rsshub 1.0.0-master.f6cb490 → 1.0.0-master.f6f0273

package/lib/routes/stcn/index.ts

@@ -1,159 +1,264 @@
-import { Route, ViewType } from '@/types';
+import { type Data, type DataItem, type Route, ViewType } from '@/types';
+
 import cache from '@/utils/cache';
-import got from '@/utils/got';
-import { load } from 'cheerio';
-import timezone from '@/utils/timezone';
+import ofetch from '@/utils/ofetch';
 import { parseDate } from '@/utils/parse-date';
+import timezone from '@/utils/timezone';
 
-export const route: Route = {
-    path: '/:id?',
-    categories: ['finance', 'popular'],
-    view: ViewType.Articles,
-    example: '/stcn/yw',
-    parameters: {
-        id: {
-            description: '栏目 id',
-            options: [
-                { value: 'kx', label: '快讯' },
-                { value: 'yw', label: '要闻' },
-                { value: 'gs', label: '股市' },
-                { value: 'company', label: '公司' },
-                { value: 'data', label: '数据' },
-                { value: 'fund', label: '基金' },
-                { value: 'finance', label: '金融' },
-                { value: 'comment', label: '评论' },
-                { value: 'cj', label: '产经' },
-                { value: 'ct', label: '创投' },
-                { value: 'kcb', label: '科创板' },
-                { value: 'xsb', label: '新三板' },
-                { value: 'tj', label: '投教' },
-                { value: 'zk', label: 'ESG' },
-                { value: 'gd', label: '滚动' },
-                { value: 'gsyl', label: '股市一览' },
-                { value: 'djjd', label: '独家解读' },
-                { value: 'gsxw', label: '公司新闻' },
-                { value: 'gsdt', label: '公司动态' },
-                { value: 'djsj', label: '独家数据' },
-                { value: 'kd', label: '看点数据' },
-                { value: 'zj', label: '资金流向' },
-                { value: 'sj_kcb', label: '科创板' },
-                { value: 'hq', label: '行情总貌' },
-                { value: 'zl', label: '专栏' },
-                { value: 'author', label: '作者' },
-                { value: 'cjhy', label: '行业' },
-                { value: 'cjqc', label: '汽车' },
-                { value: 'tjkt', label: '投教课堂' },
-                { value: 'zczs', label: '政策知识' },
-                { value: 'tjdt', label: '投教动态' },
-                { value: 'zthd', label: '专题活动' },
-            ],
-            default: 'yw',
-        },
-    },
-    features: {
-        requireConfig: false,
-        requirePuppeteer: false,
-        antiCrawler: false,
-        supportBT: false,
-        supportPodcast: false,
-        supportScihub: false,
-    },
-    name: '栏目',
-    maintainers: ['nczitzk'],
-    handler,
-    description: `| 快讯 | 要闻 | 股市 | 公司    | 数据 |
-| ---- | ---- | ---- | ------- | ---- |
-| kx   | yw   | gs   | company | data |
-
-| 基金 | 金融    | 评论    | 产经 | 创投 |
-| ---- | ------- | ------- | ---- | ---- |
-| fund | finance | comment | cj   | ct   |
-
-| 科创板 | 新三板 | 投教 | ESG | 滚动 |
-| ------ | ------ | ---- | --- | ---- |
-| kcb    | xsb    | tj   | zk  | gd   |
-
-| 股市一览 | 独家解读 |
-| -------- | -------- |
-| gsyl     | djjd     |
+import { type CheerioAPI, type Cheerio, type Element, load } from 'cheerio';
+import { type Context } from 'hono';
 
-| 公司新闻 | 公司动态 |
-| -------- | -------- |
-| gsxw     | gsdt     |
+export const handler = async (ctx: Context): Promise<Data> => {
+    const { id = 'yw' } = ctx.req.param();
+    const limit: number = Number.parseInt(ctx.req.query('limit') ?? '30', 10);
 
-| 独家数据 | 看点数据 | 资金流向 | 科创板  | 行情总貌 |
-| -------- | -------- | -------- | ------- | -------- |
-| djsj     | kd       | zj       | sj\_kcb | hq       |
+    const baseUrl: string = 'https://www.stcn.com';
+    const targetUrl: string = new URL(`article/list/${id}.html`, baseUrl).href;
 
-| 专栏 | 作者   |
-| ---- | ------ |
-| zl   | author |
+    const response = await ofetch(targetUrl);
+    const $: CheerioAPI = load(response);
+    const language = $('html').attr('lang') ?? 'zh-CN';
 
-| 行业 | 汽车 |
-| ---- | ---- |
-| cjhy | cjqc |
+    let items: DataItem[] = [];
 
-| 投教课堂 | 政策知识 | 投教动态 | 专题活动 |
-| -------- | -------- | -------- | -------- |
-| tjkt     | zczs     | tjdt     | zthd     |`,
-};
+    items = $('ul.infinite-list li')
+        .slice(0, limit)
+        .toArray()
+        .map((el): Element => {
+            const $el: Cheerio<Element> = $(el);
 
-async function handler(ctx) {
-    const id = ctx.req.param('id') ?? 'yw';
+            const $aEl: Cheerio<Element> = $el.find('div.tt a');
 
-    const rootUrl = 'https://www.stcn.com';
-    const currentUrl = `${rootUrl}/article/list/${id}.html`;
-    const apiUrl = `${rootUrl}/article/list.html?type=${id}`;
+            const title: string = $aEl.text();
+            const description: string = $el.find('div.text').html();
+            const pubDateStr: string | undefined = $el.find('div.info span').last().text().trim();
+            const linkUrl: string | undefined = $aEl.attr('href');
+            const categoryEls: Element[] = $el.find('div.tags span').toArray();
+            const categories: string[] = [...new Set(categoryEls.map((el) => $(el).text()).filter(Boolean))];
+            const authors: DataItem['author'] = $el.find('div.info span').first().text();
+            const image: string | undefined = $el.find('div.side a img').attr('src');
+            const upDatedStr: string | undefined = pubDateStr;
 
-    const response = await got({
-        method: 'get',
-        url: apiUrl,
-    });
+            const processedItem: DataItem = {
+                title,
+                description,
+                pubDate: pubDateStr ? timezone(parseDate(pubDateStr, ['HH:mm', 'MM-DD HH:mm', 'YYYY-MM-DD HH:mm']), +8) : undefined,
+                link: linkUrl ? new URL(linkUrl, baseUrl).href : undefined,
+                category: categories,
+                author: authors,
+                content: {
+                    html: description,
+                    text: description,
+                },
+                image,
+                banner: image,
+                updated: upDatedStr ? timezone(parseDate(upDatedStr, ['HH:mm', 'MM-DD HH:mm', 'YYYY-MM-DD HH:mm']), +8) : undefined,
+                language,
+            };
 
-    const $ = load(response.data);
+            return processedItem;
+        });
 
-    let items = $('.t, .tt, .title')
-        .find('a')
-        .toArray()
-        .map((item) => {
-            item = $(item);
+    items = (
+        await Promise.all(
+            items.map((item) => {
+                if (!item.link) {
+                    return item;
+                }
 
-            const link = item.attr('href');
+                return cache.tryGet(item.link, async (): Promise<DataItem> => {
+                    const detailResponse = await ofetch(item.link);
+                    const $$: CheerioAPI = load(detailResponse);
 
-            return {
-                title: item.text().replaceAll(/(^【|】$)/g, ''),
-                link: link.startsWith('http') ? link : `${rootUrl}${link}`,
-            };
-        });
+                    const title: string = $$('div.detail-title').text();
+                    const description: string = $$('div.detail-content').html() ?? '';
+                    const pubDateStr: string | undefined = $$('div.detail-info span').last().text().trim();
+                    const categories: string[] = $$('meta[name="keywords"]').attr('content')?.split(/,/) ?? [];
+                    const authors: DataItem['author'] = $$('div.detail-info span').first().text().split(/：/).pop();
+                    const upDatedStr: string | undefined = pubDateStr;
 
-    items = await Promise.all(
-        items.map((item) =>
-            cache.tryGet(item.link, async () => {
-                if (/\.html$/.test(item.link)) {
-                    const detailResponse = await got({
-                        method: 'get',
-                        url: item.link,
-                    });
-
-                    const content = load(detailResponse.data);
-
-                    item.title = content('.detail-title').text();
-                    item.author = content('.detail-info span').first().text().split('：').pop();
-                    item.pubDate = timezone(parseDate(content('.detail-info span').last().text()), +8);
-                    item.category = content('.detail-content-tags div')
-                        .toArray()
-                        .map((t) => content(t).text());
-                    item.description = content('.detail-content').html();
-                }
+                    const processedItem: DataItem = {
+                        title,
+                        description,
+                        pubDate: pubDateStr ? timezone(parseDate(pubDateStr), +8) : item.pubDate,
+                        category: categories,
+                        author: authors,
+                        content: {
+                            html: description,
+                            text: description,
+                        },
+                        updated: upDatedStr ? timezone(parseDate(upDatedStr), +8) : item.updated,
+                        language,
+                    };
 
-                return item;
+                    return {
+                        ...item,
+                        ...processedItem,
+                    };
+                });
             })
         )
-    );
+    ).filter((_): _ is DataItem => true);
 
     return {
-        title: `证券时报网 - ${$('.breadcrumb a').last().text()}`,
-        link: currentUrl,
+        title: $('title').text(),
+        description: $('meta[name="description"]').attr('content'),
+        link: targetUrl,
         item: items,
+        allowEmpty: true,
+        image: $('img.stcn-logo').attr('src'),
+        author: $('meta[name="keywords"]').attr('content')?.split(/,/)[0],
+        language,
+        id: targetUrl,
     };
-}
+};
+
+export const route: Route = {
+    path: '/article/list/:id?',
+    name: '列表',
+    url: 'www.stcn.com',
+    maintainers: ['nczitzk'],
+    handler,
+    example: '/stcn/article/list/yw',
+    parameters: {
+        category: {
+            description: '分类，默认为 `yw`，即要闻，可在对应分类页 URL 中找到',
+            options: [
+                {
+                    label: '要闻',
+                    value: 'yw',
+                },
+                {
+                    label: '股市',
+                    value: 'gs',
+                },
+                {
+                    label: '公司',
+                    value: 'company',
+                },
+                {
+                    label: '基金',
+                    value: 'fund',
+                },
+                {
+                    label: '金融',
+                    value: 'finance',
+                },
+                {
+                    label: '评论',
+                    value: 'comment',
+                },
+                {
+                    label: '产经',
+                    value: 'cj',
+                },
+                {
+                    label: '科创板',
+                    value: 'kcb',
+                },
+                {
+                    label: '新三板',
+                    value: 'xsb',
+                },
+                {
+                    label: 'ESG',
+                    value: 'zk',
+                },
+                {
+                    label: '滚动',
+                    value: 'gd',
+                },
+            ],
+        },
+    },
+    description: `:::tip
+若订阅 [要闻](https://www.stcn.com/article/list/yw.html)，网址为 \`https://www.stcn.com/article/list/yw.html\`，请截取 \`https://www.stcn.com/article/list/\` 到末尾 \`.html\` 的部分 \`yw\` 作为 \`id\` 参数填入，此时目标路由为 [\`/stcn/article/list/yw\`](https://rsshub.app/stcn/article/list/yw)。
+
+:::
+
+| 要闻 | 股市 | 公司    | 基金 | 金融    | 评论    |
+| ---- | ---- | ------- | ---- | ------- | ------- |
+| yw   | gs   | company | fund | finance | comment |
+
+| 产经 | 科创板 | 新三板 | ESG | 滚动 |
+| ---- | ------ | ------ | --- | ---- |
+| cj   | kcb    | xsb    | zk  | gd   |
+`,
+    categories: ['finance'],
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportRadar: true,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['www.stcn.com/article/list.html', 'www.stcn.com/article/list/:id'],
+            target: (params, url) => {
+                const urlObj: URL = new URL(url);
+                const id: string | undefined = urlObj.searchParams.get('type') ?? params.id;
+
+                return `/stcn/article/list${id ? `/${id}` : ''}`;
+            },
+        },
+        {
+            title: '要闻',
+            source: ['www.stcn.com/article/list.html', 'www.stcn.com/article/list/yw.html'],
+            target: '/article/list/yw',
+        },
+        {
+            title: '股市',
+            source: ['www.stcn.com/article/list.html', 'www.stcn.com/article/list/gs.html'],
+            target: '/article/list/gs',
+        },
+        {
+            title: '公司',
+            source: ['www.stcn.com/article/list.html', 'www.stcn.com/article/list/company.html'],
+            target: '/article/list/company',
+        },
+        {
+            title: '基金',
+            source: ['www.stcn.com/article/list.html', 'www.stcn.com/article/list/fund.html'],
+            target: '/article/list/fund',
+        },
+        {
+            title: '金融',
+            source: ['www.stcn.com/article/list.html', 'www.stcn.com/article/list/finance.html'],
+            target: '/article/list/finance',
+        },
+        {
+            title: '评论',
+            source: ['www.stcn.com/article/list.html', 'www.stcn.com/article/list/comment.html'],
+            target: '/article/list/comment',
+        },
+        {
+            title: '产经',
+            source: ['www.stcn.com/article/list.html', 'www.stcn.com/article/list/cj.html'],
+            target: '/article/list/cj',
+        },
+        {
+            title: '科创板',
+            source: ['www.stcn.com/article/list.html', 'www.stcn.com/article/list/kcb.html'],
+            target: '/article/list/kcb',
+        },
+        {
+            title: '新三板',
+            source: ['www.stcn.com/article/list.html', 'www.stcn.com/article/list/xsb.html'],
+            target: '/article/list/xsb',
+        },
+        {
+            title: 'ESG',
+            source: ['www.stcn.com/article/list.html', 'www.stcn.com/article/list/zk.html'],
+            target: '/article/list/zk',
+        },
+        {
+            title: '滚动',
+            source: ['www.stcn.com/article/list.html', 'www.stcn.com/article/list/gd.html'],
+            target: '/article/list/gd',
+        },
+    ],
+    view: ViewType.Articles,
+};

package/lib/routes/stcn/kx.ts

@@ -0,0 +1,144 @@
+import { type Data, type DataItem, type Route, ViewType } from '@/types';
+
+import cache from '@/utils/cache';
+import ofetch from '@/utils/ofetch';
+import { parseDate } from '@/utils/parse-date';
+import timezone from '@/utils/timezone';
+
+import { type CheerioAPI, load } from 'cheerio';
+import { type Context } from 'hono';
+
+export const handler = async (ctx: Context): Promise<Data> => {
+    const limit: number = Number.parseInt(ctx.req.query('limit') ?? '30', 10);
+
+    const baseUrl: string = 'https://www.stcn.com';
+    const targetUrl: string = new URL('article/list/kx.html', baseUrl).href;
+    const apiUrl: string = new URL('article/list.html', baseUrl).href;
+
+    const targetResponse = await ofetch(targetUrl);
+
+    const response = await ofetch(apiUrl, {
+        headers: {
+            'x-requested-with': 'XMLHttpRequest',
+        },
+        query: {
+            type: 'kx',
+        },
+    });
+
+    const $: CheerioAPI = load(targetResponse);
+    const language = $('html').attr('lang') ?? 'zh-CN';
+
+    let items: DataItem[] = [];
+
+    items = response.data.slice(0, limit).map((item): DataItem => {
+        const title: string = item.title;
+        const description: string = item.content;
+        const pubDate: number | string = item.time;
+        const linkUrl: string | undefined = item.url;
+        const categories: string[] = item.tags ? item.tags.map((c) => c.name) : [];
+        const authors: DataItem['author'] = item.source;
+        const image: string | undefined = item.share?.image;
+        const updated: number | string = pubDate;
+
+        const processedItem: DataItem = {
+            title,
+            description,
+            pubDate: pubDate ? parseDate(pubDate, 'X') : undefined,
+            link: linkUrl ? new URL(linkUrl, baseUrl).href : undefined,
+            category: categories,
+            author: authors,
+            content: {
+                html: description,
+                text: item.content ?? description,
+            },
+            image,
+            banner: image,
+            updated: updated ? parseDate(updated, 'X') : undefined,
+            language,
+        };
+
+        return processedItem;
+    });
+
+    items = (
+        await Promise.all(
+            items.map((item) => {
+                if (!item.link) {
+                    return item;
+                }
+
+                return cache.tryGet(item.link, async (): Promise<DataItem> => {
+                    const detailResponse = await ofetch(item.link);
+                    const $$: CheerioAPI = load(detailResponse);
+
+                    const title: string = $$('div.detail-title').text();
+                    const description: string = $$('div.detail-content').html() ?? '';
+                    const pubDateStr: string | undefined = $$('div.detail-info span').last().text().trim();
+                    const categories: string[] = [...new Set([...(item.category as string[]), ...($$('meta[name="keywords"]').attr('content')?.split(/,/) ?? [])])];
+                    const authors: DataItem['author'] = $$('div.detail-info span').first().text().split(/：/).pop();
+                    const upDatedStr: string | undefined = pubDateStr;
+
+                    const processedItem: DataItem = {
+                        title,
+                        description,
+                        pubDate: pubDateStr ? timezone(parseDate(pubDateStr), +8) : item.pubDate,
+                        category: categories,
+                        author: authors,
+                        content: {
+                            html: description,
+                            text: description,
+                        },
+                        updated: upDatedStr ? timezone(parseDate(upDatedStr), +8) : item.updated,
+                        language,
+                    };
+
+                    return {
+                        ...item,
+                        ...processedItem,
+                    };
+                });
+            })
+        )
+    ).filter((_): _ is DataItem => true);
+
+    return {
+        title: $('title').text(),
+        description: $('meta[name="description"]').attr('content'),
+        link: targetUrl,
+        item: items,
+        allowEmpty: true,
+        image: $('img.stcn-logo').attr('src'),
+        author: $('meta[name="keywords"]').attr('content')?.split(/,/)[0],
+        language,
+        id: targetUrl,
+    };
+};
+
+export const route: Route = {
+    path: '/article/list/kx',
+    name: '快讯',
+    url: 'www.stcn.com',
+    maintainers: ['nczitzk'],
+    handler,
+    example: '/stcn/article/list/kx',
+    parameters: undefined,
+    description: undefined,
+    categories: ['finance'],
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportRadar: true,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['www.stcn.com/article/list/kx.html'],
+            target: '/article/list/kx',
+        },
+    ],
+    view: ViewType.Articles,
+};

package/lib/routes/swjtu/namespace.ts

@@ -2,6 +2,6 @@ import type { Namespace } from '@/types';
 
 export const namespace: Namespace = {
     name: '西南交通大学',
-    url: 'ctt.swjtu.edu.cn',
+    url: 'www.swjtu.edu.cn',
     lang: 'zh-CN',
 };

package/lib/routes/swjtu/{scai/bks.ts → scai.ts}

@@ -5,13 +5,12 @@ import { parseDate } from '@/utils/parse-date';
 import ofetch from '@/utils/ofetch';
 
 const rootURL = 'https://scai.swjtu.edu.cn';
-const pageURL = `${rootURL}/web/page-module.html?mid=B730BEB095B31840`;
 
 export const route: Route = {
-    path: '/scai/bks',
+    path: '/scai/:type',
     categories: ['university'],
     example: '/swjtu/scai/bks',
-    parameters: {},
+    parameters: { type: '通知类型' },
     features: {
         requireConfig: false,
         requirePuppeteer: false,
@@ -26,11 +25,32 @@ export const route: Route = {
         },
     ],
     name: '计算机与人工智能学院',
-    description: '本科生教育',
+    description: `
+| 分区              | 参数         |
+| ----------------- | ----------- |
+| 本科生教育         | bks         |
+| 研究生教育         | yjs         |
+| 学生工作           | xsgz        |
+`,
     maintainers: ['AzureG03', 'SuperJeason'],
     handler,
 };
 
+const partition = {
+    bks: {
+        title: '本科生教育',
+        url: `${rootURL}/web/page-module.html?mid=B730BEB095B31840`,
+    },
+    yjs: {
+        title: '研究生教育',
+        url: `${rootURL}/web/page-module.html?mid=6A69B0E32021446B`,
+    },
+    xsgz: {
+        title: '学生工作',
+        url: `${rootURL}/web/page-module.html?mid=F3D3909EB1861B5D`,
+    },
+};
+
 const getItem = (item, cache) => {
     const title = item.find('a').text();
     const link = `${rootURL}${item.find('a').attr('href').slice(2)}`;
@@ -39,22 +59,14 @@ const getItem = (item, cache) => {
         const res = await ofetch(link);
         const $ = load(res);
 
-        // 尝试获取日期
         let dateText = $('div.news-info span:nth-of-type(2)').text();
+        // 转教务通知时的时间获取方法
         if (!dateText) {
             dateText = $('div.news-top-bar span:nth-of-type(1)').text();
         }
-        if (!dateText) {
-            //   console.error('Date not found for', link);
-            return null; // 或者返回一个默认值
-        }
-
-        const dateMatch = dateText.match(/\d{4}(-|\/|.)\d{1,2}\1\d{1,2}/);
-        if (!dateMatch) {
-            //   console.error('Invalid date format for', link);
-            return null; // 或者返回一个默认值
-        }
-        const pubDate = parseDate(dateMatch[0]);
+        // 'date' may be undefined. and 'parseDate' will return current time.
+        const date = dateText.match(/\d{4}(-|\/|.)\d{1,2}\1\d{1,2}/)?.[0];
+        const pubDate = parseDate(date);
         const description = $('div.content-main').html();
 
         return {
@@ -66,8 +78,10 @@ const getItem = (item, cache) => {
     });
 };
 
-async function handler() {
-    const res = await ofetch(pageURL);
+async function handler(ctx) {
+    const { type } = ctx.req.param();
+    const url = partition[type].url;
+    const res = await ofetch(url);
 
     const $ = load(res);
     const $list = $('div.list-top-item, div.item-wrapper');
@@ -80,8 +94,8 @@ async function handler() {
     );
 
     return {
-        title: '西南交大计算机学院-本科生教育',
-        link: pageURL,
+        title: `西南交大计院-${partition[type].title}`,
+        link: url,
         item: items,
         allowEmpty: true,
     };