rsshub 1.0.0-master.fb6d4f3 → 1.0.0-master.fb7e42c

package/lib/routes/huggingface/daily-papers.ts

@@ -4,10 +4,10 @@ import { load } from 'cheerio';
 import { parseDate } from '@/utils/parse-date';
 
 export const route: Route = {
-    path: '/daily-papers',
+    path: '/daily-papers/:cycle?/:voteFliter?',
     categories: ['programming'],
-    example: '/huggingface/daily-papers',
-    parameters: {},
+    example: '/huggingface/daily-papers/week/50',
+    parameters: { cycle: 'The publication cycle you want to follow. Choose from: date, week, month. Default: date', voteFliter: 'Filter papers by vote count.' },
     features: {
         requireConfig: false,
         requirePuppeteer: false,
@@ -18,21 +18,57 @@ export const route: Route = {
     },
     radar: [
         {
-            source: ['huggingface.co/papers', 'huggingface.co/'],
+            source: ['huggingface.co/papers/:cycle'],
+            target: '/daily-papers/:cycle',
         },
     ],
     name: 'Daily Papers',
-    maintainers: ['zeyugao'],
+    maintainers: ['zeyugao', 'ovo-tim'],
     handler,
     url: 'huggingface.co/papers',
 };
 
-async function handler() {
-    const { body: response } = await got('https://huggingface.co/papers');
+interface Paper {
+    id: string;
+    summary: string;
+    upvotes: number;
+    authors: { name: string }[];
+}
+
+interface DailyPaperItem {
+    title: string;
+    paper: Paper;
+    publishedAt: string;
+}
+
+interface PapersData {
+    dailyPapers: DailyPaperItem[];
+}
+
+async function handler(ctx) {
+    const { cycle = 'date', voteFliter = '0' } = ctx.req.param();
+    let url: string;
+    switch (cycle) {
+        case 'date':
+            url = 'https://huggingface.co/papers';
+            break;
+        case 'week':
+            // We don't actually need to get the week number, because huggingface.co/papers/week/YYYY-W52 will redirect to the latest week
+            url = `https://huggingface.co/papers/week/${new Date().getFullYear()}-W52`;
+            break;
+        case 'month':
+            url = `https://huggingface.co/papers/month/${new Date().toISOString().slice(0, 7)}`;
+            break;
+        default:
+            throw new Error(`Invalid cycle: ${cycle}`);
+    }
+
+    const { body: response } = await got(url);
     const $ = load(response);
-    const papers = $('main > div[data-target="DailyPapers"]').data('props');
+    const papers = $('main > div[data-target="DailyPapers"]').data('props') as PapersData;
 
     const items = papers.dailyPapers
+        .filter((item) => item.paper.upvotes >= voteFliter)
         .map((item) => ({
             title: item.title,
             link: `https://arxiv.org/abs/${item.paper.id}`,
@@ -41,7 +77,7 @@ async function handler() {
             author: item.paper.authors.map((author) => author.name).join(', '),
             upvotes: item.paper.upvotes,
         }))
-        .sort((a, b) => b.upvotes - a.upvotes);
+        .toSorted((a, b) => b.upvotes - a.upvotes);
 
     return {
         title: 'Huggingface Daily Papers',

package/lib/routes/huijin-inv/namespace.ts

@@ -0,0 +1,7 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: '中央汇金投资有限责任公司',
+    url: 'www.huijin-inv.cn',
+    lang: 'zh-CN',
+};

package/lib/routes/huijin-inv/news.ts

@@ -0,0 +1,83 @@
+import { type Data, type DataItem, Route } from '@/types';
+import { type CheerioAPI, load } from 'cheerio';
+import ofetch from '@/utils/ofetch';
+import timezone from '@/utils/timezone';
+import { parseDate } from '@/utils/parse-date';
+
+const BASE_URL = 'https://www.huijin-inv.cn';
+
+/**
+ * The client-side entrypoint to redirect to the post index of the latest year.
+ */
+const ENTRY_URL = `${BASE_URL}/huijin-inv/Corporate_History/index.shtml`;
+const DEFAULT_REDIRECT_PATH = '/huijin-inv/SC20252/Information_Center.shtml';
+
+async function handler(): Promise<Data> {
+    const entryRes = await ofetch(ENTRY_URL);
+    const $entry: CheerioAPI = load(entryRes);
+    const $scripts = $entry('head script');
+    let redirectPath = DEFAULT_REDIRECT_PATH;
+    $scripts.each((_, el) => {
+        const redirectScript = $entry(el).text();
+        if (redirectScript !== null) {
+            // Get the real index page by JS redirection href. The path may change.
+            const match = redirectScript.match(/window\.location\.href\s*=\s*["']([^"']+)["']/);
+            if (match) {
+                redirectPath = match[1];
+            }
+        }
+    });
+    const redirectURL = `${BASE_URL}${redirectPath}`;
+    const indexPage = await ofetch(redirectURL);
+    const $: CheerioAPI = load(indexPage);
+    const title = $('title').text()?.trim();
+    const author = $('div.logo a').attr('title')?.trim();
+    const items: DataItem[] = $('div.infor-list-item')
+        .toArray()
+        .map((listItem) => {
+            const item = $(listItem);
+            const title = item.find('h1').text();
+            const pubDate = `${item.find('span.year').text()}.${item.find('span.day').text()}`;
+            const href = item.find('a').prop('href');
+            const link = href ? (href.startsWith('http') ? href : new URL(href, BASE_URL).href) : BASE_URL;
+            const description = item.find('p').text();
+            return {
+                title,
+                link,
+                pubDate: timezone(parseDate(pubDate), +8),
+                description,
+            };
+        });
+    return {
+        item: items,
+        title,
+        link: BASE_URL,
+        description: `${author} - ${title}`,
+        author,
+        language: 'zh-CN',
+    };
+}
+
+export const route: Route = {
+    path: '/news',
+    categories: ['finance'],
+    example: '/huijin-inv/news',
+    parameters: {},
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['www.huijin-inv.cn/'],
+        },
+    ],
+    name: '资讯中心',
+    maintainers: ['la3rence'],
+    handler,
+    description: '中央汇金投资有限责任公司 - 资讯中心',
+};

package/lib/routes/hupu/index.ts

@@ -1,9 +1,10 @@
-import { Route } from '@/types';
+import { Route, Data, DataItem } from '@/types';
 import cache from '@/utils/cache';
 import got from '@/utils/got';
 import { load } from 'cheerio';
 import timezone from '@/utils/timezone';
-import { parseDate } from '@/utils/parse-date';
+import { parseDate, parseRelativeDate } from '@/utils/parse-date';
+import { type HupuApiResponse, type HomePostItem, type NewsDataItem, isHomePostItem } from './types';
 
 const categories = {
     nba: {
@@ -18,85 +19,180 @@ const categories = {
         title: '足球',
         data: 'news',
     },
-};
+    '': {
+        title: '首页',
+        data: 'res',
+    },
+} as const;
 
 export const route: Route = {
     path: ['/dept/:category?', '/:category?'],
+    name: '手机虎扑网',
+    url: 'm.hupu.com',
+    maintainers: ['nczitzk', 'hyoban'],
+    example: '/hupu/nba',
+    parameters: {
+        category: {
+            description: '分类，可选值：nba、cba、soccer，默认为空（首页）',
+            default: '',
+            options: Object.entries(categories).map(([key, value]) => ({
+                label: value.title,
+                value: key,
+            })),
+        },
+    },
+    description: `::: tip
+电竞分类参见 [游戏热帖](https://bbs.hupu.com/all-gg) 的对应路由 [\`/hupu/all/all-gg\`](https://rsshub.app/hupu/all/all-gg)。
+:::`,
+    categories: ['bbs'],
     radar: [
         {
             source: ['m.hupu.com/:category', 'm.hupu.com/'],
             target: '/:category',
         },
     ],
-    name: 'Unknown',
-    maintainers: ['nczitzk'],
-    handler,
-    description: `| NBA | CBA | 足球   |
-| --- | --- | ------ |
-| nba | cba | soccer |
-
-::: tip
-  电竞分类参见 [游戏热帖](https://bbs.hupu.com/all-gg) 的对应路由 [\`/hupu/all/all-gg\`](https://rsshub.app/hupu/all/all-gg)。
-:::`,
-};
+    handler: async (ctx): Promise<Data> => {
+        const c = ctx.req.param('category') || '';
+        if (!(c in categories)) {
+            throw new Error('Invalid category. Valid options are: ' + Object.keys(categories).filter(Boolean).join(', '));
+        }
+        const category = c as keyof typeof categories;
+
+        const rootUrl = 'https://m.hupu.com';
+        const currentUrl = `${rootUrl}/${category}`;
+
+        const response = await got({
+            method: 'get',
+            url: currentUrl,
+        });
+
+        const scriptMatch = response.data.match(/<script id="__NEXT_DATA__" type="application\/json">(.*?)<\/script>/);
+        if (!scriptMatch || !scriptMatch[1]) {
+            throw new Error(`Failed to find __NEXT_DATA__ script tag in page: ${currentUrl}`);
+        }
+
+        const fullJsonString = scriptMatch[1];
+        let fullData;
+
+        try {
+            fullData = JSON.parse(fullJsonString);
+        } catch (error) {
+            throw new Error(`Failed to parse full JSON data: ${error instanceof Error ? error.message : String(error)}`);
+        }
+
+        const data: HupuApiResponse = fullData;
+        const { pageProps } = data.props;
+
+        const dataKey = categories[category].data;
+        if (!(dataKey in pageProps)) {
+            throw new Error(`Expected '${dataKey}' property not found in pageProps for category: ${category || 'home'}`);
+        }
+
+        const rawDataArray: (HomePostItem | NewsDataItem)[] = (() => {
+            const data = (pageProps as any)[dataKey];
+            return Array.isArray(data) ? data : [];
+        })();
 
-async function handler(ctx) {
-    const category = ctx.req.param('category') ?? 'soccer';
-
-    const rootUrl = 'https://m.hupu.com';
-    const currentUrl = `${rootUrl}/${category}`;
-
-    const response = await got({
-        method: 'get',
-        url: currentUrl,
-    });
-
-    const data = JSON.parse(response.data.match(/"props":(.*),"page":"\//)[1]);
-
-    let items = data.pageProps[categories[category].data].map((item) => ({
-        title: item.title,
-        pubDate: timezone(parseDate(item.publishTime), +8),
-        link: item.link.replace(/bbs\.hupu.com/, 'm.hupu.com/bbs'),
-    }));
-
-    items = await Promise.all(
-        items
-            .filter((item) => !/subject/.test(item.link))
-            .map((item) =>
-                cache.tryGet(item.link, async () => {
-                    try {
-                        const detailResponse = await got({
-                            method: 'get',
-                            url: item.link,
-                        });
-
-                        const content = load(detailResponse.data);
-
-                        item.author = content('.bbs-user-info-name, .bbs-user-wrapper-content-name-span').text();
-                        item.category = content('.basketballTobbs_tag > a, .tag-player-team')
-                            .toArray()
-                            .map((c) => content(c).text());
-
-                        content('.basketballTobbs_tag').remove();
-                        content('.hupu-img').each(function () {
-                            content(this)
-                                .parent()
-                                .html(`<img src="${content(this).attr('data-origin')}">`);
-                        });
-
-                        item.description = content('#bbs-thread-content, .bbs-content-font').html();
-                    } catch {
-                        // no-empty
-                    }
-
-                    return item;
-                })
-            )
-    );
-
-    return {
-        title: `虎扑 - ${categories[category].title}`,
-        link: currentUrl,
-        item: items,
-    };
-}
+        let items: DataItem[] = rawDataArray.map((item) =>
+            isHomePostItem(item)
+                ? ({
+                      title: item.title,
+                      link: item.url.replace(/bbs\.hupu.com/, 'm.hupu.com/bbs'),
+                      guid: item.tid,
+                      category: item.label ? [item.label] : undefined,
+                  } satisfies DataItem)
+                : ({
+                      title: item.title,
+                      pubDate: timezone(parseDate(item.publishTime), +8),
+                      link: item.link.replace(/bbs\.hupu.com/, 'm.hupu.com/bbs'),
+                      guid: item.tid,
+                  } satisfies DataItem)
+        );
+
+        items = await Promise.all(
+            items
+                .filter((item) => item.link && !/subject/.test(item.link))
+                .map((item) =>
+                    cache.tryGet(item.link!, async () => {
+                        try {
+                            const detailResponse = await got({
+                                method: 'get',
+                                url: item.link,
+                            });
+
+                            const content = load(detailResponse.data);
+
+                            const author = content('.bbs-user-info-name, .bbs-user-wrapper-content-name-span').text();
+                            const pubDateString = content('.second-line-user-info span:not([class])').text();
+                            // Possible formats: 10:21, 45分钟前, 09-15 19:57
+                            const currentYear = new Date().getFullYear();
+                            const currentDate = new Date();
+                            const monthDayTimePattern = /^(\d{2})-(\d{2}) (\d{2}):(\d{2})$/;
+                            const timeOnlyPattern = /^(\d{1,2}):(\d{2})$/;
+                            let processedDateString = pubDateString;
+
+                            if (monthDayTimePattern.test(pubDateString)) {
+                                processedDateString = `${currentYear}-${pubDateString}`;
+                            } else if (timeOnlyPattern.test(pubDateString)) {
+                                const month = String(currentDate.getMonth() + 1).padStart(2, '0');
+                                const day = String(currentDate.getDate()).padStart(2, '0');
+                                processedDateString = `${currentYear}-${month}-${day} ${pubDateString}`;
+                            }
+
+                            const pubDate = [item.pubDate, timezone(parseDate(processedDateString), +8), timezone(parseRelativeDate(pubDateString), +8)].find((d) => d instanceof Date && !Number.isNaN(d.getTime()));
+                            const categories = content('.basketballTobbs_tag > a, .tag-player-team')
+                                .toArray()
+                                .map((c) => content(c).text())
+                                .filter(Boolean);
+
+                            content('.basketballTobbs_tag').remove();
+                            content('.hupu-img').each(function () {
+                                const imgSrc = content(this).attr('data-gif') || content(this).attr('data-origin') || content(this).attr('src');
+                                content(this).parent().html(`<img src="${imgSrc}">`);
+                            });
+
+                            // 分别获取内容元素
+                            const descriptionParts: string[] = [];
+
+                            // 获取主要内容
+                            const mainContent = content('#bbs-thread-content, .bbs-content-font').html();
+                            if (mainContent) {
+                                descriptionParts.push(mainContent);
+                            }
+
+                            // 单独处理视频部分
+                            const videoWrapper = content('.header-video-wrapper');
+                            if (videoWrapper.length > 0) {
+                                const videoElement = videoWrapper.find('video');
+                                if (videoElement.length > 0) {
+                                    const videoHtml = videoElement.prop('outerHTML');
+                                    if (videoHtml) {
+                                        descriptionParts.push(videoHtml);
+                                    }
+                                }
+                            }
+
+                            const description = descriptionParts.length > 0 ? descriptionParts.join('') : undefined;
+
+                            return {
+                                ...item,
+                                author,
+                                category: categories.length > 0 ? categories : item.category,
+                                description,
+                                pubDate,
+                            };
+                        } catch {
+                            // no-empty
+                            return item;
+                        }
+                    })
+                )
+        );
+
+        return {
+            title: `虎扑 - ${categories[category].title}`,
+            link: currentUrl,
+            item: items,
+        } as Data;
+    },
+};

package/lib/routes/hupu/types.ts

@@ -0,0 +1,163 @@
+// 虎扑 API 响应数据类型定义
+
+/**
+ * 徽章信息
+ */
+interface Badge {
+    name: string;
+    color: string;
+    v2DayColor: string;
+    v2NightColor: string;
+    relationUrl: string | null;
+    colorBg: string;
+    v2DayColorBg: string;
+    v2NightColorBg: string;
+    iconDay: string;
+    iconNight: string;
+}
+
+/**
+ * 首页帖子数据项（res 数组中的项）
+ */
+interface HomePostItem {
+    tid: string;
+    title: string;
+    url: string;
+    label: string;
+    lights: string;
+    replies: string;
+    type: string; // e.g., "3_pic", "1_pic", "video"
+    source: string[];
+    isNews: boolean;
+    badge: Badge[];
+}
+
+/**
+ * 新闻/分类页面数据项（newsData 数组中的项）
+ */
+interface NewsDataItem {
+    nid: string;
+    title: string;
+    img: string;
+    link: string;
+    badge: Badge[];
+    type: string; // e.g., "LINK", "IMG_TEXT"
+    lights: number;
+    replies: number;
+    publishTime: string;
+    tid: string;
+}
+
+/**
+ * 推荐比赛 Toast 信息
+ */
+interface RecommendMatchToast {
+    date: string;
+    matchCount: number;
+    title: string;
+    toastTitle: string | null;
+    foldTitle: string;
+    matchListLink: string | null;
+    matchCountText: string;
+}
+
+/**
+ * 推荐比赛信息
+ */
+interface RecommendMatch {
+    projectId: string | null;
+    version: string | null;
+    traceId: string | null;
+    matchList: any[];
+    events: any[];
+    toast: RecommendMatchToast;
+    success: boolean;
+    is_login: number;
+    is_jrs: boolean;
+    night: number;
+    client: string | null;
+}
+
+/**
+ * 首页页面属性
+ */
+interface HomePageProps {
+    res: HomePostItem[];
+    totalNum: number;
+    supportHydrate: boolean;
+}
+
+/**
+ * 篮球分类页面属性 (NBA/CBA)
+ */
+interface BasketballPageProps {
+    leagueType: string;
+    newsData: NewsDataItem[];
+    recommendMatch: RecommendMatch;
+    supportHydrate: boolean;
+}
+
+/**
+ * 足球分类页面属性
+ */
+interface SoccerPageProps {
+    schedules: any[];
+    news: NewsDataItem[];
+    supportHydrate: boolean;
+}
+
+/**
+ * 所有分类页面属性的联合类型
+ */
+type CategoryPageProps = BasketballPageProps | SoccerPageProps;
+
+/**
+ * API 响应的 props 结构
+ */
+interface ApiResponseProps {
+    pageProps: HomePageProps | CategoryPageProps;
+    __N_SSP: boolean;
+}
+
+/**
+ * 完整的 API 响应结构
+ */
+export interface HupuApiResponse {
+    props: ApiResponseProps;
+    page: string;
+    query: Record<string, any>;
+    buildId: string;
+    assetPrefix: string;
+    isFallback: boolean;
+    gssp: boolean;
+    scriptLoader: any[];
+}
+
+/**
+ * 用于区分不同页面类型的类型守卫
+ */
+export function isHomePageProps(pageProps: HomePageProps | CategoryPageProps): pageProps is HomePageProps {
+    return 'res' in pageProps;
+}
+
+export function isBasketballPageProps(pageProps: CategoryPageProps): pageProps is BasketballPageProps {
+    return 'newsData' in pageProps && 'leagueType' in pageProps;
+}
+
+export function isSoccerPageProps(pageProps: CategoryPageProps): pageProps is SoccerPageProps {
+    return 'news' in pageProps && 'schedules' in pageProps;
+}
+
+/**
+ * 用于区分不同数据项类型的类型守卫
+ */
+export function isHomePostItem(item: HomePostItem | NewsDataItem): item is HomePostItem {
+    return 'url' in item && 'isNews' in item;
+}
+
+export function isNewsDataItem(item: HomePostItem | NewsDataItem): item is NewsDataItem {
+    return 'nid' in item && 'link' in item && 'publishTime' in item;
+}
+
+// 导出具体类型以供外部使用
+export type { Badge, HomePostItem, NewsDataItem, RecommendMatch, HomePageProps, BasketballPageProps, SoccerPageProps, CategoryPageProps, ApiResponseProps };

package/lib/routes/hust/gs.ts

@@ -57,7 +57,7 @@ export const handler = async (ctx) => {
         )
     );
 
-    const title = $('meta[name="keywords"]').prop('content')?.replace(/,/g, ' - ') ?? $('title').text();
+    const title = $('meta[name="keywords"]').prop('content')?.replaceAll(',', ' - ') ?? $('title').text();
     const image = new URL($('div.logo img').prop('src'), rootUrl).href;
 
     return {

package/lib/routes/hust/mse.ts

@@ -75,7 +75,7 @@ export const handler = async (ctx) => {
         )
     );
 
-    const title = $('meta[name="keywords"]').prop('content')?.replace(/,/g, ' - ') ?? $('title').text();
+    const title = $('meta[name="keywords"]').prop('content')?.replaceAll(',', ' - ') ?? $('title').text();
     const image = new URL($('div.logo img').prop('src'), rootUrl).href;
 
     return {

package/lib/routes/huxiu/util.ts

@@ -257,7 +257,7 @@ const generateSignature = () => {
 
     const appSecret = 'hUzaABtNfDE-6UiyaYhfsmjW-8dnoyVc';
     const nonce = generateNonce();
-    const r = [appSecret, timestamp, nonce].sort();
+    const r = [appSecret, timestamp, nonce].toSorted();
     return {
         nonce,
         timestamp,
@@ -363,7 +363,7 @@ const processItems = async (items, limit, tryGet) => {
             return {
                 ...audioItem,
                 ...videoItem,
-                title: (item.title ?? item.summary ?? item.content)?.replace(/<\/?(?:em|br)?>/g, ''),
+                title: (item.title ?? item.summary ?? item.content)?.replaceAll(/<\/?(?:em|br)?>/g, ''),
                 link,
                 description: art(path.join(__dirname, 'templates/description.art'), {
                     image: {