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

package/lib/routes/zhihu/question.ts

@@ -1,9 +1,6 @@
 import { Route } from '@/types';
-import cache from '@/utils/cache';
 import got from '@/utils/got';
-import utils from './utils';
-import md5 from '@/utils/md5';
-import g_encrypt from './execlib/x-zse-96-v3';
+import { header, processImage, getSignedHeader } from './utils';
 import { parseDate } from '@/utils/parse-date';
 
 export const route: Route = {
@@ -35,50 +32,25 @@ async function handler(ctx) {
         questionId,
         sortBy = 'default', // default,created,updated
     } = ctx.req.param();
-    // 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 got({
-            method: 'get',
-            url: `https://www.zhihu.com/question/${questionId}`,
-            headers: {
-                ...utils.header,
-            },
-        });
-
-        const cookie_org = response.headers['set-cookie'];
-        const cookie = cookie_org.toString();
-        const match = cookie.match(/d_c0=(\S+?)(?:;|$)/);
-        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 limit = 20;
-    const include =
-        'data%5B*%5D.is_normal%2Cadmin_closed_comment%2Creward_info%2Cis_collapsed%2Cannotation_action%2Cannotation_detail%2Ccollapse_reason%2Cis_sticky%2Ccollapsed_by%2Csuggest_edit%2Ccomment_count%2Ccan_comment%2Ccontent%2Ceditable_content%2Cattachment%2Cvoteup_count%2Creshipment_settings%2Ccomment_permission%2Ccreated_time%2Cupdated_time%2Creview_info%2Crelevant_info%2Cquestion%2Cexcerpt%2Cis_labeled%2Cpaid_info%2Cpaid_info_content%2Crelationship.is_authorized%2Cis_author%2Cvoting%2Cis_thanked%2Cis_nothelp%2Cis_recognized%3Bdata%5B*%5D.mark_infos%5B*%5D.url%3Bdata%5B*%5D.author.follower_count%2Cbadge%5B*%5D.topics%3Bdata%5B*%5D.settings.table_of_content.enabled&offset=0';
     const rootUrl = 'https://www.zhihu.com';
-    const apiPath = `/api/v4/questions/${questionId}/answers?include=${include}&limit=${limit}&sort_by=${sortBy}&platform=desktop`;
-    const url = rootUrl + apiPath;
+    const apiPath = `/api/v4/questions/${questionId}/answers?${new URLSearchParams({
+        include:
+            'data[*].is_normal,admin_closed_comment,reward_info,is_collapsed,annotation_action,annotation_detail,collapse_reason,is_sticky,collapsed_by,suggest_edit,comment_count,can_comment,content,editable_content,attachment,voteup_count,reshipment_settings,comment_permission,created_time,updated_time,review_info,relevant_info,question,excerpt,is_labeled,paid_info,paid_info_content,relationship.is_authorized,is_author,voting,is_thanked,is_nothelp,is_recognized;data[*].mark_infos[*].url;data[*].author.follower_count,badge[*].topics;data[*].settings.table_of_content.enabled&offset=0',
+        limit: '20',
+        sort_by: sortBy,
+        platform: 'desktop',
+    })}`;
 
-    // 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/question/${questionId}`, apiPath);
 
     const response = await got({
         method: 'get',
-        url,
+        url: rootUrl + apiPath,
         headers: {
-            ...utils.header,
-            ..._header,
+            ...header,
+            ...signedHeader,
             Referer: `https://www.zhihu.com/question/${questionId}`,
             // Authorization: 'oauth c3cef7c66a1843f8b3a9e6a1e3160e20', // previously hard-coded in js, outdated
         },
@@ -91,7 +63,7 @@ async function handler(ctx) {
         link: `https://www.zhihu.com/question/${questionId}`,
         item: listRes.map((item) => {
             const title = `${item.author.name}的回答：${item.excerpt}`;
-            const description = `${item.author.name}的回答<br/><br/>${utils.ProcessImage(item.content)}`;
+            const description = `${item.author.name}的回答<br/><br/>${processImage(item.content)}`;
 
             return {
                 title,

package/lib/routes/zhihu/timeline.ts

@@ -1,7 +1,7 @@
 import { Route } from '@/types';
 import got from '@/utils/got';
 import { config } from '@/config';
-import utils from './utils';
+import { processImage } from './utils';
 import { parseDate } from '@/utils/parse-date';
 import ConfigNotFoundError from '@/errors/types/config-not-found';
 
@@ -106,7 +106,7 @@ async function handler(ctx) {
         const link = buildLink(e);
         return {
             title: `${e.action_text_tpl.replace('{}', buildActors(e))}: ${getOne([e.target.title, e.target.question ? e.target.question.title : ''])}`,
-            description: utils.ProcessImage(`<div>${getOne([e.target.content_html, getContent(e.target.content), e.target.detail, e.target.excerpt, ''])}</div>`),
+            description: processImage(`<div>${getOne([e.target.content_html, getContent(e.target.content), e.target.detail, e.target.excerpt, ''])}</div>`),
             pubDate: parseDate(e.updated_time * 1000),
             link,
             author: e.target.author ? e.target.author.name : '',

package/lib/routes/zhihu/topic.ts

@@ -1,10 +1,8 @@
 import { Route } from '@/types';
 import cache from '@/utils/cache';
 import got from '@/utils/got';
-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: '/topic/:topicId/:isTop?',
@@ -34,22 +32,6 @@ async function handler(ctx) {
     const { topicId, isTop = false } = ctx.req.param();
     const link = `https://www.zhihu.com/topic/${topicId}/${isTop ? 'top-answers' : 'newest'}`;
 
-    const cookieDc0 = await cache.tryGet('zhihu:cookies:d_c0', async () => {
-        const { headers } = await got(link, {
-            headers: {
-                ...utils.header,
-            },
-        });
-        const cookie = headers['set-cookie'].toString();
-        const match = cookie.match(/d_c0=(\S+?)(?:;|$)/);
-        const cookieMatched = match?.[1];
-        if (!cookieMatched) {
-            throw new Error('Failed to extract `d_c0` from cookies');
-        }
-        return cookieMatched;
-    });
-    const cookie = `d_c0=${cookieDc0}`;
-
     const topicMeta = await cache.tryGet(`zhihu:topic:${topicId}`, async () => {
         const { data: response } = await got(`https://www.zhihu.com/api/v4/topics/${topicId}/intro`, {
             searchParams: {
@@ -60,15 +42,12 @@ async function handler(ctx) {
     });
 
     const apiPath = `/api/v5.1/topics/${topicId}/feeds/${isTop ? 'top_activity' : 'timeline_activity'}`;
-    const xzse93 = '101_3_3.0';
-    const f = `${xzse93}+${apiPath}+${cookieDc0}`;
-    const xzse96 = '2.0_' + g_encrypt(md5(f));
-    const _header = { cookie, 'x-zse-96': xzse96, 'x-zse-93': xzse93 };
+    const signedHeader = await getSignedHeader(link, apiPath);
 
     const { data: response } = await got(`https://www.zhihu.com${apiPath}`, {
         headers: {
-            ...utils.header,
-            ..._header,
+            ...header,
+            ...signedHeader,
             Referer: link,
         },
     });
@@ -78,13 +57,13 @@ async function handler(ctx) {
         let title = '';
         let description = '';
         let link = '';
-        let pubDate = '';
+        let pubDate: Date;
         let author = '';
 
         switch (type) {
             case 'answer':
                 title = `${item.question.title}-${item.author.name}的回答：${item.excerpt}`;
-                description = `<strong>${item.question.title}</strong><br>${item.author.name}的回答<br/>${utils.ProcessImage(item.content)}`;
+                description = `<strong>${item.question.title}</strong><br>${item.author.name}的回答<br/>${processImage(item.content)}`;
                 link = `https://www.zhihu.com/question/${item.question.id}/answer/${item.id}`;
                 pubDate = parseDate(item.updated_time * 1000);
                 author = item.author.name;
@@ -99,7 +78,7 @@ async function handler(ctx) {
 
             case 'article':
                 title = item.title;
-                description = utils.ProcessImage(item.content);
+                description = processImage(item.content);
                 link = item.url;
                 pubDate = parseDate(item.created * 1000);
                 break;
@@ -107,7 +86,7 @@ async function handler(ctx) {
             case 'zvideo':
                 title = item.title;
                 description = `${item.description}<br>
-                <video controls poster="${item.video.thumbnail}" preload="none">
+                <video controls poster="${item.video.thumbnail}" preload="metadata">
                     <source src="${item.video.playlist.fhd?.url ?? item.video.playlist.hd?.url ?? item.video.playlist.ld?.url ?? item.video.playlist.sd?.url}" type="video/mp4">
                 </video>`;
                 link = item.url;
@@ -116,6 +95,7 @@ async function handler(ctx) {
 
             default:
                 description = '未知类型，请点击<a href="https://github.com/DIYgod/RSSHub/issues">链接</a>提交issue';
+                pubDate = parseDate(Date.now());
         }
 
         return {

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

@@ -1,51 +1,117 @@
 import { load } from 'cheerio';
+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 default {
-    header: {
-        'x-api-version': '3.0.91',
-    },
-    ProcessImage(content) {
-        const $ = load(content, null, false);
-
-        $('noscript').remove();
-
-        $('a[data-draft-type="mcn-link-card"]').remove();
-
-        $('a').each((_, elem) => {
-            const href = $(elem).attr('href');
-            if (href?.startsWith('https://link.zhihu.com/?target=')) {
-                const url = new URL(href);
-                const target = url.searchParams.get('target') || '';
-                try {
-                    $(elem).attr('href', decodeURIComponent(target));
-                } catch {
-                    // sometimes the target is not a valid url
-                }
-            }
-        });
-
-        $('img.content_image, img.origin_image, img.content-image, img.data-actualsrc, figure>img').each((i, e) => {
-            if (e.attribs['data-actualsrc']) {
-                $(e).attr({
-                    src: e.attribs['data-actualsrc'].replace('_b.jpg', '_1440w.jpg'),
-                    width: null,
-                    height: null,
-                });
-            } else if (e.attribs['data-original']) {
-                $(e).attr({
-                    src: e.attribs['data-original'].replace('_r.jpg', '_1440w.jpg'),
-                    width: null,
-                    height: null,
-                });
-            } else {
-                $(e).attr({
-                    src: e.attribs.src.replace('_b.jpg', '_1440w.jpg'),
-                    width: null,
-                    height: null,
-                });
+export const header = {
+    'x-api-version': '3.0.91',
+};
+
+const fixImageUrl = (url: string) => url.split('?')[0].replace('_b.jpg', '.jpg').replace('_r.jpg', '.jpg').replace('_720w.jpg', '.jpg');
+
+export const processImage = (content: string) => {
+    const $ = load(content, null, false);
+
+    $('noscript, a[data-draft-type="mcn-link-card"]').remove();
+
+    $('a').each((_, elem) => {
+        const href = $(elem).attr('href');
+        if (href?.startsWith('http://link.zhihu.com/?target=') || href?.startsWith('https://link.zhihu.com/?target=')) {
+            const url = new URL(href);
+            const target = url.searchParams.get('target') || '';
+            try {
+                $(elem).attr('href', decodeURIComponent(target));
+            } catch {
+                // sometimes the target is not a valid url
             }
-        });
+        }
+    });
+
+    $('img.content_image, img.origin_image, img.content-image, img.data-actualsrc, figure>img').each((i, e) => {
+        if (e.attribs['data-actualsrc']) {
+            $(e).attr({
+                src: fixImageUrl(e.attribs['data-actualsrc']),
+                width: null,
+                height: null,
+            });
+            $(e).removeAttr('data-actualsrc');
+        } else if (e.attribs['data-original']) {
+            $(e).attr({
+                src: fixImageUrl(e.attribs['data-original']),
+                width: null,
+                height: null,
+            });
+            $(e).removeAttr('data-original');
+        } else {
+            $(e).attr({
+                src: fixImageUrl(e.attribs.src),
+                width: null,
+                height: null,
+            });
+        }
+    });
+
+    return $.html();
+};
+
+export const getCookieValueByKey = (key: string) =>
+    config.zhihu.cookies
+        ?.split(';')
+        .map((e) => e.trim())
+        .find((e) => e.startsWith(key + '='))
+        ?.slice(key.length + 1);
+
+export const getSignedHeader = async (url: string, apiPath: string) => {
+    // Because the API of zhihu.com has changed, we must use the value of `d_c0` (extracted from cookies) to calculate
+    // `x-zse-96`. So first get `d_c0`, then get the actual data of a ZhiHu question. In this way, we don't need to
+    // require users to set the cookie in environmental variables anymore.
+
+    // fisrt: get cookie(dc_0) from zhihu.com
+    const dc0 = await cache.tryGet('zhihu:cookies:d_c0', async () => {
+        if (getCookieValueByKey('d_c0')) {
+            return getCookieValueByKey('d_c0');
+        }
+        const response1 = await ofetch.raw('https://static.zhihu.com/zse-ck/v3.js');
+        const script = await response1._data.text();
+        const zseCk = script.match(/__g\.ck\|\|"([\w+/=\\]*?)",_=630e8;/)?.[1];
+
+        const response2 = zseCk
+            ? await ofetch.raw(url, {
+                  headers: {
+                      cookie: `${response1.headers
+                          .getSetCookie()
+                          .map((s) => s.split(';')[0])
+                          .join('; ')}; __zse_ck=${zseCk}`,
+                  },
+              })
+            : null;
+
+        const dc0 = (response2 || response1).headers
+            .getSetCookie()
+            .find((s) => s.startsWith('d_c0='))
+            ?.split(';')[0]
+            .trim()
+            .slice('d_c0='.length);
+        if (!dc0) {
+            throw new Error('Failed to extract `d_c0` from cookies');
+        }
+
+        return dc0;
+    });
+
+    // calculate x-zse-96, refer to https://github.com/srx-2000/spider_collection/issues/18
+    const xzse93 = '101_3_3.0';
+    const f = `${xzse93}+${apiPath}+${dc0}`;
+    const xzse96 = '2.0_' + g_encrypt(md5(f));
+
+    const zc0 = getCookieValueByKey('z_c0');
 
-        return $.html();
-    },
+    return {
+        cookie: `d_c0=${dc0}${zc0 ? `;z_c0=${zc0}` : ''}`,
+        'x-zse-96': xzse96,
+        'x-app-za': 'OS=Web',
+        'x-zse-93': xzse93,
+    };
 };

package/lib/routes/zhihu/xhu/activities.ts

@@ -1,7 +1,7 @@
 import { Route } from '@/types';
 import got from '@/utils/got';
 import auth from './auth';
-import utils from '../utils';
+import { processImage } from '../utils';
 import { parseDate } from '@/utils/parse-date';
 
 export const route: Route = {
@@ -37,7 +37,7 @@ export const route: Route = {
 };
 
 async function handler(ctx) {
-    const xhuCookie = await auth.getCookie(ctx);
+    const xhuCookie = await auth.getCookie();
     const hexId = ctx.req.param('hexId');
     const link = `https://www.zhihu.com/people/${hexId}`;
     const url = `https://api.zhihuvvv.workers.dev/people/${hexId}/activities?before_id=0&limit=20`;
@@ -62,7 +62,7 @@ async function handler(ctx) {
             let title;
             let description;
             let url;
-            const images = [];
+            const images: string[] = [];
             let text = '';
             let link = '';
             let author = '';
@@ -71,13 +71,13 @@ 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/${detail.id}`;
                     break;
                 case 'pin':
@@ -119,7 +119,7 @@ async function handler(ctx) {
                 case 'question':
                     title = detail.title;
                     author = detail.author.name;
-                    description = utils.ProcessImage(detail.detail);
+                    description = processImage(detail.detail);
                     url = `https://www.zhihu.com/question/${detail.id}`;
                     break;
                 case 'collection':
@@ -146,6 +146,8 @@ 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/xhu/answers.ts

@@ -28,7 +28,7 @@ export const route: Route = {
 };
 
 async function handler(ctx) {
-    const xhuCookie = await auth.getCookie(ctx);
+    const xhuCookie = await auth.getCookie();
     const hexId = ctx.req.param('hexId');
     const link = `https://www.zhihu.com/people/${hexId}/answers`;
     const url = `https://api.zhihuvvv.workers.dev/people/${hexId}/answers?limit=20&offset=0`;

package/lib/routes/zhihu/xhu/collection.ts

@@ -28,7 +28,7 @@ export const route: Route = {
 };
 
 async function handler(ctx) {
-    const xhuCookie = await auth.getCookie(ctx);
+    const xhuCookie = await auth.getCookie();
     const id = ctx.req.param('id');
     const link = `https://www.zhihu.com/collection/${id}`;
 

package/lib/routes/zhihu/xhu/posts.ts

@@ -22,7 +22,7 @@ export const route: Route = {
 };
 
 async function handler(ctx) {
-    const xhuCookie = await auth.getCookie(ctx);
+    const xhuCookie = await auth.getCookie();
     const hexId = ctx.req.param('hexId');
     const link = `https://www.zhihu.com/people/${hexId}/posts`;
     const url = `https://api.zhihuvvv.workers.dev/people/${hexId}/articles?limit=20&offset=0`;

package/lib/routes/zhihu/xhu/question.ts

@@ -1,7 +1,7 @@
 import { Route } from '@/types';
 import got from '@/utils/got';
 import auth from './auth';
-import utils from '../utils';
+import { processImage } from '../utils';
 import { parseDate } from '@/utils/parse-date';
 
 export const route: Route = {
@@ -29,7 +29,7 @@ export const route: Route = {
 };
 
 async function handler(ctx) {
-    const xhuCookie = await auth.getCookie(ctx);
+    const xhuCookie = await auth.getCookie();
     const {
         questionId,
         sortBy = 'default', // default,created,updated
@@ -54,7 +54,7 @@ async function handler(ctx) {
             const link = `https://www.zhihu.com/question/${questionId}/answer/${item.id}`;
             const author = item.author.name;
             const title = `${author}的回答：${item.excerpt}`;
-            const description = `${author}的回答<br/><br/>${utils.ProcessImage(item.excerpt)}`;
+            const description = `${author}的回答<br/><br/>${processImage(item.excerpt)}`;
 
             return {
                 title,

package/lib/routes/zhihu/xhu/topic.ts

@@ -1,7 +1,7 @@
 import { Route } from '@/types';
 import got from '@/utils/got';
 import auth from './auth';
-import utils from '../utils';
+import { processImage } from '../utils';
 import { parseDate } from '@/utils/parse-date';
 
 export const route: Route = {
@@ -28,7 +28,7 @@ export const route: Route = {
 };
 
 async function handler(ctx) {
-    const xhuCookie = await auth.getCookie(ctx);
+    const xhuCookie = await auth.getCookie();
     const topicId = ctx.req.param('topicId');
     const link = `https://www.zhihu.com/topic/${topicId}/newest`;
     const url = `https://api.zhihuvvv.workers.dev/topics/${topicId}/feeds/timeline_activity?before_id=0&limit=20`;
@@ -51,13 +51,13 @@ async function handler(ctx) {
             let title = '';
             let description = '';
             let link = '';
-            let pubDate = '';
+            let pubDate = new Date();
             let author = '';
 
             switch (type) {
                 case 'answer':
                     title = `${item.question.title}-${item.author.name}的回答：${item.excerpt}`;
-                    description = `<strong>${item.question.title}</strong><br>${item.author.name}的回答<br/>${utils.ProcessImage(item.content)}`;
+                    description = `<strong>${item.question.title}</strong><br>${item.author.name}的回答<br/>${processImage(item.content)}`;
                     link = `https://www.zhihu.com/question/${item.question.id}/answer/${item.id}`;
                     pubDate = parseDate(item.updated_time * 1000);
                     author = item.author.name;