rsshub 1.0.0-master.fe69819 → 1.0.0-master.fe775f5

package/lib/v2/wfu/news.js

@@ -0,0 +1,96 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+// 参考 whu/news 武汉大学页面写成
+
+const baseUrl = 'https://www.wfu.edu.cn';
+const sizeTitle = '--潍坊学院新闻';
+
+const catrgoryMap = {
+    wyyw: ['/50/list.htm', '潍院要闻'],
+    zhxw: ['/52/list.htm', '综合新闻'],
+    xszh: ['/xszh/list.htm', '学术纵横'],
+};
+
+// 专门定义一个function用于加载文章内容
+async function load(link) {
+    let description = '';
+
+    let response;
+    // 如果不是 大学的站点, 直接返回简单的标题即可
+    // 判断 是否外站链接,如果是 则直接返回页面 不做单独的解析
+    const https_reg = new RegExp('https://www.wfu.edu.cn(.*)');
+    if (!https_reg.test(link)) {
+        return { description };
+    }
+    try {
+        // 异步请求文章
+        response = await got.get(link);
+    } catch (err) {
+        // 如果网络问题 直接出错
+        if (err.name && (err.name === 'HTTPError' || err.name === 'RequestError')) {
+            description = 'Page 404 Please Check!';
+        }
+        return { description };
+    }
+    // 加载文章内容
+    const $ = cheerio.load(response.data);
+
+    // 提取文章内容
+    description = $('div.wp_articlecontent').html();
+    // 返回解析的结果
+    return { description };
+}
+
+module.exports = async (ctx) => {
+    // 默认 潍院要闻 然后获取列表页面
+    const type = ctx.params.type ?? 'wyyw';
+    const listPageUrl = baseUrl + catrgoryMap[type][0];
+    const response = await got({
+        method: 'get',
+        url: listPageUrl,
+        headers: {
+            Referer: baseUrl,
+        },
+    });
+    const $ = cheerio.load(response.data);
+
+    // 获取当前页面的 list
+    const list = $('div[frag=窗口1]>ul>li');
+
+    const result = await Promise.all(
+        // 遍历每一篇文章
+        list
+            .map(async (item) => {
+                const $ = cheerio.load(list[item]); // 将列表项加载成 html
+                const $item_url = 'https://www.wfu.edu.cn' + $('a').attr('href'); // 获取 每一项的url
+                const $title = $('a>div.txt>h1').text(); // 获取每个的标题
+                const $pubdate = timezone(parseDate($('a>div.txt>span.date').text().split('：')[1]), +8); // 获取发布时间
+
+                // 列表上提取到的信息
+                // 标题 链接
+                const single = {
+                    title: $title,
+                    pubDate: $pubdate,
+                    link: $item_url,
+                    guid: $item_url,
+                };
+
+                // 对于列表的每一项, 单独获取 时间与详细内容
+                // eslint-disable-next-line no-return-await
+                const other = await ctx.cache.tryGet($item_url, async () => await load($item_url));
+                // 合并解析后的结果集作为该篇文章最终的输出结果
+                return { ...single, ...other };
+            })
+            .get()
+    );
+
+    ctx.state.data = {
+        title: catrgoryMap[type][1] + sizeTitle,
+        link: baseUrl,
+        description: catrgoryMap[type][1] + sizeTitle,
+        item: result,
+    };
+};

package/lib/v2/wfu/radar.js

@@ -0,0 +1,21 @@
+module.exports = {
+    'wfu.edu.cn': {
+        _name: '潍坊学院',
+        jwc: [
+            {
+                title: '教务处通知',
+                docs: 'https://docs.rsshub.app/university.html#wei-fang-xue-yuan',
+                source: '/',
+                target: '/wfu/jwc',
+            },
+        ],
+        news: [
+            {
+                title: '新闻',
+                docs: 'https://docs.rsshub.app/university.html#wei-fang-xue-yuan',
+                source: '/',
+                target: '/wfu/news',
+            },
+        ],
+    },
+};

package/lib/v2/wfu/router.js

@@ -0,0 +1,4 @@
+module.exports = (router) => {
+    router.get('/jwc', require('./jwc'));
+    router.get('/news/:type?', require('./news'));
+};

package/lib/v2/whitehouse/maintainer.js

@@ -1,3 +1,4 @@
 module.exports = {
     '/briefing-room/:category?': ['nczitzk'],
+    '/ostp': ['LyleLee'],
 };

package/lib/v2/whitehouse/ostp.js

@@ -0,0 +1,50 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const rootUrl = 'https://www.whitehouse.gov';
+    const currentUrl = `${rootUrl}/ostp/news-updates`;
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = cheerio.load(response.data);
+
+    const list = $('.news-item__title')
+        .slice(0, 10)
+        .map((_, item) => {
+            item = $(item);
+            return {
+                title: item.text(),
+                link: item.attr('href'),
+            };
+        })
+        .get();
+
+    const items = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const detailResponse = await got({
+                    method: 'get',
+                    url: item.link,
+                });
+                const content = cheerio.load(detailResponse.data);
+
+                item.description = content('.body-content').html();
+                item.pubDate = parseDate(content('time.published').attr('datetime'));
+                item.author = content('span.tax-links.cat-links > a').text();
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: 'Whitehouse OSTP',
+        link: currentUrl,
+        description: 'Whitehouse Office of Science and Technology Policy',
+        item: items,
+    };
+};

package/lib/v2/whitehouse/radar.js

@@ -8,6 +8,12 @@ module.exports = {
                 source: ['/briefing-room/:category', '/'],
                 target: '/whitehouse/briefing-room/:category',
             },
+            {
+                title: '科技政策办公室',
+                docs: 'https://docs.rsshub.app/government.html#mei-guo-bai-gong-ban-gong-ting',
+                source: ['/ostp', '/'],
+                target: '/whitehouse/ostp',
+            },
         ],
     },
 };

package/lib/v2/whitehouse/router.js

@@ -1,3 +1,4 @@
 module.exports = (router) => {
     router.get('/briefing-room/:category?', require('./briefing-room'));
+    router.get('/ostp', require('./ostp'));
 };

package/lib/v2/who/maintainer.js

@@ -0,0 +1,5 @@
+module.exports = {
+    '/news/:language?': ['nczitzk'],
+    '/news-room/:category?/:language?': ['LogicJake', 'nczitzk'],
+    '/speeches/:language?': ['nczitzk'],
+};

package/lib/v2/who/news-room.js

@@ -0,0 +1,76 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const category = ctx.params.category ?? 'feature-stories';
+    const language = ctx.params.language ?? '';
+
+    const rootUrl = 'https://www.who.int';
+    const currentUrl = `${rootUrl}/${language ? `${language}/` : ''}news-room/${category}`;
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = cheerio.load(response.data);
+
+    let list = $('.list-view--item a');
+
+    if (list.length === 0) {
+        const response = await got({
+            method: 'get',
+            url: `${rootUrl}/api/hubs/${category.replace(/-/g, '')}?sf_culture=zh&$orderby=PublicationDateAndTime%20desc&$select=Title,PublicationDateAndTime,ItemDefaultUrl&$top=30`,
+        });
+
+        list = response.data.value.map((item) => ({
+            title: item.Title,
+            link: `${currentUrl}/detail/${item.ItemDefaultUrl}`,
+            pubDate: parseDate(item.PublicationDateAndTime),
+        }));
+    } else {
+        list = list
+            .map((_, item) => {
+                item = $(item);
+                const link = item.attr('href');
+
+                return {
+                    link: `${link.indexOf('http') === 0 ? '' : rootUrl}${item.attr('href')}`,
+                };
+            })
+            .get();
+    }
+
+    const items = await Promise.all(
+        list.map(
+            async (item) =>
+                await ctx.cache.tryGet(item.link, async () => {
+                    const detailResponse = await got({
+                        method: 'get',
+                        url: item.link,
+                    });
+
+                    const matches = detailResponse.data.match(/"headline":"(.*)","description":"(.*)","datePublished":"(.*)","image"/);
+
+                    if (matches) {
+                        item.title = matches[1];
+                        item.description = matches[2];
+                        item.pubDate = parseDate(matches[3]);
+                    } else {
+                        const content = cheerio.load(detailResponse.data);
+
+                        item.description = content('.sf-content-block').html();
+                    }
+
+                    return item;
+                })
+        )
+    );
+
+    ctx.state.data = {
+        title: `${$('meta[property="og:title"]').attr('content')} - WHO`,
+        link: currentUrl,
+        item: items,
+    };
+};

package/lib/{routes → v2}/who/news.js

@@ -2,7 +2,7 @@ const got = require('@/utils/got');
 const { parseDate } = require('@/utils/parse-date');
 
 module.exports = async (ctx) => {
-    const language = ctx.params.language || 'en';
+    const language = ctx.params.language ?? 'en';
 
     const rootUrl = 'https://www.who.int';
     const currentUrl = `${rootUrl}/${language === 'en' ? '' : `${language}/`}news`;

package/lib/v2/who/radar.js

@@ -0,0 +1,25 @@
+module.exports = {
+    'who.int': {
+        _name: '世界卫生组织 WHO',
+        '.': [
+            {
+                title: '新闻稿',
+                docs: 'https://docs.rsshub.app/government.html#shi-jie-wei-sheng-zu-zhi-who',
+                source: '/news',
+                target: '/who/news',
+            },
+            {
+                title: '媒体中心',
+                docs: 'https://docs.rsshub.app/government.html#shi-jie-wei-sheng-zu-zhi-who',
+                source: '/news-room/:type',
+                target: '/who/news-room/:type',
+            },
+            {
+                title: '总干事的讲话',
+                docs: 'https://docs.rsshub.app/government.html#shi-jie-wei-sheng-zu-zhi-who',
+                source: '/director-general/speeches',
+                target: '/who/speeches',
+            },
+        ],
+    },
+};

package/lib/v2/who/router.js

@@ -0,0 +1,5 @@
+module.exports = (router) => {
+    router.get('/news/:language?', require('./news'));
+    router.get('/news-room/:category?/:language?', require('./news-room'));
+    router.get('/speeches/:language?', require('./speeches'));
+};

package/lib/v2/wikinews/index.js

@@ -0,0 +1,43 @@
+const got = require('@/utils/got');
+const { parseString } = require('xml2js');
+const cheerio = require('cheerio');
+const currentURL = 'https://zh.wikinews.org/wiki/Special:%E6%96%B0%E9%97%BB%E8%AE%A2%E9%98%85';
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const resp = await got(currentURL);
+    const {
+        urlset: { url: urls },
+    } = await new Promise((resolve, reject) => {
+        parseString(resp.data, (err, result) => {
+            if (err) {
+                reject(err);
+                return;
+            }
+            resolve(result);
+        });
+    });
+
+    ctx.state.data = {
+        title: '最新新闻 - 维基新闻',
+        link: currentURL,
+        item: await Promise.all(
+            urls.map(async (url) => {
+                const loc = url.loc[0];
+                const news = url['news:news'][0];
+                const description = await ctx.cache.tryGet(loc, async () => {
+                    const resp = await got(loc);
+                    const $ = cheerio.load(resp.data);
+                    return $('#bodyContent').html();
+                });
+                return {
+                    title: news['news:title'][0],
+                    pubDate: parseDate(news['news:publication_date'][0]),
+                    category: news['news:keywords'][0].split(',').map((item) => item.trim()),
+                    link: loc,
+                    description,
+                };
+            })
+        ),
+    };
+};

package/lib/v2/wikinews/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/latest': ['KotoriK'],
+};

package/lib/v2/wikinews/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'wikinews.org': {
+        _name: '维基新闻',
+        zh: [
+            {
+                title: '最新新闻',
+                docs: 'https://docs.rsshub.app/new-media.html#wei-ji-xin-wen',
+                source: ['/wiki/Special:新闻订阅'],
+                target: '/wikinews/latest',
+            },
+        ],
+    },
+};

package/lib/v2/wikinews/router.js

@@ -0,0 +1,4 @@
+const routerIndex = require('./index');
+module.exports = function (router) {
+    router.get('/latest', routerIndex);
+};

package/lib/v2/wp-china/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/news/:category?': ['nczitzk'],
+};

package/lib/v2/wp-china/news.js

@@ -0,0 +1,57 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+const config = {
+    latest: 'list-35b3413e4c03411d8f3726162570804f',
+    stories: 'list-7224ef82da1b498eb24046a6c56ff566',
+};
+
+module.exports = async (ctx) => {
+    const category = ctx.params.category ?? 'latest';
+
+    const rootUrl = 'http://www.wp-china.com';
+    const currentUrl = `${rootUrl}/f/${config[category]}.html`;
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = cheerio.load(response.data);
+
+    const list = $('.new_list')
+        .slice(0, 10)
+        .map((_, item) => {
+            item = $(item);
+            return {
+                title: item.children('.fb').text(),
+                link: `${rootUrl}${item.parent().attr('href')}`,
+                pubDate: timezone(parseDate(item.children('.grey').text()), +8),
+            };
+        })
+        .get();
+
+    const items = await Promise.all(
+        list.map(
+            async (item) =>
+                await ctx.cache.tryGet(item.link, async () => {
+                    const detailResponse = await got({
+                        method: 'get',
+                        url: item.link,
+                    });
+                    const content = cheerio.load(detailResponse.data);
+
+                    item.description = content('.wb100').html();
+
+                    return item;
+                })
+        )
+    );
+
+    ctx.state.data = {
+        title: $('title').text().replace(' - Powered By gobestsoft', ''),
+        link: currentUrl,
+        item: items,
+    };
+};

package/lib/v2/wp-china/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'wp-china.com': {
+        _name: '中国工人出版社',
+        '.': [
+            {
+                title: '新闻中心',
+                docs: 'https://docs.rsshub.app/new-media.html#zhong-guo-gong-ren-chu-ban-she',
+                source: ['/'],
+                target: '/wp-china/news',
+            },
+        ],
+    },
+};

package/lib/v2/wp-china/router.js

@@ -0,0 +1,3 @@
+module.exports = (router) => {
+    router.get('/news/:category?', require('./news'));
+};

package/lib/{routes → v2}/wsj/index.js

@@ -1,6 +1,7 @@
 const parser = require('@/utils/rss-parser');
 const cheerio = require('cheerio');
 const got = require('@/utils/got');
+const { parseDate } = require('@/utils/parse-date');
 
 const categoryToXMLFileName = {
     opinion: 'RSSOpinion.xml',
@@ -57,7 +58,7 @@ module.exports = async (ctx) => {
                 });
                 const html = response.body;
                 const $ = cheerio.load(html);
-                const content = $('.articleBody > div[amp-access="access"]');
+                const content = $('.articleBody > section');
 
                 // Cover
                 const cover = $('.articleLead > div.is-lead-inset > div.header > .img-header > div.image-container > amp-img > img');
@@ -73,6 +74,7 @@ module.exports = async (ctx) => {
                 // Metadata (categories & updatedAt)
                 const updatedAt = $('meta[itemprop="dateModified"]').attr('content');
                 const publishedAt = $('meta[itemprop="datePublished"]').attr('content');
+                const author = $('.author > a[rel="author"]').html();
 
                 const categories = $('meta[name="keywords"]')
                     .attr('content')
@@ -113,9 +115,9 @@ module.exports = async (ctx) => {
                 return {
                     title: item.title,
                     id: item.guid,
-                    pubDate: new Date(publishedAt).toUTCString(),
-                    updated: new Date(updatedAt).toUTCString(),
-                    author: item.creator,
+                    pubDate: parseDate(publishedAt),
+                    updated: parseDate(updatedAt),
+                    author,
                     link: item.link,
                     summary,
                     description: content.html(),

package/lib/v2/wsj/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/:lang/:category?': ['oppilate'],
+};

package/lib/v2/wsj/radar.js

@@ -0,0 +1,21 @@
+module.exports = {
+    'wsj.com': {
+        _name: '华尔街日报 The Wall Street Journal (WSJ)',
+        cn: [
+            {
+                title: '新闻',
+                docs: 'https://docs.rsshub.app/traditional-media.html#hua-er-jie-ri-bao-the-wall-street-journal-wsj',
+                souce: '/',
+                target: '/wsj/zh-cn',
+            },
+        ],
+        www: [
+            {
+                title: '新闻',
+                docs: 'https://docs.rsshub.app/traditional-media.html#hua-er-jie-ri-bao-the-wall-street-journal-wsj',
+                souce: '/',
+                target: '/wsj/en-us',
+            },
+        ],
+    },
+};

package/lib/v2/wsj/router.js

@@ -0,0 +1,3 @@
+module.exports = (router) => {
+    router.get('/:lang/:category?', require('./index'));
+};

package/lib/v2/xiaoyuzhou/maintainer.js

@@ -1,4 +1,4 @@
 module.exports = {
     '/': ['prnake', 'Maecenas'],
-    '/podcast/:id': ['hondajojo'],
+    '/podcast/:id': ['hondajojo', 'jtsang4'],
 };

package/lib/v2/xiaoyuzhou/pickup.js

@@ -58,7 +58,7 @@ const ProcessFeed = async (ctx) => {
         const itunes_item_image = item.episode.image ? item.episode.image.picUrl : item.episode.podcast.image ? item.episode.podcast.image.picUrl : '';
         const link = `https://www.xiaoyuzhoufm.com/episode/${eid}`;
         const pubDate = item.pubDate;
-        const enclosure_length = item.episode.duration;
+        const itunes_duration = item.episode.duration;
         const enclosure_url = item.episode.enclosure.url;
         const desc = `<p><strong>${item.comment.author.nickname}：</strong>${item.comment.text}</p><hr>` + item.episode.shownotes;
         const author = item.episode.podcast.author;
@@ -70,7 +70,7 @@ const ProcessFeed = async (ctx) => {
             author,
             pubDate,
             enclosure_url,
-            enclosure_length,
+            itunes_duration,
             itunes_item_image,
             enclosure_type: 'audio/mpeg',
         };

package/lib/v2/xiaoyuzhou/podcast.js

@@ -16,12 +16,12 @@ module.exports = async (ctx) => {
     const episodes = page_data.props.pageProps.podcast.episodes.map((item) => ({
         title: item.title,
         enclosure_url: item.enclosure.url,
-        enclosure_length: item.duration,
+        itunes_duration: item.duration,
         enclosure_type: 'audio/mpeg',
         link: `https://www.xiaoyuzhoufm.com/episode/${item.eid}`,
         pubDate: parseDate(item.pubDate),
         description: item.shownotes,
-        itunes_item_image: item.image.smallPicUrl,
+        itunes_item_image: (item.image || item.podcast?.image)?.smallPicUrl,
     }));
 
     ctx.state.data = {