rsshub 1.0.0-master.ff3a9ce → 1.0.0-master.ff3ea5a

package/lib/routes/mathpix/blog.ts

@@ -0,0 +1,155 @@
+import { type Data, type DataItem, type Route, ViewType } from '@/types';
+
+import { art } from '@/utils/render';
+import cache from '@/utils/cache';
+import ofetch from '@/utils/ofetch';
+import { parseDate } from '@/utils/parse-date';
+
+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 handler = async (ctx: Context): Promise<Data> => {
+    const limit: number = Number.parseInt(ctx.req.query('limit') ?? '30', 10);
+
+    const baseUrl: string = 'https://mathpix.com';
+    const targetUrl: string = new URL('blog', baseUrl).href;
+
+    const response = await ofetch(targetUrl);
+    const $: CheerioAPI = load(response);
+    const language = $('html').attr('lang') ?? 'en';
+
+    const categoryMap = {};
+
+    $('div.navbar-menu a.blog-category').each((_, el) => {
+        const $el: Cheerio<Element> = $(el);
+
+        const id: string = $el.attr('data-category');
+        const name: string = $el.text()?.trim();
+
+        if (id && name) {
+            categoryMap[id] = name;
+        }
+    });
+
+    let items: DataItem[] = [];
+
+    items = $('li.articles__item')
+        .slice(0, limit)
+        .toArray()
+        .map((el): Element => {
+            const $el: Cheerio<Element> = $(el);
+
+            const title: string = $el.find('a.articles__title').text();
+            const image: string | undefined = $el.find('div.articles__image img').attr('srcset') ? new URL($el.find('div.articles__image img').attr('srcset') as string, baseUrl).href : undefined;
+            const description: string | undefined = art(path.join(__dirname, 'templates/description.art'), {
+                images: image
+                    ? [
+                          {
+                              src: image,
+                              alt: title,
+                          },
+                      ]
+                    : undefined,
+                intro: $el.find('div.articles__text').text(),
+            });
+            const pubDateStr: string | undefined = $el.find('time.articles__date').attr('datetime');
+            const linkUrl: string | undefined = $el.find('a.articles__title').attr('href');
+            const categoryIds: string[] =
+                $el
+                    .attr('data-category')
+                    ?.split(/,\s/)
+                    .map((category) => category.trim()) ?? [];
+            const categories: string[] = categoryIds.map((id) => categoryMap[id]).filter(Boolean);
+            const upDatedStr: string | undefined = pubDateStr;
+
+            const processedItem: DataItem = {
+                title,
+                description,
+                pubDate: pubDateStr ? parseDate(pubDateStr) : undefined,
+                link: linkUrl ? new URL(linkUrl, baseUrl).href : undefined,
+                category: categories,
+                content: {
+                    html: description,
+                    text: description,
+                },
+                image,
+                banner: image,
+                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 $$: CheerioAPI = load(detailResponse);
+
+                const title: string = $$('h1.article__title').text();
+                const description: string | undefined = $$('div#setText').html();
+
+                const processedItem: DataItem = {
+                    title,
+                    description,
+                    content: {
+                        html: description,
+                        text: description,
+                    },
+                    language,
+                };
+
+                return {
+                    ...item,
+                    ...processedItem,
+                };
+            });
+        })
+    );
+
+    return {
+        title: $('title').text(),
+        description: $('meta[property="og:description"]').attr('content'),
+        link: targetUrl,
+        item: items,
+        allowEmpty: true,
+        image: $('meta[property="og:image"]').attr('content'),
+        language,
+        id: $('meta[property="og:url"]').attr('content'),
+    };
+};
+
+export const route: Route = {
+    path: '/blog',
+    name: 'Blog',
+    url: 'mathpix.com',
+    maintainers: ['nczitzk'],
+    handler,
+    example: '/mathpix/blog',
+    parameters: undefined,
+    description: undefined,
+    categories: ['blog'],
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportRadar: true,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['mathpix.com/blog'],
+            target: '/blog',
+        },
+    ],
+    view: ViewType.Articles,
+};

package/lib/routes/mathpix/namespace.ts

@@ -0,0 +1,9 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: 'Mathpix',
+    url: 'mathpix.com',
+    categories: ['blog'],
+    description: '',
+    lang: 'en',
+};

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

@@ -0,0 +1,21 @@
+{{ 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 intro }}
+  <blockquote>{{ intro }}</blockquote>
+{{ /if }}
+
+{{ if description }}
+  {{@ description }}
+{{ /if }}

package/lib/routes/mckinsey/cn/category-map.ts

@@ -1,4 +1,4 @@
-const categories = {
+export const categories = {
     // https://www.mckinsey.com.cn/wp-json/wp/v2/categories?parent=25&per_page=100
     2: {
         name: '汽车',
@@ -13,7 +13,7 @@ const categories = {
         slug: 'consumers',
     },
     5: {
-        name: '医药',
+        name: '医药与医疗',
         slug: 'healthcare-pharmaceuticals',
     },
     7: {
@@ -24,17 +24,13 @@ const categories = {
         name: '制造业',
         slug: 'manufacturing',
     },
-    9: {
-        name: '私募',
-        slug: 'private-equity',
-    },
     10: {
         name: '技术，媒体与通信',
         slug: 'technology-media-and-telecom',
     },
     12: {
         name: '城市化与可持续发展',
-        slug: 'urbanization-sustainability-insights',
+        slug: 'urbanization-sustainability',
     },
     13: {
         name: '创新',
@@ -56,22 +52,20 @@ const categories = {
         name: '洞见',
         slug: 'insights',
     },
-    37: {
-        name: '麦肯锡季刊',
-        slug: 'mckinsey-season-publication',
-    },
     41: {
         name: '资本项目和基础设施',
         slug: 'capital-projects-infrastructure',
     },
     42: {
         name: '旅游、运输和物流',
-        slug: 'transport-logistics',
+        slug: '交通运输与物流',
+    },
+    19139: {
+        name: '出海与国际化、转型',
+        slug: '出海与国际化、转型',
     },
     45: {
         name: '全球基础材料',
-        slug: 'global-basic-material',
+        slug: '全球基础材料',
     },
 };
-
-export { categories };

package/lib/routes/mckinsey/cn/index.ts

@@ -2,9 +2,10 @@ import { Route, ViewType } from '@/types';
 import ofetch from '@/utils/ofetch';
 import { parseDate } from '@/utils/parse-date';
 import { categories } from './category-map';
+import { load } from 'cheerio';
+import cache from '@/utils/cache';
 
 const baseUrl = 'https://www.mckinsey.com.cn';
-const endpoint = `${baseUrl}/wp-json`;
 
 export const route: Route = {
     path: '/cn/:category?',
@@ -13,9 +14,9 @@ export const route: Route = {
     example: '/mckinsey/cn',
     parameters: {
         category: {
-            description: '分类',
-            options: Object.entries(categories).map(([value, label]) => ({ value, label: label.name })),
-            default: '25',
+            description: '分类，留空为 `最新洞见`',
+            options: Object.entries(categories).map(([, label]) => ({ value: label.slug, label: label.name })),
+            default: '最新洞见',
         },
     },
     features: {
@@ -29,52 +30,88 @@ export const route: Route = {
     name: '洞见',
     maintainers: ['laampui'],
     handler,
-    description: `| 分类 | 分类名             |
-| ---- | ------------------ |
-| 25   | 全部洞见           |
-| 2    | 汽车               |
-| 3    | 金融服务           |
-| 4    | 消费者             |
-| 5    | 医药               |
-| 7    | 数字化             |
-| 8    | 制造业             |
-| 9    | 私募               |
-| 10   | 技术，媒体与通信   |
-| 12   | 城市化与可持续发展 |
-| 13   | 创新               |
-| 16   | 人才与领导力       |
-| 18   | 宏观经济           |
-| 19   | 麦肯锡全球研究院   |
-| 37   | 麦肯锡季刊         |
-| 41   | 资本项目和基础设施 |
-| 42   | 旅游、运输和物流   |
-| 45   | 全球基础材料       |`,
+    description: `| 分类                            | 分类名             |
+| ------------------------------- | ------------------ |
+|                                 | 全部洞见           |
+| autos                           | 汽车               |
+| banking-insurance               | 金融服务           |
+| consumers                       | 消费者             |
+| healthcare-pharmaceuticals      | 医药与医疗         |
+| business-technology             | 数字化             |
+| manufacturing                   | 制造业             |
+| technology-media-and-telecom    | 技术，媒体与通信   |
+| urbanization-sustainability     | 城市化与可持续发展 |
+| innovation                      | 创新               |
+| talent-leadership               | 人才与领导力       |
+| macroeconomy                    | 宏观经济           |
+| mckinsey-global-institute       | 麦肯锡全球研究院   |
+| capital-projects-infrastructure | 资本项目和基础设施 |
+| 交通运输与物流                  | 旅游、运输和物流   |
+| 出海与国际化、转型              | 出海与国际化、转型 |
+| 全球基础材料                    | 全球基础材料       |`,
 };
 
 async function handler(ctx) {
     const { category = '25' } = ctx.req.param();
-    if (Number.isNaN(category)) {
-        categories.find((c) => c.slug === category);
+
+    let categorySlug = '';
+    if (category === '25') {
+        /* empty */
+    } else if (categories[category]) {
+        // Category number for backwards compatibility
+        categorySlug = categories[category].slug;
+    } else {
+        // Category slug
+        const c = Object.values(categories).find((c) => c.slug === category);
+        categorySlug = c ? c.slug : '';
     }
+    const link = `${baseUrl}/insights/${categorySlug ? `${categorySlug}/` : ''}`;
 
-    const posts = await ofetch(`${endpoint}/wp/v2/posts`, {
-        query: {
-            per_page: ctx.req.query('limit') ? Number(ctx.req.query('limit')) : 50,
-            categories: category,
-        },
+    const headers = {
+        'accept-language': 'en-US,en;q=0.9',
+        'sec-fetch-dest': 'document',
+        'sec-fetch-mode': 'navigate',
+        'sec-fetch-site': 'none',
+    };
+
+    const response = await ofetch(link, {
+        headers,
     });
+    const $ = load(response);
+
+    const list = $('.fl-post-grid-post')
+        .toArray()
+        .map((item) => {
+            const $item = $(item);
+            const a = $item.find('h2 a');
+            return {
+                title: a.attr('title'),
+                description: $item.find('.fl-post-grid-content').html()?.trim(),
+                link: a.attr('href'),
+                pubDate: $item.find('[itemprop="datePublished"]').length ? parseDate($item.find('[itemprop="datePublished"]').attr('content')!) : undefined,
+            };
+        });
+
+    const items = await Promise.all(
+        list.map((item) =>
+            cache.tryGet(item.link, async () => {
+                const response = await ofetch(item.link, {
+                    headers,
+                });
+                const $ = load(response);
+
+                item.description = $('.ast-post-format-').html() || item.description;
+                item.guid = $('link[rel="shortlink"]').attr('href');
 
-    const items = posts.map((item) => ({
-        title: item.title.rendered,
-        description: item.content.rendered,
-        link: item.link,
-        guid: item.guid.rendered,
-        pubDate: parseDate(item.date_gmt),
-    }));
+                return item;
+            })
+        )
+    );
 
     return {
-        title: category ? `McKinsey Greater China - ${categories[category].name}` : `McKinsey Greater China`,
-        link: `${baseUrl}/${category === '25' ? categories[category].slug : `${categories[25].slug}/${categories[category].slug}`}/`,
+        title: $('head title').text(),
+        link,
+        image: $('meta[name="msapplication-TileImage"]').attr('content'),
         item: items,
     };
 }

package/lib/routes/melonbooks/search.ts

@@ -34,6 +34,7 @@ export const route: Route = {
         supportBT: false,
         supportPodcast: false,
         supportScihub: false,
+        nsfw: true,
     },
     name: '搜索结果',
     maintainers: ['cokemine'],

package/lib/routes/metacritic/index.ts

@@ -21,7 +21,7 @@ async function handler(ctx) {
 
     const rootUrl = 'https://www.metacritic.com';
     const rootApiUrl = 'https://backend.metacritic.com';
-    const apiUrl = new URL('v1/xapi/finder/metacritic/web', rootApiUrl).href;
+    const apiUrl = new URL('finder/metacritic/web', rootApiUrl).href;
 
     const currentUrlObject = new URL(`/browse/${type}/all/all/all-time/${sort}/${filter ? `?${filter}` : ''}`, rootUrl);
     const currentUrlParams = currentUrlObject.searchParams;
@@ -112,7 +112,7 @@ async function handler(ctx) {
             description: item.description,
             score: item.criticScoreSummary?.score ?? undefined,
         }),
-        category: item.genres.map((c) => c.name),
+        category: item.genres?.map((c) => c.name),
         guid: `metacritic-${item.id}`,
         pubDate: parseDate(item.releaseDate),
         upvotes: item.criticScoreSummary?.positiveCount ? Number.parseInt(item.criticScoreSummary?.positiveCount, 10) : 0,

package/lib/routes/meteoblue/namespace.ts

@@ -0,0 +1,8 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: 'meteoblue',
+    url: 'meteoblue.com',
+    lang: 'en',
+    description: 'Weather forecasts, climate data, and meteorological news from meteoblue',
+};

package/lib/routes/meteoblue/weathernews.ts

@@ -0,0 +1,75 @@
+import { Route } from '@/types';
+import { load } from 'cheerio';
+import ofetch from '@/utils/ofetch';
+import { parseDate } from '@/utils/parse-date';
+
+export const route: Route = {
+    path: '/weathernews',
+    name: 'Weather News',
+    maintainers: ['tssujt'],
+    handler,
+    example: '/meteoblue/weathernews',
+    categories: ['blog'],
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    description: 'Weather news and articles from meteoblue',
+};
+
+async function handler() {
+    const baseUrl = 'https://www.meteoblue.com';
+    const url = `${baseUrl}/en/blog/article/weathernews`;
+
+    const response = await ofetch(url);
+    const $ = load(response);
+
+    // Extract articles from the page using the actual HTML structure
+    const articles = $('article[itemprop="blogPost"]')
+        .toArray()
+        .map((element) => {
+            const $article = $(element);
+
+            // Get title and link from h3 > a
+            const $link = $article.find('h3[itemprop="headline"] a[itemprop="mainEntityOfPage"]');
+            const title = $link.text().trim();
+            const link = $link.attr('href');
+
+            if (!title || !link) {
+                return null;
+            }
+
+            // Get date from time element
+            const $time = $article.find('time[itemprop="datePublished"]');
+            const dateText = $time.attr('datetime') || '';
+
+            // Extract author from the time element text
+            const $authorMeta = $article.find('meta[itemprop="author"]');
+            const author = $authorMeta.attr('content')?.trim() || 'meteoblue';
+
+            // Get description from itemprop="description"
+            const $description = $article.find('div[itemprop="description"]');
+            const description = $description.text().trim() || title;
+
+            return {
+                title,
+                link: link.startsWith('http') ? link : `${baseUrl}${link}`,
+                pubDate: dateText ? parseDate(dateText) : undefined,
+                author,
+                description,
+            };
+        })
+        .filter(Boolean);
+
+    return {
+        title: 'meteoblue Weather News',
+        link: url,
+        description: 'Latest weather news and articles from meteoblue',
+        item: articles,
+        allowEmpty: true,
+    };
+}

package/lib/routes/meteor/index.ts

@@ -58,7 +58,7 @@ async function handler(ctx) {
     return {
         title: `${board === 'all' ? '全部看板' : boardInfo.title} | Meteor 學生社群`,
         description: board === 'all' ? null : boardInfo.feedDescription,
-        image: board === 'all' ? null : (boardInfo.imgUrl === 'not_set' ? null : boardInfo.imgUrl),
+        image: board === 'all' ? null : boardInfo.imgUrl === 'not_set' ? null : boardInfo.imgUrl,
         link: `${board === 'all' ? `${baseUrl}/board/all` : boardInfo.link}/new`,
         item: items,
     };

package/lib/routes/microsoft/addon.ts

@@ -1,5 +1,5 @@
 import { Route } from '@/types';
-import got from '@/utils/got';
+import ofetch from '@/utils/ofetch';
 
 export const route: Route = {
     path: '/edge/addon/:crxid',
@@ -27,21 +27,19 @@ export const route: Route = {
 async function handler(ctx) {
     const crxid = ctx.req.param('crxid');
 
-    const page_url = `https://microsoftedge.microsoft.com/addons/detail/${crxid}`;
+    const pageUrl = `https://microsoftedge.microsoft.com/addons/detail/${crxid}`;
 
-    const { data } = await got({
-        method: 'get',
-        url: `https://microsoftedge.microsoft.com/addons/getproductdetailsbycrxid/${crxid}?hl=en`,
+    const data = await ofetch(`https://microsoftedge.microsoft.com/addons/getproductdetailsbycrxid/${crxid}?hl=en`, {
         headers: {
-            Referer: page_url,
+            Referer: pageUrl,
         },
     });
 
     return {
         title: `${data.name} - Microsoft Edge Addons`,
         description: data.shortDescription,
-        image: `https:${data.thumbnail}`,
-        link: page_url,
+        image: `https:${data.logoUrl}`,
+        link: pageUrl,
         item: [
             {
                 title: 'v' + data.version,
@@ -49,7 +47,7 @@ async function handler(ctx) {
                 description: data.description,
                 pubDate: new Date(data.lastUpdateDate * 1000),
                 guid: `edge::${crxid}::${data.version}`,
-                link: page_url,
+                link: pageUrl,
             },
         ],
     };

package/lib/routes/mihoyo/zzz/news.ts

@@ -0,0 +1,106 @@
+import { Route } from '@/types';
+import got from '@/utils/got';
+import { parseDate } from '@/utils/parse-date';
+
+const categories = {
+    'zh-cn': {
+        'news-all': {
+            id: '273',
+            title: '最新',
+        },
+        news: {
+            id: '278',
+            title: '新闻',
+        },
+        notice: {
+            id: '279',
+            title: '公告',
+        },
+        activity: {
+            id: '280',
+            title: '活动',
+        },
+        title: '绝区零',
+        link: 'https://zzz.mihoyo.com/news',
+        apiRootUrl: 'https://api-takumi-static.mihoyo.com/content_v2_user/app/706fd13a87294881/getContentList',
+    },
+    'zh-tw': {
+        'news-all': {
+            id: '288',
+            title: '最新',
+        },
+        news: {
+            id: '295',
+            title: '新聞',
+        },
+        notice: {
+            id: '296',
+            title: '公告',
+        },
+        activity: {
+            id: '297',
+            title: '活動',
+        },
+        title: '絕區零',
+        link: 'https://zenless.hoyoverse.com/zh-tw/news',
+        apiRootUrl: 'https://api-os-takumi-static.hoyoverse.com/content_v2_user/app/3e9196a4b9274bd7/getContentList',
+    },
+};
+
+export const route: Route = {
+    path: '/zzz/:location?/:category?',
+    categories: ['game'],
+    example: '/mihoyo/zzz',
+    parameters: { location: '区域，可选 `zh-cn`（国服，简中）或 `zh-tw`（国际服，繁中）', category: '分类，见下表，默认为最新' },
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['zzz.mihoyo.com/news'],
+            target: '/zzz',
+        },
+    ],
+    name: '绝区零',
+    maintainers: ['Yeye-0426'],
+    handler,
+    url: 'zzz.mihoyo.com/news',
+    description: `#### 新闻 {#mi-ha-you-jue-qu-ling-xin-wen}
+
+| 最新     | 新闻 | 公告   | 活动     |
+| -------- | ---- | ------ | -------- |
+| news-all | news | notice | activity |`,
+};
+
+async function handler(ctx) {
+    // location 地区 category 类型
+    const { location = 'zh-cn', category = 'news-all' } = ctx.req.param();
+    const limit = ctx.req.query('limit') ? Number.parseInt(ctx.req.query('limit')) : 50;
+
+    const params = {
+        'zh-cn': `iPageSize=${limit}&iPage=1&sLangKey=zh-cn&isPreview=0&iChanId=${categories[location][category].id}`,
+        'zh-tw': `iPageSize=${limit}&iPage=1&sLangKey=zh-tw&isPreview=0&iChanId=${categories[location][category].id}`,
+    };
+
+    const apiUrl = `${categories[location].apiRootUrl}?${params[location]}`;
+    const response = await got(apiUrl);
+    const list = response.data.data.list;
+    const items = list.map((item) => ({
+        title: item.sTitle,
+        description: item.sContent,
+        link: `${categories[location].link}/${item.iInfoId}`,
+        pubDate: parseDate(item.dtStartTime),
+        category: item.sCategoryName,
+    }));
+
+    return {
+        title: `${categories[location][category].title}-${categories[location].title}`,
+        link: `${categories[location].link}?category=${categories[location][category].id}`,
+        item: items,
+    };
+}