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

package/lib/routes/zhihu/zhuanlan.ts

@@ -1,6 +1,6 @@
 import { Route } from '@/types';
 import got from '@/utils/got';
-import utils from './utils';
+import { getSignedHeader, header } from './utils';
 import { load } from 'cheerio';
 import { parseDate } from '@/utils/parse-date';
 
@@ -29,12 +29,19 @@ export const route: Route = {
 
 async function handler(ctx) {
     const id = ctx.req.param('id');
+    // 知乎专栏链接存在两种格式, 一种以 'zhuanlan.' 开头, 另一种新增的以 'c_' 结尾
+    let url = `https://zhuanlan.zhihu.com/${id}`;
+    if (id.search('c_') === 0) {
+        url = `https://www.zhihu.com/column/${id}`;
+    }
 
+    const signedHeader = await getSignedHeader(url, `https://www.zhihu.com/api/v4/columns/${id}/items`);
     const listRes = await got({
         method: 'get',
         url: `https://www.zhihu.com/api/v4/columns/${id}/items`,
         headers: {
-            ...utils.header,
+            ...header,
+            ...signedHeader,
             Referer: `https://zhuanlan.zhihu.com/${id}`,
         },
     });
@@ -43,20 +50,20 @@ async function handler(ctx) {
         method: 'get',
         url: `https://www.zhihu.com/api/v4/columns/${id}/pinned-items`,
         headers: {
-            ...utils.header,
+            ...header,
+            ...signedHeader,
             Referer: `https://zhuanlan.zhihu.com/${id}`,
         },
     });
 
     listRes.data.data = [...listRes.data.data, ...pinnedRes.data.data];
 
-    // 知乎专栏链接存在两种格式, 一种以 'zhuanlan.' 开头, 另一种新增的以 'c_' 结尾
-    let url = `https://zhuanlan.zhihu.com/${id}`;
-    if (id.search('c_') === 0) {
-        url = `https://www.zhihu.com/column/${id}`;
-    }
-
-    const infoRes = await got(url);
+    const infoRes = await got(url, {
+        headers: {
+            ...signedHeader,
+            Referer: url,
+        },
+    });
     const $ = load(infoRes.data);
     const title = $('.css-zyehvu').text();
     const description = $('.css-1bnklpv').text();
@@ -73,7 +80,7 @@ async function handler(ctx) {
         let title = '';
         let link = '';
         let author = '';
-        let pubDate = '';
+        let pubDate: Date;
 
         switch (item.type) {
             case 'answer':

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/zju/cst/index.ts

@@ -37,7 +37,7 @@ async function getPage(id) {
                 return {
                     title: item.find('a').text(),
                     pubDate: parseDate(item.find('.fr').text()),
-                    link: new URL(item.find('a').attr('href'), res.url).href,
+                    link: new URL(item.find('a').attr('href'), host).href,
                 };
             })
             .get()

package/lib/routes/zsxq/group.ts

@@ -0,0 +1,63 @@
+import type { Data, Route } from '@/types';
+import type { Context } from 'hono';
+import { config } from '@/config';
+import ConfigNotFoundError from '@/errors/types/config-not-found';
+import type { GroupInfoResponse, TopicsResponse } from './types';
+import { customFetch, generateTopicDataItem } from './utils';
+
+export const route: Route = {
+    name: '星球',
+    categories: ['social-media'],
+    path: '/group/:id/:scope?',
+    example: '/zsxq/group/88855458825252',
+    parameters: {
+        id: '星球id，从网页端url中获取',
+        scope: '栏目分类，默认为"all"，见下表',
+    },
+    maintainers: ['KarasuShin'],
+    radar: [
+        {
+            source: ['wx.zsxq.com/dweb2/index/group/:id'],
+        },
+    ],
+    features: {
+        requireConfig: [
+            {
+                name: 'ZSXQ_ACCESS_TOKEN',
+                description:
+                    '知识星球访问令牌,获取方式：\n1. 登录知识星球网页版\n2. 打开浏览器开发者工具，切换到 Application 面板\n3. 点击侧边栏中的Storage -> Cookies -> https://wx.zsxq.com\n4. 复制 Cookie 中的 zsxq_access_token 值',
+            },
+        ],
+    },
+    handler,
+    description: `| all  | digests | by_owner | questions | tasks |
+    | ---- | ------ | --------- | -------- | ------ |
+    | 最新 | 精华    | 只看星主    | 问答      | 作业   |`,
+};
+
+async function handler(ctx: Context): Promise<Data> {
+    const groupId = ctx.req.param('id');
+    const scope = ctx.req.param('scope') ?? 'all';
+    const accessToken = config.zsxq.accessToken;
+
+    if (!accessToken) {
+        throw new ConfigNotFoundError('该 RSS 源由于配置不正确而被禁用：令牌丢失。');
+    }
+
+    let count = Number(ctx.req.query('limit')) || 20;
+    if (count > 30) {
+        count = 30;
+    }
+
+    const { group } = await customFetch<GroupInfoResponse>(`/groups/${groupId}`);
+
+    const { topics } = await customFetch<TopicsResponse>(`/groups/${groupId}/topics?scope=${scope}&count=${count}`);
+
+    return {
+        title: `知识星球 - ${group.name}`,
+        description: group.description,
+        image: group.background_url,
+        link: `https://wx.zsxq.com/dweb2/index/group/${groupId}`,
+        item: generateTopicDataItem(topics),
+    };
+}

package/lib/routes/zsxq/namespace.ts

@@ -0,0 +1,6 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: '知识星球',
+    url: 'zsxq.com',
+};

package/lib/routes/zsxq/types.ts

@@ -0,0 +1,158 @@
+export interface BasicResponse<T> {
+    succeeded: boolean;
+    resp_data: T;
+    code?: number;
+    info?: string;
+    error?: string;
+}
+
+interface BasicTopic {
+    comments_count: number;
+    create_time: string;
+    digested: boolean;
+    group: {
+        group_id: number;
+        name: string;
+        type: string;
+        background_url: string;
+    };
+    latest_likes: {
+        create_time: string;
+        owner: {
+            avatar_url: string;
+            name: string;
+            number: number;
+            user_id: number;
+        };
+    }[];
+    likes_count: number;
+    readers_count: number;
+    reading_count: number;
+    rewards_count: number;
+    topic_id: number;
+    type: string;
+    user_specific: {
+        liked: false;
+        subscribed: false;
+    };
+    title?: string;
+}
+
+interface TalkTopic extends BasicTopic {
+    type: 'talk';
+    talk: {
+        text?: string;
+        images?: TopicImage[];
+    };
+}
+
+export interface QATopic extends BasicTopic {
+    type: 'q&a';
+    answer?: {
+        owner: {
+            avatar_url: string;
+            description: string;
+            location: string;
+            name: string;
+            user_id: number;
+        };
+        text?: string;
+        images?: TopicImage[];
+    };
+    answered: boolean;
+    question: {
+        images?: TopicImage[];
+        text?: string;
+        owner?: {
+            avatar_url: string;
+            description: string;
+            location: string;
+            name: string;
+            user_id: number;
+        };
+    };
+}
+
+interface TaskTopic extends BasicTopic {
+    type: 'task';
+    task: {
+        images?: TopicImage[];
+        text?: string;
+    };
+}
+
+interface SolutionTopic extends BasicTopic {
+    type: 'solution';
+    solution: {
+        task_id: number;
+        text?: string;
+        images?: TopicImage[];
+    };
+}
+
+interface UserInfo {
+    chat: {
+        identifier: string;
+    };
+    user: {
+        avatar_url: string;
+        introduction: string;
+        location: string;
+        name: string;
+        unique_id: string;
+        user_id: number;
+    };
+    user_specific: {
+        followed: boolean;
+    };
+}
+
+interface GroupInfo {
+    group: {
+        admin_ids: number[];
+        alive_time: string;
+        background_url: string;
+        category: {
+            category_id: number;
+            title: string;
+        };
+        create_time: string;
+        description: string;
+        group_id: number;
+        guest_ids: number[];
+        name: string;
+    };
+}
+
+export interface TopicImage {
+    image_id: number;
+    large: {
+        height: number;
+        width: number;
+        url: string;
+    };
+    original: {
+        height: number;
+        width: number;
+        url: string;
+        size: number;
+    };
+    thumbnail: {
+        height: number;
+        width: number;
+        url: string;
+    };
+    type: string;
+}
+
+export type Topic = TalkTopic | QATopic | TaskTopic | SolutionTopic;
+
+export type ResponseData = UserInfo | GroupInfo | Topic[];
+
+export type UserInfoResponse = BasicResponse<UserInfo>;
+
+export type GroupInfoResponse = BasicResponse<GroupInfo>;
+
+export type TopicsResponse = BasicResponse<{
+    topics: Topic[];
+}>;

package/lib/routes/zsxq/user.ts

@@ -0,0 +1,58 @@
+import type { Data, Route } from '@/types';
+import type { Context } from 'hono';
+import { config } from '@/config';
+import ConfigNotFoundError from '@/errors/types/config-not-found';
+import type { TopicsResponse, UserInfoResponse } from './types';
+import { customFetch, generateTopicDataItem } from './utils';
+
+export const route: Route = {
+    name: '用户足迹',
+    categories: ['social-media'],
+    path: '/user/:id',
+    example: '/zsxq/user/2414218251',
+    parameters: {
+        id: '用户id，从网页端url中获取',
+    },
+    maintainers: ['KarasuShin'],
+    radar: [
+        {
+            source: ['wx.zsxq.com/dweb2/index/footprint/:id'],
+        },
+    ],
+    features: {
+        requireConfig: [
+            {
+                name: 'ZSXQ_ACCESS_TOKEN',
+                description:
+                    '知识星球访问令牌,获取方式：\n1. 登录知识星球网页版\n2. 打开浏览器开发者工具，切换到 Application 面板\n3. 点击侧边栏中的Storage -> Cookies -> https://wx.zsxq.com\n4. 复制 Cookie 中的 zsxq_access_token 值',
+            },
+        ],
+    },
+    handler,
+};
+
+async function handler(ctx: Context): Promise<Data> {
+    const uid = ctx.req.param('id');
+    const accessToken = config.zsxq.accessToken;
+
+    if (!accessToken) {
+        throw new ConfigNotFoundError('该 RSS 源由于配置不正确而被禁用：令牌丢失。');
+    }
+
+    let count = Number(ctx.req.query('limit')) || 20;
+    if (count > 30) {
+        count = 30;
+    }
+
+    const userInfo = await customFetch<UserInfoResponse>(`/users/${uid}`);
+
+    const { topics } = await customFetch<TopicsResponse>(`/users/${uid}/topics/footprint?count=${count}`);
+
+    return {
+        title: `知识星球 - ${userInfo.user.name}`,
+        description: userInfo.user.introduction,
+        image: userInfo.user.avatar_url,
+        link: `https://wx.zsxq.com/dweb2/index/footprint/${uid}`,
+        item: generateTopicDataItem(topics),
+    };
+}

package/lib/routes/zsxq/utils.ts

@@ -0,0 +1,71 @@
+import got from '@/utils/got';
+import type { TopicImage, Topic, BasicResponse, ResponseData } from './types';
+import { parseDate } from '@/utils/parse-date';
+import { config } from '@/config';
+import type { DataItem } from '@/types';
+
+export async function customFetch<T extends BasicResponse<ResponseData>>(path: string, retryCount = 0): Promise<T['resp_data']> {
+    const apiUrl = 'https://api.zsxq.com/v2';
+
+    const response = await got(apiUrl + path, {
+        headers: {
+            cookie: `zsxq_access_token=${config.zsxq.accessToken};`,
+        },
+    });
+    const { succeeded, code, resp_data } = response.data as T;
+    if (succeeded) {
+        return resp_data;
+    }
+    // sometimes the request will fail with code 1059, retry will solve the problem
+    if (code === 1059 && retryCount < 3) {
+        return customFetch(path, ++retryCount);
+    }
+    throw new Error('something wrong');
+}
+
+function parseTopicContent(text: string = '', images: TopicImage[] = []) {
+    let result = text.replaceAll('\n', '<br>');
+    result = result.replaceAll(/<e type="web" href="(.*?)" title="(.*?)" \/>/g, (_, p1, p2) => `<a href=${decodeURIComponent(p1)}>${decodeURIComponent(p2)}</a>`);
+    result = result.replaceAll(/<e type="hashtag".*?title="(.*?)" \/>/g, (_, p1) => {
+        const title = decodeURIComponent(p1);
+        return `<span>${title}</span>`;
+    });
+    result += images.map((image) => `<img src="${image.original?.url ?? image.large?.url ?? image.thumbnail?.url}">`).join('<br>');
+    return result;
+}
+
+export function generateTopicDataItem(topics: Topic[]): DataItem[] {
+    return topics.map((topic) => {
+        let description: string | undefined;
+        let title = '';
+        switch (topic.type) {
+            case 'talk':
+                title = topic.talk?.text?.split('\n')[0] ?? '文章';
+                description = parseTopicContent(topic.talk?.text, topic.talk?.images);
+                break;
+            case 'q&a':
+                title = topic.question?.text?.split('\n')[0] ?? '问答';
+                description = parseTopicContent(topic.question?.text, topic.question?.images);
+                description = `<blockquote>${String(topic.question?.owner?.name ?? '匿名用户')} 提问：${description}</blockquote>`;
+                if (topic.answered) {
+                    description += '<br>' + topic.answer?.owner.name + ' 回答：<br><br>';
+                    description += parseTopicContent(topic.answer?.text, topic.answer?.images);
+                }
+                break;
+            case 'task':
+                title = topic.task?.text?.split('\n')[0] ?? '作业';
+                description = parseTopicContent(topic.task?.text, topic.task?.images);
+                break;
+            case 'solution':
+                title = topic.solution?.text?.split('\n')[0] ?? '写作业';
+                description = parseTopicContent(topic.solution?.text, topic.solution?.images);
+                break;
+            default:
+        }
+        return {
+            title: topic.title ?? title,
+            description,
+            pubDate: parseDate(topic.create_time),
+        };
+    });
+}

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);
+            }
+        );
     }
 });