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

package/lib/routes/nankai/cc-notice.ts

@@ -0,0 +1,114 @@
+import { Route } from '@/types';
+import got from '@/utils/got';
+import { load } from 'cheerio';
+import { parseDate } from '@/utils/parse-date';
+import cache from '@/utils/cache';
+import timezone from '@/utils/timezone';
+
+export const route: Route = {
+    path: '/cc/:type?',
+    categories: ['university'],
+    example: '/nankai/cc/13291',
+    parameters: { type: '栏目编号（若为空则默认为"最新动态"）' },
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['cc.nankai.edu.cn', 'cc.nankai.edu.cn/:type/list.htm'],
+            target: '/cc/:type?',
+        },
+    ],
+    name: '计算机学院',
+    maintainers: ['vicguo0724'],
+    description: `| 最新动态 | 学院公告 | 学生工作通知 | 科研信息 | 本科生教学 | 党团园地 | 研究生招生 | 研究生教学 | 境外交流 |
+| -------- | -------- | ---------- | -------- | ---------- | -------- | ---------- | ---------- | -------- |
+| 13291    | 13292    | 13293      | 13294    | 13295      | 13296    | 13297      | 13298      | 13299    |`,
+    url: 'cc.nankai.edu.cn',
+    handler: async (ctx) => {
+        // 从 URL 参数中获取通知分类
+        const { type = '13291' } = ctx.req.param();
+        const baseUrl = 'https://cc.nankai.edu.cn';
+        const { data: response } = await got(`${baseUrl}/${type}/list.htm`);
+        const $ = load(response);
+
+        // 获取分类名称
+        const categoryName = $('.accordion .selected a').text().trim() || '最新动态';
+
+        // 解析新闻列表
+        const list = $('#wp_news_w49 table tr')
+            .slice(1) // 跳过表头
+            .toArray()
+            .map((tr) => {
+                const $tr = $(tr);
+                const cells = $tr.find('td');
+
+                if (cells.length < 5) {
+                    return null;
+                }
+
+                const title = cells.eq(1).text().trim();
+                const publisher = cells.eq(2).text().trim();
+                const dateStr = cells.eq(3).text().trim();
+
+                // 从onclick属性中提取链接
+                const onclick = $tr.attr('onclick');
+                let link = '';
+                if (onclick) {
+                    const match = onclick.match(/window\.location\.href='([^']+)';/);
+                    if (match) {
+                        link = match[1];
+                        if (!link.startsWith('http')) {
+                            link = `${baseUrl}${link}`;
+                        }
+                    }
+                }
+
+                return {
+                    title,
+                    link,
+                    pubDate: timezone(parseDate(dateStr), +8),
+                    author: publisher,
+                };
+            })
+            .filter((item) => item && item.link); // 过滤掉空项目和没有链接的项目
+
+        // 获取每篇文章的详细内容
+        const items = await Promise.all(
+            list.map((item) =>
+                item
+                    ? cache.tryGet(item.link, async () => {
+                          try {
+                              const { data: response } = await got(item.link);
+                              const $ = load(response);
+
+                              // 优化内容选择器逻辑，避免重复选择
+                              let description = $('.wp_articlecontent').html() || $('.body-news-detail').html();
+
+                              description = description || item.title;
+                              item.description = description;
+                          } catch {
+                              // 如果获取详细内容失败，返回基本信息
+                              item.description = item.title + ' (获取详细内容失败)';
+                          }
+                          return item;
+                      })
+                    : null
+            )
+        );
+
+        return {
+            // 源标题
+            title: `南开大学计算机学院-${categoryName}`,
+            // 源链接
+            link: `${baseUrl}/${type}/list.htm`,
+            // 源文章
+            item: items,
+        };
+    },
+};

package/lib/routes/nankai/jwc.ts

@@ -0,0 +1,131 @@
+import { Route } from '@/types';
+import got from '@/utils/got';
+import { load } from 'cheerio';
+import { parseDate } from '@/utils/parse-date';
+import cache from '@/utils/cache';
+import timezone from '@/utils/timezone';
+
+export const route: Route = {
+    path: '/jwc',
+    categories: ['university'],
+    example: '/nankai/jwc',
+    parameters: {},
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['jwc.nankai.edu.cn/tzgg/list.htm'],
+            target: '/jwc',
+        },
+    ],
+    name: '教务处通知公告',
+    maintainers: ['vicguo0724'],
+    description: '南开大学教务处通知公告',
+    url: 'jwc.nankai.edu.cn',
+    handler: async () => {
+        const baseUrl = 'https://jwc.nankai.edu.cn';
+        const { data: response } = await got(`${baseUrl}/tzgg/list.htm`);
+        const $ = load(response);
+
+        // 解析列表页面中的所有通知项
+        const list = $('.page-con-list-news .item')
+            .toArray()
+            .map((item) => {
+                const $item = $(item);
+                const $link = $item.find('.t a');
+                const $dateDay = $item.find('.d .d-d');
+                const $dateMonth = $item.find('.d .d-m');
+
+                // 构建完整的日期
+                const day = $dateDay.text().trim(); // 格式：04
+                const monthYear = $dateMonth.text().trim(); // 格式：2025/06
+                const fullDate = `${monthYear}/${day}`; // 2025/06/04
+
+                let linkStr = $link.attr('href');
+                // 处理相对链接
+                if (linkStr && !linkStr.startsWith('http')) {
+                    linkStr = `${baseUrl}${linkStr}`;
+                }
+
+                return {
+                    title: $link.text().trim(),
+                    link: linkStr,
+                    pubDate: timezone(parseDate(fullDate, 'YYYY/MM/DD'), +8),
+                };
+            })
+            .filter((item) => item.link); // 过滤掉没有链接的项目
+
+        // 获取每个通知的详细内容
+        const items = await Promise.all(
+            list.map((item) =>
+                cache.tryGet(item.link, async () => {
+                    try {
+                        const { data: response } = await got(item.link);
+                        const $ = load(response);
+
+                        // 获取发布时间（更精确的时间）
+                        const publishTimeText = $('.page-news-souse').text();
+                        const timeMatch = publishTimeText.match(/发布时间：(\d{4}-\d{2}-\d{2})/);
+                        if (timeMatch) {
+                            item.pubDate = timezone(parseDate(timeMatch[1]), +8);
+                        }
+
+                        // 获取文章内容
+                        const content = $('.page-news-con .wp_articlecontent');
+                        if (content.length > 0) {
+                            // 处理PDF链接，转换为绝对链接
+                            content.find('a').each((i, el) => {
+                                const $el = $(el);
+                                const href = $el.attr('href');
+                                if (href && !href.startsWith('http')) {
+                                    if (href.startsWith('/')) {
+                                        $el.attr('href', `${baseUrl}${href}`);
+                                    } else {
+                                        $el.attr('href', `${baseUrl}/${href}`);
+                                    }
+                                }
+                            });
+
+                            // 处理PDF播放器div，提取PDF链接
+                            content.find('.wp_pdf_player').each((i, el) => {
+                                const $el = $(el);
+
+                                const pdfSrc = $el.attr('pdfsrc');
+                                const sudyfileAttr = ($(el).attr('sudyfile-attr') || '{}').replaceAll("'", '"');
+                                const sudyfileAttrJson = JSON.parse(sudyfileAttr);
+                                const fileName = sudyfileAttrJson.title || '未命名文件.pdf';
+                                if (pdfSrc) {
+                                    let pdfUrl = pdfSrc;
+                                    if (!pdfUrl.startsWith('http')) {
+                                        pdfUrl = `${baseUrl}${pdfUrl}`;
+                                    }
+                                    // 替换PDF播放器为下载链接
+                                    $el.replaceWith(`<p><a href="${pdfUrl}" target="_blank">${fileName}</a></p>`);
+                                }
+                            });
+
+                            item.description = content.html();
+                        } else {
+                            item.description = '无法获取内容详情';
+                        }
+                    } catch {
+                        item.description = '获取内容失败';
+                    }
+                    return item;
+                })
+            )
+        );
+
+        return {
+            title: '南开大学教务处-通知公告',
+            link: `${baseUrl}/tzgg/list.htm`,
+            item: items,
+        };
+    },
+};

package/lib/routes/nankai/namespace.ts

@@ -0,0 +1,7 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: '南开大学',
+    url: 'yzb.nankai.edu.cn',
+    lang: 'zh-CN',
+};

package/lib/routes/nankai/notice.ts

@@ -0,0 +1,84 @@
+import { Route } from '@/types';
+import got from '@/utils/got';
+import { load } from 'cheerio';
+import { parseDate } from '@/utils/parse-date';
+import cache from '@/utils/cache';
+import timezone from '@/utils/timezone';
+
+export const route: Route = {
+    path: '/notice',
+    categories: ['university'],
+    example: '/nankai/notice',
+    parameters: {},
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: true,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['www.nankai.edu.cn', 'www.nankai.edu.cn/157/list.htm'],
+        },
+    ],
+    name: '通知公告',
+    maintainers: ['vicguo0724'],
+    handler: async () => {
+        const baseUrl = 'https://www.nankai.edu.cn';
+        const listUrl = `${baseUrl}/157/list.htm`;
+        const { data: response } = await got(listUrl);
+        const $ = load(response);
+
+        const list = $('ul.newslist li')
+            .toArray()
+            .map((item) => {
+                const $item = $(item);
+                const $time = $item.find('.time');
+                const day = $time.find('.time-d').text().trim();
+                const monthYear = $time.contents().last().text().trim();
+                const pubDate = timezone(parseDate(`${monthYear}-${day}`, 'YYYY-MM-DD'), +8);
+
+                const $link = $item.find('.tit a');
+                let href = $link.attr('href') || '';
+                href = href.startsWith('http') ? href : new URL(href, baseUrl).href;
+
+                return {
+                    title: $link.text().trim(),
+                    link: href,
+                    pubDate,
+                };
+            });
+
+        const items = await Promise.all(
+            list.map((item) =>
+                cache.tryGet(item.link, async () => {
+                    try {
+                        // 判断link如果是https://xb.nankai.edu.cn/的则为校内访问的
+                        if (item.link.includes('xb.nankai.edu.cn')) {
+                            item.description = '该通知可能需要校内访问权限';
+                        } else {
+                            const { data: detailResponse } = await got(item.link);
+                            const $detail = load(detailResponse);
+
+                            // 提取正文内容
+                            const content = $detail('.wp_articlecontent').html() || '';
+                            item.description = content;
+                        }
+                    } catch {
+                        // 如果提取正文内容失败，则返回默认内容
+                        item.description = '正文内容获取失败';
+                    }
+                    return item;
+                })
+            )
+        );
+
+        return {
+            title: '南开大学通知公告',
+            link: listUrl,
+            item: items,
+        };
+    },
+};

package/lib/routes/nankai/yzb.ts

@@ -0,0 +1,97 @@
+import { Route } from '@/types';
+import got from '@/utils/got';
+import { load } from 'cheerio';
+import { parseDate } from '@/utils/parse-date';
+import cache from '@/utils/cache';
+import timezone from '@/utils/timezone';
+
+export const route: Route = {
+    path: '/yzb/:type?',
+    categories: ['university'],
+    example: '/nankai/yzb/5509',
+    parameters: { type: '栏目名（若为空则默认为“硕士招生”）' },
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['yzb.nankai.edu.cn', 'yzb.nankai.edu.cn/:type/list.htm'],
+            target: '/yzb/:type?',
+        },
+    ],
+    name: '研究生招生网',
+    maintainers: ['sddzhyc'],
+    description: `| 硕士招生 | 博士招生 | 港澳台研究生最新信息 |
+| -------- | -------- | -------- |
+| 5509     | 2552    | 2562   |`,
+    url: 'yzb.nankai.edu.cn',
+    handler: async (ctx) => {
+        // 从 URL 参数中获取通知分类
+        const { type = '5509' } = ctx.req.param();
+        const baseUrl = 'https://yzb.nankai.edu.cn';
+        const { data: response } = await got(`${baseUrl}/${type}/list.htm`);
+        const $ = load(response);
+
+        // 先转换为数组，再使用原生JavaScript的map
+        const dateList = $('#wp_news_w9')
+            .find('span.col_news_date')
+            .toArray()
+            .map((item) => $(item).text());
+
+        const list = $('#wp_news_w9')
+            .find('a[title]')
+            .toArray()
+            .map((element, index) => {
+                const $a = $(element);
+                let linkStr = $a.attr('href') || '';
+
+                // 使用三元表达式处理链接
+                linkStr = linkStr.startsWith('http://') ? linkStr.replace('http://', 'https://') : `${baseUrl}${linkStr}`;
+
+                return {
+                    title: $a.text(),
+                    link: linkStr,
+                    pubDate: timezone(parseDate(dateList[index]), +8),
+                };
+            });
+
+        const items = await Promise.all(
+            list.map((item) =>
+                cache.tryGet(item.link.toString(), async () => {
+                    const { data: response } = await got(item.link);
+                    const $ = load(response);
+                    item.description = $('.read').first().html();
+
+                    // 提取PDF链接，先转换为数组再使用map
+                    const pdfLinks = $('div[pdfsrc$=".pdf"]')
+                        .toArray()
+                        .map((el) => {
+                            const pdfUrl = $(el).attr('pdfsrc') || '';
+                            const sudyfileAttr = ($(el).attr('sudyfile-attr') || '{}').replaceAll("'", '"');
+                            const sudyfileAttrJson = JSON.parse(sudyfileAttr);
+                            const fileName = sudyfileAttrJson.title || '未命名文件.pdf';
+                            return `<a href="${new URL(pdfUrl, baseUrl).href}">${fileName}</a>`;
+                        })
+                        .join('<br>');
+
+                    if (pdfLinks) {
+                        item.description += `<h4>相关附件：</h4>${pdfLinks}`;
+                    }
+
+                    return item;
+                })
+            )
+        );
+
+        return {
+            title: `南开大学研究生招生网-${$('.column-title').text()}`,
+            link: `${baseUrl}/${type}/list.htm`,
+            item: items as any[],
+        };
+    },
+};

package/lib/routes/neu/yz.ts

@@ -0,0 +1,172 @@
+import { DataItem, Route } from '@/types';
+import cache from '@/utils/cache';
+import got from '@/utils/got';
+import { load } from 'cheerio';
+import { parseDate } from '@/utils/parse-date';
+import timezone from '@/utils/timezone';
+
+const BASE_URL = 'http://yz.neu.edu.cn';
+const DOWNLOAD_ID = 5792;
+const MASTER1_ID = 5932;
+const MASTER2_ID = 5933;
+const PhD1_ID = 5945;
+const PhD2_ID = 5946;
+const DOWNLOAD_AUTHOR = '研招办';
+
+const idMp = {
+    download: DOWNLOAD_ID,
+    master1: MASTER1_ID,
+    master2: MASTER2_ID,
+    phd1: PhD1_ID,
+    phd2: PhD2_ID,
+};
+
+const parsePage = async (items, type) => {
+    const results = await Promise.all(
+        items.map(async (item) => {
+            const $ = load(item);
+            const aTag = $('span.Article_Title > a');
+            const title = aTag.attr('title') ?? '';
+            const url = BASE_URL + aTag.attr('href');
+            const resultItem: DataItem = {
+                title,
+                link: url,
+                description: '',
+                pubDate: (() => {
+                    const sTag = $('span.Article_PublishDate');
+                    const pubDate = sTag.text() ? timezone(parseDate(sTag.text()), +8) : undefined;
+                    return pubDate;
+                })(),
+                author: type === DOWNLOAD_ID ? DOWNLOAD_AUTHOR : '',
+            };
+            if (type === DOWNLOAD_ID && /\.(pdf|docx?|xlsx?|zip|rar|7z)$/i.test(url)) {
+                resultItem.description = `
+                        <p>${title}</p><br/>
+                        <a href="${url}">点击进入下载地址传送门～</a>
+                    `;
+                return resultItem;
+            } else {
+                return await cache.tryGet(url, async () => {
+                    const result = await got(url);
+                    const $ = load(result.data);
+                    const description = cleanEntryContent($);
+                    resultItem.description = description;
+                    if (type !== DOWNLOAD_ID) {
+                        const dateText = $('.arti_update')
+                            .text()
+                            .match(/(\d{4}-\d{2}-\d{2})/);
+                        const date = dateText ? dateText[1] : '';
+                        const authorText = $('.arti_publisher')
+                            .text()
+                            .match(/[:：]?\s*(.+)/);
+                        const author = authorText ? authorText[1].trim() : '';
+                        resultItem.pubDate = timezone(parseDate(date), +8);
+                        resultItem.author = author;
+                    }
+                    return resultItem;
+                });
+            }
+        })
+    );
+    return results;
+};
+
+const handler = async (ctx) => {
+    let type = ctx.req.param('type');
+    if (idMp[type]) {
+        type = idMp[type];
+    }
+    const newsUrl = `${BASE_URL}/${type}/list.htm`;
+    const response = await got(newsUrl);
+    const data = response.data;
+    const $ = load(data);
+    const title = $('title').text();
+    const items = $('div.col_news_list ul.wp_article_list li').toArray();
+    const results = await parsePage(items, type);
+    return {
+        title: `${title}-东北大学研究生招生信息网`,
+        description: title,
+        link: BASE_URL,
+        item: results,
+    };
+};
+
+const cleanEntryContent = ($) => {
+    const entry = $('.entry');
+    if (!entry || entry.length === 0) {
+        return '';
+    }
+    entry.find('span').removeAttr('style').removeAttr('class');
+    entry.find('div').each((_, el) => {
+        const html = $(el).html();
+        $(el).replaceWith(html);
+    });
+    entry.find('p').removeAttr('style').removeAttr('class');
+    entry.find('a').removeAttr('style').removeAttr('class');
+    entry.find('td').removeAttr('bgcolor');
+    entry.find('img').each((_, el) => {
+        const src = $(el).attr('src');
+        const alt = $(el).attr('alt') || '';
+        $(el).replaceWith(`<img src="${src}" alt="${alt}" />`);
+    });
+    entry.find('.wp_pdf_player').each((_, el) => {
+        const url = BASE_URL + $(el).attr('pdfsrc');
+        $(el).replaceWith(
+            `
+            <p>点击进入文件传送门～：<a href="${url}">查看文件</a></p>
+            `
+        );
+    });
+    entry.find('.wp_video_player').each((_, el) => {
+        const div = $(el);
+        const src = div.attr('sudy-wp-src');
+        if (src) {
+            const videoUrl = BASE_URL + src;
+            const widthMatch = div.attr('style')?.match(/width:\s*(\d+)px/);
+            const width = widthMatch && widthMatch[1] ? widthMatch[1] : '600';
+            const heightMatch = div.attr('style')?.match(/height:\s*(\d+)px/);
+            const height = heightMatch && heightMatch[1] ? heightMatch[1] : '400';
+            const videoTag = `
+                <video controls width="${width}" height="${height}" style="max-width: 100%;margin-left: auto;margin-right: auto;">
+                    <source src="${videoUrl}" type="video/mp4">
+                    您的浏览器不支持 video 标签。
+                </video>
+            `;
+            div.replaceWith(videoTag);
+        }
+    });
+    return entry.html();
+};
+
+export const route: Route = {
+    path: '/yz/:type',
+    categories: ['university'],
+    example: '/neu/yz/master1',
+    parameters: { type: '分类id,见下表' },
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['yz.neu.edu.cn/:type/list.htm'],
+            target: '/yz/:type',
+        },
+    ],
+    name: '研究生招生信息网',
+    url: 'yz.neu.edu.cn',
+    maintainers: ['paintstar'],
+    handler,
+    description: `
+| 分类名                     | 分类id      |
+| ------------------------- | ---------- |
+| 硕士公告                   | master1     |
+| 硕士简章                   | master2     |
+| 博士公告                   | phd1        |
+| 博士简章                   | phd2        |
+| 下载中心                   | download    |`,
+};

package/lib/routes/newslaundry/explainer.ts

@@ -0,0 +1,30 @@
+import { Data, Route, ViewType } from '@/types';
+import { fetchCollection, rootUrl } from './utils';
+
+export const route: Route = {
+    path: '/explainer',
+    view: ViewType.Articles,
+    categories: ['new-media'],
+    example: '/newslaundry/explainer',
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['newslaundry.com/explainer'],
+            target: '/explainer',
+        },
+    ],
+    name: 'Explainer',
+    maintainers: ['Rjnishant530'],
+    handler,
+};
+
+async function handler(): Promise<Data> {
+    return await fetchCollection('explainer', `${rootUrl}/explainer`);
+}

package/lib/routes/newslaundry/namespace.ts

@@ -0,0 +1,8 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: 'Newslaundry',
+    url: 'newslaundry.com',
+    categories: ['new-media'],
+    lang: 'en',
+};

package/lib/routes/newslaundry/nl-cheatsheet.ts

@@ -0,0 +1,30 @@
+import { Data, Route, ViewType } from '@/types';
+import { fetchCollection, rootUrl } from './utils';
+
+export const route: Route = {
+    path: '/nl-cheatsheet',
+    view: ViewType.Articles,
+    categories: ['new-media'],
+    example: '/newslaundry/nl-cheatsheet',
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['newslaundry.com/collection/nl-cheatsheet'],
+            target: '/nl-cheatsheet',
+        },
+    ],
+    name: 'Explains',
+    maintainers: ['Rjnishant530'],
+    handler,
+};
+
+async function handler(): Promise<Data> {
+    return await fetchCollection('nl-cheatsheet', `${rootUrl}/collection/nl-cheatsheet`);
+}