rsshub 1.0.0-master.f9003f9 → 1.0.0-master.f917bdc

package/lib/routes/zaobao/{index.ts → other.ts}

@@ -3,18 +3,10 @@ import { parseList } from './util';
 const baseUrl = 'https://www.zaobao.com';
 
 export const route: Route = {
-    path: '/:type?/:section?',
+    path: '/other/:type?/:section?',
     categories: ['traditional-media'],
-    example: '/zaobao/lifestyle/health',
+    example: '/zaobao/other/lifestyle/health',
     parameters: { type: 'https://www.zaobao.com/**lifestyle**/health 中的 **lifestyle**', section: 'https://www.zaobao.com/lifestyle/**health** 中的 **health**' },
-    features: {
-        requireConfig: false,
-        requirePuppeteer: false,
-        antiCrawler: false,
-        supportBT: false,
-        supportPodcast: false,
-        supportScihub: false,
-    },
     name: '其他栏目',
     maintainers: ['shunf4'],
     handler,

package/lib/routes/zaobao/realtime.ts

@@ -7,14 +7,6 @@ export const route: Route = {
     categories: ['traditional-media'],
     example: '/zaobao/realtime/china',
     parameters: { section: '分类，缺省为 china' },
-    features: {
-        requireConfig: false,
-        requirePuppeteer: false,
-        antiCrawler: false,
-        supportBT: false,
-        supportPodcast: false,
-        supportScihub: false,
-    },
     name: '即时新闻',
     maintainers: ['shunf4'],
     handler,

package/lib/routes/zaobao/util.ts

@@ -22,19 +22,23 @@ const got_ins = got.extend({
  *
  * @param {*} ctx RSSHub 的 ctx 参数，用来设置缓存
  * @param {string} sectionUrl 形如 /realtime/china 的字符串
- * @returns {Promise<{
- *  title: string;
- *  resultList: {
- *    title: string;
- *    description: string;
- *    pubDate: string;
- *    link: string;
- *  }[];}>} 新闻标题以及新闻列表
+ * @returns 新闻标题以及新闻列表
  */
-const parseList = async (ctx, sectionUrl) => {
+const parseList = async (
+    ctx,
+    sectionUrl: string
+): Promise<{
+    title: string;
+    resultList: {
+        title: string;
+        description: string;
+        pubDate: string;
+        link: string;
+    }[];
+}> => {
     const response = await got_ins.get(baseUrl + sectionUrl);
     const $ = load(response.data);
-    let data = $('.article-list').find('.article-type');
+    let data = /realtime/.test(sectionUrl) ? $('.on-listing-pages') : $('.article-list').find('.article-type');
     if (data.length === 0) {
         // for HK version
         data = $('.clearfix').find('.list-block');

package/lib/routes/zcool/discover.ts

@@ -219,7 +219,8 @@ async function handler(ctx) {
                 queries.subCate = '809824';
                 queries.recommendLevel = '2';
                 break;
-            case 'editor' || 'edit':
+            case 'editor':
+            case 'edit':
                 queries.recommendLevel = '2';
                 break;
             default:

package/lib/routes/zhihu/activities.ts

@@ -1,4 +1,4 @@
-import { Route } from '@/types';
+import { Route, ViewType } from '@/types';
 import got from '@/utils/got';
 import { header, processImage, getSignedHeader } from './utils';
 import { parseDate } from '@/utils/parse-date';
@@ -6,6 +6,7 @@ import { parseDate } from '@/utils/parse-date';
 export const route: Route = {
     path: '/people/activities/:id',
     categories: ['social-media', 'popular'],
+    view: ViewType.Articles,
     example: '/zhihu/people/activities/diygod',
     parameters: { id: '作者 id，可在用户主页 URL 中找到' },
     features: {

package/lib/routes/zhihu/execlib/x-zse-96-v3.ts

@@ -2,6 +2,8 @@
 // https://blog.csdn.net/zjq592767809/article/details/126512798
 // https://blog.csdn.net/zhoumi_/article/details/126659351
 
+/* eslint-disable @typescript-eslint/no-unused-expressions */
+
 function i(e, t, n) {
     (t[n] = 255 & (e >>> 24)), (t[n + 1] = 255 & (e >>> 16)), (t[n + 2] = 255 & (e >>> 8)), (t[n + 3] = 255 & e);
 }

package/lib/routes/zhihu/hot.ts

@@ -1,4 +1,4 @@
-import { Route } from '@/types';
+import { Route, ViewType } from '@/types';
 import got from '@/utils/got';
 import { parseDate } from '@/utils/parse-date';
 
@@ -18,9 +18,61 @@ const titles = {
 
 export const route: Route = {
     path: '/hot/:category?',
-    categories: ['social-media'],
+    categories: ['social-media', 'popular'],
     example: '/zhihu/hot',
-    parameters: { category: '分类，见下表，默认为全站' },
+    view: ViewType.Articles,
+    parameters: {
+        category: {
+            description: '分类',
+            default: 'total',
+            options: [
+                {
+                    value: 'total',
+                    label: '全站',
+                },
+                {
+                    value: 'focus',
+                    label: '国际',
+                },
+                {
+                    value: 'science',
+                    label: '科学',
+                },
+                {
+                    value: 'car',
+                    label: '汽车',
+                },
+                {
+                    value: 'zvideo',
+                    label: '视频',
+                },
+                {
+                    value: 'fashion',
+                    label: '时尚',
+                },
+                {
+                    value: 'depth',
+                    label: '时事',
+                },
+                {
+                    value: 'digital',
+                    label: '数码',
+                },
+                {
+                    value: 'sport',
+                    label: '体育',
+                },
+                {
+                    value: 'school',
+                    label: '校园',
+                },
+                {
+                    value: 'film',
+                    label: '影视',
+                },
+            ],
+        },
+    },
     features: {
         requireConfig: false,
         requirePuppeteer: false,
@@ -29,12 +81,9 @@ export const route: Route = {
         supportPodcast: false,
         supportScihub: false,
     },
-    name: '知乎分类热榜',
+    name: '知乎热榜',
     maintainers: ['nczitzk'],
     handler,
-    description: `| 全站  | 国际  | 科学    | 汽车 | 视频   | 时尚    | 时事  | 数码    | 体育  | 校园   | 影视 |
-  | ----- | ----- | ------- | ---- | ------ | ------- | ----- | ------- | ----- | ------ | ---- |
-  | total | focus | science | car  | zvideo | fashion | depth | digital | sport | school | film |`,
 };
 
 async function handler(ctx) {

package/lib/routes/zhihu/namespace.ts

@@ -3,4 +3,7 @@ import type { Namespace } from '@/types';
 export const namespace: Namespace = {
     name: '知乎',
     url: 'www.zhihu.com',
+    description: `:::tip
+自2024年7月，未登录状态下大部分路由[无法获取全文](https://github.com/DIYgod/RSSHub/issues/16260)。若有需要请在登陆知乎后寻找并添加包含\`z_c0\`的Cookies至环境变量\`ZHIHU_COOKIES\`。
+:::`,
 };

package/lib/routes/zhihu/posts.ts

@@ -1,8 +1,9 @@
 import { Route } from '@/types';
 import ofetch from '@/utils/ofetch';
+import cache from '@/utils/cache';
 import { header, getSignedHeader, processImage } from './utils';
-import { load } from 'cheerio';
 import { parseDate } from '@/utils/parse-date';
+import { Articles, Profile } from './types';
 
 export const route: Route = {
     path: '/posts/:usertype/:id',
@@ -19,7 +20,7 @@ export const route: Route = {
     },
     radar: [
         {
-            source: ['www.zhihu.com/:usertype/:id/posts'],
+            source: ['www.zhihu.com/:usertype/:id/posts', 'www.zhihu.com/:usertype/:id'],
         },
     ],
     name: '用户文章',
@@ -34,15 +35,19 @@ async function handler(ctx) {
     const id = ctx.req.param('id');
     const usertype = ctx.req.param('usertype');
 
-    const data = await ofetch(`https://www.zhihu.com/${usertype}/${id}/posts`, {
-        headers: {
-            ...header,
-            Referer: `https://www.zhihu.com/${usertype}/${id}/`,
-        },
+    const userProfile = await cache.tryGet(`zhihu:posts:profile:${id}`, async () => {
+        const userAPIPath = `/api/v4/${usertype === 'people' ? 'members' : 'org'}/${id}?${new URLSearchParams({
+            include: 'allow_message,is_followed,is_following,is_org,is_blocking,employments,answer_count,follower_count,articles_count,gender,badge[?(type=best_answerer)].topics',
+        })}`;
+
+        return await ofetch<Profile>(`https://www.zhihu.com${userAPIPath}`, {
+            headers: {
+                ...header,
+                ...(await getSignedHeader(`https://www.zhihu.com/${usertype}/${id}/`, userAPIPath)),
+                Referer: `https://www.zhihu.com/${usertype}/${id}/`,
+            },
+        });
     });
-    const $ = load(data);
-    const initialData = JSON.parse($('#js-initialData').text());
-    const userData = initialData?.initialState?.entities?.users?.[id];
 
     const apiPath = `/api/v4/${usertype === 'people' ? 'members' : 'org'}/${id}/articles?${new URLSearchParams({
         include:
@@ -54,7 +59,7 @@ async function handler(ctx) {
 
     const signedHeader = await getSignedHeader(`https://www.zhihu.com/${usertype}/${id}/posts`, apiPath);
 
-    const articleResponse = await ofetch(`https://www.zhihu.com${apiPath}`, {
+    const articleResponse = await ofetch<Articles>(`https://www.zhihu.com${apiPath}`, {
         headers: {
             ...header,
             ...signedHeader,
@@ -72,11 +77,11 @@ async function handler(ctx) {
     }));
 
     return {
-        title: `${userData?.name} 的知乎文章`,
+        title: `${userProfile.name} 的知乎文章`,
         link: `https://www.zhihu.com/${usertype}/${id}/posts`,
-        description: userData?.headline,
-        image: userData?.avatarUrlTemplate?.split('?')[0],
-        banner: userData?.coverUrl?.split('?')[0],
+        description: userProfile.headline,
+        image: userProfile.avatar_url.split('?')[0],
+        // banner: userData?.coverUrl?.split('?')[0],
         item: items,
     };
 }

package/lib/routes/zhihu/types.ts

@@ -0,0 +1,104 @@
+export interface Profile {
+    id: string;
+    url_token: string;
+    name: string;
+    use_default_avatar: boolean;
+    avatar_url: string;
+    avatar_url_template: string;
+    is_org: boolean;
+    type: string;
+    url: string;
+    user_type: string;
+    headline: string;
+    headline_render: string;
+    gender: number;
+    is_advertiser: boolean;
+    ip_info: string;
+    vip_info: {
+        is_vip: boolean;
+        vip_type: number;
+        rename_days: string;
+        entrance_v2: null;
+        rename_frequency: number;
+        rename_await_days: number;
+    };
+    badge: any[];
+    badge_v2: {
+        title: string;
+        merged_badges: any[];
+        detail_badges: any[];
+        icon: string;
+        night_icon: string;
+    };
+    allow_message: boolean;
+    is_following: boolean;
+    is_followed: boolean;
+    is_blocking: boolean;
+    follower_count: number;
+    answer_count: number;
+    articles_count: number;
+    available_medals_count: number;
+    employments: any[];
+    is_realname: boolean;
+    has_applying_column: boolean;
+}
+
+interface Article {
+    image_url: string;
+    updated: number;
+    is_jump_native: boolean;
+    is_labeled: boolean;
+    copyright_permission: string;
+    vessay_info: {
+        enable_video_translate: boolean;
+    };
+    excerpt: string;
+    admin_closed_comment: boolean;
+    article_type: string;
+    excerpt_title: string;
+    reaction_instruction: {
+        REACTION_CONTENT_SEGMENT_LIKE: string;
+    };
+    id: number;
+    voteup_count: number;
+    upvoted_followees: [];
+    can_comment: {
+        status: boolean;
+        reason: string;
+    };
+    author: Profile;
+    url: string;
+    comment_permission: string;
+    created: number;
+    image_width: number;
+    content: string;
+    comment_count: number;
+    linkbox: {
+        url: string;
+        category: string;
+        pic: string;
+        title: string;
+    };
+    title: string;
+    voting: number;
+    type: string;
+    suggest_edit: {
+        status: boolean;
+        url: string;
+        reason: string;
+        tip: string;
+        title: string;
+    };
+    is_normal: boolean;
+}
+
+export interface Articles {
+    paging: {
+        is_end: boolean;
+        totals: number;
+        previous: string;
+        is_start: boolean;
+        next: string;
+    };
+    data: Article[];
+}

package/lib/routes/zhihu/utils.ts

@@ -3,6 +3,7 @@ import cache from '@/utils/cache';
 import ofetch from '@/utils/ofetch';
 import g_encrypt from './execlib/x-zse-96-v3';
 import md5 from '@/utils/md5';
+import { config } from '@/config';
 
 export const header = {
     'x-api-version': '3.0.91',
@@ -62,12 +63,26 @@ export const getSignedHeader = async (url: string, apiPath: string) => {
 
     // fisrt: get cookie(dc_0) from zhihu.com
     const dc0 = await cache.tryGet('zhihu:cookies:d_c0', async () => {
-        const response = await ofetch.raw(url);
+        const response1 = await ofetch.raw(url);
+        const zseCk = response1._data.match(/var e="__zse_ck",t=\(typeof __g\.ck == 'string' && __g\.ck\)\|\|"([\w+/=]*?)",_=6048e5;/)?.[1];
 
-        const dc0 = response.headers
+        const response2 = zseCk
+            ? await ofetch.raw(url, {
+                  headers: {
+                      cookie: `${response1.headers
+                          .getSetCookie()
+                          .map((s) => s.split(';')[0])
+                          .join('; ')}; __zse_ck=${zseCk}`,
+                  },
+              })
+            : null;
+
+        const dc0 = (response2 || response1).headers
             .getSetCookie()
             .find((s) => s.startsWith('d_c0='))
-            ?.split(';')[0];
+            ?.split(';')[0]
+            .trim()
+            .slice('d_c0='.length);
         if (!dc0) {
             throw new Error('Failed to extract `d_c0` from cookies');
         }
@@ -79,8 +94,14 @@ export const getSignedHeader = async (url: string, apiPath: string) => {
     const xzse93 = '101_3_3.0';
     const f = `${xzse93}+${apiPath}+${dc0}`;
     const xzse96 = '2.0_' + g_encrypt(md5(f));
+
+    const zc0 = config.zhihu.cookies
+        ?.split(';')
+        .map((e) => e.trim())
+        .find((e) => e.includes('z_c0'))
+        ?.slice('z_c0='.length);
     return {
-        cookie: `d_c0=${dc0}`,
+        cookie: `d_c0=${dc0}${zc0 ? `;z_c0=${zc0}` : ''}`,
         'x-zse-96': xzse96,
         'x-app-za': 'OS=Web',
         'x-zse-93': xzse93,

package/lib/routes/zhihu/zhuanlan.ts

@@ -1,6 +1,6 @@
 import { Route } from '@/types';
 import got from '@/utils/got';
-import { header } from './utils';
+import { getSignedHeader, header } from './utils';
 import { load } from 'cheerio';
 import { parseDate } from '@/utils/parse-date';
 
@@ -56,7 +56,13 @@ async function handler(ctx) {
         url = `https://www.zhihu.com/column/${id}`;
     }
 
-    const infoRes = await got(url);
+    const signedHeader = await getSignedHeader(url, `https://www.zhihu.com/api/v4/columns/${id}/items`);
+    const infoRes = await got(url, {
+        headers: {
+            ...signedHeader,
+            Referer: url,
+        },
+    });
     const $ = load(infoRes.data);
     const title = $('.css-zyehvu').text();
     const description = $('.css-1bnklpv').text();

package/lib/routes/zhubai/top20.ts

@@ -51,7 +51,7 @@ async function handler(ctx) {
 
     let items = response.data.slice(0, limit).map((item) => ({
         title: item.pn,
-        link: item.fp ?? item.pq ?? item.pu,
+        link: item.pu ?? item.pq ?? item.fp,
         description: item.pa,
         author: item.zn,
         pubDate: parseRelativeDate(item.lu.replace(/\.\d+/, '')),
@@ -60,7 +60,7 @@ async function handler(ctx) {
     items = await Promise.all(
         items.map((item) =>
             cache.tryGet(item.link, async () => {
-                const matches = item.link.match(/\/(?:fp|pq|pu)\/([\w-]+)\/(\d+)/);
+                const matches = item.link.match(/\/(?:pl|pq|fp)\/([\w-]+)\/(\d+)/);
 
                 const { data } = await got(`https://${matches[1]}.zhubai.love/api/posts/${matches[2]}`);
 

package/lib/routes/zhuwang/index.ts

@@ -26,7 +26,7 @@ export const route: Route = {
 };
 
 async function handler() {
-    const baseUrl = 'https://zhujia.zhuwang.cc/';
+    const baseUrl = 'https://zhujia.zhuwang.com.cn/';
     const now = new Date();
     const date = `${now.getFullYear()}-${now.getMonth() + 1}-${now.getDate()}`;
     const response = await got(`${baseUrl}/api/chartData`, {

package/lib/routes/zjgtjy/index.ts

@@ -1,6 +1,6 @@
 import { Route } from '@/types';
 import cache from '@/utils/cache';
-import got from '@/utils/got';
+import ofetch from '@/utils/ofetch';
 
 export const route: Route = {
     path: '/:type?',
@@ -13,10 +13,7 @@ async function handler(ctx) {
     const type = ctx.req.param('type') === 'all' ? '' : ctx.req.param('type').toUpperCase();
     const host = `https://td.zjgtjy.cn:8553/devops/noticeInfo/queryNoticeInfoList?pageSize=10&pageNumber=1&noticeType=${type}&sort=DESC`;
 
-    const response = await got({
-        method: 'get',
-        url: host,
-    }).json();
+    const response = await ofetch(host);
     const data = response.data;
 
     const items = await Promise.all(
@@ -25,10 +22,7 @@ async function handler(ctx) {
             const pageLink = `https://td.zjgtjy.cn/view/trade/announcement/detail?id=${item.GGID}&category=${item.ZYLB}&type=${item.JYFS}`;
 
             const desc = await cache.tryGet(pageUrl, async () => {
-                let desc = await got({
-                    method: 'get',
-                    url: pageUrl,
-                }).json();
+                let desc = await ofetch(pageUrl);
                 desc = desc.queryNoticeContent.GGNR;
 
                 desc = desc.replaceAll('&lt;', '<').replaceAll('&gt;', '>').replaceAll('&quot;', '"');

package/lib/routes.test.ts

@@ -69,10 +69,16 @@ async function checkRSS(response) {
 
 describe('routes', () => {
     for (const route in routes) {
-        it.concurrent(route, async () => {
-            const response = await app.request(routes[route]);
-            expect(response.status).toBe(200);
-            await checkRSS(response);
-        });
+        it.concurrent(
+            route,
+            {
+                timeout: 60000,
+            },
+            async () => {
+                const response = await app.request(routes[route]);
+                expect(response.status).toBe(200);
+                await checkRSS(response);
+            }
+        );
     }
 });

package/lib/types.ts

@@ -35,7 +35,13 @@ export type DataItem = {
     pubDate?: number | string | Date;
     link?: string;
     category?: string[];
-    author?: string | { name: string }[];
+    author?:
+        | string
+        | {
+              name: string;
+              url?: string;
+              avatar?: string;
+          }[];
     doi?: string;
     guid?: string;
     id?: string;
@@ -119,6 +125,15 @@ interface Namespace extends NamespaceItem {
 
 export type { Namespace };
 
+export enum ViewType {
+    Articles = 0,
+    SocialMedia = 1,
+    Pictures = 2,
+    Videos = 3,
+    Audios = 4,
+    Notifications = 5,
+}
+
 // route
 interface RouteItem {
     /**
@@ -155,7 +170,18 @@ interface RouteItem {
     /**
      * The description of the route parameters
      */
-    parameters?: Record<string, string>;
+    parameters?: Record<
+        string,
+        | string
+        | {
+              description: string;
+              default?: string;
+              options?: {
+                  value: string;
+                  label: string;
+              }[];
+          }
+    >;
 
     /**
      * Hints and additional explanations for users using this route, it will be appended after the route component, supports markdown
@@ -206,6 +232,11 @@ interface RouteItem {
      * The [RSSHub-Radar](https://github.com/DIYgod/RSSHub-Radar) rule of the route
      */
     radar?: RadarItem[];
+
+    /**
+     * The [Follow](https://github.com/RSSNext/follow) default view of the route, default to `ViewType.Articles`
+     */
+    view?: ViewType;
 }
 
 interface Route extends RouteItem {
@@ -242,6 +273,7 @@ export type RadarItem = {
      *
      * Using `target` as a function is deprecated in RSSHub-Radar 2.0.19
      * @see https://github.com/DIYgod/RSSHub-Radar/commit/5a97647f900bb2bca792787a322b2b1ca512e40b#diff-f84e3c1e16af314bc4ed7c706d7189844663cde9b5142463dc5c0db34c2e8d54L10
+     * @see https://github.com/DIYgod/RSSHub-Radar/issues/692
      */
     target?:
         | string

package/lib/utils/cache/index.ts

@@ -63,6 +63,14 @@ if (config.cache.type === 'redis') {
 // plz, write these tips in comments!
 export default {
     ...cacheModule,
+    /**
+     * Try to get the cache. If the cache does not exist, the `getValueFunc` function will be called to get the data, and the data will be cached.
+     * @param key The key used to store and retrieve the cache. You can use `:` as a separator to create a hierarchy.
+     * @param getValueFunc A function that returns data to be cached when a cache miss occurs.
+     * @param maxAge The maximum age of the cache in seconds. This should left to the default value in most cases which is `CACHE_CONTENT_EXPIRE`.
+     * @param refresh Whether to renew the cache expiration time when the cache is hit. `true` by default.
+     * @returns
+     */
     tryGet: async (key: string, getValueFunc: () => Promise<string | Record<string, any>>, maxAge = config.cache.contentExpire, refresh = true) => {
         if (typeof key !== 'string') {
             throw new TypeError('Cache key must be a string');

package/lib/utils/ofetch.ts

@@ -3,9 +3,23 @@ import { config } from '@/config';
 import logger from '@/utils/logger';
 
 const rofetch = createFetch().create({
+    retryStatusCodes: [400, 408, 409, 425, 429, 500, 502, 503, 504],
     retry: config.requestRetry,
     retryDelay: 1000,
-    timeout: config.requestTimeout,
+    // timeout: config.requestTimeout,
+    onResponseError({ request, response, options }) {
+        if (options.retry) {
+            logger.warn(`Request ${request} with error ${response.status} remaining retry attempts: ${options.retry}`);
+            if (!options.headers) {
+                options.headers = {};
+            }
+            if (options.headers instanceof Headers) {
+                options.headers.set('x-prefer-proxy', '1');
+            } else {
+                options.headers['x-prefer-proxy'] = '1';
+            }
+        }
+    },
     onRequestError({ request, error }) {
         logger.error(`Request ${request} fail: ${error}`);
     },

package/lib/utils/otel/index.ts

@@ -0,0 +1,2 @@
+export * from './metric';
+export * from './trace';