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

package/lib/routes/taobao/mysql.ts

@@ -0,0 +1,157 @@
+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 { type CheerioAPI, type Cheerio, load } from 'cheerio';
+import type { Element } from 'domhandler';
+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 = 'http://mysql.taobao.org';
+    const targetUrl: string = new URL('monthly/', baseUrl).href;
+
+    const response = await ofetch(targetUrl);
+    const $: CheerioAPI = load(response);
+    const language = $('html').attr('lang') ?? 'zh';
+
+    let items: DataItem[] = [];
+    let count = 0;
+
+    items = await Promise.all(
+        $('h3 a.main')
+            .toArray()
+            .map(async (monthlyEl): Promise<Element[] | undefined> => {
+                const $monthlyEl: Cheerio<Element> = $(monthlyEl);
+
+                const monthlyUrl: string | undefined = $monthlyEl.attr('href') ? new URL($monthlyEl.attr('href') as string, baseUrl).href : undefined;
+
+                if (!monthlyUrl) {
+                    return undefined;
+                }
+
+                const monthlyResponse = await ofetch(monthlyUrl);
+
+                const $$: CheerioAPI = load(monthlyResponse);
+
+                return $$('h3 a.main')
+                    .toArray()
+                    .map((el): Element => {
+                        if (count < limit) {
+                            const $$el: Cheerio<Element> = $$(el);
+
+                            const title: string = $$el.text();
+                            const linkUrl: string | undefined = $$el.attr('href');
+                            const pubDateStr: string | undefined = linkUrl?.split(/monthly\//).pop();
+                            const upDatedStr: string | undefined = pubDateStr;
+
+                            const processedItem: DataItem = {
+                                title,
+                                pubDate: pubDateStr ? parseDate(pubDateStr) : undefined,
+                                link: linkUrl ? new URL(linkUrl, baseUrl).href : undefined,
+                                updated: upDatedStr ? parseDate(upDatedStr) : undefined,
+                                language,
+                            };
+                            count++;
+                            return processedItem;
+                        }
+                        return undefined;
+                    })
+                    .filter(Boolean);
+            })
+    );
+
+    items = await Promise.all(
+        items
+            .filter(Boolean)
+            .flat()
+            .slice(0, limit)
+            .map((item) => {
+                if (!item.link) {
+                    return item;
+                }
+
+                return cache.tryGet(item.link, async (): Promise<DataItem> => {
+                    const detailResponse = await ofetch(item.link);
+                    const $$: CheerioAPI = load(detailResponse);
+
+                    const title: string = $$('h2').first().text()?.trim() || item.title;
+                    const description: string | undefined = $$('div.content').html() ?? undefined;
+                    const pubDateStr: string | undefined = item.link.split(/monthly\//).pop();
+                    const authorEls: Element[] = $$('div.block p').toArray();
+                    const authors: DataItem['author'] = authorEls.map((authorEl) => {
+                        const $$authorEl: Cheerio<Element> = $$(authorEl);
+
+                        return {
+                            name: $$authorEl.text().split(/:/).pop()?.trim() ?? '',
+                            url: undefined,
+                            avatar: undefined,
+                        };
+                    });
+                    const upDatedStr: string | undefined = pubDateStr;
+
+                    const processedItem: DataItem = {
+                        title,
+                        description,
+                        pubDate: pubDateStr ? parseDate(pubDateStr) : item.pubDate,
+                        author: authors,
+                        content: {
+                            html: description,
+                            text: description,
+                        },
+                        updated: upDatedStr ? parseDate(upDatedStr) : item.updated,
+                        language,
+                    };
+
+                    return {
+                        ...item,
+                        ...processedItem,
+                    };
+                });
+            })
+    );
+
+    const title: string = $('title').text();
+
+    return {
+        title,
+        description: $('meta[name="description"]').attr('content'),
+        link: targetUrl,
+        item: items,
+        allowEmpty: true,
+        author: title,
+        language,
+        id: targetUrl,
+    };
+};
+
+export const route: Route = {
+    path: '/mysql/monthly',
+    name: '数据库内核月报',
+    url: 'mysql.taobao.org',
+    maintainers: ['nczitzk'],
+    handler,
+    example: '/taobao/mysql/monthly',
+    parameters: undefined,
+    description: undefined,
+    categories: ['programming'],
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportRadar: true,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['mysql.taobao.org/monthly/'],
+            target: '/mysql/monthly',
+        },
+    ],
+    view: ViewType.Articles,
+};

package/lib/routes/taobao/namespace.ts

@@ -1,7 +1,7 @@
 import type { Namespace } from '@/types';
 
 export const namespace: Namespace = {
-    name: '淘宝众筹',
-    url: 'izhongchou.taobao.com',
+    name: '淘宝网',
+    url: 'taobao.com',
     lang: 'zh-CN',
 };

package/lib/routes/techcrunch/category.ts

@@ -0,0 +1,48 @@
+import { Route } from '@/types';
+import got from '@/utils/got';
+import { load } from 'cheerio';
+import { parseDate } from '@/utils/parse-date';
+import { art } from '@/utils/render';
+import path from 'node:path';
+
+const host = 'https://techcrunch.com';
+export const route: Route = {
+    path: '/category/:categoryId',
+    categories: ['new-media'],
+    example: '/techcrunch/category/577047203',
+    parameters: {
+        categoryId: '分类id',
+    },
+    name: 'Category',
+    maintainers: ['MilliumOrion'],
+    handler,
+    description: `Use the category ID to retrieve a list of articles, category ID.  
+From the page source of \`https://techcrunch.com/category/***\`, locate the \`{category_id}\`  
+Example:  
+\`html\` -> \`head\` -> \`<link rel="alternate" title="JSON" type="application/json" href="https://techcrunch.com/wp-json/wp/v2/categories/{category_id}">\``,
+};
+
+async function handler(ctx) {
+    const categoryId = ctx.req.param('categoryId');
+    const { data } = await got(`${host}/wp-json/wp/v2/posts?categories=${categoryId}`);
+    const items = data.map((item) => {
+        const head = item.yoast_head_json;
+        const $ = load(item.content.rendered, null, false);
+        return {
+            title: item.title.rendered,
+            description: art(path.join(__dirname, 'templates/description.art'), {
+                head,
+                rendered: $.html(),
+            }),
+            link: item.link,
+            pubDate: parseDate(item.date_gmt),
+        };
+    });
+
+    return {
+        title: 'TechCrunch',
+        link: host,
+        description: 'Reporting on the business of technology, startups, venture capital funding, and Silicon Valley.',
+        item: items,
+    };
+}

package/lib/routes/telegram/channel.ts

@@ -515,7 +515,7 @@ async function handler(ctx) {
                         const mapBackground = locationObj.find('.tgme_widget_message_location').css('background-image');
                         const mapBackgroundUrl = mapBackground && mapBackground.match(/url\('(.*)'\)/);
                         const mapBackgroundUrlSrc = mapBackgroundUrl && mapBackgroundUrl[1];
-                        const mapImgHtml = mapBackgroundUrlSrc ? `<img src="${mapBackgroundUrlSrc}">` : (showMediaTagAsEmoji ? mediaTagDict[LOCATION][1] : mediaTagDict[LOCATION][0]);
+                        const mapImgHtml = mapBackgroundUrlSrc ? `<img src="${mapBackgroundUrlSrc}">` : showMediaTagAsEmoji ? mediaTagDict[LOCATION][1] : mediaTagDict[LOCATION][0];
                         return locationLink ? `<a href="${locationLink}">${mapImgHtml}</a>` : mapImgHtml;
                     } else {
                         return '';
@@ -752,6 +752,6 @@ async function handler(ctx) {
                 };
             })
             .filter(Boolean)
-            .reverse(),
+            .toReversed(),
     };
 }

package/lib/routes/tesla/cx.ts

@@ -132,7 +132,7 @@ async function handler(ctx) {
             pageSize: limit,
             pageNumber: 0,
             benefitCategoryId: categoryObject?.id ?? undefined,
-            category: categoryObject ? undefined : (category === '充电免停' ? 2 : undefined),
+            category: categoryObject ? undefined : category === '充电免停' ? 2 : undefined,
             city,
         },
     });

package/lib/routes/test/index.ts

@@ -416,6 +416,6 @@ async function handler(ctx) {
         item,
         allowEmpty: ctx.req.param('id') === 'allow_empty',
         description:
-            ctx.req.param('id') === 'complicated' ? '<img src="http://mock.com/DIYgod/DIYgod/RSSHub">' : (ctx.req.param('id') === 'multimedia' ? '<video src="http://mock.com/DIYgod/DIYgod/RSSHub"></video>' : 'A test route for RSSHub'),
+            ctx.req.param('id') === 'complicated' ? '<img src="http://mock.com/DIYgod/DIYgod/RSSHub">' : ctx.req.param('id') === 'multimedia' ? '<video src="http://mock.com/DIYgod/DIYgod/RSSHub"></video>' : 'A test route for RSSHub',
     };
 }

package/lib/routes/themoviedb/episodes.ts

@@ -36,7 +36,7 @@ async function handler(ctx) {
         description: seasonDetails.overview.trim(),
         image: `https://image.tmdb.org/t/p/original${seasonDetails.poster_path}`,
         link: `https://www.themoviedb.org/tv/${tvShowDetails.id}/season/${seasonDetails.season_number}`,
-        item: seasonDetails.episodes.reverse().map((item) => ({
+        item: seasonDetails.episodes.toReversed().map((item) => ({
             title: `${item.episode_number} ${item.name}`,
             link: `https://www.themoviedb.org/tv/${tvShowDetails.id}/season/${item.season_number}/episode/${item.episode_number}`,
             description: handleDescription(item),

package/lib/routes/thewirehindi/category.ts

@@ -0,0 +1,78 @@
+import { Data, Route } from '@/types';
+import got from '@/utils/got';
+import { mapPostToItem } from './utils';
+
+const categories = [
+    { value: 'interview', label: 'इंटरव्यू', id: 92820 },
+    { value: 'audio', label: 'ऑडियो', id: 5153 },
+    { value: 'kala-sahitya', label: 'कला-साहित्य', id: 101061 },
+    { value: 'campus', label: 'कैंपस', id: 5261 },
+    { value: 'covid-19', label: 'कोविड-19', id: 73034 },
+    { value: 'jan-ki-baat', label: 'जन की बात', id: 985 },
+    { value: 'duniya', label: 'दुनिया', id: 33 },
+    { value: 'north-east', label: 'नॉर्थ ईस्ट', id: 5834 },
+    { value: 'prasangik', label: 'प्रासंगिक', id: 3394 },
+    { value: 'bharat', label: 'भारत', id: 30 },
+    { value: 'media', label: 'मीडिया', id: 3338 },
+    { value: 'media-bol', label: 'मीडिया बोल', id: 7963 },
+    { value: 'rajneeti', label: 'राजनीति', id: 31 },
+    { value: 'vichar', label: 'विचार', id: 73061 },
+    { value: 'vigyan', label: 'विज्ञान', id: 32 },
+    { value: 'vishesh', label: 'विशेष', id: 2494 },
+    { value: 'video', label: 'वीडियो', id: 34 },
+    { value: 'samaj', label: 'समाज', id: 28 },
+    { value: 'ham-bhi-bharat', label: 'हम भी भारत', id: 14383 },
+    { value: 'hamare-bare-mein', label: 'हमारे बारे में', id: 29 },
+];
+
+export const route: Route = {
+    path: '/category/:category',
+    categories: ['new-media'],
+    example: '/thewirehindi/category/bharat',
+    parameters: {
+        category: {
+            description: 'Category name',
+            options: categories.map(({ value, label }) => ({ value, label })),
+        },
+    },
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['thewirehindi.com/category/*'],
+        },
+    ],
+    name: 'Category',
+    maintainers: ['Rjnishant530'],
+    handler,
+    url: 'thewirehindi.com/',
+};
+
+async function handler(ctx) {
+    const { category } = ctx.req.param();
+    const categoryData = categories.find((cat) => cat.value === category);
+
+    if (!categoryData) {
+        throw new Error(`Category "${category}" not found`);
+    }
+
+    const apiUrl = `https://thewirehindi.com/wp-json/wp/v2/posts?categories=${categoryData.id}&_embed`;
+    const { data } = await got(apiUrl);
+
+    const items = data.map((v) => mapPostToItem(v));
+
+    return {
+        title: `The Wire Hindi - ${categoryData.label}`,
+        link: `https://thewirehindi.com/category/${category}/`,
+        item: items,
+        description: `Latest news from The Wire Hindi - ${categoryData.label} category`,
+        logo: 'https://thewirehindi.com/wp-content/uploads/2023/05/cropped-The-wire-32x32.jpeg',
+        language: 'hi',
+    } as Data;
+}

package/lib/routes/thewirehindi/index.ts

@@ -0,0 +1,43 @@
+import { Data, Route } from '@/types';
+import got from '@/utils/got';
+import { mapPostToItem } from './utils';
+
+export const route: Route = {
+    path: '/',
+    categories: ['new-media'],
+    example: '/thewirehindi',
+    parameters: {},
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['thewirehindi.com/'],
+        },
+    ],
+    name: 'Latest News',
+    maintainers: ['Rjnishant530'],
+    handler,
+    url: 'thewirehindi.com/',
+};
+
+async function handler() {
+    const apiUrl = 'https://thewirehindi.com/wp-json/wp/v2/posts?_embed';
+    const { data } = await got(apiUrl);
+
+    const items = data.map((v) => mapPostToItem(v));
+
+    return {
+        title: 'The Wire Hindi - Latest News',
+        link: 'https://thewirehindi.com/',
+        item: items,
+        description: 'Latest news from The Wire Hindi',
+        logo: 'https://thewirehindi.com/wp-content/uploads/2023/05/cropped-The-wire-32x32.jpeg',
+        language: 'hi',
+    } as Data;
+}

package/lib/routes/thewirehindi/namespace.ts

@@ -0,0 +1,7 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: 'The Wire Hindi',
+    url: 'thewirehindi.com',
+    lang: 'hi',
+};

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

@@ -0,0 +1,9 @@
+{{@ excerpt }}
+<br><br>
+{{ if image }}
+<img src="{{ image }}" alt="{{ altText }}" style="max-width: 100%; height: auto;">
+<br><br>
+{{ /if }}
+{{ if content }}
+{{@ content }}
+{{ /if }}

package/lib/routes/thewirehindi/utils.ts

@@ -0,0 +1,33 @@
+import { DataItem } from '@/types';
+import { parseDate } from '@/utils/parse-date';
+import { art } from '@/utils/render';
+import path from 'node:path';
+
+export function mapPostToItem(post): DataItem {
+    const featuredMedia = post._embedded?.['wp:featuredmedia']?.find((v) => v.id === post.featured_media);
+    const image = featuredMedia?.source_url;
+    const altText = featuredMedia?.alt_text || featuredMedia?.title?.rendered || 'Featured Image';
+    return {
+        title: post.title.rendered,
+        description: art(path.join(__dirname, 'templates/description.art'), {
+            excerpt: post.excerpt.rendered,
+            content: post.content.rendered,
+            image,
+            altText,
+        }),
+        link: post.link,
+        pubDate: parseDate(post.date_gmt),
+        updated: parseDate(post.modified_gmt),
+        itunes_item_image: image,
+        image,
+        author:
+            post._embedded?.author
+                ?.map((v) => ({
+                    name: v.name,
+                    url: v.link,
+                    avatar: v.avatar_urls?.['96'] || v.avatar_urls?.['48'] || v.avatar_urls?.['24'],
+                }))
+                .join(', ') || 'The Wire Hindi',
+        category: post._embedded?.['wp:term']?.flat().map((v) => v.name),
+    } as DataItem;
+}

package/lib/routes/threads/utils.ts

@@ -6,8 +6,8 @@ import ofetch from '@/utils/ofetch';
 import { JSDOM } from 'jsdom';
 import { JSONPath } from 'jsonpath-plus';
 
-const profileUrl = (user: string) => `https://www.threads.net/@${user}`;
-const threadUrl = (code: string) => `https://www.threads.net/t/${code}`;
+const profileUrl = (user: string) => `https://www.threads.com/@${user}`;
+const threadUrl = (code: string) => `https://www.threads.com/t/${code}`;
 
 const USER_AGENT = 'Mozilla/5.0 (iPhone; CPU iPhone OS 16_6 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/16.6 Mobile/15E148 Safari/604.1';