feat(route): add yahoo japan covid19 news collection 日本疫情消息汇总 (#8099)

docs/en/other.md

@@ -51,6 +51,12 @@ Official Website: [https://www.ssm.gov.mo/apps1/PreventWuhanInfection/en.aspx](h
 
 <RouteEn author="Gnnng" example="/coronavirus/sg-moh" path="/coronavirus/sg-moh"/>
 
+### Yahoo Japan COVID19 news collection
+
+Official Website: <https://news.yahoo.co.jp/pages/article/20200207>
+
+<RouteEn author="sgqy" example="/coronavirus/yahoo-japan" path="/coronavirus/yahoo-japan/:tdfk?" :paramsDesc="['Romaji of Todofuken. Can be got from URLs on area detail page. Example: kyoto']"/>
+
 ## Darwin Awards
 
 ### Articles

docs/other.md

@@ -870,6 +870,12 @@ type 为 all 时，category 参数不支持 cost 和 free
 
 <Route author="Gnnng" example="/coronavirus/sg-moh" path="/coronavirus/sg-moh"/>
 
+### Yahoo Japan 新型コロナウイルス感染症まとめ
+
+新闻主页：<https://news.yahoo.co.jp/pages/article/20200207>
+
+<Route author="sgqy" example="/coronavirus/yahoo-japan" path="/coronavirus/yahoo-japan/:tdfk?" :paramsDesc="['都道府県的拼音，可从地图详情页的链接中获取。例如：京都府 = kyoto']"/>
+
 ## 新趣集
 
 > 官方 Feed 地址为: <https://xinquji.com/rss>

lib/router.js

@@ -2566,6 +2566,7 @@ router.get('/coronavirus/nhc', lazyloadRouteHandler('./routes/coronavirus/nhc'))
 router.get('/coronavirus/mogov-2019ncov/:lang', lazyloadRouteHandler('./routes/coronavirus/mogov-2019ncov'));
 router.get('/coronavirus/qq/fact', lazyloadRouteHandler('./routes/tencent/factcheck'));
 router.get('/coronavirus/sg-moh', lazyloadRouteHandler('./routes/coronavirus/sg-moh'));
+router.get('/coronavirus/yahoo-japan/:tdfk?', lazyloadRouteHandler('./routes/coronavirus/yahoo-japan'));
 
 // 南京林业大学教务处
 router.get('/njfu/jwc/:category?', lazyloadRouteHandler('./routes/universities/njfu/jwc'));

lib/routes/coronavirus/yahoo-japan.js

@@ -0,0 +1,77 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const dayjs = require('dayjs');
+
+module.exports = async (ctx) => {
+    const tdfk = ctx.params.tdfk || false;
+    const uri = tdfk ? `https://news.yahoo.co.jp/pages/article/covid19${tdfk}` : `https://news.yahoo.co.jp/pages/article/20200207`;
+    const req_header = { 'User-Agent': 'Mozilla/5.0 (Linux; Android 9; SM-G960F Build/PPR1.180610.011; wv) AppleWebKit/537.36 (KHTML, like Gecko) Version/4.0 Chrome/74.0.3729.157 Mobile Safari/537.36' };
+
+    const resp = await got({
+        method: 'get',
+        url: uri,
+        headers: req_header,
+    });
+
+    const $ = cheerio.load(resp.data);
+
+    const this_year = dayjs().year();
+    const this_month = dayjs().month() + 1;
+
+    const art_uri = [];
+    $('#layoutFooter ul.dlpThumbLink a').each((i, e) => {
+        const link = $(e).attr('href');
+        const text = $(e).find('.dlpThumbText span').eq(0).text();
+        const author = $(e).find('.dlpQuote').text();
+        const date = $(e).find('.dlpDate').text() + ' +9'; // explicit timezone
+
+        let date_obj = dayjs(date).year(this_year);
+
+        if (date_obj.month() + 1 > this_month) {
+            // if the article is from the last year
+            date_obj = date_obj.year(this_year - 1);
+        }
+
+        art_uri.push({ l: link, t: text, a: author, d: date_obj.toString() });
+    });
+
+    const getNews = async (uri) => {
+        const ret = {
+            link: uri.l,
+            title: uri.t,
+            author: uri.a,
+            pubDate: uri.d,
+            description: null,
+        };
+
+        if (!uri.l.includes('//news.yahoo.co.jp')) {
+            return ret; // do not process uncertain pages
+        }
+
+        const page_data = await ctx.cache.tryGet(uri.l, async () => {
+            const resp = await got({
+                method: 'get',
+                url: uri.l,
+                headers: req_header,
+            });
+            return resp.data;
+        });
+
+        const $doc = cheerio.load(page_data);
+
+        const iso_date = $doc('meta[name="pubdate"]').attr('content');
+        ret.pubDate = dayjs(iso_date).toString();
+        ret.description = $doc('div.article_body').html() || $doc('meta[name="description"]').attr('content');
+
+        return ret;
+    };
+
+    const items = await Promise.all(art_uri.map(getNews));
+
+    ctx.state.data = {
+        title: $('title').text(),
+        link: uri,
+        description: $('meta[name="description"]').attr('content'),
+        item: items,
+    };
+};