rsshub 1.0.0-master.f9b85f5 → 1.0.0-master.f9c381a

package/lib/routes/surfshark/blog.ts

@@ -1,101 +1,297 @@
-import { Route } from '@/types';
+import { type Data, type DataItem, type Route, ViewType } from '@/types';
 
+import { art } from '@/utils/render';
 import cache from '@/utils/cache';
-import got from '@/utils/got';
-import { load } from 'cheerio';
+import ofetch from '@/utils/ofetch';
 import { parseDate } from '@/utils/parse-date';
-import { art } from '@/utils/render';
+
+import { type CheerioAPI, type Cheerio, load } from 'cheerio';
+import type { Element } from 'domhandler';
+import { type Context } from 'hono';
 import path from 'node:path';
 
-export const route: Route = {
-    path: '/blog/:category{.+}?',
-    name: 'Unknown',
-    maintainers: [],
-    handler,
-};
+export const handler = async (ctx: Context): Promise<Data> => {
+    const { category } = ctx.req.param();
+    const limit: number = Number.parseInt(ctx.req.query('limit') ?? '30', 10);
 
-async function handler(ctx) {
-    const category = ctx.req.param('category');
-    const limit = ctx.req.query('limit') ? Number.parseInt(ctx.req.query('limit'), 10) : 15;
+    const domain: string = 'surfshark.com';
+    const baseUrl: string = `https://${domain}`;
+    const targetUrl: string = new URL(`blog${category ? `/${category}` : ''}`, baseUrl).href;
 
-    const rootUrl = 'https://surfshark.com';
-    const currentUrl = new URL(`blog${category ? `/${category}` : ''}`, rootUrl).href;
+    const headers = {
+        host: 'surfshark.com',
+        origin: baseUrl,
+        referer: targetUrl,
+    };
 
-    const { data: response } = await got(currentUrl);
+    const response = await ofetch(targetUrl, {
+        headers,
+    });
+    const $: CheerioAPI = load(response);
+    const language = $('html').attr('lang') ?? 'en';
 
-    const $ = load(response);
+    let items: DataItem[] = [];
 
-    let items = $('div.article-info')
+    items = $('div.dg-article-single-card')
         .slice(0, limit)
         .toArray()
-        .map((item) => {
-            item = $(item);
-
-            const a = item.find('a');
-
-            return {
-                title: a.prop('title'),
-                link: a.prop('href'),
-                author: item.find('div.author, div.name').text().split('in')[0].trim(),
-                category: item
-                    .find('div.author, div.name')
-                    .find('a')
-                    .toArray()
-                    .map((c) => $(c).text()),
-                pubDate: parseDate(item.find('div.date, div.time').text().split('·')[0].trim(), 'YYYY, MMMM D'),
+        .map((el): Element => {
+            const $el: Cheerio<Element> = $(el);
+            const $aEl: Cheerio<Element> = $el.find('div.dg-article-content__info a[title]');
+
+            const title: string = $aEl.text();
+            const image: string | undefined = $el.find('a.dg-article-image img').attr('src');
+            const description: string | undefined = art(path.join(__dirname, 'templates/description.art'), {
+                images: image
+                    ? [
+                          {
+                              src: image,
+                              alt: title,
+                          },
+                      ]
+                    : undefined,
+            });
+
+            const pubDateEl: Cheerio<Element> = $el.find('span.js-date');
+            const pubDateStr: string | undefined = `${pubDateEl.attr('data-year')}-${pubDateEl.attr('data-month')}-${pubDateEl.attr('data-day')}`;
+
+            const linkUrl: string | undefined = $aEl.attr('href');
+            const categoryEls: Element[] = $el.find('div.dg-blog-breadcrumbs a').toArray();
+            const categories: string[] = [
+                ...new Set(
+                    categoryEls
+                        .map((el) => $(el).text())
+                        .slice(1)
+                        .filter(Boolean)
+                ),
+            ];
+            const authorEls: Element[] = $el.find('a.author-link ').toArray();
+            const authors: DataItem['author'] = authorEls.map((authorEl) => {
+                const $authorEl: Cheerio<Element> = $(authorEl);
+
+                return {
+                    name: $authorEl.text(),
+                    url: $authorEl.attr('href'),
+                    avatar: undefined,
+                };
+            });
+            const upDatedStr: string | undefined = pubDateStr;
+
+            const processedItem: DataItem = {
+                title,
+                description,
+                pubDate: parseDate(pubDateStr),
+                link: linkUrl,
+                category: categories,
+                author: authors,
+                content: {
+                    html: description,
+                    text: description,
+                },
+                image,
+                banner: image,
+                updated: parseDate(upDatedStr),
+                language,
             };
+
+            return processedItem;
         });
 
     items = await Promise.all(
-        items.map((item) =>
-            cache.tryGet(item.link, async () => {
-                const { data: detailResponse } = await got(item.link);
-
-                const content = load(detailResponse);
-
-                content('div.post-main-img').each(function () {
-                    const image = content(this).find('img');
-
-                    content(this).replaceWith(
-                        art(path.join(__dirname, 'templates/description.art'), {
-                            image: {
-                                src: image
-                                    .prop('srcset')
-                                    .match(/(https?:.*?\.\w+\s)/g)
-                                    .pop()
-                                    .trim(),
-                                alt: image.prop('alt'),
-                            },
-                        })
-                    );
+        items.map((item) => {
+            if (!item.link) {
+                return item;
+            }
+
+            return cache.tryGet(item.link, async (): Promise<DataItem> => {
+                const detailResponse = await ofetch(item.link, {
+                    headers,
                 });
+                const $$: CheerioAPI = load(detailResponse);
 
-                item.title = content('div.blog-post-title').text();
-                item.description = content('div.post-content').html();
-                item.author = content('meta[name="author"]').prop('content');
-                item.category = content('div.name')
-                    .find('a')
-                    .toArray()
-                    .map((c) => content(c).text());
-                item.pubDate = parseDate(content('meta[property="article:published_time"]').prop('content'));
+                const title: string = $$('meta[property="og:title"]').attr('content') ?? item.title;
+                const image: string | undefined = $$('div.dg-featured-img-container img').attr('src');
+                const description: string | undefined = art(path.join(__dirname, 'templates/description.art'), {
+                    images: image
+                        ? [
+                              {
+                                  src: image,
+                                  alt: title,
+                              },
+                          ]
+                        : undefined,
+                    description: $$('div.dg-blog-post-blocks').html(),
+                });
+                const pubDateStr: string | undefined = $$('meta[property="article:published_time"]').attr('content');
+                const authorEls: Element[] = $$('div.dg-blog-post-author-top').toArray();
+                const authors: DataItem['author'] = authorEls.map((authorEl) => {
+                    const $$authorEl: Cheerio<Element> = $$(authorEl);
 
-                return item;
-            })
-        )
-    );
+                    return {
+                        name: $$authorEl.find('a.author-link').last().text(),
+                        url: $$authorEl.find('a.author-link').last().attr('href'),
+                        avatar: $$authorEl.find('a.author-avatar img').attr('src'),
+                    };
+                });
+                const upDatedStr: string | undefined = $$('meta[property="article:modified_time"]').attr('content');
+
+                const processedItem: DataItem = {
+                    title,
+                    description,
+                    pubDate: pubDateStr ? parseDate(pubDateStr) : item.pubDate,
+                    author: authors,
+                    content: {
+                        html: description,
+                        text: description,
+                    },
+                    image,
+                    banner: image,
+                    updated: upDatedStr ? parseDate(upDatedStr) : item.updated,
+                    language,
+                };
 
-    const icon = new URL($('link[rel="apple-touch-icon"]').prop('href'), rootUrl).href;
+                return {
+                    ...item,
+                    ...processedItem,
+                };
+            });
+        })
+    );
 
     return {
-        item: items,
         title: $('title').text(),
-        link: currentUrl,
-        description: $('meta[property="og:description"]').prop('content'),
-        language: $('html').prop('lang'),
-        image: $('meta[property="og:image"]').prop('content'),
-        icon,
-        logo: icon,
-        subtitle: $('meta[property="og:title"]').prop('content'),
-        author: $('meta[property="og:site_name"]').prop('content'),
+        description: $('meta[property="og:description"]').attr('content'),
+        link: targetUrl,
+        item: items,
+        allowEmpty: true,
+        image: $('meta[property="og:image"]').attr('content'),
+        author: $('meta[property="og:site_name"]').attr('content'),
+        language,
+        id: $('meta[property="og:url"]').attr('content'),
     };
-}
+};
+
+export const route: Route = {
+    path: '/blog/:category{.+}?',
+    name: 'Blog',
+    url: 'surfshark.com',
+    maintainers: ['nczitzk'],
+    handler,
+    example: '/surfshark/blog',
+    parameters: {
+        category: {
+            description: 'Category, All by default',
+            options: [
+                {
+                    label: 'All',
+                    value: '',
+                },
+                {
+                    label: 'Cybersecurity',
+                    value: 'cybersecurity',
+                },
+                {
+                    label: 'All things VPN',
+                    value: 'all-things-vpn',
+                },
+                {
+                    label: 'Internet censorship',
+                    value: 'internet-censorship',
+                },
+                {
+                    label: 'Entertainment',
+                    value: 'entertainment',
+                },
+                {
+                    label: 'Expert Insights',
+                    value: 'expert-insights',
+                },
+                {
+                    label: 'Video',
+                    value: 'video',
+                },
+                {
+                    label: 'News',
+                    value: 'news',
+                },
+            ],
+        },
+    },
+    description: `:::tip
+To subscribe to [Cybersecurity](https://surfshark.com/blog/cybersecurity), where the source URL is \`https://surfshark.com/blog/cybersecurity\`, extract the certain parts from this URL to be used as parameters, resulting in the route as [\`/surfshark/blog/cybersecurity\`](https://rsshub.app/surfshark/blog/cybersecurity).
+:::
+
+<details>
+  <summary>More categories</summary>
+
+| Category                                                              | ID                                                                           |
+| --------------------------------------------------------------------- | ---------------------------------------------------------------------------- |
+| [All](https://surfshark.com/blog)                                     | (empty)                                                                      |
+| [Cybersecurity](https://surfshark.com/blog/cybersecurity)             | [cybersecurity](https://rsshub.app/surfshark/blog/cybersecurity)             |
+| [All things VPN](https://surfshark.com/blog/all-things-vpn)           | [all-things-vpn](https://rsshub.app/surfshark/blog/all-things-vpn)           |
+| [Internet censorship](https://surfshark.com/blog/internet-censorship) | [internet-censorship](https://rsshub.app/surfshark/blog/internet-censorship) |
+| [Entertainment](https://surfshark.com/blog/entertainment)             | [entertainment](https://rsshub.app/surfshark/blog/entertainment)             |
+| [Expert Insights](https://surfshark.com/blog/expert-insights)         | [expert-insights](https://rsshub.app/surfshark/blog/expert-insights)         |
+| [Video](https://surfshark.com/blog/video)                             | [video](https://rsshub.app/surfshark/blog/video)                             |
+| [News](https://surfshark.com/blog/news)                               | [news](https://rsshub.app/surfshark/blog/news)                               |
+
+</details>
+`,
+    categories: ['new-media'],
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportRadar: true,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['surfshark.com/blog/:category'],
+            target: '/blog/:category',
+        },
+        {
+            title: 'All',
+            source: ['surfshark.com/blog'],
+            target: '/blog',
+        },
+        {
+            title: 'Cybersecurity',
+            source: ['surfshark.com/blog/cybersecurity'],
+            target: '/blog/cybersecurity',
+        },
+        {
+            title: 'All things VPN',
+            source: ['surfshark.com/blog/all-things-vpn'],
+            target: '/blog/all-things-vpn',
+        },
+        {
+            title: 'Internet censorship',
+            source: ['surfshark.com/blog/internet-censorship'],
+            target: '/blog/internet-censorship',
+        },
+        {
+            title: 'Entertainment',
+            source: ['surfshark.com/blog/entertainment'],
+            target: '/blog/entertainment',
+        },
+        {
+            title: 'Expert Insights',
+            source: ['surfshark.com/blog/expert-insights'],
+            target: '/blog/expert-insights',
+        },
+        {
+            title: 'Video',
+            source: ['surfshark.com/blog/video'],
+            target: '/blog/video',
+        },
+        {
+            title: 'News',
+            source: ['surfshark.com/blog/news'],
+            target: '/blog/news',
+        },
+    ],
+    view: ViewType.Articles,
+};

package/lib/routes/surfshark/templates/description.art

@@ -1,5 +1,17 @@
-{{ if image }}
-  <figure>
-    <img src="{{ image.src }}" alt="{{ image.alt }}">
-  </figure>
+{{ if images }}
+  {{ each images image }}
+    {{ if image?.src }}
+      <figure>
+        <img
+          {{ if image.alt }}
+            alt="{{ image.alt }}"
+          {{ /if }}
+        src="{{ image.src }}">
+      </figure>
+    {{ /if }}
+  {{ /each }}
+{{ /if }}
+
+{{ if description }}
+  {{@ description }}
 {{ /if }}

package/lib/routes/sustainabilitymag/articles.ts

@@ -31,7 +31,7 @@ export const route: Route = {
 const findLargestImgKey = (images) =>
     Object.keys(images)
         .filter((key) => key.startsWith('inline_free_') || key.startsWith('hero_landscape_'))
-        .sort((a, b) => Number.parseInt(b.split('_')[2]) - Number.parseInt(a.split('_')[2]))[0];
+        .toSorted((a, b) => Number.parseInt(b.split('_')[2]) - Number.parseInt(a.split('_')[2]))[0];
 
 const renderFigure = (url, caption) => `<figure><img src="${url}" alt="${caption}" /><figcaption>${caption}</figcaption></figure>`;
 

package/lib/routes/syosetu/dev.ts

@@ -49,7 +49,7 @@ async function handler(): Promise<Data> {
     const updates = dates
         .map((date, index) => ({
             date,
-            content: contents[index]?.replace(/\n/g, '<br>') ?? '',
+            content: contents[index]?.replaceAll('\n', '<br>') ?? '',
         }))
         .filter((update) => update.content);
 

package/lib/routes/syosetu/ranking-isekai.ts

@@ -70,7 +70,7 @@ export async function handleIsekaiRanking(type: string, limit: number): Promise<
     }
 
     const items = uniqueNovels
-        .sort((a, b) => (b[pointField] || 0) - (a[pointField] || 0))
+        .toSorted((a, b) => (b[pointField] || 0) - (a[pointField] || 0))
         .map((novel, index) => ({
             title: `#${index + 1} ${novel.title}`,
             link: `https://ncode.syosetu.com/${String(novel.ncode).toLowerCase()}`,

package/lib/routes/szse/disclosure/listed-notice.ts

@@ -3,14 +3,52 @@ import { type Data, type DataItem, type Route, ViewType } from '@/types';
 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';
 
+function isValidDate(dateString: string): boolean {
+    // 正则表达式检查格式：YYYY-MM-DD
+    const regex = /^\d{4}-(0[1-9]|1[0-2])-(0[1-9]|[12][0-9]|3[01])$/;
+    if (!regex.test(dateString)) {
+        return false;
+    }
+
+    // 解析日期并验证有效性（避免像 2025-02-30 这样的无效日期）
+    const [year, month, day] = dateString.split('-').map(Number);
+    const date = new Date(year, month - 1, day); // 月份从 0 开始
+
+    // 检查解析后的日期是否与输入一致
+    return date.getFullYear() === year && date.getMonth() === month - 1 && date.getDate() === day;
+}
+
 export const handler = async (ctx: Context): Promise<Data> => {
     const { category = '' } = ctx.req.param();
     const limit: number = Number.parseInt(ctx.req.query('limit') ?? '50', 10);
-
+    const query: string = ctx.req.param('query') ?? '';
+    const queries: Record<string, string> = {
+        stock: '',
+        beginDate: '',
+        endDate: '',
+    };
+    if (query) {
+        for (const pair of query.split('&')) {
+            const [key, value] = pair.split('=');
+            if (key) {
+                queries[key] = value;
+            }
+        }
+    }
+    if (queries.beginDate && !isValidDate(queries.beginDate)) {
+        throw new Error('Invalid beginDate format. Expected YYYY-MM-DD');
+    }
+    if (queries.endDate) {
+        if (!isValidDate(queries.endDate)) {
+            throw new Error('Invalid endDate format. Expected YYYY-MM-DD');
+        }
+    } else if (queries.beginDate) {
+        // 如果只提供了开始日期，则将结束日期设置为开始日期
+        queries.endDate = queries.beginDate;
+    }
     const baseUrl: string = 'https://www.szse.cn';
     const staticBaseUrl: string = 'https://disc.static.szse.cn';
     const apiUrl: string = new URL('api/disc/announcement/annList', baseUrl).href;
@@ -19,11 +57,11 @@ export const handler = async (ctx: Context): Promise<Data> => {
     const targetResponse = await ofetch(targetUrl);
     const $: CheerioAPI = load(targetResponse);
     const language = $('html').attr('lang') ?? 'zh-CN';
-
     const response = await ofetch(apiUrl, {
         method: 'POST',
         body: {
-            seDate: ['', ''],
+            stock: queries.stock ? [queries.stock] : [],
+            seDate: [queries.beginDate, queries.endDate],
             channelCode: ['listedNotice_disc'],
             pageSize: limit,
             pageNum: 1,
@@ -83,13 +121,13 @@ export const handler = async (ctx: Context): Promise<Data> => {
 };
 
 export const route: Route = {
-    path: '/disclosure/listed/notice',
+    path: '/disclosure/listed/notice/:query?',
     name: '上市公司公告',
     url: 'www.szse.cn',
     maintainers: ['nczitzk'],
     handler,
     example: '/szse/disclosure/listed/notice',
-    parameters: undefined,
+    parameters: { query: 'Filter options. can filte by "stock","beginDate","endDate". example:"stock=000001&beginDate=2025-07-01&endDate=2025-08-30"' },
     description: undefined,
     categories: ['finance'],
     features: {