feat(route/baidu): add BAIDU_COOKIE support and extract shared …

lib/config.ts

@@ -85,6 +85,7 @@ type ConfigEnvKeys =
     | 'BITBUCKET_USERNAME'
     | 'BITBUCKET_PASSWORD'
     | 'BTBYR_HOST'
+    | 'BAIDU_COOKIE'
     | 'BTBYR_COOKIE'
     | 'BUPT_PORTAL_COOKIE'
     | 'CAIXIN_COOKIE'
@@ -348,6 +349,9 @@ export type Config = {
     };
 
     // Route-specific Configurations
+    baidu: {
+        cookie?: string;
+    };
     bilibili: {
         cookies: Record<string, string | undefined>;
         dmImgList?: string;
@@ -763,7 +767,7 @@ const calculateValue = () => {
             type: envs.CACHE_TYPE || (envs.CACHE_TYPE === '' ? '' : 'memory'), // 缓存类型，支持 'memory' 和 'redis'，设为空可以禁止缓存
             requestTimeout: toInt(envs.CACHE_REQUEST_TIMEOUT, 60),
             routeExpire: toInt(envs.CACHE_EXPIRE, 5 * 60), // 路由缓存时间，单位为秒
-            contentExpire: toInt(envs.CACHE_CONTENT_EXPIRE, 1 * 60 * 60), // 不变内容缓存时间，单位为秒
+            contentExpire: toInt(envs.CACHE_CONTENT_EXPIRE, 60 * 60), // 不变内容缓存时间，单位为秒
         },
         memory: {
             max: toInt(envs.MEMORY_MAX, Math.pow(2, 8)), // The maximum number of items that remain in the cache. This must be a positive finite intger.
@@ -843,6 +847,9 @@ const calculateValue = () => {
         },
 
         // Route-specific Configurations
+        baidu: {
+            cookie: envs.BAIDU_COOKIE,
+        },
         bilibili: {
             cookies: bilibili_cookies,
             dmImgList: envs.BILIBILI_DM_IMG_LIST,

lib/routes/baidu/tieba/common.ts

@@ -0,0 +1,122 @@
+import { config } from '@/config';
+import ConfigNotFoundError from '@/errors/types/config-not-found';
+import cache from '@/utils/cache';
+
+/**
+ * 解析百度 cookie 字符串为 Puppeteer 可用的 cookie 对象数组
+ * 正确处理包含 '=' 的 cookie 值
+ */
+export function parseBaiduCookies(cookieStr: string): Array<{ name: string; value: string; domain: string }> {
+    return cookieStr
+        .split(';')
+        .map((c) => c.trim())
+        .filter((c) => c.length > 0)
+        .map((c) => {
+            const firstEqualIndex = c.indexOf('=');
+            if (firstEqualIndex === -1) {
+                return { name: c, value: '', domain: '.tieba.baidu.com' };
+            }
+            const name = c.slice(0, firstEqualIndex).trim();
+            const value = c.slice(firstEqualIndex + 1).trim();
+            return { name, value, domain: '.tieba.baidu.com' };
+        });
+}
+
+/**
+ * 检查 HTML 内容是否包含百度安全验证页面
+ */
+export function checkSecurityVerification(html: string): void {
+    if (html.includes('安全验证') || html.includes('百度安全验证')) {
+        throw new Error('Baidu security verification required. The cookie may be expired or invalid. Please update your BAIDU_COOKIE.');
+    }
+}
+
+/**
+ * 使用 Puppeteer 获取贴吧页面内容
+ * 包含统一的 cookie 设置、安全验证检查和缓存逻辑
+ * 带有重试机制处理瞬态错误
+ */
+export async function getTiebaPageContent(
+    url: string,
+    cacheKey: string,
+    options: {
+        waitForSelector?: string;
+        timeout?: number;
+        retries?: number;
+    } = {}
+): Promise<string> {
+    const cookie = config.baidu.cookie;
+
+    if (!cookie) {
+        throw new ConfigNotFoundError('Baidu Tieba RSS is disabled due to the lack of <a href="https://docs.rsshub.app/deploy/config#baidu">BAIDU_COOKIE</a>');
+    }
+
+    const { getPuppeteerPage } = await import('@/utils/puppeteer');
+    const { waitForSelector = '.thread-card-wrapper, .virtual-list-item, .thread-content-box, .thread-card', timeout = 3000, retries = 3 } = options;
+
+    const data = await cache.tryGet(
+        cacheKey,
+        async () => {
+            let lastError: Error | undefined;
+
+            /* eslint-disable no-await-in-loop -- Intentional sequential retry logic */
+            for (let attempt = 0; attempt < retries; attempt++) {
+                const { page, destroy } = await getPuppeteerPage(url, { noGoto: true });
+
+                try {
+                    // 设置 Cookie（在访问页面前设置，减少一次导航）
+                    const cookies = parseBaiduCookies(cookie);
+                    await page.setCookie(...cookies);
+
+                    // 访问目标页面 - 使用更宽松的等待条件
+                    await page.goto(url, { waitUntil: 'domcontentloaded', timeout: 60000 });
+
+                    // 等待页面稳定
+                    await new Promise((resolve) => setTimeout(resolve, 2000));
+
+                    // 动态等待内容加载
+                    try {
+                        await page.waitForSelector(waitForSelector, { timeout });
+                    } catch {
+                        // 如果超时，继续执行
+                    }
+
+                    return await page.content();
+                } catch (error) {
+                    lastError = error as Error;
+                    // 如果是最后一次尝试，抛出错误
+                    if (attempt === retries - 1) {
+                        throw lastError;
+                    }
+                    // 等待后重试
+                    await new Promise((resolve) => setTimeout(resolve, 1000 * (attempt + 1)));
+                } finally {
+                    await destroy();
+                }
+            }
+            /* eslint-enable no-await-in-loop */
+
+            throw lastError || new Error('Failed to fetch page content');
+        },
+        config.cache.routeExpire,
+        false
+    );
+
+    const html = data as string;
+    checkSecurityVerification(html);
+    return html;
+}
+
+/**
+ * 规范化 URL 为绝对地址
+ */
+export function normalizeUrl(href: string, base: string = 'https://tieba.baidu.com'): string {
+    if (!href) {
+        return '';
+    }
+    if (href.startsWith('http')) {
+        return href;
+    }
+    const path = href.startsWith('/') ? href : `/${href}`;
+    return `${base}${path}`;
+}

lib/routes/baidu/tieba/forum.tsx

@@ -1,85 +1,101 @@
 import { load } from 'cheerio';
-import { raw } from 'hono/html';
 import { renderToString } from 'hono/jsx/dom/server';
 
 import type { Route } from '@/types';
-import got from '@/utils/got';
-import { parseDate } from '@/utils/parse-date';
 import timezone from '@/utils/timezone';
 
+import { getTiebaPageContent, normalizeUrl } from './common';
+import { parseRelativeTime, parseThreads } from './utils';
+
 export const route: Route = {
     path: ['/tieba/forum/good/:kw/:cid?/:sortBy?', '/tieba/forum/:kw/:sortBy?'],
     categories: ['bbs'],
     example: '/baidu/tieba/forum/good/女图',
     parameters: { kw: '吧名', cid: '精品分类，默认为 `0`（全部分类），如果不传 `cid` 则获取全部分类', sortBy: '排序方式：`created`, `replied`。默认为 `created`' },
     features: {
-        requireConfig: false,
-        requirePuppeteer: false,
-        antiCrawler: false,
+        requireConfig: [
+            {
+                name: 'BAIDU_COOKIE',
+                optional: false,
+                description: '百度 cookie 值，用于需要登录的贴吧页面',
+            },
+        ],
+        requirePuppeteer: true,
+        antiCrawler: true,
         supportBT: false,
         supportPodcast: false,
         supportScihub: false,
     },
     name: '精品帖子',
-    maintainers: ['u3u'],
+    maintainers: ['u3u', 'FlanChanXwO'],
     handler,
 };
 
 async function handler(ctx) {
     // sortBy: created, replied
     const { kw, cid = '0', sortBy = 'created' } = ctx.req.param();
+    const sortParam = sortBy === 'replied' ? '&sc=67108864' : '';
 
-    // PC端：https://tieba.baidu.com/f?kw=${encodeURIComponent(kw)}
-    // 移动端接口：https://tieba.baidu.com/mo/q/m?kw=${encodeURIComponent(kw)}&lp=5024&forum_recommend=1&lm=0&cid=0&has_url_param=1&pn=0&is_ajax=1
-    const params = { kw: encodeURIComponent(kw) };
-    ctx.req.path.includes('good') && (params.tab = 'good');
-    cid && (params.cid = cid);
-    const { data } = await got(`https://tieba.baidu.com/f`, {
-        headers: {
-            Referer: 'https://tieba.baidu.com/',
-        },
-        searchParams: params,
-    });
+    // 固定抓取3页，约30条帖子
+    const maxPages = 3;
+
+    // 并发获取所有页面
+    const pagePromises = [];
+    for (let pageNum = 0; pageNum < maxPages; pageNum++) {
+        const pageUrl = `https://tieba.baidu.com/f?kw=${encodeURIComponent(kw)}&pn=${pageNum * 50}${cid === '0' ? '' : `&cid=${cid}`}${ctx.req.path.includes('good') ? '&tab=good' : ''}${pageNum === 0 ? '' : '&ie=utf-8'}${sortParam}`;
+
+        const promise = getTiebaPageContent(pageUrl, `tieba:forum:${kw}:${cid}:${sortBy}:page${pageNum}`, { waitForSelector: '.thread-card-wrapper', timeout: 3000 });
+        pagePromises.push(promise);
+    }
 
-    const threadListHTML = load(data)('code[id="pagelet_html_frs-list/pagelet/thread_list"]')
-        .contents()
-        .filter((e) => e.nodeType === '8');
+    // 等待所有页面获取完成
+    const pageResults = await Promise.all(pagePromises);
 
-    const $ = load(threadListHTML.prevObject[0].data);
-    const list = $('#thread_list > .j_thread_list[data-field]')
-        .toArray()
-        .map((element) => {
-            const item = $(element);
-            const { id, author_name } = item.data('field');
-            const time = sortBy === 'created' ? item.find('.is_show_create_time').text().trim() : item.find('.threadlist_reply_date').text().trim();
-            const title = item.find('a.j_th_tit').text().trim();
-            const details = item.find('.threadlist_abs').text().trim();
-            const medias = item
-                .find('.threadlist_media img')
-                .toArray()
-                .map((element) => {
-                    const item = $(element);
-                    return `<img src="${item.attr('bpic')}">`;
-                })
-                .join('');
+    // 解析所有页面数据并去重
+    const threadMap = new Map();
+    for (const html of pageResults) {
+        if (html && html.length > 0) {
+            const $ = load(html);
+            const threads = parseThreads($);
+            for (const thread of threads) {
+                // 使用帖子ID去重，只保留第一次出现的
+                if (!threadMap.has(thread.id)) {
+                    threadMap.set(thread.id, thread);
+                }
+            }
+        }
+    }
 
-            return {
-                title,
-                description: renderToString(
-                    <>
-                        <p>{details}</p>
-                        <p>{raw(medias)}</p>
-                        <p>作者：{author_name}</p>
-                    </>
-                ),
-                pubDate: timezone(parseDate(time, ['HH:mm', 'M-D', 'YYYY-MM'], true), +8),
-                link: `https://tieba.baidu.com/p/${id}`,
-            };
-        });
+    const allThreads = [...threadMap.values()];
+
+    if (allThreads.length === 0) {
+        throw new Error('No threads found. The cookie may be expired or invalid. Please check your BAIDU_COOKIE.');
+    }
+
+    const list = allThreads.map((thread) => {
+        const parsedDate = parseRelativeTime(thread.time);
+        return {
+            title: thread.title,
+            link: normalizeUrl(thread.link) || `https://tieba.baidu.com/p/${thread.id}`,
+            pubDate: parsedDate ? timezone(parsedDate, +8) : undefined,
+            author: thread.author,
+            description: renderToString(
+                <>
+                    {thread.content ? <p>{thread.content}</p> : null}
+                    {thread.images && thread.images.length > 0 ? (
+                        <div>
+                            {thread.images.map((img) => (
+                                <img src={img} alt="" style={{ maxWidth: '100%', margin: '5px 0' }} />
+                            ))}
+                        </div>
+                    ) : null}
+                </>
+            ),
+        };
+    });
 
     return {
         title: `${kw}吧`,
-        description: load(data)('meta[name="description"]').attr('content'),
         link: `https://tieba.baidu.com/f?kw=${encodeURIComponent(kw)}`,
         item: list,
     };