fix(route/xsijishe): update rank list parsing logic with puppeteer support

lib/routes/xsijishe/rank.ts

@@ -1,9 +1,10 @@
 import InvalidParameterError from '@/errors/types/invalid-parameter';
 import { Route } from '@/types';
 import cache from '@/utils/cache';
-import got from '@/utils/got';
 import { load } from 'cheerio';
-import { config } from '@/config';
+import { puppeteerGet } from './utils';
+import puppeteer from '@/utils/puppeteer';
+
 const baseUrl = 'https://xsijishe.com';
 
 export const route: Route = {
@@ -20,18 +21,8 @@ export const route: Route = {
         },
     },
     features: {
-        requireConfig: [
-            {
-                name: 'XSIJISHE_COOKIE',
-                description: '',
-            },
-            {
-                name: 'XSIJISHE_USER_AGENT',
-                description: '',
-            },
-        ],
-        requirePuppeteer: false,
-        antiCrawler: false,
+        requirePuppeteer: true,
+        antiCrawler: true,
         supportBT: false,
         supportPodcast: false,
         supportScihub: false,
@@ -42,6 +33,8 @@ export const route: Route = {
 };
 
 async function handler(ctx) {
+    const browser = await puppeteer();
+
     const rankType = ctx.req.param('type');
     let title;
     let index; // 用于选择第几个 li
@@ -57,28 +50,9 @@ async function handler(ctx) {
     }
 
     const url = `${baseUrl}/portal.php`;
-    const headers = {
-        'Accept-Encoding': 'gzip, deflate, br',
-        'Accept-Language': 'zh-CN,zh;q=0.9,en;q=0.8',
-        Cookie: config.xsijishe.cookie,
-        'User-Agent': config.xsijishe.user_agent,
-    };
-
-    const resp = await got(url, {
-        headers,
-    });
-
-    const redirectMatch = resp.data.match(/window\.location\.href\s*=\s*"([^"]+)"/);
-    if (redirectMatch) {
-        const redirectUrl = `${baseUrl}${redirectMatch[1]}`;
-        // 使用提取到的地址重新请求
-        const realResp = await got(redirectUrl, {
-            headers,
-        });
-        resp.data = realResp.data;
-    }
+    const data = await cache.tryGet(url, () => puppeteerGet(url, browser));
+    const $ = load(data);
 
-    const $ = load(resp.data);
     // 根据 index 选择对应的 li，然后获取其中的 dd 元素
     let items = $('.nex_recon_lists ul li')
         .eq(index)
@@ -97,10 +71,8 @@ async function handler(ctx) {
     items = await Promise.all(
         items.map((item) =>
             cache.tryGet(item.link, async () => {
-                const resp = await got(item.link, {
-                    headers,
-                });
-                const $ = load(resp.data);
+                const data = await puppeteerGet(item.link, browser);
+                const $ = load(data);
                 const firstViewBox = $('.t_f').first();
 
                 firstViewBox.find('img').each((_, img) => {
@@ -118,6 +90,9 @@ async function handler(ctx) {
             })
         )
     );
+
+    await browser.close();
+
     return {
         title,
         link: url,

lib/routes/xsijishe/utils.ts

@@ -0,0 +1,19 @@
+const puppeteerGet = async (url, browser) => {
+    const isIndex = url.endsWith('/portal.php');
+    const page = await browser.newPage();
+    const expectResourceTypes = new Set(['document', 'script']);
+    await page.setRequestInterception(true);
+    page.on('request', (request) => {
+        expectResourceTypes.has(request.resourceType()) ? request.continue() : request.abort();
+    });
+    await page.goto(url, {
+        waitUntil: 'domcontentloaded',
+    });
+
+    await (isIndex ? page.waitForSelector('.nex_recon_lists') : page.waitForSelector('.t_f'));
+    const html = await page.evaluate(() => document.documentElement.innerHTML);
+    await page.close();
+    return html;
+};
+
+export { puppeteerGet };