rsshub 1.0.0-master.fdb1381 → 1.0.0-master.fdb4325

package/lib/routes/xjtu/yz.ts

@@ -0,0 +1,74 @@
+import type { Data, DataItem, Route } from '@/types';
+import ofetch from '@/utils/ofetch';
+import { load } from 'cheerio';
+import { parseDate } from '@/utils/parse-date';
+import cache from '@/utils/cache';
+import logger from '@/utils/logger';
+import timezone from '@/utils/timezone';
+
+export const route: Route = {
+    path: '/yz/:category?',
+    categories: ['university'],
+    example: '/xjtu/yz/zsdt',
+    parameters: { category: '栏目类型，默认请求`zsdt`，详见下方表格' },
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['yz.xjtu.edu.cn/index/:category.htm'],
+            target: '/yz/:category',
+        },
+    ],
+    name: '研究生招生信息网',
+    maintainers: ['YoghurtGuy'],
+    handler,
+    description: `栏目类型
+
+| 招生动态 | 通知公告 | 政策法规 | 招生统计 | 历年复试线 | 博士招生 | 硕士招生 | 推免生 | 其他招生 |
+| -------- | -------- | -------- | -------- | ---------- | -------- | -------- | ------ | -------- |
+| zsdt     | tzgg     | zcfg     | zstj     | lnfsx      | bszs     | sszs     | tms    | qtzs     |`,
+};
+async function handler(ctx) {
+    const { category = 'zsdt' } = ctx.req.param();
+    const baseUrl = 'https://yz.xjtu.edu.cn';
+
+    const response = await ofetch(`${baseUrl}/index/${category}.htm`);
+    const $ = load(response);
+    const list = $('div.list-con ul li')
+        .toArray()
+        .map((item_) => {
+            const item = $(item_);
+            const a = item.find('a');
+            return {
+                title: a.attr('title'),
+                link: new URL(a.attr('href')!, baseUrl).href,
+                pubDate: timezone(parseDate(item.find('span.date.fr').text()), +8),
+            } as DataItem;
+        });
+    const items = await Promise.all(
+        list.map((item) =>
+            cache.tryGet(item.link!, async () => {
+                try {
+                    const res = await ofetch(item.link!);
+                    const content = load(res);
+                    item.description = content('#vsb_content').html()! + (content('form ul').length > 0 ? content('form ul').html() : '');
+                    return item;
+                } catch (error) {
+                    logger.error(`Fetch failed for ${item.link}:`, error);
+                    return item;
+                }
+            })
+        )
+    );
+    return {
+        title: '西安交通大学研究生招生信息网',
+        link: 'https://yz.xjtu.edu.cn',
+        item: items,
+    } as Data;
+}

package/lib/routes/xjtu/zs.ts

@@ -0,0 +1,170 @@
+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, type Cheerio, type Element, load } from 'cheerio';
+import { type Context } from 'hono';
+
+export const handler = async (ctx: Context): Promise<Data> => {
+    const { category = 'zsxx1/zskx' } = ctx.req.param();
+    const limit: number = Number.parseInt(ctx.req.query('limit') ?? '30', 10);
+
+    const baseUrl: string = 'https://zs.xjtu.edu.cn';
+    const targetUrl: string = new URL(`${category}.htm`, baseUrl).href;
+
+    const response = await ofetch(targetUrl);
+    const $: CheerioAPI = load(response);
+    const language = $('html').attr('lang') ?? 'zh';
+
+    let items: DataItem[] = [];
+
+    items = $('section.TextList ul li')
+        .slice(0, limit)
+        .toArray()
+        .map((el): Element => {
+            const $el: Cheerio<Element> = $(el);
+            const $aEl: Cheerio<Element> = $el.find('a.flex');
+
+            const title: string = $aEl.text();
+            const pubDateStr: string | undefined = $el.find('b').text();
+            const linkUrl: string | undefined = $aEl.attr('href');
+            const categoryEls: Element[] = $el.find('i.zc').toArray();
+            const categories: string[] = [...new Set(categoryEls.map((el) => $(el).text()).filter(Boolean))];
+            const upDatedStr: string | undefined = pubDateStr;
+
+            const processedItem: DataItem = {
+                title,
+                pubDate: pubDateStr ? parseDate(pubDateStr) : undefined,
+                link: linkUrl ? new URL(linkUrl, targetUrl).href : undefined,
+                category: categories,
+                updated: upDatedStr ? parseDate(upDatedStr) : 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.show01 h5').text();
+                    const description: string | undefined = $$('div.v_news_content').html();
+                    const pubDateStr: string | undefined = $$('div.show01 i')
+                        .text()
+                        ?.match(/(\d{4}-\d{2}-\d{2}\s\d{2}:\d{2}:\d{2})/)?.[1];
+                    const categoryEls: Element[] = $$('div.mianbao a').toArray().slice(1);
+                    const categories: string[] = [...new Set(categoryEls.map((el) => $$(el).text()).filter(Boolean))];
+                    const upDatedStr: string | undefined = pubDateStr;
+
+                    const processedItem: DataItem = {
+                        title,
+                        description,
+                        pubDate: pubDateStr ? timezone(parseDate(pubDateStr), +8) : item.pubDate,
+                        category: categories,
+                        content: {
+                            html: description,
+                            text: description,
+                        },
+                        updated: upDatedStr ? timezone(parseDate(upDatedStr), +8) : item.updated,
+                        language,
+                    };
+
+                    return {
+                        ...item,
+                        ...processedItem,
+                    };
+                });
+            })
+        )
+    ).filter((_): _ is DataItem => true);
+
+    const title: string = $('title').text();
+
+    return {
+        title,
+        description: title.split(/-/)[0],
+        link: targetUrl,
+        item: items,
+        allowEmpty: true,
+        image: $('div.logoimg img').attr('src'),
+        author: $('META[Name="keywords"]').attr('Content'),
+        language,
+        id: targetUrl,
+    };
+};
+
+export const route: Route = {
+    path: '/zs/:category{.+}?',
+    name: '本科招生网',
+    url: 'zs.xjtu.edu.cn',
+    maintainers: ['nczitzk'],
+    handler,
+    example: '/xjtu/zs/zsxx1/zskx',
+    parameters: {
+        category: {
+            description: '分类，默认为 zsxx1/zskx，可在对应分类页 URL 中找到',
+            options: [
+                {
+                    label: '招生快讯',
+                    value: 'zsxx1/zskx',
+                },
+                {
+                    label: '招生政策',
+                    value: 'zsxx1/zszc',
+                },
+                {
+                    label: '招生计划',
+                    value: 'zsxx1/zsjh',
+                },
+                {
+                    label: '阳光公告',
+                    value: 'zsxx1/yggg',
+                },
+                {
+                    label: '历年录取',
+                    value: 'zsxx1/lnlq',
+                },
+            ],
+        },
+    },
+    description: `:::tip
+若订阅 [招生快讯](https://zs.xjtu.edu.cn/zsxx1/zskx.htm)，网址为 \`https://zs.xjtu.edu.cn/zsxx1/zskx.htm\`，请截取 \`https://zs.xjtu.edu.cn/\` 到末尾 \`.htm\` 的部分 \`zsxx1/zskx\` 作为 \`category\` 参数填入，此时目标路由为 [\`/xjtu/zs/zsxx1/zskx\`](https://rsshub.app/xjtu/zs/zsxx1/zskx)。
+:::
+
+| [招生快讯](https://zs.xjtu.edu.cn/zsxx1/zskx.htm)   | [招生政策](https://zs.xjtu.edu.cn/zsxx1/zszc.htm)   | [招生计划](https://zs.xjtu.edu.cn/zsxx1/zsjh.htm)   | [阳光公告](https://zs.xjtu.edu.cn/zsxx1/yggg.htm)   | [历年录取](https://zs.xjtu.edu.cn/zsxx1/lnlq.htm)   |
+| --------------------------------------------------- | --------------------------------------------------- | --------------------------------------------------- | --------------------------------------------------- | --------------------------------------------------- |
+| [zsxx1/zskx](https://rsshub.app/xjtu/zs/zsxx1/zskx) | [zsxx1/zszc](https://rsshub.app/xjtu/zs/zsxx1/zszc) | [zsxx1/zsjh](https://rsshub.app/xjtu/zs/zsxx1/zsjh) | [zsxx1/yggg](https://rsshub.app/xjtu/zs/zsxx1/yggg) | [zsxx1/lnlq](https://rsshub.app/xjtu/zs/zsxx1/lnlq) |
+`,
+    categories: ['university'],
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportRadar: true,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['zs.xjtu.edu.cn/:category'],
+            target: (params) => {
+                const category: string = params.category;
+
+                return `/xjtu/zs${category ? `/${category}` : ''}`;
+            },
+        },
+    ],
+    view: ViewType.Articles,
+};

package/lib/routes/xkb/index.ts

@@ -27,13 +27,13 @@ export const route: Route = {
     handler,
     description: `常用栏目 ID：
 
-  | 栏目名 | ID  |
-  | ------ | --- |
-  | 首页   | 350 |
-  | 重点   | 359 |
-  | 广州   | 353 |
-  | 湾区   | 360 |
-  | 天下   | 355 |`,
+| 栏目名 | ID  |
+| ------ | --- |
+| 首页   | 350 |
+| 重点   | 359 |
+| 广州   | 353 |
+| 湾区   | 360 |
+| 天下   | 355 |`,
 };
 
 async function handler(ctx) {
@@ -44,7 +44,7 @@ async function handler(ctx) {
         method: 'get',
         url: currentUrl,
         headers: {
-            siteId: '35',
+            siteId: 35,
         },
     });
 

package/lib/routes/xmnn/epaper.ts

@@ -28,8 +28,8 @@ export const route: Route = {
     maintainers: ['nczitzk'],
     handler,
     description: `| 厦门日报 | 厦门晚报 | 海西晨报 | 城市捷报 |
-  | -------- | -------- | -------- | -------- |
-  | xmrb     | xmwb     | hxcb     | csjb     |`,
+| -------- | -------- | -------- | -------- |
+| xmrb     | xmwb     | hxcb     | csjb     |`,
 };
 
 async function handler(ctx) {

package/lib/routes/xsijishe/rank.ts

@@ -4,6 +4,9 @@ import cache from '@/utils/cache';
 import got from '@/utils/got';
 import { load } from 'cheerio';
 import { config } from '@/config';
+import { puppeteerGet } from './utils';
+import puppeteer from '@/utils/puppeteer';
+
 const baseUrl = 'https://xsijishe.com';
 
 export const route: Route = {
@@ -30,8 +33,8 @@ export const route: Route = {
                 description: '',
             },
         ],
-        requirePuppeteer: false,
-        antiCrawler: false,
+        requirePuppeteer: true,
+        antiCrawler: true,
         supportBT: false,
         supportPodcast: false,
         supportScihub: false,
@@ -56,6 +59,9 @@ async function handler(ctx) {
         throw new InvalidParameterError('Invalid rank type');
     }
 
+    const browser = await puppeteer();
+    let usePuppeteer = false;
+
     const url = `${baseUrl}/portal.php`;
     const headers = {
         'Accept-Encoding': 'gzip, deflate, br',
@@ -88,36 +94,62 @@ async function handler(ctx) {
             item = $(item);
             const title = item.find('h5').text().trim();
             const link = item.find('a').attr('href');
+            const description = item.find('img').prop('outerHTML');
             return {
                 title,
                 link: `${baseUrl}/${link}`,
+                description,
             };
         });
 
+    if (items.length > 0) {
+        const firstItem = items[0];
+        const resp = await got(firstItem.link, {
+            headers,
+        });
+        const $ = load(resp.data);
+        const firstViewBox = $('.t_f').first();
+        if (firstViewBox.length === 0) {
+            usePuppeteer = true;
+        }
+    }
+
     items = await Promise.all(
         items.map((item) =>
             cache.tryGet(item.link, async () => {
-                const resp = await got(item.link, {
-                    headers,
-                });
-                const $ = load(resp.data);
+                let data;
+                if (usePuppeteer) {
+                    data = await puppeteerGet(item.link, browser);
+                } else {
+                    const resp = await got(item.link, {
+                        headers,
+                    });
+                    data = resp.data;
+                }
+                const $ = load(data);
                 const firstViewBox = $('.t_f').first();
 
-                firstViewBox.find('img').each((_, img) => {
-                    img = $(img);
-                    if (img.attr('zoomfile')) {
-                        img.attr('src', img.attr('zoomfile'));
-                        img.removeAttr('zoomfile');
-                        img.removeAttr('file');
-                    }
-                    img.removeAttr('onmouseover');
-                });
-
-                item.description = firstViewBox.html();
+                if (firstViewBox.length === 1) {
+                    firstViewBox.find('img').each((_, img) => {
+                        img = $(img);
+                        if (img.attr('zoomfile')) {
+                            img.attr('src', img.attr('zoomfile'));
+                            img.removeAttr('zoomfile');
+                            img.removeAttr('file');
+                        }
+                        img.removeAttr('onmouseover');
+                    });
+
+                    item.description = firstViewBox.html();
+                }
+
                 return item;
             })
         )
     );
+
+    await browser.close();
+
     return {
         title,
         link: url,

package/lib/routes/xsijishe/utils.ts

@@ -0,0 +1,25 @@
+const puppeteerGet = async (url, browser) => {
+    const page = await browser.newPage();
+    const expectResourceTypes = new Set(['document', 'script']);
+    await page.setRequestInterception(true);
+    page.on('request', (request) => {
+        expectResourceTypes.has(request.resourceType()) ? request.continue() : request.abort();
+    });
+    await page.goto(url, {
+        waitUntil: 'domcontentloaded',
+    });
+
+    let html;
+    html = await page.evaluate(() => document.documentElement.innerHTML);
+    if (html.includes('抱歉，您尚未登录，没有权限访问该版块')) {
+        await page.close();
+        return html;
+    }
+
+    await page.waitForSelector('.t_f');
+    html = await page.evaluate(() => document.documentElement.innerHTML);
+    await page.close();
+    return html;
+};
+
+export { puppeteerGet };

package/lib/routes/xueqiu/stock-info.ts

@@ -29,8 +29,8 @@ export const route: Route = {
     maintainers: ['YuYang'],
     handler,
     description: `| 公告         | 新闻 | 研报     |
-  | ------------ | ---- | -------- |
-  | announcement | news | research |`,
+| ------------ | ---- | -------- |
+| announcement | news | research |`,
 };
 
 async function handler(ctx) {

package/lib/routes/xueqiu/timeline.ts

@@ -30,9 +30,9 @@ export const route: Route = {
   用户关注动态需要登录后的 Cookie 值，所以只能自建，详情见部署页面的配置模块。
 :::
 
-  | -1   | -2       | 1             |
-  | ---- | -------- | ------------- |
-  | 全部 | 关注精选 | 自定义第 1 组 |`,
+| -1   | -2       | 1             |
+| ---- | -------- | ------------- |
+| 全部 | 关注精选 | 自定义第 1 组 |`,
 };
 
 async function handler(ctx) {

package/lib/routes/xueqiu/user.ts

@@ -31,8 +31,8 @@ export const route: Route = {
     maintainers: ['imlonghao'],
     handler,
     description: `| 原发布 | 长文 | 问答 | 热门 | 交易 |
-  | ------ | ---- | ---- | ---- | ---- |
-  | 0      | 2    | 4    | 9    | 11   |`,
+| ------ | ---- | ---- | ---- | ---- |
+| 0      | 2    | 4    | 9    | 11   |`,
 };
 
 async function handler(ctx) {

package/lib/routes/yande/post.ts

@@ -27,8 +27,8 @@ export const route: Route = {
     name: 'Popular Recent Posts',
     maintainers: ['magic-akari', 'SettingDust', 'fashioncj', 'NekoAria'],
     description: `| 最近 24 小时    | 最近一周     | 最近一月    | 最近一年     |
-  | ------- | -------- | ------- | -------- |
-  | 1d | 1w | 1m | 1y |`,
+| ------- | -------- | ------- | -------- |
+| 1d | 1w | 1m | 1y |`,
     handler,
 };
 

package/lib/routes/ycwb/index.ts

@@ -30,13 +30,13 @@ export const route: Route = {
 
   常用栏目节点：
 
-  | 首页 | 中国 | 国际 | 体育 | 要闻 | 珠江评论 | 民生观察 | 房产 | 金羊教育 | 金羊财富 | 金羊文化 | 金羊健康 | 金羊汽车 |
-  | ---- | ---- | ---- | ---- | ---- | -------- | -------- | ---- | -------- | -------- | -------- | -------- | -------- |
-  | 1    | 14   | 15   | 16   | 22   | 1875     | 21773    | 222  | 5725     | 633      | 5281     | 21692    | 223      |
+| 首页 | 中国 | 国际 | 体育 | 要闻 | 珠江评论 | 民生观察 | 房产 | 金羊教育 | 金羊财富 | 金羊文化 | 金羊健康 | 金羊汽车 |
+| ---- | ---- | ---- | ---- | ---- | -------- | -------- | ---- | -------- | -------- | -------- | -------- | -------- |
+| 1    | 14   | 15   | 16   | 22   | 1875     | 21773    | 222  | 5725     | 633      | 5281     | 21692    | 223      |
 
-  | 广州 | 广州 - 广州要闻 | 广州 - 社会百态 | 广州 - 深读广州 | 广州 - 生活服务 | 今日大湾区 | 广东 - 政经热闻 | 广东 - 民生视点 | 广东 - 滚动新闻 |
-  | ---- | --------------- | --------------- | --------------- | --------------- | ---------- | --------------- | --------------- | --------------- |
-  | 18   | 5261            | 6030            | 13352           | 83422           | 100418     | 13074           | 12252           | 12212           |`,
+| 广州 | 广州 - 广州要闻 | 广州 - 社会百态 | 广州 - 深读广州 | 广州 - 生活服务 | 今日大湾区 | 广东 - 政经热闻 | 广东 - 民生视点 | 广东 - 滚动新闻 |
+| ---- | --------------- | --------------- | --------------- | --------------- | ---------- | --------------- | --------------- | --------------- |
+| 18   | 5261            | 6030            | 13352           | 83422           | 100418     | 13074           | 12252           | 12212           |`,
 };
 
 async function handler(ctx) {

package/lib/routes/yicai/dt.ts

@@ -33,45 +33,45 @@ export const route: Route = {
     handler,
     description: `#### [文章](https://dt.yicai.com/article)
 
-  | 分类     | ID         |
-  | -------- | ---------- |
-  | 全部     | article/0  |
-  | 新流行   | article/31 |
-  | 新趋势   | article/32 |
-  | 商业黑马 | article/33 |
-  | 新品     | article/34 |
-  | 营销     | article/35 |
-  | 大公司   | article/36 |
-  | 城市生活 | article/38 |
-
-  #### [报告](https://dt.yicai.com/report)
-
-  | 分类       | ID        |
-  | ---------- | --------- |
-  | 全部       | report/0  |
-  | 人群观念   | report/9  |
-  | 人群行为   | report/22 |
-  | 美妆个护   | report/23 |
-  | 3C 数码    | report/24 |
-  | 营销趋势   | report/25 |
-  | 服饰鞋包   | report/27 |
-  | 互联网     | report/28 |
-  | 城市与居住 | report/29 |
-  | 消费趋势   | report/30 |
-  | 生活趋势   | report/37 |
-
-  #### [可视化](https://dt.yicai.com/visualization)
-
-  | 分类     | ID               |
-  | -------- | ---------------- |
-  | 全部     | visualization/0  |
-  | 新流行   | visualization/39 |
-  | 新趋势   | visualization/40 |
-  | 商业黑马 | visualization/41 |
-  | 新品     | visualization/42 |
-  | 营销     | visualization/43 |
-  | 大公司   | visualization/44 |
-  | 城市生活 | visualization/45 |`,
+| 分类     | ID         |
+| -------- | ---------- |
+| 全部     | article/0  |
+| 新流行   | article/31 |
+| 新趋势   | article/32 |
+| 商业黑马 | article/33 |
+| 新品     | article/34 |
+| 营销     | article/35 |
+| 大公司   | article/36 |
+| 城市生活 | article/38 |
+
+#### [报告](https://dt.yicai.com/report)
+
+| 分类       | ID        |
+| ---------- | --------- |
+| 全部       | report/0  |
+| 人群观念   | report/9  |
+| 人群行为   | report/22 |
+| 美妆个护   | report/23 |
+| 3C 数码    | report/24 |
+| 营销趋势   | report/25 |
+| 服饰鞋包   | report/27 |
+| 互联网     | report/28 |
+| 城市与居住 | report/29 |
+| 消费趋势   | report/30 |
+| 生活趋势   | report/37 |
+
+#### [可视化](https://dt.yicai.com/visualization)
+
+| 分类     | ID               |
+| -------- | ---------------- |
+| 全部     | visualization/0  |
+| 新流行   | visualization/39 |
+| 新趋势   | visualization/40 |
+| 商业黑马 | visualization/41 |
+| 新品     | visualization/42 |
+| 营销     | visualization/43 |
+| 大公司   | visualization/44 |
+| 城市生活 | visualization/45 |`,
 };
 
 async function handler(ctx) {

package/lib/routes/yicai/news.ts

@@ -27,32 +27,32 @@ export const route: Route = {
     maintainers: ['nczitzk'],
     handler,
     description: `| Id                       | 名称       |
-  | ------------------------ | ---------- |
-  | gushi                    | A 股       |
-  | kechuangban              | 科创板     |
-  | hongguan                 | 大政       |
-  | jinrong                  | 金融       |
-  | quanqiushichang          | 海外市场   |
-  | gongsi                   | 产经       |
-  | shijie                   | 全球       |
-  | kechuang                 | 科技       |
-  | quyu                     | 区域       |
-  | comment                  | 评论       |
-  | dafengwenhua             | 商业人文   |
-  | books                    | 阅读周刊   |
-  | loushi                   | 地产       |
-  | automobile               | 汽车       |
-  | china\_financial\_herald | 对话陆家嘴 |
-  | fashion                  | 时尚       |
-  | ad                       | 商业资讯   |
-  | info                     | 资讯       |
-  | jzfxb                    | 价值风向标 |
-  | shuducaijing             | 数读财经   |
-  | shujujiepan              | 数据解盘   |
-  | shudushenghuo            | 数读生活   |
-  | cbndata                  | CBNData    |
-  | dtcj                     | DT 财经    |
-  | xfsz                     | 消费数知   |`,
+| ------------------------ | ---------- |
+| gushi                    | A 股       |
+| kechuangban              | 科创板     |
+| hongguan                 | 大政       |
+| jinrong                  | 金融       |
+| quanqiushichang          | 海外市场   |
+| gongsi                   | 产经       |
+| shijie                   | 全球       |
+| kechuang                 | 科技       |
+| quyu                     | 区域       |
+| comment                  | 评论       |
+| dafengwenhua             | 商业人文   |
+| books                    | 阅读周刊   |
+| loushi                   | 地产       |
+| automobile               | 汽车       |
+| china\_financial\_herald | 对话陆家嘴 |
+| fashion                  | 时尚       |
+| ad                       | 商业资讯   |
+| info                     | 资讯       |
+| jzfxb                    | 价值风向标 |
+| shuducaijing             | 数读财经   |
+| shujujiepan              | 数据解盘   |
+| shudushenghuo            | 数读生活   |
+| cbndata                  | CBNData    |
+| dtcj                     | DT 财经    |
+| xfsz                     | 消费数知   |`,
 };
 
 async function handler(ctx) {