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

package/lib/routes/hit/jwc.ts

@@ -1,86 +0,0 @@
-import { Route } from '@/types';
-import cache from '@/utils/cache';
-import got from '@/utils/got';
-import { load } from 'cheerio';
-import { parseDate } from '@/utils/parse-date';
-
-const baseUrl = 'https://jwc.hit.edu.cn';
-const type = (filename) => filename.split('.').pop();
-
-export const route: Route = {
-    path: '/jwc',
-    categories: ['university'],
-    example: '/hit/jwc',
-    parameters: {},
-    features: {
-        requireConfig: false,
-        requirePuppeteer: false,
-        antiCrawler: false,
-        supportBT: false,
-        supportPodcast: false,
-        supportScihub: false,
-    },
-    radar: [
-        {
-            source: ['jwc.hit.edu.cn/*'],
-        },
-    ],
-    name: '教务处通知公告',
-    maintainers: ['lty96117'],
-    handler,
-    url: 'jwc.hit.edu.cn/*',
-};
-
-async function handler() {
-    const response = await got(`${baseUrl}/2591/list.htm`);
-
-    const { data } = response;
-    const $ = load(data);
-    const links = $('.news_list li')
-        .toArray()
-        .map((el) => {
-            el = $(el);
-            return {
-                pubDate: parseDate(el.find('span.fbll').children().first().text().replaceAll('[', '')),
-                link: new URL(el.find('a').attr('href'), baseUrl).href,
-                title: el.find('a').attr('title'),
-            };
-        });
-
-    const items = await Promise.all(
-        links.map((item) =>
-            cache.tryGet(item.link, async () => {
-                if (type(item.link) === 'htm') {
-                    try {
-                        const { data } = await got(item.link);
-                        const $ = load(data);
-
-                        const author = $('p.arti_metas>span:nth-child(3)').text().trim();
-                        const description =
-                            $('div.wp_articlecontent').html() &&
-                            $('div.wp_articlecontent')
-                                .html()
-                                .replaceAll('src="/', `src="${new URL('.', baseUrl).href}`)
-                                .replaceAll('href="/', `href="${new URL('.', baseUrl).href}`)
-                                .trim();
-
-                        item.author = author;
-                        item.description = description;
-                    } catch {
-                        // intranet
-                    }
-                } else {
-                    // file to download
-                    item.description = '此链接为文件，点击以下载';
-                }
-                return item;
-            })
-        )
-    );
-
-    return {
-        title: '哈尔滨工业大学教务处通知公告',
-        link: `${baseUrl}/2591/list.htm`,
-        item: items,
-    };
-}

package/lib/routes/picuki/utils.ts

@@ -1,19 +0,0 @@
-const puppeteerGet = async (url, browser) => {
-    const page = await browser.newPage();
-    const expectResourceTypes = new Set(['document', 'script', 'xhr', 'fetch']);
-    await page.setRequestInterception(true);
-    page.on('request', (request) => {
-        expectResourceTypes.has(request.resourceType()) ? request.continue() : request.abort();
-    });
-    await page.goto(url, {
-        waitUntil: 'domcontentloaded',
-    });
-
-    await page.waitForSelector('.content');
-
-    const html = await page.content();
-    await page.close();
-    return html;
-};
-
-export { puppeteerGet };

package/lib/routes/qweather/templates/3days.art

@@ -1,22 +0,0 @@
-<text>白天：{{item.textDay}}——夜间：{{item.textNight}}</text>
-<br>
-<text>气温：{{item.tempMin}}℃~{{item.tempMax}}℃</text>
-<br>
-<text>相对湿度：{{item.humidity}}%</text>
-<br>
-<text>空气质量指数：{{item.aqi}} ({{item.aqiCategory}})</text>
-<br>
-<text>大气压强：{{item.pressure}}百帕</text>
-<br>
-<text>紫外线强度：{{item.uvIndex}}</text>
-<br>
-<text>白天风向：{{item.windDirDay}}  风力：{{item.windScaleDay}}级 风速：{{item.windSpeedDay}}公里/小时</text>
-<br>
-<text>夜间风向：{{item.windDirNight}}  风力：{{item.windScaleNight}}级 风速：{{item.windSpeedNight}}公里/小时</text>
-<br>
-<text>能见度：{{item.vis}}公里</text>
-<br>
-<text>日出：{{item.sunrise}}    日落： {{item.sunset}}</text>
-<br>
-<text>月相：{{item.moonPhase}}  月出：{{item.sunrise}}  月落：{{item.moonset}}
-<br>

package/lib/routes/qweather/templates/now.art

@@ -1,16 +0,0 @@
-<text>天气：{{item.text}}</text>
-<br>
-<text>气温：{{item.temp}}℃</text>
-<br>
-<text>体感温度：{{item.feelsLike}}℃</text>
-<br>
-<text>风向：{{item.windDir}}</text>
-<br>
-<text>风力：{{item.windScale}}级  风速：{{item.windSpeed}}km/h</text>
-<br>
-<text>湿度：{{item.humidity}}%  大气压强：{{item.pressure}}hPa</text>
-<br>
-<text>本小时降水量：{{item.precip}}mm</text>
-<br>
-<text>能见度：{{item.vis}}km</text>
-<br>

package/lib/routes/xinhuanet/app.ts

@@ -1,109 +0,0 @@
-import { Route } 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';
-
-export const handler = async (ctx) => {
-    const limit = ctx.req.query('limit') ? Number.parseInt(ctx.req.query('limit'), 10) : 7;
-
-    const rootUrl = 'https://app.xinhuanet.com';
-    const currentUrl = new URL('news/index.html', rootUrl).href;
-
-    const { data: response } = await got(currentUrl);
-
-    const $ = load(response);
-
-    const language = $('html').prop('lang');
-
-    let items = $('a.article-item')
-        .slice(0, limit)
-        .toArray()
-        .map((item) => {
-            item = $(item);
-
-            const title = item.find('div.article-title').text();
-            const guid = `xinhuanet-${item.prop('data-uuid')}`;
-
-            return {
-                title,
-                link: item.prop('href'),
-                guid,
-                id: guid,
-                language,
-            };
-        });
-
-    items = await Promise.all(
-        items.map((item) =>
-            cache.tryGet(item.link, async () => {
-                const { data: detailResponse } = await got(item.link);
-
-                const $$ = load(detailResponse);
-
-                const title = $$('div.article_title').text();
-                const description = $$('#detail-content').html();
-
-                item.title = title;
-                item.description = description;
-
-                const authorEl = $$('div.article_auth div').first();
-                item.author = authorEl.text();
-
-                authorEl.remove();
-
-                item.pubDate = timezone(parseDate($$('div.article_auth div').first().text()), +8);
-                item.content = {
-                    html: description,
-                    text: $$('#detail-content').text(),
-                };
-                item.language = language;
-
-                return item;
-            })
-        )
-    );
-
-    const image = $('meta[itemprop="image"]').prop('content');
-
-    return {
-        title: $('title').text(),
-        description: $('meta[name="description"]').prop('content'),
-        link: currentUrl,
-        item: items,
-        allowEmpty: true,
-        image,
-        author: $('meta[itemprop="name"]').prop('content'),
-        language,
-    };
-};
-
-export const route: Route = {
-    path: '/app',
-    name: '客户端',
-    url: 'app.xinhuanet.com',
-    maintainers: ['nczitzk'],
-    handler,
-    example: '/xinhuanet/app',
-    parameters: undefined,
-    description: '',
-    categories: ['traditional-media'],
-
-    features: {
-        requireConfig: false,
-        requirePuppeteer: false,
-        antiCrawler: false,
-        supportRadar: true,
-        supportBT: false,
-        supportPodcast: false,
-        supportScihub: false,
-    },
-    radar: [
-        {
-            source: ['app.xinhuanet.com'],
-            target: '/app',
-        },
-    ],
-};

package/lib/routes/zaobao/templates/zaobao.art

@@ -1,17 +0,0 @@
-{{ each imageDataArray imageData }}
-    {{ if imageData.type == 'normalHTML' }}
-        {{@ imageData.html }}
-    {{ /if }}
-    {{ if imageData.type == 'data' }}
-        <figure>
-            <img src="{{ imageData.src }}" />
-            <figcaption>
-                {{ imageData.title }}
-            </figcaption>
-        </figure>
-    {{ /if }}
-{{ /each }}
-
-{{ if articleBody }}
-    {{@ articleBody }}
-{{ /if }}

package/lib/routes/zaobao/util.ts

@@ -1,198 +0,0 @@
-import cache from '@/utils/cache';
-import got from '@/utils/got';
-import { load } from 'cheerio';
-import { parseRelativeDate } from '@/utils/parse-date';
-import timezone from '@/utils/timezone';
-import { art } from '@/utils/render';
-import path from 'node:path';
-import { base32 } from 'rfc4648';
-
-const baseUrl = 'https://www.zaobao.com';
-const got_ins = got.extend({
-    headers: {
-        Referer: baseUrl,
-    },
-});
-
-/**
- * 通用解析页面类似 https://www.zaobao.com/realtime/china 的网站
- *
- * @param {*} ctx RSSHub 的 ctx 参数，用来设置缓存
- * @param {string} sectionUrl 形如 /realtime/china 的字符串
- * @returns 新闻标题以及新闻列表
- */
-const parseList = async (
-    ctx,
-    sectionUrl: string
-): Promise<{
-    title: string;
-    resultList: {
-        title: string;
-        description: string;
-        pubDate: string;
-        link: string;
-    }[];
-}> => {
-    const response = await got_ins.get(baseUrl + sectionUrl);
-    const $ = load(response.data);
-    let data = $('.card-listing .card .content-header a');
-    if (data.length === 0) {
-        // for HK version
-        data = $('[data-testid="article-list"] article a.article-link');
-    }
-
-    const title = $('meta[property="og:title"]').attr('content');
-
-    const resultList = await Promise.all(
-        data.toArray().map((item) => {
-            const $item = $(item);
-            // addBack: for HK version
-            let link = $item.attr('href');
-
-            if (link[0] !== '/') {
-                // https://www.zaobao.com/interactive-graphics
-                const title = $item.find('a').text();
-                let dateNodes = $item.find('.meta-published-date');
-                if (dateNodes.length === 0) {
-                    dateNodes = $item.find('.datestamp');
-                }
-                let dateString;
-                let pubDate;
-                if (dateNodes.length !== 0) {
-                    dateString = dateNodes.text().trim();
-                    const dateParts = dateString.split('/');
-                    dateParts.reverse();
-                    dateString = dateParts.join('-');
-                    pubDate = parseRelativeDate(dateString);
-                    // if conversion was no success, pubDate === dateString
-                    // zaobao seems always UTC+8
-                    pubDate = pubDate === dateString ? undefined : timezone(pubDate, +8);
-                }
-
-                return {
-                    title,
-                    description: title,
-                    pubDate,
-                    link,
-                };
-            }
-
-            link = baseUrl + link;
-
-            return cache.tryGet(link, async () => {
-                const article = await got_ins.get(link);
-                const $1 = load(article.data);
-
-                let title, time;
-                if ($1('#seo-article-page').text() === '') {
-                    // HK
-                    title = $1('h1.article-title').text();
-                    const jsonText = $1("head script[type='application/ld+json']")
-                        .eq(0)
-                        .text()
-                        .replaceAll(/[\u0000-\u001F\u007F-\u009F]/g, '');
-                    time = new Date(JSON.parse(jsonText)?.datePublished);
-                } else {
-                    // SG
-                    const jsonText = $1('#seo-article-page')
-                        .text()
-                        .replaceAll(/[\u0000-\u001F\u007F-\u009F]/g, '');
-                    const json = JSON.parse(jsonText);
-                    title = json['@graph'][0]?.headline;
-                    time = new Date(json['@graph'][0]?.datePublished);
-                }
-
-                $1('.overlay-microtransaction').remove();
-                $1('#video-freemium-player').remove();
-                $1('script').remove();
-                $1('.bff-google-ad').remove();
-                $1('.bff-recommend-article').remove();
-
-                let articleBodyNode = $1('.articleBody');
-                if (articleBodyNode.length === 0) {
-                    // for HK version
-                    $1('.read-on-app-cover').remove();
-                    $1('astro-island').remove();
-                    $1('.further-reading').remove();
-                    articleBodyNode = $1('.article-body');
-                }
-
-                const articleBody = articleBodyNode.html();
-
-                const imageDataArray = processImageData($1);
-
-                return {
-                    // <- for HK version  -> for SG version
-                    title,
-                    description: art(path.join(__dirname, 'templates/zaobao.art'), {
-                        articleBody,
-                        imageDataArray,
-                    }),
-                    pubDate: time,
-                    link,
-                };
-            });
-        })
-    );
-    return {
-        title,
-        resultList,
-    };
-};
-
-const orderContent = (parent) => {
-    for (const [i, e] of parent
-        .children()
-        .toArray()
-        .sort((a, b) => {
-            const index = Buffer.from(base32.parse('GM======')).toString(); // substring(3)
-            a = Buffer.from(base32.parse(parent.find(a).data('s').slice(index))).toString();
-            b = Buffer.from(base32.parse(parent.find(b).data('s').slice(index))).toString();
-            return a - b;
-        })
-        .entries()) {
-        parent.find(e).attr('s', i);
-        parent.append(e);
-    }
-};
-
-interface ImageData {
-    type: string;
-    html: string;
-    src?: string;
-    title?: string;
-}
-
-const processImageData = ($1) => {
-    const imageDataArray: ImageData[] = [];
-
-    const imageSelectors = [
-        '.inline-figure-img', // for SG version
-        '.body-content .loadme picture img', // Unused?
-        '.inline-figure-gallery', // for SG version
-        '#carousel-article', // for HK version, HK version of multi images use same selector as single image, so g is needed for all pages
-    ];
-
-    for (const selector of imageSelectors) {
-        if ($1(selector).length) {
-            let html = $1(selector === '#carousel-article' ? '#carousel-article .carousel-inner' : selector).html();
-
-            if (html) {
-                html = html.replaceAll(/\/\/.*\.com\/s3fs-public/g, '//static.zaobao.com/s3fs-public').replaceAll('s3/files', 's3fs-public');
-
-                imageDataArray.push({
-                    type: selector === '.body-content .loadme picture img' ? 'data' : 'normalHTML',
-                    html,
-                    ...(selector === '.body-content .loadme picture img' && {
-                        src: $1('.body-content .loadme picture source').attr('data-srcset'),
-                        title: $1(selector).attr('title'),
-                    }),
-                });
-            }
-        }
-    }
-
-    return imageDataArray;
-};
-
-export { parseList, orderContent };