rsshub 1.0.0-master.f9ec483 → 1.0.0-master.fa1dcc5

package/lib/routes/thegradient/index.ts

@@ -0,0 +1,80 @@
+import { Route, DataItem } from '@/types';
+import got from '@/utils/got';
+import { load } from 'cheerio';
+import { parseDate } from '@/utils/parse-date';
+import cache from '@/utils/cache';
+
+export const route: Route = {
+    path: '/posts',
+    categories: ['blog'],
+    example: '/thegradient/posts',
+    radar: [
+        {
+            source: ['thegradient.pub/'],
+        },
+    ],
+    url: 'thegradient.pub/',
+    name: 'Posts',
+    maintainers: ['liyaozhong'],
+    handler,
+    description: 'The Gradient Blog Posts',
+};
+
+async function handler() {
+    const rootUrl = 'https://thegradient.pub';
+    const currentUrl = rootUrl;
+
+    const response = await got(currentUrl);
+    const $ = load(response.data);
+
+    let items = $('.c-post-card-wrap')
+        .toArray()
+        .map((item) => {
+            const $item = $(item);
+            const $link = $item.find('.c-post-card__title-link').first();
+            const $meta = $item.find('.c-post-card__meta');
+
+            const href = $link.attr('href');
+            const title = $link.text().trim();
+            const dateStr = $meta.find('time').attr('datetime');
+
+            if (!href || !title || !dateStr) {
+                return null;
+            }
+
+            const link = new URL(href, rootUrl).href;
+            const pubDate = parseDate(dateStr);
+
+            return {
+                title,
+                link,
+                pubDate,
+            } as DataItem;
+        })
+        .filter((item): item is DataItem => item !== null);
+
+    items = (
+        await Promise.all(
+            items.map((item) =>
+                cache.tryGet(item.link as string, async () => {
+                    try {
+                        const detailResponse = await got(item.link);
+                        const $detail = load(detailResponse.data);
+
+                        item.description = $detail('.c-content').html() || '';
+
+                        return item as DataItem;
+                    } catch {
+                        return item;
+                    }
+                })
+            )
+        )
+    ).filter((item): item is DataItem => item !== null);
+
+    return {
+        title: 'The Gradient Blog',
+        link: rootUrl,
+        item: items,
+    };
+}

package/lib/routes/thegradient/namespace.ts

@@ -0,0 +1,7 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: 'The Gradient',
+    url: 'thegradient.pub',
+    lang: 'en',
+};

package/lib/routes/theinitium/app.ts

@@ -131,6 +131,10 @@ async function handler(ctx) {
                 });
                 const article = $('.pp-article__body');
                 article.find('.block-related-articles').remove();
+                article.find('figure.wp-block-pullquote').children().unwrap();
+                article.find('div.block-explanation-note').wrapInner('<blockquote></blockquote>');
+                article.find('div.wp-block-tcc-author-note').wrapInner('<em></em>').after('<hr>');
+                article.find('p.has-small-font-size').wrapInner('<small></small>');
                 item.description = art(path.join(__dirname, 'templates/description.art'), {
                     standfirst: $('.pp-header-group__standfirst').html(),
                     coverImage: $('.pp-media__image').attr('src'),

package/lib/routes/thepaper/channel.ts

@@ -5,7 +5,7 @@ import got from '@/utils/got';
 
 export const route: Route = {
     path: '/channel/:id',
-    categories: ['traditional-media'],
+    categories: ['new-media'],
     example: '/thepaper/channel/25950',
     parameters: { id: '频道 id，可在频道页 URL 中找到' },
     features: {

package/lib/routes/thepaper/factpaper.ts

@@ -11,7 +11,7 @@ import path from 'node:path';
 
 export const route: Route = {
     path: '/factpaper/:status?',
-    categories: ['traditional-media'],
+    categories: ['new-media'],
     example: '/thepaper/factpaper',
     parameters: { status: '状态 id，可选 `1` 即 有定论 或 `0` 即 核查中，默认为 `1`' },
     features: {

package/lib/routes/thepaper/featured.ts

@@ -5,7 +5,7 @@ import got from '@/utils/got';
 
 export const route: Route = {
     path: '/featured',
-    categories: ['traditional-media'],
+    categories: ['new-media'],
     example: '/thepaper/featured',
     parameters: {},
     features: {

package/lib/routes/thepaper/list.ts

@@ -5,7 +5,7 @@ import got from '@/utils/got';
 
 export const route: Route = {
     path: '/list/:id',
-    categories: ['traditional-media'],
+    categories: ['new-media'],
     example: '/thepaper/list/25457',
     parameters: { id: '栏目 id，可在栏目页 URL 中找到' },
     features: {

package/lib/routes/thepaper/user.ts

@@ -0,0 +1,174 @@
+import { Route } from '@/types';
+import * as cheerio from 'cheerio';
+import ofetch from '@/utils/ofetch';
+import cache from '@/utils/cache';
+import { parseDate } from '@/utils/parse-date';
+
+export const route: Route = {
+    path: '/user/:pphId',
+    categories: ['new-media'],
+    example: '/thepaper/user/4221423',
+    parameters: { pphId: '澎湃号 id，可在澎湃号页 URL 中找到' },
+    name: '澎湃号',
+    maintainers: ['TonyRL'],
+    handler,
+};
+
+interface AuthorInfo {
+    userId: number;
+    sname: string;
+    pic: string;
+    isAuth: string;
+    userType: string;
+    perDesc: string;
+    mobile: null;
+    isOrder: string;
+    sex: string;
+    area: string;
+    attentionNum: string;
+    fansNum: string;
+    praiseNum: null;
+    pph: boolean;
+    normalUser: boolean;
+    mobForwardType: number;
+    authInfo: string;
+    mail: string;
+    pphImpactNum: string;
+    wonderfulCommentCount: string;
+    location: string;
+    lastLoginDate: null;
+}
+
+interface PPHContentResponse {
+    code: number;
+    data: Data;
+    desc: string;
+    time: number;
+}
+
+interface Data {
+    hasNext: boolean;
+    startTime: number;
+    list: ListItem[];
+    nodeInfo: null;
+    tagInfo: null;
+    moreNodeInfo: null;
+    pageNum: null;
+    pageSize: null;
+    pages: null;
+    total: null;
+    prevPageNum: null;
+    nextPageNum: null;
+    excludeContIds: null;
+    contCont: null;
+    filterIds: null;
+    updateCount: null;
+}
+
+interface ListItem {
+    contId: string;
+    isOutForword: string;
+    isOutForward: string;
+    forwardType: string;
+    mobForwardType: number;
+    interactionNum: string;
+    praiseTimes: string;
+    pic: string;
+    imgCardMode: number;
+    smallPic: string;
+    sharePic: string;
+    pubTime: string;
+    pubTimeNew: string;
+    name: string;
+    closePraise: string;
+    authorInfo: AuthorInfo;
+    nodeId: number;
+    contType: number;
+    pubTimeLong: number;
+    specialNodeId: number;
+    cardMode: string;
+    dataObjId: number;
+    closeFrontComment: boolean;
+    isSupInteraction: boolean;
+    hideVideoFlag: boolean;
+    praiseStyle: number;
+    isSustainedFly: number;
+    softLocType: number;
+    closeComment: boolean;
+    voiceInfo: VoiceInfo;
+    softAdTypeStr: string;
+}
+
+interface VoiceInfo {
+    imgSrc: string;
+    isHaveVoice: string;
+}
+
+async function handler(ctx) {
+    const { pphId } = ctx.req.param();
+
+    const mobileBuildId = (await cache.tryGet('thepaper:m:buildId', async () => {
+        const response = await ofetch('https://m.thepaper.cn');
+        const $ = cheerio.load(response);
+        const nextData = JSON.parse($('script#__NEXT_DATA__').text());
+        return nextData.buildId;
+    })) as string;
+
+    const userInfo = (await cache.tryGet(`thepaper:user:${pphId}`, async () => {
+        const response = await ofetch(`https://api.thepaper.cn/userservice/user/homePage/${pphId}`, {
+            headers: {
+                'Client-Type': '2',
+                Origin: 'https://m.thepaper.cn',
+                Referer: 'https://m.thepaper.cn/',
+            },
+        });
+        return response.userInfo;
+    })) as AuthorInfo;
+
+    const response = await ofetch<PPHContentResponse>('https://api.thepaper.cn/contentapi/cont/pph/user', {
+        method: 'POST',
+        body: {
+            pageSize: 10,
+            pageNum: 1,
+            contType: 0,
+            excludeContIds: [],
+            pphId,
+            startTime: 0,
+        },
+    });
+
+    const list = response.data.list.map((item) => ({
+        title: item.name,
+        link: `https://www.thepaper.cn/newsDetail_forward_${item.contId}`,
+        pubDate: parseDate(item.pubTimeLong),
+        author: item.authorInfo.sname,
+        contId: item.contId,
+        image: item.pic,
+    }));
+
+    const items = await Promise.all(
+        list.map((item) =>
+            cache.tryGet(item.link, async () => {
+                const response = await ofetch(`https://m.thepaper.cn/_next/data/${mobileBuildId}/detail/${item.contId}.json`, {
+                    query: {
+                        id: item.contId,
+                    },
+                });
+
+                item.description = response.pageProps.detailData.contentDetail.content;
+                item.updated = parseDate(response.pageProps.detailData.contentDetail.updateTime);
+
+                return item;
+            })
+        )
+    );
+
+    return {
+        title: userInfo.sname,
+        description: userInfo.perDesc,
+        link: `https://www.thepaper.cn/user_${pphId}`,
+        item: items,
+        itunes_author: userInfo.sname,
+        image: userInfo.pic,
+    };
+}

package/lib/routes/tongji/sem/_utils.ts

@@ -0,0 +1,65 @@
+import got from '@/utils/got';
+import { load } from 'cheerio';
+import { parseDate } from '@/utils/parse-date';
+import { config } from '@/config';
+
+export async function getNotifByPage(url): Promise<{ title: string; link: string; pubDate: Date }[]> {
+    const pageUrl: string = url;
+
+    try {
+        const response = await got.get(pageUrl, {
+            headers: {
+                'User-Agent': config.ua,
+            },
+        });
+
+        const html = response.body;
+        const $ = load(html);
+
+        const notifListElements = $('#page-wrap > div.maim_pages > div > div.leftmain_page > div > ul > li');
+
+        return notifListElements.toArray().map((Element) => {
+            const aTagFirst = $(Element).find('a.bt');
+            const aTagSecond = $(Element).find('a.time');
+
+            const title = aTagFirst.attr('title');
+            const href = aTagFirst.attr('href');
+            const time = aTagSecond.text().trim();
+
+            return {
+                title,
+                link: href,
+                pubDate: parseDate(time, 'YYYY-MM-DD'),
+            };
+        });
+    } catch {
+        // console.error(error);
+    }
+    return [];
+}
+
+export async function getArticle(item) {
+    const articleUrl: string = item.link;
+
+    if (articleUrl.includes('sem.tongji.edu.cn/semch')) {
+        // console.log(articleUrl);
+
+        try {
+            const response = await got.get(articleUrl, {
+                headers: {
+                    'User-Agent': config.ua,
+                },
+            });
+
+            const html = response.body;
+            const $ = load(html);
+
+            const articleContentElement = $('#page-wrap > div.maim_pages > div > div.leftmain_page > div');
+            item.description = articleContentElement ? articleContentElement.html() : '';
+        } catch {
+            // console.error(error);
+        }
+    }
+
+    return item;
+}

package/lib/routes/tongji/sem/notice.ts

@@ -0,0 +1,58 @@
+// Warning: The author still knows nothing about javascript!
+import { Route } from '@/types';
+import { getNotifByPage, getArticle } from './_utils';
+import cache from '@/utils/cache';
+
+export const route: Route = {
+    path: '/sem/:type?',
+    categories: ['university'],
+    example: '/tongji/sem/notice',
+    parameters: { type: '通知类型，默认为 `notice`' },
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    name: '经济与管理学院通知',
+    maintainers: ['sitdownkevin'],
+    url: 'sem.tongji.edu.cn/semch',
+    handler,
+    description: `| 学院通知 | 招生通知 | 学术观点 | 新闻 | 活动 | 视点 | 教师与行政人员招聘 |
+| -------- | -------------- | ------------------ | ---- | ---------- | --------- | ------------------ |
+| notice   | enrollment     | academic-paper     | news | events     | focus     | collegerecruitment |
+`,
+};
+
+async function handler(ctx) {
+    const type = ctx.req.param('type') || 'notice';
+    const subType = new Set(['enrollment', 'academic-paper', 'news', 'events', 'focus', 'collegerecruitment']);
+    const subTypeName = {
+        notice: '学院通知',
+        enrollment: '招生通知',
+        'academic-paper': '学术观点',
+        news: '新闻',
+        events: '活动',
+        focus: '视点',
+        collegerecruitment: '教师与行政人员招聘',
+    };
+
+    const url = `https://sem.tongji.edu.cn/semch/category/frontpage/${subType.has(type) ? type : 'notice'}`;
+
+    const results: { title: string; link: string; pubDate: Date }[] = await getNotifByPage(url);
+
+    const resultsWithContent = await Promise.all(results.map((item) => cache.tryGet(item.link, () => getArticle(item))));
+
+    // feed the data to rss
+    return {
+        title: '同济大学经济与管理学院',
+        description: String(subType.has(type) ? subTypeName[type] : '学院通知'),
+        image: 'https://upload.wikimedia.org/wikipedia/zh/f/f8/Tongji_University_Emblem.svg',
+        icon: 'https://upload.wikimedia.org/wikipedia/zh/f/f8/Tongji_University_Emblem.svg',
+        logo: 'https://upload.wikimedia.org/wikipedia/zh/f/f8/Tongji_University_Emblem.svg',
+        link: 'https://sem.tongji.edu.cn/semch',
+        item: resultsWithContent,
+    };
+}