rsshub 1.0.0-master.f71451d → 1.0.0-master.f72af1b

package/lib/routes/yystv/docs.ts

@@ -1,7 +1,8 @@
-import { Route } from '@/types';
-import got from '@/utils/got';
+import type { Route, DataItem } from '@/types';
+import ofetch from '@/utils/ofetch';
 import { load } from 'cheerio';
 import { parseRelativeDate } from '@/utils/parse-date';
+import cache from '@/utils/cache';
 
 export const route: Route = {
     path: '/docs',
@@ -22,34 +23,42 @@ export const route: Route = {
         },
     ],
     name: '游研社 - 全部文章',
-    maintainers: ['HaitianLiu'],
+    maintainers: ['HaitianLiu', 'yy4382'],
     handler,
     url: 'yystv.cn/docs',
 };
 
 async function handler() {
     const url = `https://www.yystv.cn/docs`;
-    const response = await got({
-        method: 'get',
-        url,
-    });
+    const response = await ofetch(url);
 
-    const data = response.data;
-    const $ = load(data);
+    const $ = load(response);
 
-    const items = $('.list-container li')
-        .slice(0, 18)
-        .map(function () {
+    const itemList = $('.list-container li')
+        .toArray()
+        .map((item) => {
+            const itemElement = $(item);
             const info = {
-                title: $('.list-article-title', this).text(),
-                link: 'https://www.yystv.cn' + $('a', this).attr('href'),
-                pubDate: parseRelativeDate($('.c-999', this).text()),
-                author: $('.handler-author-link', this).text(),
-                description: $('.list-article-intro', this).text(),
+                title: itemElement.find('.list-article-title').text(),
+                link: 'https://www.yystv.cn' + itemElement.find('a').attr('href'),
+                pubDate: parseRelativeDate(itemElement.find('.c-999').text()),
+                author: itemElement.find('.handler-author-link').text(),
+                description: itemElement.find('.list-article-intro').text(),
             };
             return info;
-        })
-        .get();
+        }) satisfies DataItem[];
+
+    const items = (await Promise.all(
+        itemList.map(
+            (item) =>
+                cache.tryGet(item.link, async () => {
+                    const resp = await ofetch(item.link);
+                    const $ = load(resp);
+                    item.description = $('#main section.article-section .doc-content > div').html() || item.description;
+                    return item;
+                }) as Promise<DataItem>
+        )
+    )) satisfies DataItem[];
 
     return {
         title: '游研社-' + $('title').text(),

package/lib/routes/zaobao/interactive.ts

@@ -6,27 +6,18 @@ export const route: Route = {
     path: '/interactive-graphics',
     categories: ['traditional-media'],
     example: '/zaobao/interactive-graphics',
-    parameters: {},
-    features: {
-        requireConfig: false,
-        requirePuppeteer: false,
-        antiCrawler: false,
-        supportBT: false,
-        supportPodcast: false,
-        supportScihub: false,
-    },
     name: '互动新闻',
     maintainers: ['shunf4'],
     handler,
 };
 
 async function handler(ctx) {
-    const sectionLink = `/interactive-graphics`;
+    const sectionLink = '/interactive-graphics';
 
     const { resultList } = await parseList(ctx, sectionLink);
 
     return {
-        title: `《联合早报》互动新闻`,
+        title: '《联合早报》互动新闻',
         link: baseUrl + sectionLink,
         description: '新加坡、中国、亚洲和国际的即时、评论、商业、体育、生活、科技与多媒体新闻，尽在联合早报。',
         item: resultList,

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) ? $('.card-listing .card') : $('.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

@@ -1,15 +1,12 @@
-import { Route } from '@/types';
-import cache from '@/utils/cache';
+import { Route, ViewType } from '@/types';
 import got from '@/utils/got';
-import ofetch from '@/utils/ofetch';
-import utils from './utils';
+import { header, processImage, getSignedHeader } from './utils';
 import { parseDate } from '@/utils/parse-date';
-import g_encrypt from './execlib/x-zse-96-v3';
-import md5 from '@/utils/md5';
 
 export const route: Route = {
     path: '/people/activities/:id',
-    categories: ['social-media'],
+    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: {
@@ -33,43 +30,17 @@ export const route: Route = {
 async function handler(ctx) {
     const id = ctx.req.param('id');
 
-    // Because the API of zhihu.com has changed, we must use the value of `d_c0` (extracted from cookies) to calculate
-    // `x-zse-96`. So first get `d_c0`, then get the actual data of a ZhiHu question. In this way, we don't need to
-    // require users to set the cookie in environmental variables anymore.
-
-    // fisrt: get cookie(dc_0) from zhihu.com
-    const cookie_mes = await cache.tryGet('zhihu:cookies:d_c0', async () => {
-        const response = await ofetch.raw(`https://www.zhihu.com/people/${id}`, {
-            headers: {
-                ...utils.header,
-            },
-        });
-
-        const cookie_org = response.headers.get('set-cookie');
-        const cookie = cookie_org?.toString();
-        const match = cookie?.match(/d_c0=(\S+?)(?:;|$)/);
-        const cookie_mes = match && match[1];
-        if (!cookie_mes) {
-            throw new Error('Failed to extract `d_c0` from cookies');
-        }
-        return cookie_mes;
-    });
-    const cookie = `d_c0=${cookie_mes}`;
-
     // second: get real data from zhihu
     const apiPath = `/api/v3/moments/${id}/activities?limit=7&desktop=true`;
 
-    // calculate x-zse-96, refer to https://github.com/srx-2000/spider_collection/issues/18
-    const f = `101_3_3.0+${apiPath}+${cookie_mes}`;
-    const xzse96 = '2.0_' + g_encrypt(md5(f));
-    const _header = { cookie, 'x-zse-96': xzse96, 'x-app-za': 'OS=Web', 'x-zse-93': '101_3_3.0' };
+    const signedHeader = await getSignedHeader(`https://www.zhihu.com/people/${id}`, apiPath);
 
     const response = await got(`https://www.zhihu.com${apiPath}`, {
         headers: {
-            ...utils.header,
-            ..._header,
+            ...header,
+            ...signedHeader,
             Referer: `https://www.zhihu.com/people/${id}/activities`,
-            Authorization: 'oauth c3cef7c66a1843f8b3a9e6a1e3160e20', // hard-coded in js
+            // Authorization: 'oauth c3cef7c66a1843f8b3a9e6a1e3160e20', // hard-coded in js
         },
     });
 
@@ -85,7 +56,7 @@ async function handler(ctx) {
             let title;
             let description;
             let url;
-            const images = [];
+            const images: string[] = [];
             let text = '';
             let link = '';
             let author = '';
@@ -94,13 +65,13 @@ async function handler(ctx) {
                 case 'answer':
                     title = detail.question.title;
                     author = detail.author.name;
-                    description = utils.ProcessImage(detail.content);
+                    description = processImage(detail.content);
                     url = `https://www.zhihu.com/question/${detail.question.id}/answer/${detail.id}`;
                     break;
                 case 'article':
                     title = detail.title;
                     author = detail.author.name;
-                    description = utils.ProcessImage(detail.content);
+                    description = processImage(detail.content);
                     url = `https://zhuanlan.zhihu.com/p/${detail.id}`;
                     break;
                 case 'pin':
@@ -142,7 +113,7 @@ async function handler(ctx) {
                 case 'question':
                     title = detail.title;
                     author = detail.author.name;
-                    description = utils.ProcessImage(detail.detail);
+                    description = processImage(detail.detail);
                     url = `https://www.zhihu.com/question/${detail.id}`;
                     break;
                 case 'collection':
@@ -169,6 +140,8 @@ async function handler(ctx) {
                     description = detail.description;
                     url = `https://www.zhihu.com/roundtable/${detail.id}`;
                     break;
+                default:
+                    description = `未知类型 ${item.target.type}，请点击<a href="https://github.com/DIYgod/RSSHub/issues">链接</a>提交issue`;
             }
 
             return {

package/lib/routes/zhihu/answers.ts

@@ -1,7 +1,7 @@
 import { Route } from '@/types';
 import cache from '@/utils/cache';
 import got from '@/utils/got';
-import utils from './utils';
+import { getCookieValueByKey, header, processImage } from './utils';
 import { parseDate } from '@/utils/parse-date';
 
 export const route: Route = {
@@ -29,9 +29,13 @@ export const route: Route = {
 
 async function handler(ctx) {
     const id = ctx.req.param('id');
+
+    const zc0 = getCookieValueByKey('z_c0');
+
     const headers = {
         'User-Agent': 'ZhihuHybrid com.zhihu.android/Futureve/6.59.0 Mozilla/5.0 (Linux; Android 10; GM1900 Build/QKQ1.190716.003; wv) AppleWebKit/537.36 (KHTML, like Gecko) Version/4.0 Chrome/85.0.4183.127 Mobile Safari/537.36',
         Referer: `https://www.zhihu.com/people/${id}/answers`,
+        Cookie: zc0 ? `z_c0=${zc0}` : '',
     };
 
     const response = await got({
@@ -51,7 +55,7 @@ async function handler(ctx) {
                 method: 'get',
                 url: `https://www.zhihu.com${apiPath}`,
                 headers: {
-                    ...utils.header,
+                    ...header,
                     Referer: `https://www.zhihu.com/people/${id}`,
                 },
             });
@@ -71,7 +75,7 @@ async function handler(ctx) {
                     url: `https://api.zhihu.com/appview/api/v4/answers/${item.id}?include=content&is_appview=true`,
                     headers,
                 });
-                description = utils.ProcessImage(detail.data.content);
+                description = processImage(detail.data.content);
             } catch {
                 description = `<a href="${link}" target="_blank">${title}</a>`;
             }

package/lib/routes/zhihu/collection.ts

@@ -2,7 +2,7 @@ import { Route } from '@/types';
 import cache from '@/utils/cache';
 import got from '@/utils/got';
 import { load } from 'cheerio';
-import utils from './utils';
+import { header } from './utils';
 import { generateData } from './pin/utils';
 import { parseDate } from '@/utils/parse-date';
 
@@ -38,7 +38,7 @@ async function handler(ctx) {
         method: 'get',
         url: `https://www.zhihu.com/api/v4/collections/${id}/items?offset=0&limit=20`,
         headers: {
-            ...utils.header,
+            ...header,
             Referer: `https://www.zhihu.com/collection/${id}`,
         },
     });
@@ -56,7 +56,7 @@ async function handler(ctx) {
                         method: 'get',
                         url: `https://www.zhihu.com/api/v4/collections/${id}/items?offset=${offset}&limit=20`,
                         headers: {
-                            ...utils.header,
+                            ...header,
                             Referer: `https://www.zhihu.com/collection/${id}`,
                         },
                     });
@@ -72,7 +72,7 @@ async function handler(ctx) {
         method: 'get',
         url: `https://www.zhihu.com/collection/${id}`,
         headers: {
-            ...utils.header,
+            ...header,
             Referer: `https://www.zhihu.com/collection/${id}`,
         },
     });

package/lib/routes/zhihu/daily-section.ts

@@ -1,7 +1,7 @@
 import { Route } from '@/types';
 import cache from '@/utils/cache';
 import got from '@/utils/got';
-import utils from './utils';
+import { header, processImage } from './utils';
 import { parseDate } from '@/utils/parse-date';
 
 // 参考：https://github.com/izzyleung/ZhihuDailyPurify/wiki/%E7%9F%A5%E4%B9%8E%E6%97%A5%E6%8A%A5-API-%E5%88%86%E6%9E%90
@@ -38,7 +38,7 @@ async function handler(ctx) {
         method: 'get',
         url: `https://news-at.zhihu.com/api/7/section/${sectionId}`,
         headers: {
-            ...utils.header,
+            ...header,
             Referer: `https://news-at.zhihu.com/api/7/section/${sectionId}`,
         },
     });
@@ -61,7 +61,7 @@ async function handler(ctx) {
                         Referer: url,
                     },
                 });
-                item.description = utils.ProcessImage(storyDetail.data.body.replaceAll(/<div class="meta">([\S\s]*?)<\/div>/g, '<strong>$1</strong>').replaceAll(/<\/?h2.*?>/g, ''));
+                item.description = processImage(storyDetail.data.body.replaceAll(/<div class="meta">([\S\s]*?)<\/div>/g, '<strong>$1</strong>').replaceAll(/<\/?h2.*?>/g, ''));
 
                 return item;
             });

package/lib/routes/zhihu/daily.ts

@@ -1,7 +1,7 @@
 import { Route } from '@/types';
 import cache from '@/utils/cache';
 import got from '@/utils/got';
-import utils from './utils';
+import { header, processImage } from './utils';
 import { parseDate } from '@/utils/parse-date';
 
 // 参考：https://github.com/izzyleung/ZhihuDailyPurify/wiki/%E7%9F%A5%E4%B9%8E%E6%97%A5%E6%8A%A5-API-%E5%88%86%E6%9E%90
@@ -58,7 +58,7 @@ async function handler() {
         method: 'get',
         url: `${api}/latest`,
         headers: {
-            ...utils.header,
+            ...header,
             Referer: `${api}/latest`,
             Host: address,
         },
@@ -89,7 +89,7 @@ async function handler() {
                     },
                     // host: address,
                 });
-                return utils.ProcessImage(storyDetail.data.body.replaceAll(/<div class="meta">([\S\s]*?)<\/div>/g, '<strong>$1</strong>').replaceAll(/<\/?h2.*?>/g, ''));
+                return processImage(storyDetail.data.body.replaceAll(/<div class="meta">([\S\s]*?)<\/div>/g, '<strong>$1</strong>').replaceAll(/<\/?h2.*?>/g, ''));
             });
             item.description = description;
             return item;

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 got from '@/utils/got';
-import utils from './utils';
-import { load } from 'cheerio';
+import ofetch from '@/utils/ofetch';
+import cache from '@/utils/cache';
+import { header, getSignedHeader, processImage } from './utils';
 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',
@@ -34,47 +35,53 @@ async function handler(ctx) {
     const id = ctx.req.param('id');
     const usertype = ctx.req.param('usertype');
 
-    const { data } = await got(`https://www.zhihu.com/${usertype}/${id}/posts`, {
-        headers: {
-            ...utils.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 jsondata = $('#js-initialData');
-    const authorname = $('.ProfileHeader-name')
-        .contents()
-        .filter((_index, element) => element.type === 'text')
-        .text();
-    const authordescription = $('.ProfileHeader-headline').text();
 
-    const parsed = JSON.parse(jsondata.html());
-    const articlesdata = parsed.initialState.entities.articles;
+    const apiPath = `/api/v4/${usertype === 'people' ? 'members' : 'org'}/${id}/articles?${new URLSearchParams({
+        include:
+            'data[*].comment_count,suggest_edit,is_normal,thumbnail_extra_info,thumbnail,can_comment,comment_permission,admin_closed_comment,content,voteup_count,created,updated,upvoted_followees,voting,review_info,reaction_instruction,is_labeled,label_info;data[*].vessay_info;data[*].author.badge[?(type=best_answerer)].topics;data[*].author.vip_info;',
+        offset: '0',
+        limit: '20',
+        sort_by: 'created',
+    })}`;
 
-    const list = Object.keys(articlesdata).map((key) => {
-        const $ = load(articlesdata[key].content, null, false);
-        $('noscript').remove();
-        $('img').each((_, item) => {
-            if (item.attribs['data-actualsrc'] || item.attribs['data-original']) {
-                item.attribs['data-actualsrc'] = item.attribs['data-actualsrc'] ? item.attribs['data-actualsrc'].split('?source')[0] : null;
-                item.attribs['data-original'] = item.attribs['data-original'] ? item.attribs['data-original'].split('?source')[0] : null;
-                item.attribs.src = item.attribs['data-original'] || item.attribs['data-actualsrc'];
-                delete item.attribs['data-actualsrc'];
-                delete item.attribs['data-original'];
-            }
-        });
-        return {
-            title: articlesdata[key].title,
-            description: $.html(),
-            link: articlesdata[key].url,
-            pubDate: parseDate(articlesdata[key].created, 'X'),
-        };
+    const signedHeader = await getSignedHeader(`https://www.zhihu.com/${usertype}/${id}/posts`, apiPath);
+
+    const articleResponse = await ofetch<Articles>(`https://www.zhihu.com${apiPath}`, {
+        headers: {
+            ...header,
+            ...signedHeader,
+            Referer: `https://www.zhihu.com/${usertype}/${id}/posts`,
+        },
     });
 
+    const items = articleResponse.data.map((item) => ({
+        title: item.title,
+        description: processImage(item.content),
+        link: `https://zhuanlan.zhihu.com/p/${item.id}`,
+        pubDate: parseDate(item.created, 'X'),
+        updated: parseDate(item.updated, 'X'),
+        author: item.author.name,
+    }));
+
     return {
-        title: `${authorname} 的知乎文章`,
+        title: `${userProfile.name} 的知乎文章`,
         link: `https://www.zhihu.com/${usertype}/${id}/posts`,
-        description: authordescription,
-        item: list,
+        description: userProfile.headline,
+        image: userProfile.avatar_url.split('?')[0],
+        // banner: userData?.coverUrl?.split('?')[0],
+        item: items,
     };
 }