]*post-entry[^"]*"[^>]*>(.*?)<\/div>/is'
|
||
];
|
||
|
||
foreach ($contentPatterns as $pattern) {
|
||
if (preg_match($pattern, $html, $matches) && isset($matches[1])) {
|
||
$extracted = $matches[1];
|
||
|
||
// 移除脚本和样式
|
||
$extracted = preg_replace('/<script[^>]*>.*?<\/script>/is', '', $extracted);
|
||
$extracted = preg_replace('/<style[^>]*>.*?<\/style>/is', '', $extracted);
|
||
$extracted = preg_replace('/<!--.*?-->/s', '', $extracted);
|
||
|
||
// 提取纯文本，但保留段落结构
|
||
$extracted = strip_tags($extracted, '<p><br><div><h1><h2><h3><h4><h5><h6><ul><ol><li>');
|
||
$extracted = preg_replace('/\s+/', ' ', $extracted);
|
||
$extracted = trim($extracted);
|
||
|
||
if (strlen($extracted) > 300) {
|
||
$fullContent = $extracted;
|
||
error_log("UrlNav: 从页面提取主要内容，长度: " . strlen($fullContent));
|
||
break;
|
||
}
|
||
}
|
||
}
|
||
}
|
||
|
||
// 方法4：提取所有段落文本
|
||
if (empty($fullContent) || strlen($fullContent) < 500) {
|
||
if (preg_match_all('/<p[^>]*>(.*?)<\/p>/is', $html, $paragraphs)) {
|
||
$allText = '';
|
||
foreach ($paragraphs[1] as $para) {
|
||
$paraText = strip_tags($para);
|
||
$paraText = preg_replace('/\s+/', ' ', $paraText);
|
||
$paraText = trim($paraText);
|
||
|
||
if (strlen($paraText) > 50) { // 只保留有意义的段落
|
||
$allText .= $paraText . "\n\n";
|
||
}
|
||
}
|
||
|
||
if (strlen($allText) > 500) {
|
||
$fullContent = $allText;
|
||
error_log("UrlNav: 从段落提取文本，长度: " . strlen($fullContent));
|
||
}
|
||
}
|
||
}
|
||
|
||
// 方法5：最后尝试提取整个body文本
|
||
if (empty($fullContent) || strlen($fullContent) < 500) {
|
||
if (preg_match('/<body[^>]*>(.*?)<\/body>/is', $html, $matches)) {
|
||
$bodyText = $matches[1];
|
||
|
||
// 清理
|
||
$bodyText = preg_replace('/<script[^>]*>.*?<\/script>/is', '', $bodyText);
|
||
$bodyText = preg_replace('/<style[^>]*>.*?<\/style>/is', '', $bodyText);
|
||
$bodyText = preg_replace('/<!--.*?-->/s', '', $bodyText);
|
||
$bodyText = preg_replace('/<header[^>]*>.*?<\/header>/is', '', $bodyText);
|
||
$bodyText = preg_replace('/<footer[^>]*>.*?<\/footer>/is', '', $bodyText);
|
||
$bodyText = preg_replace('/<nav[^>]*>.*?<\/nav>/is', '', $bodyText);
|
||
|
||
// 提取文本
|
||
$bodyText = strip_tags($bodyText);
|
||
$bodyText = preg_replace('/\s+/', ' ', $bodyText);
|
||
$bodyText = trim($bodyText);
|
||
|
||
if (strlen($bodyText) > 500) {
|
||
// 尝试提取核心部分（去掉导航、页脚等）
|
||
$bodyText = preg_replace('/首页|关于|联系我们|版权|©|Copyright|all rights reserved/i', '', $bodyText);
|
||
$fullContent = $bodyText;
|
||
error_log("UrlNav: 从body提取文本，长度: " . strlen($fullContent));
|
||
}
|
||
}
|
||
}
|
||
|
||
// 内容清理和截断
|
||
if (!empty($fullContent)) {
|
||
// 移除过多的空白字符
|
||
$fullContent = preg_replace('/\s+/', ' ', $fullContent);
|
||
$fullContent = trim($fullContent);
|
||
|
||
// 移除常见噪音
|
||
$noisePatterns = [
|
||
'/分享到.*?(?:微信|微博|QQ|Twitter|Facebook)/',
|
||
'/阅读\s*\d+\s*次/',
|
||
'/点赞\s*\d+\s*次/',
|
||
'/本文由.*?创作/',
|
||
'/版权声明.*/',
|
||
'/转载请注明出处.*/',
|
||
'/相关文章推荐.*/',
|
||
'/你可能也喜欢.*/'
|
||
];
|
||
|
||
foreach ($noisePatterns as $pattern) {
|
||
$fullContent = preg_replace($pattern, '', $fullContent);
|
||
}
|
||
|
||
// 截断到合理长度
|
||
if (strlen($fullContent) > 15000) {
|
||
$fullContent = substr($fullContent, 0, 15000) . '... [页面内容已截断]';
|
||
}
|
||
|
||
$finalLength = strlen($fullContent);
|
||
error_log("UrlNav: 页面抓取成功，获得 {$finalLength} 字符内容");
|
||
return $fullContent;
|
||
}
|
||
|
||
error_log("UrlNav: 无法从页面提取有效内容");
|
||
return null;
|
||
|
||
} catch (Exception $e) {
|
||
error_log("UrlNav: 页面抓取异常: " . $e->getMessage());
|
||
return null;
|
||
}
|
||
}
|
||
|
||
|
||
/**
|
||
* 智能全文抓取（带选择器）
|
||
* @param string $articleUrl 文章链接
|
||
* @param string $selector CSS选择器
|
||
* @param int $timeout 超时时间（秒）
|
||
* @return string|null 抓取到的内容，失败返回null
|
||
*/
|
||
public static function fetchFullContentWithSelector($articleUrl, $selector, $timeout = 10)
|
||
{
|
||
error_log("UrlNav: 使用选择器抓取全文: {$articleUrl} [选择器: {$selector}]");
|
||
|
||
try {
|
||
// 获取页面HTML
|
||
$html = self::fetchHtmlContent($articleUrl, $timeout);
|
||
if (empty($html)) {
|
||
error_log("UrlNav: 无法获取页面HTML");
|
||
return null;
|
||
}
|
||
|
||
$htmlLength = strlen($html);
|
||
error_log("UrlNav: 获取页面成功，长度: {$htmlLength} 字节");
|
||
|
||
// 解析选择器并提取内容
|
||
$content = self::extractContentBySelector($html, $selector);
|
||
|
||
if (!empty($content)) {
|
||
$contentLength = strlen($content);
|
||
error_log("UrlNav: ✓ 使用选择器 {$selector} 抓取成功，长度: {$contentLength}");
|
||
return $content;
|
||
}
|
||
|
||
// 选择器失败，尝试备用方法
|
||
error_log("UrlNav: 选择器 {$selector} 未找到内容，尝试备用方法");
|
||
$fallbackContent = self::extractContentFallback($html);
|
||
|
||
if (!empty($fallbackContent)) {
|
||
$fallbackLength = strlen($fallbackContent);
|
||
error_log("UrlNav: ✓ 备用方法抓取成功，长度: {$fallbackLength}");
|
||
return $fallbackContent;
|
||
}
|
||
|
||
error_log("UrlNav: ✗ 所有提取方法都失败");
|
||
return null;
|
||
|
||
} catch (Exception $e) {
|
||
error_log("UrlNav: 智能抓取失败: " . $e->getMessage());
|
||
return null;
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 获取HTML内容
|
||
* @param string $url 页面URL
|
||
* @param int $timeout 超时时间
|
||
* @return string|null 页面HTML内容
|
||
*/
|
||
private static function fetchHtmlContent($url, $timeout)
|
||
{
|
||
$context = stream_context_create([
|
||
'http' => [
|
||
'timeout' => $timeout,
|
||
'ignore_errors' => true,
|
||
'header' => "User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36\r\n" .
|
||
"Accept: text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8\r\n" .
|
||
"Accept-Language: zh-CN,zh;q=0.9,en;q=0.8\r\n" .
|
||
"Accept-Encoding: gzip\r\n" .
|
||
"Connection: close\r\n" .
|
||
"Upgrade-Insecure-Requests: 1",
|
||
'method' => 'GET'
|
||
],
|
||
'ssl' => [
|
||
'verify_peer' => false,
|
||
'verify_peer_name' => false,
|
||
'allow_self_signed' => true
|
||
]
|
||
]);
|
||
|
||
// 只读取前100KB，减少内存占用
|
||
$content = @file_get_contents($url, false, $context, 0, 102400);
|
||
|
||
if ($content === false) {
|
||
$error = error_get_last();
|
||
error_log("UrlNav: file_get_contents失败: " . ($error['message'] ?? '未知错误'));
|
||
return null;
|
||
}
|
||
|
||
return $content;
|
||
}
|
||
|
||
/**
|
||
* 根据CSS选择器提取内容
|
||
* @param string $html HTML内容
|
||
* @param string $selector CSS选择器（.class, #id, tag）
|
||
* @return string|null 提取的内容
|
||
*/
|
||
private static function extractContentBySelector($html, $selector)
|
||
{
|
||
// 去除首尾空格
|
||
$selector = trim($selector);
|
||
|
||
// 类选择器 .class
|
||
if (strpos($selector, '.') === 0) {
|
||
$className = substr($selector, 1);
|
||
|
||
// 模式1: class="... className ..."
|
||
$pattern1 = '/<([a-zA-Z][a-zA-Z0-9]*)[^>]*\sclass="[^"]*' . preg_quote($className, '/') . '[^"]*"[^>]*>(.*?)<\/\1>/is';
|
||
|
||
// 模式2: class="className"
|
||
$pattern2 = '/<([a-zA-Z][a-zA-Z0-9]*)[^>]*\sclass="' . preg_quote($className, '/') . '"[^>]*>(.*?)<\/\1>/is';
|
||
|
||
// 尝试第一个模式
|
||
if (preg_match($pattern1, $html, $matches)) {
|
||
$content = $matches[2];
|
||
error_log("UrlNav: 使用类选择器模式1找到内容，标签: {$matches[1]}");
|
||
}
|
||
// 尝试第二个模式
|
||
elseif (preg_match($pattern2, $html, $matches)) {
|
||
$content = $matches[2];
|
||
error_log("UrlNav: 使用类选择器模式2找到内容，标签: {$matches[1]}");
|
||
} else {
|
||
error_log("UrlNav: 类选择器 {$selector} 未匹配到内容");
|
||
return null;
|
||
}
|
||
}
|
||
// ID选择器 #id
|
||
elseif (strpos($selector, '#') === 0) {
|
||
$id = substr($selector, 1);
|
||
$pattern = '/<([a-zA-Z][a-zA-Z0-9]*)[^>]*\sid="' . preg_quote($id, '/') . '"[^>]*>(.*?)<\/\1>/is';
|
||
|
||
if (preg_match($pattern, $html, $matches)) {
|
||
$content = $matches[2];
|
||
error_log("UrlNav: 使用ID选择器找到内容，标签: {$matches[1]}");
|
||
} else {
|
||
error_log("UrlNav: ID选择器 {$selector} 未匹配到内容");
|
||
return null;
|
||
}
|
||
}
|
||
// 标签选择器
|
||
else {
|
||
$pattern = '/<' . preg_quote($selector, '/') . '[^>]*>(.*?)<\/' . preg_quote($selector, '/') . '>/is';
|
||
|
||
if (preg_match($pattern, $html, $matches)) {
|
||
$content = $matches[1];
|
||
error_log("UrlNav: 使用标签选择器找到内容");
|
||
} else {
|
||
error_log("UrlNav: 标签选择器 {$selector} 未匹配到内容");
|
||
return null;
|
||
}
|
||
}
|
||
|
||
// 清理内容
|
||
$content = preg_replace('/<script[^>]*>.*?<\/script>/is', '', $content);
|
||
$content = preg_replace('/<style[^>]*>.*?<\/style>/is', '', $content);
|
||
$content = preg_replace('/<!--.*?-->/s', '', $content);
|
||
|
||
// 保留基本的HTML标签
|
||
$content = strip_tags($content, '<p><br><div><span><h1><h2><h3><h4><h5><h6><ul><ol><li><strong><em><b><i><a><img>');
|
||
|
||
// 清理空白字符
|
||
$content = preg_replace('/\s+/', ' ', $content);
|
||
$content = trim($content);
|
||
|
||
// 如果内容太短，可能不是真正的文章内容
|
||
if (strlen($content) < 100) {
|
||
error_log("UrlNav: 选择器提取的内容太短（" . strlen($content) . "字符），可能不是文章内容");
|
||
return null;
|
||
}
|
||
|
||
return $content;
|
||
}
|
||
|
||
/**
|
||
* 备用提取方法（原有的提取逻辑）
|
||
* @param string $html HTML内容
|
||
* @return string|null 提取的内容
|
||
*/
|
||
private static function extractContentFallback($html)
|
||
{
|
||
$content = '';
|
||
|
||
// 方法1：尝试提取Open Graph描述
|
||
if (preg_match('/<meta\s+property="og:description"\s+content="([^"]+)"/i', $html, $matches)) {
|
||
$content = html_entity_decode($matches[1], ENT_QUOTES | ENT_HTML5, 'UTF-8');
|
||
error_log("UrlNav 备用方法: 从og:description提取内容，长度: " . strlen($content));
|
||
}
|
||
|
||
// 方法2：尝试提取meta description
|
||
if (empty($content) && preg_match('/<meta\s+name="description"\s+content="([^"]+)"/i', $html, $matches)) {
|
||
$content = html_entity_decode($matches[1], ENT_QUOTES | ENT_HTML5, 'UTF-8');
|
||
error_log("UrlNav 备用方法: 从meta description提取内容，长度: " . strlen($content));
|
||
}
|
||
|
||
// 方法3：尝试提取常见内容区域
|
||
if (empty($content) || strlen($content) < 500) {
|
||
$contentPatterns = [
|
||
'/<article[^>]*>(.*?)<\/article>/is',
|
||
'/<div\s+class="[^"]*post-content[^"]*"[^>]*>(.*?)<\/div>/is',
|
||
'/<div\s+class="[^"]*entry-content[^"]*"[^>]*>(.*?)<\/div>/is',
|
||
'/<div\s+class="[^"]*content[^"]*"[^>]*>(.*?)<\/div>/is',
|
||
'/<div\s+id="content"[^>]*>(.*?)<\/div>/is',
|
||
'/<div\s+class="[^"]*post[^"]*"[^>]*>(.*?)<\/div>/is',
|
||
'/<div\s+class="[^"]*post-body[^"]*"[^>]*>(.*?)<\/div>/is',
|
||
'/<div\s+class="[^"]*article-content[^"]*"[^>]*>(.*?)<\/div>/is',
|
||
'/<div\s+class="[^"]*blog-content[^"]*"[^>]*>(.*?)<\/div>/is',
|
||
'/<div\s+class="[^"]*post-entry[^"]*"[^>]*>(.*?)<\/div>/is',
|
||
'/<div\s+class="[^"]*markdown-body[^"]*"[^>]*>(.*?)<\/div>/is', // 新增：GitHub风格
|
||
'/<div\s+class="[^"]*article-body[^"]*"[^>]*>(.*?)<\/div>/is', // 新增：通用文章体
|
||
'/<main[^>]*>(.*?)<\/main>/is', // 新增：main标签
|
||
'/<section[^>]*>(.*?)<\/section>/is' // 新增：section标签
|
||
];
|
||
|
||
foreach ($contentPatterns as $pattern) {
|
||
if (preg_match($pattern, $html, $matches) && isset($matches[1])) {
|
||
$extracted = $matches[1];
|
||
|
||
// 清理
|
||
$extracted = preg_replace('/<script[^>]*>.*?<\/script>/is', '', $extracted);
|
||
$extracted = preg_replace('/<style[^>]*>.*?<\/style>/is', '', $extracted);
|
||
$extracted = preg_replace('/<!--.*?-->/s', '', $extracted);
|
||
|
||
// 提取纯文本，保留段落结构
|
||
$extracted = strip_tags($extracted, '<p><br><div><h1><h2><h3><h4><h5><h6><ul><ol><li>');
|
||
$extracted = preg_replace('/\s+/', ' ', $extracted);
|
||
$extracted = trim($extracted);
|
||
|
||
if (strlen($extracted) > 300) {
|
||
$content = $extracted;
|
||
error_log("UrlNav 备用方法: 从页面提取主要内容，模式: " . substr($pattern, 0, 50) . "...，长度: " . strlen($content));
|
||
break;
|
||
}
|
||
}
|
||
}
|
||
}
|
||
|
||
// 方法4：提取所有段落文本
|
||
if (empty($content) || strlen($content) < 500) {
|
||
if (preg_match_all('/<p[^>]*>(.*?)<\/p>/is', $html, $paragraphs)) {
|
||
$allText = '';
|
||
foreach ($paragraphs[1] as $para) {
|
||
$paraText = strip_tags($para);
|
||
$paraText = preg_replace('/\s+/', ' ', $paraText);
|
||
$paraText = trim($paraText);
|
||
|
||
if (strlen($paraText) > 50) {
|
||
$allText .= $paraText . "\n\n";
|
||
}
|
||
}
|
||
|
||
if (strlen($allText) > 500) {
|
||
$content = $allText;
|
||
error_log("UrlNav 备用方法: 从段落提取文本，长度: " . strlen($content));
|
||
}
|
||
}
|
||
}
|
||
|
||
// 方法5：提取整个body文本
|
||
if (empty($content) || strlen($content) < 500) {
|
||
if (preg_match('/<body[^>]*>(.*?)<\/body>/is', $html, $matches)) {
|
||
$bodyText = $matches[1];
|
||
|
||
// 清理
|
||
$bodyText = preg_replace('/<script[^>]*>.*?<\/script>/is', '', $bodyText);
|
||
$bodyText = preg_replace('/<style[^>]*>.*?<\/style>/is', '', $bodyText);
|
||
$bodyText = preg_replace('/<!--.*?-->/s', '', $bodyText);
|
||
$bodyText = preg_replace('/<header[^>]*>.*?<\/header>/is', '', $bodyText);
|
||
$bodyText = preg_replace('/<footer[^>]*>.*?<\/footer>/is', '', $bodyText);
|
||
$bodyText = preg_replace('/<nav[^>]*>.*?<\/nav>/is', '', $bodyText);
|
||
|
||
// 提取文本
|
||
$bodyText = strip_tags($bodyText);
|
||
$bodyText = preg_replace('/\s+/', ' ', $bodyText);
|
||
$bodyText = trim($bodyText);
|
||
|
||
if (strlen($bodyText) > 500) {
|
||
// 尝试提取核心部分（去掉导航、页脚等）
|
||
$bodyText = preg_replace('/首页|关于|联系我们|版权|©|Copyright|all rights reserved/i', '', $bodyText);
|
||
$content = $bodyText;
|
||
error_log("UrlNav 备用方法: 从body提取文本，长度: " . strlen($content));
|
||
}
|
||
}
|
||
}
|
||
|
||
// 内容清理
|
||
if (!empty($content)) {
|
||
// 移除过多的空白字符
|
||
$content = preg_replace('/\s+/', ' ', $content);
|
||
$content = trim($content);
|
||
|
||
// 移除常见噪音
|
||
$noisePatterns = [
|
||
'/分享到.*?(?:微信|微博|QQ|Twitter|Facebook)/',
|
||
'/阅读\s*\d+\s*次/',
|
||
'/点赞\s*\d+\s*次/',
|
||
'/本文由.*?创作/',
|
||
'/版权声明.*/',
|
||
'/转载请注明出处.*/',
|
||
'/相关文章推荐.*/',
|
||
'/你可能也喜欢.*/'
|
||
];
|
||
|
||
foreach ($noisePatterns as $pattern) {
|
||
$content = preg_replace($pattern, '', $content);
|
||
}
|
||
|
||
// 截断到合理长度
|
||
if (strlen($content) > 15000) {
|
||
$content = substr($content, 0, 15000) . '... [页面内容已截断]';
|
||
}
|
||
|
||
$finalLength = strlen($content);
|
||
error_log("UrlNav 备用方法: 提取成功，获得 {$finalLength} 字符内容");
|
||
return $content;
|
||
}
|
||
|
||
error_log("UrlNav 备用方法: 无法从页面提取有效内容");
|
||
return null;
|
||
}
|
||
|
||
/**
|
||
* 解析全文抓取白名单
|
||
* @return array 白名单数组 [rss_url => selector]
|
||
*/
|
||
public static function parseFullTextWhitelist()
|
||
{
|
||
$config = self::getConfig();
|
||
|
||
if (empty($config->fullTextWhitelist)) {
|
||
return [];
|
||
}
|
||
|
||
$whitelist = [];
|
||
$lines = explode("\n", trim($config->fullTextWhitelist));
|
||
|
||
foreach ($lines as $line) {
|
||
$line = trim($line);
|
||
if (empty($line) || strpos($line, '|') === false) {
|
||
continue;
|
||
}
|
||
|
||
list($rssUrl, $selector) = explode('|', $line, 2);
|
||
$rssUrl = trim($rssUrl);
|
||
$selector = trim($selector);
|
||
|
||
if (!empty($rssUrl) && !empty($selector)) {
|
||
$whitelist[$rssUrl] = $selector;
|
||
}
|
||
}
|
||
|
||
// 记录白名单数量用于调试
|
||
error_log("UrlNav: 解析到 " . count($whitelist) . " 个白名单条目");
|
||
foreach ($whitelist as $url => $selector) {
|
||
error_log("UrlNav: 白名单 - {$url} => {$selector}");
|
||
}
|
||
|
||
return $whitelist;
|
||
}
|
||
|
||
/**
|
||
* 检查RSS源是否在全文抓取白名单中
|
||
* @param string $rssUrl RSS地址
|
||
* @return string|false 返回选择器，不在白名单返回false
|
||
*/
|
||
public static function isInFullTextWhitelist($rssUrl)
|
||
{
|
||
static $whitelist = null;
|
||
|
||
if ($whitelist === null) {
|
||
$whitelist = self::parseFullTextWhitelist();
|
||
}
|
||
|
||
// 精确匹配
|
||
if (isset($whitelist[$rssUrl])) {
|
||
error_log("UrlNav: {$rssUrl} 在白名单中，选择器: " . $whitelist[$rssUrl]);
|
||
return $whitelist[$rssUrl];
|
||
}
|
||
|
||
// 通配符匹配（支持简单的通配符）
|
||
foreach ($whitelist as $pattern => $selector) {
|
||
// 简单的通配符匹配：* 匹配任意字符
|
||
if (strpos($pattern, '*') !== false) {
|
||
$regex = str_replace('\*', '.*', preg_quote($pattern, '/'));
|
||
if (preg_match('/^' . $regex . '$/', $rssUrl)) {
|
||
error_log("UrlNav: {$rssUrl} 匹配通配符模式 {$pattern}，选择器: {$selector}");
|
||
return $selector;
|
||
}
|
||
}
|
||
}
|
||
|
||
return false;
|
||
}
|
||
|
||
/**
|
||
* 修复常见的XML问题
|
||
*/
|
||
private static function fixCommonXmlIssues($content)
|
||
{
|
||
// 1. 移除非法控制字符
|
||
$content = preg_replace('/[\x00-\x08\x0B\x0C\x0E-\x1F\x7F]/', '', $content);
|
||
|
||
// 2. 修复未关闭的标签（简单修复）
|
||
$content = str_replace('<br>', '<br/>', $content);
|
||
$content = str_replace('<hr>', '<hr/>', $content);
|
||
|
||
// 3. 修复未转义的&符号
|
||
$content = preg_replace('/&(?!(amp|lt|gt|quot|apos|#\d+);)/', '&', $content);
|
||
|
||
// 4. 移除多余的空白字符
|
||
$content = preg_replace('/>\s+</', '><', $content);
|
||
|
||
return $content;
|
||
}
|
||
|
||
/**
|
||
* 手动刷新RSS（供管理界面调用）
|
||
*/
|
||
public static function manualRefreshRss()
|
||
{
|
||
return self::refreshAllRssFeeds(false);
|
||
}
|
||
/**
|
||
* 记录刷新日志 - 增强版，自动处理缺失字段
|
||
*/
|
||
private static function logRefresh($type, $successCount, $totalFeeds, $urlCount, $newArticles,
|
||
$errorMessage = null, $duration = 0, $cronType = 'rss',
|
||
$successRssUrls = null, $failedRssUrls = null)
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
// 🔴 新增：构建详细信息
|
||
$details = array();
|
||
if ($successRssUrls !== null) {
|
||
$details['success_rss_urls'] = $successRssUrls;
|
||
}
|
||
if ($failedRssUrls !== null) {
|
||
$details['failed_rss_urls'] = $failedRssUrls;
|
||
}
|
||
|
||
// 🔴 新增：构建详细消息
|
||
$message = "刷新完成：成功 {$successCount} 个，失败 " . ($urlCount - $successCount) . " 个";
|
||
|
||
// 如果有成功和失败的RSS地址列表，添加到消息中
|
||
if ($successRssUrls !== null && !empty($successRssUrls)) {
|
||
// 截取部分RSS地址显示（避免过长）
|
||
$shortSuccessUrls = array_map(function($url) {
|
||
if (strlen($url) > 50) {
|
||
return substr($url, 0, 50) . '...';
|
||
}
|
||
return $url;
|
||
}, array_slice($successRssUrls, 0, 5)); // 最多显示5个
|
||
|
||
$successList = implode('; ', $shortSuccessUrls);
|
||
if (count($successRssUrls) > 5) {
|
||
$successList .= '... (共' . count($successRssUrls) . '个)';
|
||
}
|
||
$message .= "\n成功的RSS: " . $successList;
|
||
}
|
||
|
||
if ($failedRssUrls !== null && !empty($failedRssUrls)) {
|
||
// 截取部分RSS地址显示（避免过长）
|
||
$shortFailedUrls = array_map(function($url) {
|
||
if (strlen($url) > 50) {
|
||
return substr($url, 0, 50) . '...';
|
||
}
|
||
return $url;
|
||
}, array_slice($failedRssUrls, 0, 5)); // 最多显示5个
|
||
|
||
$failedList = implode('; ', $shortFailedUrls);
|
||
if (count($failedRssUrls) > 5) {
|
||
$failedList .= '... (共' . count($failedRssUrls) . '个)';
|
||
}
|
||
$message .= "\n失败的RSS: " . $failedList;
|
||
}
|
||
|
||
$detailsJson = !empty($details) ? json_encode($details, JSON_UNESCAPED_UNICODE) : null;
|
||
|
||
// 方法1：首先尝试使用完整字段插入
|
||
try {
|
||
// 检查字段是否存在
|
||
$tableInfo = $db->query("PRAGMA table_info(urlnav_refresh_log)");
|
||
$columns = $tableInfo->fetchAll(PDO::FETCH_ASSOC);
|
||
|
||
$hasMessage = false;
|
||
$hasDetails = false;
|
||
foreach ($columns as $column) {
|
||
if ($column['name'] === 'message') {
|
||
$hasMessage = true;
|
||
}
|
||
if ($column['name'] === 'details') {
|
||
$hasDetails = true;
|
||
}
|
||
}
|
||
|
||
if ($hasMessage && $hasDetails) {
|
||
// 如果两个字段都存在，使用完整插入
|
||
$stmt = $db->prepare("
|
||
INSERT INTO urlnav_refresh_log
|
||
(refresh_type, success_count, total_feeds, url_count, new_articles, error_message,
|
||
refresh_time, duration, cron_type, message, details)
|
||
VALUES (?, ?, ?, ?, ?, ?, CURRENT_TIMESTAMP, ?, ?, ?, ?)
|
||
");
|
||
|
||
$stmt->execute(array(
|
||
$type,
|
||
$successCount,
|
||
$totalFeeds,
|
||
$urlCount,
|
||
$newArticles,
|
||
$errorMessage,
|
||
$duration,
|
||
$cronType,
|
||
$message,
|
||
$detailsJson
|
||
));
|
||
|
||
error_log("UrlNav: 刷新日志记录成功（使用完整字段）");
|
||
|
||
} elseif ($hasMessage) {
|
||
// 只有message字段存在
|
||
$stmt = $db->prepare("
|
||
INSERT INTO urlnav_refresh_log
|
||
(refresh_type, success_count, total_feeds, url_count, new_articles, error_message,
|
||
refresh_time, duration, cron_type, message)
|
||
VALUES (?, ?, ?, ?, ?, ?, CURRENT_TIMESTAMP, ?, ?, ?)
|
||
");
|
||
|
||
$stmt->execute(array(
|
||
$type,
|
||
$successCount,
|
||
$totalFeeds,
|
||
$urlCount,
|
||
$newArticles,
|
||
$errorMessage,
|
||
$duration,
|
||
$cronType,
|
||
$message
|
||
));
|
||
|
||
error_log("UrlNav: 刷新日志记录成功（使用message字段）");
|
||
|
||
} else {
|
||
// 两个字段都不存在，使用旧的方式
|
||
$stmt = $db->prepare("
|
||
INSERT INTO urlnav_refresh_log
|
||
(refresh_type, success_count, total_feeds, url_count, new_articles, error_message,
|
||
refresh_time, duration, cron_type)
|
||
VALUES (?, ?, ?, ?, ?, ?, CURRENT_TIMESTAMP, ?, ?)
|
||
");
|
||
|
||
$stmt->execute(array(
|
||
$type,
|
||
$successCount,
|
||
$totalFeeds,
|
||
$urlCount,
|
||
$newArticles,
|
||
$errorMessage,
|
||
$duration,
|
||
$cronType
|
||
));
|
||
|
||
error_log("UrlNav: 刷新日志记录成功（旧方式）");
|
||
}
|
||
|
||
} catch (Exception $e) {
|
||
// 如果字段缺失，尝试修复并重试
|
||
if (strpos($e->getMessage(), 'has no column named') !== false) {
|
||
error_log("UrlNav: 检测到字段缺失，尝试修复: " . $e->getMessage());
|
||
|
||
// 运行数据库迁移
|
||
self::migrateDatabase();
|
||
|
||
// 重试插入（使用更简单的SQL）
|
||
try {
|
||
$stmt = $db->prepare("
|
||
INSERT INTO urlnav_refresh_log
|
||
(refresh_type, success_count, total_feeds, url_count, error_message, refresh_time, duration, cron_type)
|
||
VALUES (?, ?, ?, ?, ?, CURRENT_TIMESTAMP, ?, ?)
|
||
");
|
||
|
||
$stmt->execute(array(
|
||
$type,
|
||
$successCount,
|
||
$totalFeeds,
|
||
$urlCount,
|
||
$errorMessage,
|
||
$duration,
|
||
$cronType
|
||
));
|
||
|
||
error_log("UrlNav: 刷新日志记录成功（修复后）");
|
||
} catch (Exception $retryError) {
|
||
// 如果还不行，尝试更简化的插入
|
||
try {
|
||
$stmt = $db->prepare("
|
||
INSERT INTO urlnav_refresh_log
|
||
(refresh_type, success_count, refresh_time)
|
||
VALUES (?, ?, CURRENT_TIMESTAMP)
|
||
");
|
||
|
||
$stmt->execute(array($type, $successCount));
|
||
error_log("UrlNav: 刷新日志记录成功（最小字段）");
|
||
} catch (Exception $minimalError) {
|
||
error_log("UrlNav: 最小字段插入也失败: " . $minimalError->getMessage());
|
||
}
|
||
}
|
||
} else {
|
||
// 其他错误，直接抛出
|
||
throw $e;
|
||
}
|
||
}
|
||
|
||
// 如果需要，记录到error_log中
|
||
error_log("UrlNav: 刷新统计 - 类型: {$type}, 成功: {$successCount}, 总文章: {$totalFeeds}, 网址数: {$urlCount}, 新文章: {$newArticles}, 耗时: {$duration}秒");
|
||
if ($successRssUrls !== null && !empty($successRssUrls)) {
|
||
error_log("UrlNav: 成功的RSS地址: " . implode(', ', array_slice($successRssUrls, 0, 3)));
|
||
}
|
||
if ($failedRssUrls !== null && !empty($failedRssUrls)) {
|
||
error_log("UrlNav: 失败的RSS地址: " . implode(', ', array_slice($failedRssUrls, 0, 3)));
|
||
}
|
||
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 记录刷新日志失败: ' . $e->getMessage());
|
||
|
||
// 终极备用方案：只记录到错误日志
|
||
error_log("UrlNav: 刷新统计（无法写入数据库） - 类型: {$type}, 成功: {$successCount}, 总文章: {$totalFeeds}, 网址数: {$urlCount}, 新文章: {$newArticles}, 耗时: {$duration}秒");
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 记录定时任务日志
|
||
*/
|
||
private static function logCron($type, $result)
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
$errorMessage = null;
|
||
if (is_array($result)) {
|
||
if (isset($result['error'])) {
|
||
$errorMessage = $result['error'];
|
||
}
|
||
$result = json_encode($result);
|
||
}
|
||
|
||
$stmt = $db->prepare("
|
||
INSERT INTO urlnav_cron_log
|
||
(cron_type, result, error_message, executed_time)
|
||
VALUES (?, ?, ?, CURRENT_TIMESTAMP)
|
||
");
|
||
|
||
$stmt->execute(array(
|
||
$type,
|
||
$result,
|
||
$errorMessage
|
||
));
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 记录定时任务日志失败: ' . $e->getMessage());
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 验证定时任务密钥
|
||
*/
|
||
public static function validateCronSecret($secret, $type = 'rss')
|
||
{
|
||
$config = self::getConfig();
|
||
if ($type === 'rss') {
|
||
return $secret === $config->rssCronSecret;
|
||
} elseif ($type === 'status') {
|
||
return $secret === $config->statusCronSecret;
|
||
}
|
||
return false;
|
||
}
|
||
|
||
/**
|
||
* 执行公开的RSS定时任务 - 防502版本
|
||
*/
|
||
public static function executePublicRssCron($secret = null)
|
||
{
|
||
// 关键：立即设置响应头并输出内容
|
||
if (!headers_sent()) {
|
||
header('Content-Type: application/json; charset=utf-8');
|
||
header('Cache-Control: no-cache, no-store, must-revalidate');
|
||
header('Pragma: no-cache');
|
||
header('Expires: 0');
|
||
|
||
// 立即输出JSON开始标记，让Nginx知道脚本在运行
|
||
echo '{"status":"processing","message":"任务开始执行...","timestamp":' . time() . '}';
|
||
flush();
|
||
ob_flush();
|
||
}
|
||
|
||
error_log('UrlNav RSS Cron: 请求开始 ' . date('Y-m-d H:i:s'));
|
||
|
||
try {
|
||
// 验证密钥
|
||
if ($secret && !self::validateCronSecret($secret, 'rss')) {
|
||
error_log('UrlNav RSS Cron: 无效的Cron密钥');
|
||
return array(
|
||
'success' => false,
|
||
'message' => '无效的访问密钥',
|
||
'timestamp' => time()
|
||
);
|
||
}
|
||
|
||
// 执行定时任务
|
||
$result = self::executeRssCronTask();
|
||
|
||
// 最终输出完整结果
|
||
if (!headers_sent()) {
|
||
echo json_encode($result);
|
||
}
|
||
|
||
return $result;
|
||
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav RSS Cron: 异常: ' . $e->getMessage());
|
||
$errorResult = array(
|
||
'success' => false,
|
||
'error' => $e->getMessage(),
|
||
'timestamp' => time(),
|
||
'message' => 'RSS定时任务执行失败'
|
||
);
|
||
|
||
if (!headers_sent()) {
|
||
echo json_encode($errorResult);
|
||
}
|
||
|
||
return $errorResult;
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 执行公开的状态检查定时任务 - 无锁版
|
||
*/
|
||
public static function executePublicStatusCron($secret = null)
|
||
{
|
||
// 立即设置响应头，避免502
|
||
if (!headers_sent()) {
|
||
header('Content-Type: application/json; charset=utf-8');
|
||
header('Cache-Control: no-cache, no-store, must-revalidate');
|
||
header('Pragma: no-cache');
|
||
header('Expires: 0');
|
||
}
|
||
|
||
error_log('UrlNav Status Cron: 请求 ' . date('Y-m-d H:i:s'));
|
||
|
||
try {
|
||
// 验证密钥
|
||
if ($secret && !self::validateCronSecret($secret, 'status')) {
|
||
error_log('UrlNav Status Cron: 无效的Cron密钥');
|
||
return array(
|
||
'success' => false,
|
||
'message' => '无效的访问密钥',
|
||
'timestamp' => time()
|
||
);
|
||
}
|
||
|
||
// 直接执行定时任务，不进行锁检查
|
||
$result = self::executeStatusCronTask();
|
||
|
||
return $result;
|
||
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav Status Cron: 异常: ' . $e->getMessage());
|
||
return array(
|
||
'success' => false,
|
||
'error' => $e->getMessage(),
|
||
'timestamp' => time(),
|
||
'message' => '状态检查定时任务执行失败'
|
||
);
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 获取解锁URL
|
||
*/
|
||
private static function getUnlockUrl($secret, $type = 'rss')
|
||
{
|
||
$options = Typecho_Widget::widget('Widget_Options');
|
||
$baseUrl = $options->siteUrl;
|
||
|
||
if (substr($baseUrl, -1) == '/') {
|
||
$baseUrl = substr($baseUrl, 0, -1);
|
||
}
|
||
|
||
return $baseUrl . '/action/urlnav?do=unlockCron&type=' . $type . '&secret=' . urlencode($secret);
|
||
}
|
||
|
||
private static function cleanExpiredCache()
|
||
{
|
||
try {
|
||
$config = self::getConfig();
|
||
$keepTime = intval($config->rssKeepTime ?? 0); // ← 默认值改为 0
|
||
|
||
// 如果设置为0，则不清理（默认行为）
|
||
if ($keepTime <= 0) {
|
||
return 0;
|
||
}
|
||
|
||
$db = self::getDbConnection();
|
||
|
||
// 计算清理时间点 - 按照文章发布时间
|
||
$cutoffTime = date('Y-m-d H:i:s', time() - $keepTime);
|
||
|
||
// 先获取需要清理的文章数量用于日志
|
||
$countStmt = $db->prepare("
|
||
SELECT COUNT(*) as count
|
||
FROM urlnav_rss_cache
|
||
WHERE pub_date < ?
|
||
AND is_fresh = 1
|
||
AND id NOT IN (
|
||
SELECT DISTINCT feed_id FROM urlnav_favorites
|
||
)
|
||
");
|
||
$countStmt->execute([$cutoffTime]);
|
||
$totalCount = $countStmt->fetchColumn();
|
||
|
||
if ($totalCount == 0) {
|
||
return 0;
|
||
}
|
||
|
||
// 标记过期的文章为非新鲜状态
|
||
$stmt = $db->prepare("
|
||
UPDATE urlnav_rss_cache
|
||
SET is_fresh = 0
|
||
WHERE pub_date < ?
|
||
AND is_fresh = 1
|
||
AND id NOT IN (
|
||
SELECT DISTINCT feed_id FROM urlnav_favorites
|
||
)
|
||
");
|
||
$stmt->execute([$cutoffTime]);
|
||
|
||
$markedCount = $stmt->rowCount();
|
||
|
||
if ($markedCount > 0) {
|
||
$days = round($keepTime / 86400, 1);
|
||
error_log("UrlNav: 自动清理 - 标记了 {$markedCount} 条超过{$days}天的文章为非新鲜状态");
|
||
}
|
||
|
||
return $markedCount;
|
||
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 清理过期文章失败: ' . $e->getMessage());
|
||
return 0;
|
||
}
|
||
}
|
||
|
||
|
||
public static function cleanAllRssCache()
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
// 清理所有非新鲜缓存（排除已收藏的文章）
|
||
$stmt = $db->prepare("
|
||
DELETE FROM urlnav_rss_cache
|
||
WHERE is_fresh = 0
|
||
AND id NOT IN (
|
||
SELECT DISTINCT feed_id FROM urlnav_favorites
|
||
)
|
||
");
|
||
$stmt->execute();
|
||
|
||
$deletedCount = $stmt->rowCount();
|
||
|
||
// 获取当前保留时间配置用于日志
|
||
$config = self::getConfig();
|
||
$keepTime = intval($config->rssKeepTime ?? 0);
|
||
|
||
if ($keepTime > 0) {
|
||
$days = round($keepTime / 86400, 1);
|
||
$logMsg = "清理了 {$deletedCount} 条非新鲜RSS文章（当前保留策略：{$days}天内的文章保持新鲜）";
|
||
} else {
|
||
$logMsg = "清理了 {$deletedCount} 条非新鲜RSS文章（当前保留策略：不自动清理，仅清理已标记文章）";
|
||
}
|
||
|
||
// 重置所有URL的统计信息
|
||
$db->exec("UPDATE urlnav_urls SET
|
||
refresh_count = 0,
|
||
success_count = 0,
|
||
failure_count = 0,
|
||
last_error = NULL,
|
||
last_refresh = NULL
|
||
");
|
||
|
||
if ($deletedCount > 0) {
|
||
error_log("UrlNav: {$logMsg}");
|
||
}
|
||
|
||
return $deletedCount;
|
||
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 清理所有缓存失败: ' . $e->getMessage());
|
||
throw $e;
|
||
}
|
||
}
|
||
/**
|
||
* 获取定时任务日志 - 修复版，确保RSS地址信息正确返回
|
||
*/
|
||
public static function getCronLogs($type = null, $limit = 20)
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
$sql = "SELECT * FROM urlnav_cron_log";
|
||
$params = [];
|
||
|
||
if ($type !== null) {
|
||
$sql .= " WHERE cron_type LIKE ?";
|
||
$params[] = $type . '%';
|
||
}
|
||
|
||
$sql .= " ORDER BY executed_time DESC LIMIT ?";
|
||
$params[] = $limit;
|
||
|
||
$stmt = $db->prepare($sql);
|
||
$stmt->execute($params);
|
||
$logs = $stmt->fetchAll(PDO::FETCH_ASSOC);
|
||
|
||
// 🆕 关键修复：确保每个日志项的result字段包含RSS地址信息
|
||
foreach ($logs as &$log) {
|
||
if (!empty($log['result'])) {
|
||
try {
|
||
$result = json_decode($log['result'], true);
|
||
if ($result) {
|
||
// 🔴 检查是否是RSS相关日志
|
||
$isRssLog = $log['cron_type'] && (
|
||
strpos($log['cron_type'], 'rss') !== false ||
|
||
strpos($log['cron_type'], 'refresh') !== false
|
||
);
|
||
|
||
if ($isRssLog) {
|
||
// 确保RSS地址字段存在
|
||
if (!isset($result['successRssUrls'])) {
|
||
$result['successRssUrls'] = [];
|
||
}
|
||
if (!isset($result['failedRssUrls'])) {
|
||
$result['failedRssUrls'] = [];
|
||
}
|
||
|
||
// 重新编码为JSON
|
||
$log['result'] = json_encode($result, JSON_UNESCAPED_UNICODE);
|
||
}
|
||
}
|
||
} catch (Exception $e) {
|
||
// JSON解析失败，保持原样
|
||
error_log('UrlNav: 解析cron_logs的result字段失败: ' . $e->getMessage());
|
||
}
|
||
}
|
||
}
|
||
|
||
return $logs;
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 获取定时任务日志失败: ' . $e->getMessage());
|
||
return array();
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 获取定时任务统计
|
||
*/
|
||
public static function getCronStats($type = null)
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
// 构建查询条件
|
||
$whereClause = "";
|
||
$params = [];
|
||
|
||
if ($type !== null) {
|
||
$whereClause = " WHERE cron_type LIKE ?";
|
||
$params[] = $type . '%';
|
||
}
|
||
|
||
// 获取总执行次数
|
||
$stmt = $db->prepare("SELECT COUNT(*) as total FROM urlnav_cron_log $whereClause");
|
||
$stmt->execute($params);
|
||
$totalResult = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
$total = $totalResult['total'] ?? 0;
|
||
|
||
// 获取成功次数
|
||
$stmt = $db->prepare("SELECT COUNT(*) as success FROM urlnav_cron_log $whereClause AND (error_message IS NULL OR error_message = '')");
|
||
$stmt->execute($params);
|
||
$successResult = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
$success = $successResult['success'] ?? 0;
|
||
|
||
// 获取失败次数
|
||
$stmt = $db->prepare("SELECT COUNT(*) as failed FROM urlnav_cron_log $whereClause AND error_message IS NOT NULL AND error_message != ''");
|
||
$stmt->execute($params);
|
||
$failedResult = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
$failed = $failedResult['failed'] ?? 0;
|
||
|
||
// 获取最后执行时间
|
||
$stmt = $db->prepare("SELECT MAX(executed_time) as last_executed FROM urlnav_cron_log $whereClause");
|
||
$stmt->execute($params);
|
||
$lastResult = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
$lastExecuted = $lastResult['last_executed'] ?? null;
|
||
|
||
// 获取各种类型的统计
|
||
$stmt = $db->prepare("SELECT cron_type, COUNT(*) as count FROM urlnav_cron_log GROUP BY cron_type ORDER BY cron_type");
|
||
$typeResults = $stmt->fetchAll(PDO::FETCH_ASSOC);
|
||
$typeStats = array();
|
||
|
||
foreach ($typeResults as $row) {
|
||
$typeStats[$row['cron_type']] = $row['count'];
|
||
}
|
||
|
||
return array(
|
||
'total' => $total,
|
||
'success' => $success,
|
||
'failed' => $failed,
|
||
'last_executed' => $lastExecuted,
|
||
'type_stats' => $typeStats,
|
||
'success_rate' => $total > 0 ? round(($success / $total) * 100, 2) : 0
|
||
);
|
||
|
||
} catch (Exception $e) {
|
||
return array(
|
||
'total' => 0,
|
||
'success' => 0,
|
||
'failed' => 0,
|
||
'last_executed' => null,
|
||
'type_stats' => array(),
|
||
'success_rate' => 0
|
||
);
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 获取RSS定时任务统计
|
||
*/
|
||
public static function getRssCronStats()
|
||
{
|
||
return self::getCronStats('rss');
|
||
}
|
||
|
||
/**
|
||
* 获取状态检查定时任务统计
|
||
*/
|
||
public static function getStatusCronStats()
|
||
{
|
||
return self::getCronStats('status');
|
||
}
|
||
|
||
/**
|
||
* 获取有RSS地址的网址总数
|
||
*/
|
||
public static function getTotalUrlsWithRss()
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
$stmt = $db->query("
|
||
SELECT COUNT(*) as total FROM urlnav_urls
|
||
WHERE is_active = 1
|
||
AND rss_url IS NOT NULL
|
||
AND TRIM(rss_url) != ''
|
||
");
|
||
$result = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
return $result['total'] ?? 0;
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 获取有RSS地址的网址总数失败: ' . $e->getMessage());
|
||
return 0;
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 获取从未刷新过的网址数量
|
||
*/
|
||
public static function getUnrefreshedUrlsCount()
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
$stmt = $db->query("
|
||
SELECT COUNT(*) as total FROM urlnav_urls
|
||
WHERE is_active = 1
|
||
AND rss_url IS NOT NULL
|
||
AND TRIM(rss_url) != ''
|
||
AND last_refresh IS NULL
|
||
");
|
||
$result = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
return $result['total'] ?? 0;
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 获取从未刷新过的网址数量失败: ' . $e->getMessage());
|
||
return 0;
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 获取最久的刷新时间
|
||
*/
|
||
public static function getOldestRefreshTime()
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
$stmt = $db->query("
|
||
SELECT MIN(last_refresh) as oldest FROM urlnav_urls
|
||
WHERE is_active = 1
|
||
AND rss_url IS NOT NULL
|
||
AND TRIM(rss_url) != ''
|
||
AND last_refresh IS NOT NULL
|
||
");
|
||
$result = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
return $result['oldest'] ?? null;
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 获取最久的刷新时间失败: ' . $e->getMessage());
|
||
return null;
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 获取所有分类
|
||
*/
|
||
public static function getAllCategories()
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
$stmt = $db->query("SELECT * FROM urlnav_categories WHERE is_active = 1 ORDER BY sort_order, created_at DESC");
|
||
$categories = $stmt->fetchAll(PDO::FETCH_ASSOC);
|
||
|
||
return $categories;
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 获取分类失败: ' . $e->getMessage());
|
||
return array();
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 获取所有网址（支持分页和分类筛选、状态筛选）
|
||
*/
|
||
public static function getAllUrls($categoryId = null, $page = 1, $pageSize = 20, $search = '', $status = '', $hasRss = '', $starRating = null)
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
$whereClause = "WHERE u.is_active = 1";
|
||
$params = array();
|
||
|
||
if ($categoryId !== null && $categoryId !== '') {
|
||
$whereClause .= " AND u.category_id = ?";
|
||
$params[] = $categoryId;
|
||
}
|
||
|
||
// 添加状态筛选 - 新增
|
||
if ($status !== null && $status !== '') {
|
||
switch ($status) {
|
||
case 'online':
|
||
$whereClause .= " AND u.is_online = 1 AND u.last_status_check IS NOT NULL";
|
||
break;
|
||
case 'offline':
|
||
$whereClause .= " AND u.is_online = 0 AND u.last_status_check IS NOT NULL";
|
||
break;
|
||
case 'unchecked':
|
||
$whereClause .= " AND u.last_status_check IS NULL";
|
||
break;
|
||
}
|
||
}
|
||
|
||
// RSS筛选 - 简化版（与状态筛选保持一致）
|
||
if ($hasRss !== null && $hasRss !== '') {
|
||
switch ($hasRss) {
|
||
case 'yes':
|
||
// 有RSS地址
|
||
$whereClause .= " AND u.rss_url IS NOT NULL AND u.rss_url != ''";
|
||
break;
|
||
case 'no':
|
||
// 无RSS地址
|
||
$whereClause .= " AND (u.rss_url IS NULL OR u.rss_url = '')";
|
||
break;
|
||
}
|
||
}
|
||
// 新增：星级筛选
|
||
if ($starRating !== null && $starRating !== '') {
|
||
switch ($starRating) {
|
||
case '1':
|
||
$whereClause .= " AND u.star_rating = 1";
|
||
break;
|
||
case '2':
|
||
$whereClause .= " AND u.star_rating = 2";
|
||
break;
|
||
case '3':
|
||
$whereClause .= " AND u.star_rating = 3";
|
||
break;
|
||
case '0':
|
||
$whereClause .= " AND u.star_rating = 0";
|
||
break;
|
||
case 'starred':
|
||
$whereClause .= " AND u.star_rating > 0";
|
||
break;
|
||
}
|
||
}
|
||
if ($search) {
|
||
$whereClause .= " AND (u.title LIKE ? OR u.url LIKE ? OR u.description LIKE ? OR u.rss_url LIKE ?)";
|
||
$searchTerm = "%{$search}%";
|
||
$params[] = $searchTerm;
|
||
$params[] = $searchTerm;
|
||
$params[] = $searchTerm;
|
||
$params[] = $searchTerm;
|
||
}
|
||
|
||
// 计算总数
|
||
$countStmt = $db->prepare("SELECT COUNT(*) as total FROM urlnav_urls u $whereClause");
|
||
$countStmt->execute($params);
|
||
$countResult = $countStmt->fetch(PDO::FETCH_ASSOC);
|
||
$total = $countResult['total'];
|
||
|
||
// 计算分页
|
||
$offset = ($page - 1) * $pageSize;
|
||
|
||
// 获取数据（包含分类名称）
|
||
$sql = "SELECT u.*, c.name as category_name
|
||
FROM urlnav_urls u
|
||
LEFT JOIN urlnav_categories c ON u.category_id = c.id
|
||
$whereClause
|
||
ORDER BY u.sort_order, u.created_at DESC
|
||
LIMIT ? OFFSET ?";
|
||
|
||
$stmt = $db->prepare($sql);
|
||
$stmtParams = array_merge($params, array($pageSize, $offset));
|
||
$stmt->execute($stmtParams);
|
||
$urls = $stmt->fetchAll(PDO::FETCH_ASSOC);
|
||
|
||
return array(
|
||
'total' => $total,
|
||
'totalPages' => ceil($total / $pageSize),
|
||
'currentPage' => $page,
|
||
'pageSize' => $pageSize,
|
||
'data' => $urls
|
||
);
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 获取网址失败: ' . $e->getMessage());
|
||
return array(
|
||
'total' => 0,
|
||
'totalPages' => 0,
|
||
'currentPage' => $page,
|
||
'pageSize' => $pageSize,
|
||
'data' => array()
|
||
);
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 通过ID获取分类
|
||
*/
|
||
public static function getCategoryById($id)
|
||
{
|
||
if (empty($id)) {
|
||
return null;
|
||
}
|
||
|
||
try {
|
||
$db = self::getDbConnection();
|
||
$stmt = $db->prepare("SELECT * FROM urlnav_categories WHERE id = ?");
|
||
$stmt->execute(array($id));
|
||
$category = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
|
||
return $category;
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 获取分类信息失败: ' . $e->getMessage());
|
||
return null;
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 通过ID获取网址
|
||
*/
|
||
public static function getUrlById($id)
|
||
{
|
||
if (empty($id)) {
|
||
return null;
|
||
}
|
||
|
||
try {
|
||
$db = self::getDbConnection();
|
||
$stmt = $db->prepare("SELECT * FROM urlnav_urls WHERE id = ?");
|
||
$stmt->execute(array($id));
|
||
$url = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
|
||
return $url;
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 获取网址信息失败: ' . $e->getMessage());
|
||
return null;
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 验证URL是否有效
|
||
*/
|
||
public static function validateUrl($url)
|
||
{
|
||
if (empty($url)) {
|
||
return false;
|
||
}
|
||
|
||
// 检查URL格式
|
||
if (!filter_var($url, FILTER_VALIDATE_URL)) {
|
||
return false;
|
||
}
|
||
|
||
// 检查URL协议
|
||
$parsedUrl = parse_url($url);
|
||
if (!in_array($parsedUrl['scheme'] ?? '', array('http', 'https'))) {
|
||
return false;
|
||
}
|
||
|
||
return true;
|
||
}
|
||
|
||
/**
|
||
* 获取所有RSS信息（支持分页和分类筛选、搜索、星级筛选）
|
||
*/
|
||
public static function getAllRssFeeds($categoryId = null, $page = 1, $pageSize = 20, $search = '', $starRating = '')
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
// 构建查询条件
|
||
$whereClause = "WHERE c.is_fresh = 1";
|
||
$params = array();
|
||
|
||
if ($categoryId !== null && $categoryId !== '' && $categoryId !== 'all') {
|
||
$whereClause .= " AND u.category_id = ?";
|
||
$params[] = $categoryId;
|
||
}
|
||
|
||
// 星级筛选
|
||
if ($starRating !== null && $starRating !== '') {
|
||
if ($starRating === '1') {
|
||
$whereClause .= " AND u.star_rating = 1";
|
||
} elseif ($starRating === '2') {
|
||
$whereClause .= " AND u.star_rating = 2";
|
||
} elseif ($starRating === '3') {
|
||
$whereClause .= " AND u.star_rating = 3";
|
||
} elseif ($starRating === '0') {
|
||
$whereClause .= " AND u.star_rating = 0";
|
||
} elseif ($starRating === 'starred') {
|
||
$whereClause .= " AND u.star_rating > 0";
|
||
}
|
||
}
|
||
|
||
if ($search) {
|
||
$whereClause .= " AND (c.feed_title LIKE ? OR c.feed_description LIKE ? OR c.full_content LIKE ? OR u.title LIKE ?)";
|
||
$searchTerm = "%{$search}%";
|
||
$params[] = $searchTerm;
|
||
$params[] = $searchTerm;
|
||
$params[] = $searchTerm;
|
||
$params[] = $searchTerm;
|
||
}
|
||
|
||
// 计算总数
|
||
$countSql = "SELECT COUNT(*) as total
|
||
FROM urlnav_rss_cache c
|
||
LEFT JOIN urlnav_urls u ON c.url_id = u.id
|
||
$whereClause";
|
||
|
||
$countStmt = $db->prepare($countSql);
|
||
$countStmt->execute($params);
|
||
$countResult = $countStmt->fetch(PDO::FETCH_ASSOC);
|
||
$total = $countResult['total'];
|
||
|
||
// ==== 修复3385行的错误 ====
|
||
// 确保page和pageSize是数字类型
|
||
$page = is_numeric($page) ? (int)$page : 1;
|
||
$pageSize = is_numeric($pageSize) ? (int)$pageSize : 20;
|
||
|
||
// 现在计算偏移量 - 这应该是3385行
|
||
$offset = ($page - 1) * $pageSize;
|
||
// ==== 修复结束 ====
|
||
|
||
// 获取数据
|
||
$sql = "SELECT c.*, u.title as site_title, u.url as site_url, u.category_id, u.star_rating,
|
||
cat.name as category_name
|
||
FROM urlnav_rss_cache c
|
||
LEFT JOIN urlnav_urls u ON c.url_id = u.id
|
||
LEFT JOIN urlnav_categories cat ON u.category_id = cat.id
|
||
$whereClause
|
||
ORDER BY c.pub_date DESC
|
||
LIMIT ? OFFSET ?";
|
||
|
||
$stmt = $db->prepare($sql);
|
||
$stmtParams = array_merge($params, array($pageSize, $offset));
|
||
$stmt->execute($stmtParams);
|
||
$feeds = $stmt->fetchAll(PDO::FETCH_ASSOC);
|
||
|
||
// 计算总页数
|
||
$totalPages = $pageSize > 0 ? ceil($total / $pageSize) : 0;
|
||
|
||
return array(
|
||
'total' => $total,
|
||
'totalPages' => $totalPages,
|
||
'currentPage' => $page,
|
||
'pageSize' => $pageSize,
|
||
'data' => $feeds
|
||
);
|
||
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 获取RSS信息失败: ' . $e->getMessage());
|
||
return array(
|
||
'total' => 0,
|
||
'totalPages' => 0,
|
||
'currentPage' => is_numeric($page) ? (int)$page : 1,
|
||
'pageSize' => is_numeric($pageSize) ? (int)$pageSize : 20,
|
||
'data' => array()
|
||
);
|
||
}
|
||
}
|
||
/**
|
||
* 获取星级选项
|
||
*/
|
||
public static function getStarRatingOptions()
|
||
{
|
||
return array(
|
||
'0' => '无星级',
|
||
'1' => '★',
|
||
'2' => '★★',
|
||
'3' => '★★★'
|
||
);
|
||
}
|
||
|
||
/**
|
||
* 获取星级显示文本
|
||
*/
|
||
public static function getStarRatingText($rating)
|
||
{
|
||
$options = self::getStarRatingOptions();
|
||
return isset($options[$rating]) ? $options[$rating] : '无星级';
|
||
}
|
||
|
||
|
||
/**
|
||
* 获取缓存统计信息
|
||
*/
|
||
public static function getCacheStats()
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
// 获取总缓存数
|
||
$stmt = $db->query("SELECT COUNT(*) as total FROM urlnav_rss_cache");
|
||
$totalResult = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
$total = $totalResult['total'] ?? 0;
|
||
|
||
// 获取新鲜缓存数
|
||
$stmt = $db->query("SELECT COUNT(*) as fresh FROM urlnav_rss_cache WHERE is_fresh = 1");
|
||
$freshResult = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
$fresh = $freshResult['fresh'] ?? 0;
|
||
|
||
// 获取陈旧缓存数
|
||
$stmt = $db->query("SELECT COUNT(*) as stale FROM urlnav_rss_cache WHERE is_fresh = 0");
|
||
$staleResult = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
$stale = $staleResult['stale'] ?? 0;
|
||
|
||
// 获取最新缓存时间
|
||
$stmt = $db->query("SELECT MAX(cached_at) as last_cached FROM urlnav_rss_cache");
|
||
$lastResult = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
$lastCached = $lastResult['last_cached'] ?? null;
|
||
|
||
// 获取最早缓存时间
|
||
$stmt = $db->query("SELECT MIN(cached_at) as first_cached FROM urlnav_rss_cache");
|
||
$firstResult = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
$firstCached = $firstResult['first_cached'] ?? null;
|
||
|
||
// 获取最后一次自动刷新时间
|
||
$stmt = $db->query("SELECT MAX(refresh_time) as last_auto_refresh FROM urlnav_refresh_log WHERE cron_type = 'rss'");
|
||
$refreshResult = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
$lastAutoRefresh = $refreshResult['last_auto_refresh'] ?? null;
|
||
|
||
// 获取URL统计
|
||
$stmt = $db->query("
|
||
SELECT
|
||
COUNT(DISTINCT url_id) as url_count,
|
||
AVG((SELECT COUNT(*) FROM urlnav_rss_cache c2 WHERE c2.url_id = c.url_id)) as avg_feeds_per_url,
|
||
MAX((SELECT COUNT(*) FROM urlnav_rss_cache c2 WHERE c2.url_id = c.url_id)) as max_feeds_per_url
|
||
FROM urlnav_rss_cache c
|
||
");
|
||
$urlStats = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
|
||
return array(
|
||
'total' => $total,
|
||
'fresh' => $fresh,
|
||
'stale' => $stale,
|
||
'last_cached' => $lastCached,
|
||
'first_cached' => $firstCached,
|
||
'last_auto_refresh' => $lastAutoRefresh,
|
||
'url_stats' => $urlStats
|
||
);
|
||
} catch (Exception $e) {
|
||
return array(
|
||
'total' => 0,
|
||
'fresh' => 0,
|
||
'stale' => 0,
|
||
'last_cached' => null,
|
||
'first_cached' => null,
|
||
'last_auto_refresh' => null,
|
||
'url_stats' => array('url_count' => 0, 'avg_feeds_per_url' => 0, 'max_feeds_per_url' => 0)
|
||
);
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 获取刷新统计信息
|
||
*/
|
||
public static function getRefreshStats($cronType = 'rss', $limit = 10)
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
$stmt = $db->prepare("
|
||
SELECT * FROM urlnav_refresh_log
|
||
WHERE cron_type = ?
|
||
ORDER BY refresh_time DESC
|
||
LIMIT ?
|
||
");
|
||
$stmt->execute(array($cronType, $limit));
|
||
$logs = $stmt->fetchAll(PDO::FETCH_ASSOC);
|
||
|
||
// 计算总体统计
|
||
$totalStmt = $db->prepare("
|
||
SELECT
|
||
COUNT(*) as total_refreshes,
|
||
SUM(success_count) as total_success,
|
||
SUM(new_articles) as total_new_articles,
|
||
AVG(duration) as avg_duration
|
||
FROM urlnav_refresh_log
|
||
WHERE cron_type = ?
|
||
");
|
||
$totalStmt->execute(array($cronType));
|
||
$totalStats = $totalStmt->fetch(PDO::FETCH_ASSOC);
|
||
|
||
return array(
|
||
'logs' => $logs,
|
||
'total_stats' => $totalStats
|
||
);
|
||
} catch (Exception $e) {
|
||
return array(
|
||
'logs' => array(),
|
||
'total_stats' => array(
|
||
'total_refreshes' => 0,
|
||
'total_success' => 0,
|
||
'total_new_articles' => 0,
|
||
'avg_duration' => 0
|
||
)
|
||
);
|
||
}
|
||
}
|
||
|
||
public static function manualCheckStatus($urlIds = null, $isBatchCheck = false, $batchInfo = null)
|
||
{
|
||
// 增加执行时间限制
|
||
@set_time_limit(300); // 5分钟
|
||
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
// 解析批次信息
|
||
$batchNumber = 1;
|
||
$totalBatches = 1;
|
||
$batchSize = 10;
|
||
$selectedIds = []; // 存储所有选中的ID
|
||
|
||
if ($batchInfo) {
|
||
$batchData = json_decode($batchInfo, true);
|
||
if ($batchData) {
|
||
$batchNumber = intval($batchData['batch'] ?? 1);
|
||
$totalBatches = intval($batchData['total'] ?? 1);
|
||
$batchSize = intval($batchData['size'] ?? 10);
|
||
|
||
// 如果有存储的选中ID，优先使用
|
||
if (!empty($batchData['selected_ids'])) {
|
||
if (is_string($batchData['selected_ids'])) {
|
||
$selectedIds = array_filter(explode(',', $batchData['selected_ids']), 'is_numeric');
|
||
} else if (is_array($batchData['selected_ids'])) {
|
||
$selectedIds = array_filter($batchData['selected_ids'], 'is_numeric');
|
||
}
|
||
}
|
||
}
|
||
}
|
||
|
||
// ==== 修改开始：自动检查逻辑 ====
|
||
// 如果是自动检查（$urlIds为null且不是批量检查）
|
||
if ($urlIds === null && !$isBatchCheck && $batchInfo === null) {
|
||
// 获取配置中的每次检查数量
|
||
$config = self::getConfig();
|
||
$maxCheck = intval($config->statusCheckMax ?? 80);
|
||
|
||
// 安全限制
|
||
if ($maxCheck > 200) $maxCheck = 200;
|
||
if ($maxCheck < 1) $maxCheck = 20;
|
||
|
||
// 自动检查逻辑：优先检查未检查过的，然后按检查时间排序
|
||
$sql = "SELECT * FROM urlnav_urls WHERE is_active = 1
|
||
ORDER BY
|
||
CASE WHEN last_status_check IS NULL THEN 0 ELSE 1 END,
|
||
last_status_check ASC NULLS FIRST
|
||
LIMIT ?";
|
||
|
||
$params = [$maxCheck];
|
||
$stmt = $db->prepare($sql);
|
||
$stmt->execute($params);
|
||
$urls = $stmt->fetchAll(PDO::FETCH_ASSOC);
|
||
|
||
if (empty($urls)) {
|
||
return [
|
||
'success' => true,
|
||
'message' => '没有需要检查的网址',
|
||
'total' => 0,
|
||
'success_count' => 0,
|
||
'failed_count' => 0
|
||
];
|
||
}
|
||
|
||
// 执行检查
|
||
$successCount = 0;
|
||
$failedCount = 0;
|
||
$results = [];
|
||
$totalResponseTime = 0;
|
||
|
||
// 🔴 新增：收集成功和失败的网址
|
||
$successUrls = [];
|
||
$failedUrls = [];
|
||
|
||
$logStmt = $db->prepare("
|
||
INSERT INTO urlnav_status_log
|
||
(url_id, is_online, status_code, response_time, check_time, error_message)
|
||
VALUES (?, ?, ?, ?, CURRENT_TIMESTAMP, ?)
|
||
");
|
||
|
||
$updateStmt = $db->prepare("
|
||
UPDATE urlnav_urls SET
|
||
is_online = ?,
|
||
last_status_check = CURRENT_TIMESTAMP,
|
||
status_check_count = status_check_count + 1,
|
||
last_status_code = ?
|
||
WHERE id = ?
|
||
");
|
||
|
||
foreach ($urls as $url) {
|
||
$status = self::checkSingleWebsite($url['url']);
|
||
$urlId = $url['id'];
|
||
|
||
$results[$urlId] = $status;
|
||
|
||
if ($status['success']) {
|
||
$totalResponseTime += $status['response_time'];
|
||
$successCount++;
|
||
|
||
// 🔴 新增：收集成功网址
|
||
$successUrls[] = [
|
||
'id' => $urlId,
|
||
'url' => $url['url'],
|
||
'title' => $url['title'] ?? $url['url'],
|
||
'response_time' => $status['response_time']
|
||
];
|
||
} else {
|
||
$failedCount++;
|
||
|
||
// 🔴 新增：收集失败网址
|
||
$failedUrls[] = [
|
||
'id' => $urlId,
|
||
'url' => $url['url'],
|
||
'title' => $url['title'] ?? $url['url'],
|
||
'error' => $status['message'],
|
||
'status_code' => $status['status_code']
|
||
];
|
||
}
|
||
|
||
// 记录日志
|
||
$logStmt->execute([
|
||
$urlId,
|
||
$status['success'] ? 1 : 0,
|
||
$status['status_code'],
|
||
$status['response_time'],
|
||
$status['message']
|
||
]);
|
||
|
||
// 更新状态
|
||
$updateStmt->execute([
|
||
$status['success'] ? 1 : 0,
|
||
$status['status_code'],
|
||
$urlId
|
||
]);
|
||
|
||
// 短暂休息，避免服务器压力过大
|
||
if (count($results) % 5 == 0) {
|
||
usleep(50000); // 50ms
|
||
}
|
||
}
|
||
|
||
$avgResponseTime = $successCount > 0 ? round($totalResponseTime / $successCount, 2) : 0;
|
||
|
||
return [
|
||
'success' => true,
|
||
'message' => sprintf("自动检查完成: %d成功, %d失败", $successCount, $failedCount),
|
||
'total' => count($urls),
|
||
'success_count' => $successCount,
|
||
'failed_count' => $failedCount,
|
||
'avg_response_time' => $avgResponseTime,
|
||
'results' => $results,
|
||
|
||
// 🔴 新增：返回成功和失败的网址列表
|
||
'successUrls' => $successUrls,
|
||
'failedUrls' => $failedUrls
|
||
];
|
||
}
|
||
// ==== 修改结束：自动检查逻辑 ====
|
||
|
||
// 以下保持原样，手动检查和批量检查逻辑不变
|
||
// 构建查询 - 修复关键逻辑
|
||
$sql = "SELECT * FROM urlnav_urls WHERE is_active = 1";
|
||
$params = [];
|
||
$totalUrls = 0;
|
||
|
||
// 确定要检查的网址ID
|
||
$idsToCheck = [];
|
||
|
||
// 情况1：直接传入ID数组或字符串
|
||
if ($urlIds !== null) {
|
||
if (is_array($urlIds)) {
|
||
if (empty($urlIds)) {
|
||
return [
|
||
'success' => true,
|
||
'message' => '没有选择要检查的网址',
|
||
'total' => 0,
|
||
'success_count' => 0,
|
||
'failed_count' => 0,
|
||
'has_more' => false,
|
||
'batch_number' => $batchNumber,
|
||
'total_batches' => $totalBatches
|
||
];
|
||
}
|
||
$idArray = array_filter($urlIds, 'is_numeric');
|
||
} else if (is_string($urlIds) && $urlIds !== '') {
|
||
if (strpos($urlIds, ',') !== false) {
|
||
$idArray = array_filter(explode(',', $urlIds), 'is_numeric');
|
||
} else if (is_numeric($urlIds)) {
|
||
$idArray = [$urlIds];
|
||
}
|
||
}
|
||
}
|
||
// 情况2：从批次信息中获取选中的ID
|
||
else if (!empty($selectedIds)) {
|
||
$idsToCheck = $selectedIds;
|
||
}
|
||
|
||
// 如果有要检查的ID，构建IN查询
|
||
if (!empty($idsToCheck)) {
|
||
$totalUrls = count($idsToCheck);
|
||
|
||
// 重新计算批次信息
|
||
$totalBatches = ceil($totalUrls / $batchSize);
|
||
|
||
// 如果批次号超过总批次，返回完成
|
||
if ($batchNumber > $totalBatches) {
|
||
return [
|
||
'success' => true,
|
||
'message' => '所有选中的网址已检查完成',
|
||
'total' => 0,
|
||
'success_count' => 0,
|
||
'failed_count' => 0,
|
||
'has_more' => false,
|
||
'batch_number' => $batchNumber,
|
||
'total_batches' => $totalBatches,
|
||
'completed' => true
|
||
];
|
||
}
|
||
|
||
// 计算当前批次的ID范围
|
||
$offset = ($batchNumber - 1) * $batchSize;
|
||
$currentBatchIds = array_slice($idsToCheck, $offset, $batchSize);
|
||
|
||
if (empty($currentBatchIds)) {
|
||
return [
|
||
'success' => true,
|
||
'message' => '没有需要检查的网址',
|
||
'total' => 0,
|
||
'success_count' => 0,
|
||
'failed_count' => 0,
|
||
'has_more' => false,
|
||
'batch_number' => $batchNumber,
|
||
'total_batches' => $totalBatches
|
||
];
|
||
}
|
||
|
||
// 构建IN查询
|
||
$placeholders = implode(',', array_fill(0, count($currentBatchIds), '?'));
|
||
$sql .= " AND id IN ($placeholders) ORDER BY id";
|
||
$params = $currentBatchIds;
|
||
}
|
||
// 情况3：没有指定ID，检查全部网址
|
||
else {
|
||
$sql .= " ORDER BY last_status_check ASC NULLS FIRST, id ASC LIMIT ? OFFSET ?";
|
||
|
||
// 获取总网址数用于计算批次
|
||
$totalStmt = $db->query("SELECT COUNT(*) as total FROM urlnav_urls WHERE is_active = 1");
|
||
$totalResult = $totalStmt->fetch(PDO::FETCH_ASSOC);
|
||
$totalUrls = $totalResult['total'] ?? 0;
|
||
$totalBatches = ceil($totalUrls / $batchSize);
|
||
|
||
// 如果批次号超过总批次，返回完成
|
||
if ($batchNumber > $totalBatches) {
|
||
return [
|
||
'success' => true,
|
||
'message' => '所有网址已检查完成',
|
||
'total' => 0,
|
||
'success_count' => 0,
|
||
'failed_count' => 0,
|
||
'has_more' => false,
|
||
'batch_number' => $batchNumber,
|
||
'total_batches' => $totalBatches,
|
||
'completed' => true
|
||
];
|
||
}
|
||
|
||
$offset = ($batchNumber - 1) * $batchSize;
|
||
$params = [$batchSize, $offset];
|
||
}
|
||
|
||
$stmt = $db->prepare($sql);
|
||
$stmt->execute($params);
|
||
$urls = $stmt->fetchAll(PDO::FETCH_ASSOC);
|
||
|
||
if (empty($urls)) {
|
||
return [
|
||
'success' => true,
|
||
'message' => '没有需要检查的网址',
|
||
'total' => 0,
|
||
'success_count' => 0,
|
||
'failed_count' => 0,
|
||
'has_more' => false,
|
||
'batch_number' => $batchNumber,
|
||
'total_batches' => $totalBatches
|
||
];
|
||
}
|
||
|
||
$successCount = 0;
|
||
$failedCount = 0;
|
||
$results = [];
|
||
$totalResponseTime = 0;
|
||
$checkedIds = [];
|
||
|
||
// 准备更新语句
|
||
$logStmt = $db->prepare("
|
||
INSERT INTO urlnav_status_log
|
||
(url_id, is_online, status_code, response_time, check_time, error_message)
|
||
VALUES (?, ?, ?, ?, CURRENT_TIMESTAMP, ?)
|
||
");
|
||
|
||
$updateStmt = $db->prepare("
|
||
UPDATE urlnav_urls SET
|
||
is_online = ?,
|
||
last_status_check = CURRENT_TIMESTAMP,
|
||
status_check_count = status_check_count + 1,
|
||
last_status_code = ?
|
||
WHERE id = ?
|
||
");
|
||
|
||
foreach ($urls as $url) {
|
||
$status = self::checkSingleWebsite($url['url']);
|
||
$urlId = $url['id'];
|
||
|
||
$results[$urlId] = $status;
|
||
$checkedIds[] = $urlId;
|
||
|
||
if ($status['success']) {
|
||
$totalResponseTime += $status['response_time'];
|
||
$successCount++;
|
||
} else {
|
||
$failedCount++;
|
||
}
|
||
|
||
// 记录日志
|
||
$logStmt->execute([
|
||
$urlId,
|
||
$status['success'] ? 1 : 0,
|
||
$status['status_code'],
|
||
$status['response_time'],
|
||
$status['message']
|
||
]);
|
||
|
||
// 更新状态
|
||
$updateStmt->execute([
|
||
$status['success'] ? 1 : 0,
|
||
$status['status_code'],
|
||
$urlId
|
||
]);
|
||
|
||
// 短暂休息，避免服务器压力过大
|
||
if (count($checkedIds) % 5 == 0) {
|
||
usleep(100000); // 100ms
|
||
}
|
||
}
|
||
|
||
// 判断是否还有更多批次
|
||
$hasMore = false;
|
||
$processedCount = ($batchNumber - 1) * $batchSize + count($checkedIds);
|
||
|
||
// 如果有选中的ID
|
||
if (!empty($idsToCheck)) {
|
||
$hasMore = ($processedCount < count($idsToCheck));
|
||
}
|
||
// 如果是检查全部
|
||
else {
|
||
$hasMore = ($processedCount < $totalUrls);
|
||
}
|
||
|
||
$avgResponseTime = $successCount > 0 ? round($totalResponseTime / $successCount, 2) : 0;
|
||
|
||
// 准备批次信息用于下一批
|
||
$nextBatchInfo = [
|
||
'batch' => $batchNumber + 1,
|
||
'total' => $totalBatches,
|
||
'size' => $batchSize,
|
||
'selected_ids' => !empty($idsToCheck) ? implode(',', $idsToCheck) : ''
|
||
];
|
||
|
||
return [
|
||
'success' => true,
|
||
'message' => sprintf("第%d批检查完成: %d成功, %d失败", $batchNumber, $successCount, $failedCount),
|
||
'total' => count($urls),
|
||
'success_count' => $successCount,
|
||
'failed_count' => $failedCount,
|
||
'avg_response_time' => $avgResponseTime,
|
||
'results' => $results,
|
||
'has_more' => $hasMore,
|
||
'checked_ids' => $checkedIds,
|
||
'batch_number' => $batchNumber,
|
||
'total_batches' => $totalBatches,
|
||
'next_batch_info' => json_encode($nextBatchInfo),
|
||
'processed_count' => $processedCount,
|
||
'total_to_process' => !empty($idsToCheck) ? count($idsToCheck) : $totalUrls
|
||
];
|
||
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 检查网站状态失败: ' . $e->getMessage());
|
||
return [
|
||
'success' => false,
|
||
'message' => '检查失败: ' . $e->getMessage(),
|
||
'has_more' => false,
|
||
'batch_number' => isset($batchNumber) ? $batchNumber : 1,
|
||
'total_batches' => isset($totalBatches) ? $totalBatches : 1
|
||
];
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 检查单个网站状态 - 优化版，解决HEAD请求兼容性问题
|
||
*/
|
||
private static function checkSingleWebsite($url)
|
||
{
|
||
$startTime = microtime(true);
|
||
|
||
try {
|
||
// 获取配置的超时时间
|
||
$config = self::getConfig();
|
||
$timeout = intval($config->statusCheckTimeout ?? 15);
|
||
|
||
// 记录调试信息
|
||
error_log("UrlNav: 开始检查网站状态: {$url} [使用GET方法]");
|
||
|
||
$ch = curl_init();
|
||
|
||
// 修改：使用GET请求而不是HEAD请求
|
||
curl_setopt_array($ch, [
|
||
CURLOPT_URL => $url,
|
||
CURLOPT_RETURNTRANSFER => true, // 改为true以获取响应
|
||
CURLOPT_FOLLOWLOCATION => true,
|
||
CURLOPT_MAXREDIRS => 5,
|
||
CURLOPT_TIMEOUT => $timeout,
|
||
CURLOPT_CONNECTTIMEOUT => $timeout,
|
||
CURLOPT_SSL_VERIFYPEER => false,
|
||
CURLOPT_SSL_VERIFYHOST => false,
|
||
CURLOPT_USERAGENT => 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36',
|
||
CURLOPT_ENCODING => '',
|
||
CURLOPT_IPRESOLVE => CURL_IPRESOLVE_V4,
|
||
CURLOPT_FAILONERROR => false,
|
||
CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_1_1,
|
||
CURLOPT_NOBODY => false, // 重要：使用GET而不是HEAD
|
||
CURLOPT_HEADER => true, // 获取头部信息
|
||
CURLOPT_RANGE => '0-1024', // 只获取前1KB内容，减少带宽使用
|
||
]);
|
||
|
||
$response = curl_exec($ch);
|
||
$httpCode = curl_getinfo($ch, CURLINFO_HTTP_CODE);
|
||
$totalTime = curl_getinfo($ch, CURLINFO_TOTAL_TIME);
|
||
$connectTime = curl_getinfo($ch, CURLINFO_CONNECT_TIME);
|
||
$namelookupTime = curl_getinfo($ch, CURLINFO_NAMELOOKUP_TIME);
|
||
$redirectCount = curl_getinfo($ch, CURLINFO_REDIRECT_COUNT);
|
||
$effectiveUrl = curl_getinfo($ch, CURLINFO_EFFECTIVE_URL);
|
||
$error = curl_error($ch);
|
||
$errorNo = curl_errno($ch);
|
||
|
||
curl_close($ch);
|
||
|
||
$endTime = microtime(true);
|
||
$responseTime = round(($endTime - $startTime) * 1000, 2);
|
||
|
||
// 详细记录
|
||
error_log("UrlNav: GET检查结果 - URL: {$url}, HTTP: {$httpCode}, 耗时: {$totalTime}s");
|
||
|
||
if ($errorNo !== 0) {
|
||
// cURL错误
|
||
$errorMessages = [
|
||
CURLE_COULDNT_CONNECT => '无法连接服务器',
|
||
CURLE_COULDNT_RESOLVE_HOST => '无法解析主机名',
|
||
CURLE_OPERATION_TIMEDOUT => '请求超时',
|
||
CURLE_SSL_CONNECT_ERROR => 'SSL连接错误',
|
||
CURLE_GOT_NOTHING => '服务器无响应',
|
||
];
|
||
|
||
$errorMessage = $errorMessages[$errorNo] ?? '连接失败: ' . $error;
|
||
|
||
error_log("UrlNav: cURL错误 - {$errorMessage}");
|
||
|
||
return [
|
||
'success' => false,
|
||
'status_code' => 0,
|
||
'message' => $errorMessage,
|
||
'response_time' => $responseTime
|
||
];
|
||
}
|
||
|
||
// 更宽松的成功判断
|
||
$isSuccess = false;
|
||
$statusMessage = '';
|
||
|
||
// 判断是否成功
|
||
if ($httpCode >= 200 && $httpCode < 500) {
|
||
// 2xx, 3xx, 4xx都算成功（网站可访问）
|
||
// 4xx表示客户端错误，但网站本身是可访问的
|
||
$isSuccess = true;
|
||
|
||
if ($httpCode >= 200 && $httpCode < 300) {
|
||
$statusMessage = "正常 (HTTP {$httpCode})";
|
||
} elseif ($httpCode >= 300 && $httpCode < 400) {
|
||
$statusMessage = "重定向 (HTTP {$httpCode})";
|
||
} else {
|
||
$statusMessage = "客户端错误但网站可访问 (HTTP {$httpCode})";
|
||
}
|
||
|
||
// 添加重定向信息
|
||
if ($redirectCount > 0 && $url != $effectiveUrl) {
|
||
$statusMessage .= "，经过 {$redirectCount} 次重定向";
|
||
}
|
||
} elseif ($httpCode >= 500) {
|
||
// 5xx服务器错误
|
||
$isSuccess = false;
|
||
$statusMessage = "服务器错误 (HTTP {$httpCode})";
|
||
} elseif ($httpCode == 0) {
|
||
// 没有获取到HTTP状态码
|
||
$isSuccess = false;
|
||
$statusMessage = "未获取到HTTP状态码";
|
||
}
|
||
|
||
// 特殊处理：如果总时间超过阈值但仍返回成功码，标记为慢速但成功
|
||
if ($isSuccess && $totalTime > 3) {
|
||
$statusMessage = "慢速但可访问 ({$totalTime}s, HTTP {$httpCode})";
|
||
}
|
||
|
||
return [
|
||
'success' => $isSuccess,
|
||
'status_code' => $httpCode,
|
||
'message' => $statusMessage,
|
||
'response_time' => $responseTime,
|
||
'response_time_curl' => round($totalTime * 1000, 2),
|
||
'redirect_count' => $redirectCount,
|
||
'effective_url' => $effectiveUrl,
|
||
'method' => 'GET', // 记录使用的方法
|
||
'timing' => [
|
||
'total' => $totalTime,
|
||
'connect' => $connectTime,
|
||
'dns' => $namelookupTime
|
||
]
|
||
];
|
||
|
||
} catch (Exception $e) {
|
||
error_log("UrlNav: 检查异常 - {$url}: " . $e->getMessage());
|
||
return [
|
||
'success' => false,
|
||
'status_code' => 0,
|
||
'message' => '检查异常: ' . $e->getMessage(),
|
||
'response_time' => round((microtime(true) - $startTime) * 1000, 2)
|
||
];
|
||
}
|
||
}
|
||
|
||
|
||
/**
|
||
* 检查网站状态 - 修复版
|
||
*/
|
||
public function checkStatus()
|
||
{
|
||
try {
|
||
// 检查是否是批量检查 - 接收两种可能的参数名
|
||
$urlIds = $this->request->get('url_ids');
|
||
if (empty($urlIds)) {
|
||
$urlIds = $this->request->get('urlIds'); // 尝试另一种参数名
|
||
}
|
||
|
||
$batchInfo = $this->request->get('batch_info');
|
||
|
||
// 解析URL IDs
|
||
$idArray = null;
|
||
if ($urlIds && $urlIds !== '') {
|
||
// 处理不同的ID格式
|
||
if (is_array($urlIds)) {
|
||
$idArray = $urlIds;
|
||
} else if (strpos($urlIds, ',') !== false) {
|
||
// 逗号分隔的字符串
|
||
$idArray = array_map('trim', explode(',', $urlIds));
|
||
$idArray = array_filter($idArray, function($id) {
|
||
return is_numeric($id) && $id > 0;
|
||
});
|
||
} else if (is_numeric($urlIds)) {
|
||
// 单个ID
|
||
$idArray = [$urlIds];
|
||
}
|
||
}
|
||
|
||
// 解析批次信息
|
||
$batchData = null;
|
||
if ($batchInfo && is_string($batchInfo)) {
|
||
$batchData = json_decode($batchInfo, true);
|
||
} else if (is_array($batchInfo)) {
|
||
$batchData = $batchInfo;
|
||
}
|
||
|
||
// 如果是选中的网址，需要特殊处理批次信息
|
||
if (!empty($idArray) && $batchData) {
|
||
// 重新计算批次信息，因为选中的网址总数可能和全部网址不同
|
||
$batchSize = $batchData['size'] ?? 10;
|
||
$totalSelected = count($idArray);
|
||
$totalBatches = ceil($totalSelected / $batchSize);
|
||
$batchNumber = $batchData['batch'] ?? 1;
|
||
|
||
// 确保批次号不超过总批次
|
||
if ($batchNumber > $totalBatches) {
|
||
return $this->response->throwJson([
|
||
'success' => true,
|
||
'message' => '所有选中的网址已检查完成',
|
||
'total' => 0,
|
||
'success_count' => 0,
|
||
'failed_count' => 0,
|
||
'has_more' => false
|
||
]);
|
||
}
|
||
|
||
// 获取当前批次要检查的ID
|
||
$offset = ($batchNumber - 1) * $batchSize;
|
||
$batchIds = array_slice($idArray, $offset, $batchSize);
|
||
|
||
// 记录调试信息
|
||
error_log("UrlNav Action: 选中检查，批次: $batchNumber/$totalBatches, 本批ID: " . implode(',', $batchIds));
|
||
|
||
// 调用检查函数 - 传递当前批次的ID，不再传递批次信息让后端分页
|
||
$result = UrlNav_Plugin::manualCheckStatus($batchIds, true, null);
|
||
|
||
// 更新批次信息
|
||
$result['batch_number'] = $batchNumber;
|
||
$result['total_batches'] = $totalBatches;
|
||
$result['has_more'] = ($batchNumber < $totalBatches);
|
||
$result['batch_info'] = json_encode([
|
||
'batch' => $batchNumber,
|
||
'total' => $totalBatches,
|
||
'size' => $batchSize,
|
||
'selected_ids' => implode(',', $idArray) // 记录所有选中的ID
|
||
]);
|
||
|
||
$this->response->throwJson($result);
|
||
return;
|
||
}
|
||
|
||
// 如果没有选中网址，检查全部（已有的逻辑）
|
||
$result = UrlNav_Plugin::manualCheckStatus(null, true, $batchInfo);
|
||
$this->response->throwJson($result);
|
||
|
||
} catch (Exception $e) {
|
||
$this->response->throwJson([
|
||
'success' => false,
|
||
'message' => '检查失败: ' . $e->getMessage()
|
||
]);
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 获取网站总数
|
||
*/
|
||
private static function getTotalUrlCount()
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
$stmt = $db->query("SELECT COUNT(*) as total FROM urlnav_urls WHERE is_active = 1");
|
||
$result = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
return $result['total'] ?? 0;
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 获取网站总数失败: ' . $e->getMessage());
|
||
return 0;
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 更新状态检查统计
|
||
*/
|
||
private static function updateStatusStats($checkResult)
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
// 获取当前统计
|
||
$stmt = $db->query("SELECT * FROM urlnav_status_stats LIMIT 1");
|
||
$currentStats = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
|
||
if ($currentStats) {
|
||
// 更新统计
|
||
$totalChecks = $currentStats['total_checks'] + $checkResult['total'];
|
||
$successChecks = $currentStats['success_checks'] + $checkResult['success_count'];
|
||
$failedChecks = $currentStats['failed_checks'] + $checkResult['failed_count'];
|
||
|
||
// 计算平均响应时间（加权平均）
|
||
$currentAvg = $currentStats['avg_response_time'] * $currentStats['total_checks'];
|
||
$newAvg = $checkResult['avg_response_time'] * $checkResult['total'];
|
||
$avgResponseTime = $totalChecks > 0 ? ($currentAvg + $newAvg) / $totalChecks : 0;
|
||
|
||
$stmt = $db->prepare("
|
||
UPDATE urlnav_status_stats SET
|
||
total_checks = ?,
|
||
success_checks = ?,
|
||
failed_checks = ?,
|
||
avg_response_time = ?,
|
||
last_check_time = CURRENT_TIMESTAMP,
|
||
updated_at = CURRENT_TIMESTAMP
|
||
WHERE id = ?
|
||
");
|
||
|
||
$stmt->execute([
|
||
$totalChecks,
|
||
$successChecks,
|
||
$failedChecks,
|
||
round($avgResponseTime, 2),
|
||
$currentStats['id']
|
||
]);
|
||
} else {
|
||
// 插入新统计
|
||
$stmt = $db->prepare("
|
||
INSERT INTO urlnav_status_stats
|
||
(total_checks, success_checks, failed_checks, avg_response_time, last_check_time)
|
||
VALUES (?, ?, ?, ?, CURRENT_TIMESTAMP)
|
||
");
|
||
|
||
$stmt->execute([
|
||
$checkResult['total'],
|
||
$checkResult['success_count'],
|
||
$checkResult['failed_count'],
|
||
round($checkResult['avg_response_time'], 2)
|
||
]);
|
||
}
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 更新状态检查统计失败: ' . $e->getMessage());
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 获取状态检查统计
|
||
*/
|
||
public static function getStatusStats()
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
// 获取总数
|
||
$totalQuery = $db->query("SELECT COUNT(*) as total FROM urlnav_urls WHERE is_active = 1");
|
||
$totalResult = $totalQuery->fetch(PDO::FETCH_ASSOC);
|
||
$total = $totalResult ? $totalResult['total'] : 0;
|
||
|
||
// 获取在线数
|
||
$onlineQuery = $db->query("SELECT COUNT(*) as online FROM urlnav_urls WHERE is_active = 1 AND is_online = 1");
|
||
$onlineResult = $onlineQuery->fetch(PDO::FETCH_ASSOC);
|
||
$online = $onlineResult ? $onlineResult['online'] : 0;
|
||
|
||
// 获取离线数
|
||
$offlineQuery = $db->query("SELECT COUNT(*) as offline FROM urlnav_urls WHERE is_active = 1 AND is_online = 0 AND last_status_check IS NOT NULL");
|
||
$offlineResult = $offlineQuery->fetch(PDO::FETCH_ASSOC);
|
||
$offline = $offlineResult ? $offlineResult['offline'] : 0;
|
||
|
||
// 获取未检查数
|
||
$uncheckedQuery = $db->query("SELECT COUNT(*) as unchecked FROM urlnav_urls WHERE is_active = 1 AND last_status_check IS NULL");
|
||
$uncheckedResult = $uncheckedQuery->fetch(PDO::FETCH_ASSOC);
|
||
$unchecked = $uncheckedResult ? $uncheckedResult['unchecked'] : 0;
|
||
|
||
// 添加RSS统计
|
||
$rssQuery = $db->query("SELECT
|
||
SUM(CASE WHEN rss_url IS NOT NULL AND rss_url != '' THEN 1 ELSE 0 END) as has_rss,
|
||
SUM(CASE WHEN rss_url IS NULL OR rss_url = '' THEN 1 ELSE 0 END) as no_rss
|
||
FROM urlnav_urls WHERE is_active = 1");
|
||
$rssResult = $rssQuery->fetch(PDO::FETCH_ASSOC);
|
||
$has_rss = $rssResult ? $rssResult['has_rss'] : 0;
|
||
$no_rss = $rssResult ? $rssResult['no_rss'] : 0;
|
||
|
||
// 计算在线率
|
||
$online_rate = $total > 0 ? round(($online / $total) * 100, 1) : 0;
|
||
|
||
return [
|
||
'total' => (int)$total,
|
||
'online' => (int)$online,
|
||
'offline' => (int)$offline,
|
||
'unchecked' => (int)$unchecked,
|
||
'online_rate' => (float)$online_rate,
|
||
'has_rss' => (int)$has_rss,
|
||
'no_rss' => (int)$no_rss
|
||
];
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav getStatusStats error: ' . $e->getMessage());
|
||
return [
|
||
'total' => 0,
|
||
'online' => 0,
|
||
'offline' => 0,
|
||
'unchecked' => 0,
|
||
'online_rate' => 0,
|
||
'has_rss' => 0,
|
||
'no_rss' => 0
|
||
];
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 获取RSS刷新统计
|
||
*/
|
||
public static function getRssRefreshStats()
|
||
{
|
||
try {
|
||
$stats = array();
|
||
|
||
// 获取有RSS的网址总数
|
||
$stats['total_with_rss'] = self::getTotalUrlsWithRss();
|
||
|
||
// 获取从未刷新的数量
|
||
$stats['never_refreshed'] = self::getUnrefreshedUrlsCount();
|
||
|
||
// 获取最久刷新时间
|
||
$stats['oldest_refresh'] = self::getOldestRefreshTime();
|
||
|
||
// 计算进度
|
||
if ($stats['total_with_rss'] > 0) {
|
||
$stats['refreshed_percentage'] = round(($stats['total_with_rss'] - $stats['never_refreshed']) / $stats['total_with_rss'] * 100, 1);
|
||
} else {
|
||
$stats['refreshed_percentage'] = 0;
|
||
}
|
||
|
||
// 获取配置的每批处理数量
|
||
$config = self::getConfig();
|
||
$stats['refresh_limit'] = intval($config->rssRefreshLimit ?? 20);
|
||
|
||
// 计算预计完成所需任务次数
|
||
if ($stats['refresh_limit'] > 0 && $stats['never_refreshed'] > 0) {
|
||
$stats['estimated_tasks'] = ceil($stats['never_refreshed'] / $stats['refresh_limit']);
|
||
} else {
|
||
$stats['estimated_tasks'] = 0;
|
||
}
|
||
|
||
// 计算最久未刷新的时间
|
||
if ($stats['oldest_refresh']) {
|
||
$hoursAgo = round((time() - strtotime($stats['oldest_refresh'])) / 3600, 1);
|
||
$stats['oldest_hours_ago'] = $hoursAgo;
|
||
}
|
||
|
||
return $stats;
|
||
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 获取RSS刷新统计失败: ' . $e->getMessage());
|
||
return array(
|
||
'total_with_rss' => 0,
|
||
'never_refreshed' => 0,
|
||
'refreshed_percentage' => 0,
|
||
'refresh_limit' => 20,
|
||
'estimated_tasks' => 0
|
||
);
|
||
}
|
||
}
|
||
|
||
// ============ 收藏功能相关方法 ============
|
||
|
||
/**
|
||
* 添加收藏
|
||
*/
|
||
public static function addFavorite($feedId, $userId = 0)
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
// 获取完整的文章信息
|
||
$stmt = $db->prepare("
|
||
SELECT c.*, u.title as site_title, u.url as site_url, cat.name as category_name
|
||
FROM urlnav_rss_cache c
|
||
LEFT JOIN urlnav_urls u ON c.url_id = u.id
|
||
LEFT JOIN urlnav_categories cat ON u.category_id = cat.id
|
||
WHERE c.id = ?
|
||
");
|
||
$stmt->execute([$feedId]);
|
||
$feed = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
|
||
if (!$feed) {
|
||
return ['success' => false, 'message' => '文章不存在'];
|
||
}
|
||
|
||
// 检查是否已收藏
|
||
$stmt = $db->prepare("SELECT id FROM urlnav_favorites WHERE user_id = ? AND feed_id = ?");
|
||
$stmt->execute([$userId, $feedId]);
|
||
$existing = $stmt->fetch();
|
||
|
||
if ($existing) {
|
||
return ['success' => false, 'message' => '已收藏'];
|
||
}
|
||
|
||
// 添加收藏（保存完整信息）
|
||
$stmt = $db->prepare("
|
||
INSERT INTO urlnav_favorites
|
||
(user_id, feed_id, feed_title, feed_link, feed_description, full_content, pub_date,
|
||
site_title, site_url, category_name)
|
||
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
|
||
");
|
||
$stmt->execute([
|
||
$userId,
|
||
$feedId,
|
||
$feed['feed_title'],
|
||
$feed['feed_link'],
|
||
$feed['feed_description'] ?? '',
|
||
$feed['full_content'] ?? '', // 保存完整内容
|
||
$feed['pub_date'],
|
||
$feed['site_title'] ?? '',
|
||
$feed['site_url'] ?? '',
|
||
$feed['category_name'] ?? ''
|
||
]);
|
||
|
||
// 标记文章为新鲜状态，避免被清理
|
||
$stmt = $db->prepare("UPDATE urlnav_rss_cache SET is_fresh = 1 WHERE id = ?");
|
||
$stmt->execute([$feedId]);
|
||
|
||
return ['success' => true, 'message' => '收藏成功'];
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 添加收藏失败: ' . $e->getMessage());
|
||
return ['success' => false, 'message' => '收藏失败: ' . $e->getMessage()];
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 取消收藏
|
||
*/
|
||
public static function removeFavorite($feedId, $userId = 0)
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
$stmt = $db->prepare("DELETE FROM urlnav_favorites WHERE user_id = ? AND feed_id = ?");
|
||
$stmt->execute([$userId, $feedId]);
|
||
|
||
if ($stmt->rowCount() > 0) {
|
||
return ['success' => true, 'message' => '已取消收藏'];
|
||
} else {
|
||
return ['success' => false, 'message' => '未收藏'];
|
||
}
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 取消收藏失败: ' . $e->getMessage());
|
||
return ['success' => false, 'message' => '取消收藏失败: ' . $e->getMessage()];
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 检查是否已收藏
|
||
*/
|
||
public static function isFavorite($feedId, $userId = 0)
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
$stmt = $db->prepare("SELECT id FROM urlnav_favorites WHERE user_id = ? AND feed_id = ?");
|
||
$stmt->execute([$userId, $feedId]);
|
||
$result = $stmt->fetch();
|
||
|
||
return $result ? true : false;
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 检查收藏失败: ' . $e->getMessage());
|
||
return false;
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 获取用户的收藏列表
|
||
*/
|
||
public static function getFavorites($userId = 0, $page = 1, $pageSize = 20, $search = '')
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
// 构建查询条件
|
||
$whereClause = "WHERE f.user_id = ?";
|
||
$params = [$userId];
|
||
|
||
if ($search) {
|
||
$whereClause .= " AND (f.feed_title LIKE ? OR f.feed_description LIKE ? OR f.full_content LIKE ? OR f.site_title LIKE ?)";
|
||
$searchTerm = "%{$search}%";
|
||
$params[] = $searchTerm;
|
||
$params[] = $searchTerm;
|
||
$params[] = $searchTerm;
|
||
$params[] = $searchTerm;
|
||
}
|
||
|
||
// 计算总数
|
||
$countSql = "SELECT COUNT(*) as total FROM urlnav_favorites f $whereClause";
|
||
|
||
$countStmt = $db->prepare($countSql);
|
||
$countStmt->execute($params);
|
||
$countResult = $countStmt->fetch(PDO::FETCH_ASSOC);
|
||
$total = $countResult['total'];
|
||
|
||
// 计算分页
|
||
$offset = ($page - 1) * $pageSize;
|
||
|
||
// 获取数据
|
||
$sql = "SELECT
|
||
f.id,
|
||
f.feed_id as original_feed_id,
|
||
f.feed_title,
|
||
f.feed_link,
|
||
f.feed_description,
|
||
f.full_content,
|
||
f.pub_date,
|
||
f.site_title,
|
||
f.site_url,
|
||
f.category_name,
|
||
f.favorited_at
|
||
FROM urlnav_favorites f
|
||
$whereClause
|
||
ORDER BY f.favorited_at DESC
|
||
LIMIT ? OFFSET ?";
|
||
|
||
$stmt = $db->prepare($sql);
|
||
$stmtParams = array_merge($params, [$pageSize, $offset]);
|
||
$stmt->execute($stmtParams);
|
||
$feeds = $stmt->fetchAll(PDO::FETCH_ASSOC);
|
||
|
||
return [
|
||
'success' => true,
|
||
'total' => $total,
|
||
'totalPages' => ceil($total / $pageSize),
|
||
'currentPage' => $page,
|
||
'pageSize' => $pageSize,
|
||
'data' => $feeds
|
||
];
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 获取收藏列表失败: ' . $e->getMessage());
|
||
return [
|
||
'success' => false,
|
||
'message' => '获取收藏失败',
|
||
'total' => 0,
|
||
'totalPages' => 0,
|
||
'currentPage' => $page,
|
||
'pageSize' => $pageSize,
|
||
'data' => []
|
||
];
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 获取收藏统计
|
||
*/
|
||
public static function getFavoriteStats($userId = 0)
|
||
{
|
||
try {
|
||
$db = self::getDbConnection();
|
||
|
||
$stmt = $db->prepare("SELECT COUNT(*) as total FROM urlnav_favorites WHERE user_id = ?");
|
||
$stmt->execute([$userId]);
|
||
$result = $stmt->fetch(PDO::FETCH_ASSOC);
|
||
|
||
return [
|
||
'total' => $result['total'] ?? 0
|
||
];
|
||
} catch (Exception $e) {
|
||
error_log('UrlNav: 获取收藏统计失败: ' . $e->getMessage());
|
||
return ['total' => 0];
|
||
}
|
||
}
|
||
}
|
||
|
||
/**
|
||
* RSS管理器类 - 封装RSS相关操作
|
||
*/
|
||
class UrlNav_RssManager
|
||
{
|
||
private $db;
|
||
|
||
public function __construct()
|
||
{
|
||
$this->db = UrlNav_Plugin::getDbConnection();
|
||
}
|
||
|
||
/**
|
||
* 智能刷新策略
|
||
*/
|
||
public function smartRefresh($urlId, $force = false)
|
||
{
|
||
$url = $this->getUrlInfo($urlId);
|
||
|
||
if (!$url) {
|
||
throw new Exception('URL不存在');
|
||
}
|
||
|
||
// 检查是否需要刷新
|
||
if (!$force && !$this->shouldRefresh($url)) {
|
||
return array('skipped' => true, 'reason' => '未到刷新时间');
|
||
}
|
||
|
||
// 执行刷新
|
||
return $this->refreshUrl($url);
|
||
}
|
||
|
||
private function getUrlInfo($urlId)
|
||
{
|
||
$stmt = $this->db->prepare("SELECT * FROM urlnav_urls WHERE id = ?");
|
||
$stmt->execute(array($urlId));
|
||
return $stmt->fetch(PDO::FETCH_ASSOC);
|
||
}
|
||
|
||
private function shouldRefresh($url)
|
||
{
|
||
$config = UrlNav_Plugin::getConfig();
|
||
$refreshInterval = intval($config->rssRefresh ?? 3600);
|
||
|
||
// 如果从未刷新过，需要刷新
|
||
if (empty($url['last_refresh'])) {
|
||
return true;
|
||
}
|
||
|
||
$lastRefresh = strtotime($url['last_refresh']);
|
||
$currentTime = time();
|
||
|
||
// 检查是否达到刷新间隔
|
||
if (($currentTime - $lastRefresh) >= $refreshInterval) {
|
||
return true;
|
||
}
|
||
|
||
// 如果最近失败次数多，降低刷新频率
|
||
$failureRate = $url['failure_count'] / max(1, $url['refresh_count']);
|
||
if ($failureRate > 0.5) {
|
||
// 失败率超过50%，延长刷新间隔
|
||
$extendedInterval = $refreshInterval * 3;
|
||
return ($currentTime - $lastRefresh) >= $extendedInterval;
|
||
}
|
||
|
||
return false;
|
||
}
|
||
|
||
private function refreshUrl($url)
|
||
{
|
||
// 这里可以实现单个URL的刷新逻辑
|
||
// 实际实现可以调用UrlNav_Plugin::parseRssFeed等方法
|
||
return array('success' => true);
|
||
}
|
||
}
|
||
?>
|