fix(route): mofcom (#10641)

* fix(route): mofcom

* fix: typo

docs/government.md

@@ -1247,7 +1247,7 @@ pageClass: routes
 
 ### 政务公开
 
-<Route author="LogicJake" example="/mofcom/article/b" path="/mofcom/article/:suffix" :paramsDesc="['支持形如`http://www.mofcom.gov.cn/article/*`的网站，传入 article 之后的后缀']" />
+<Route author="LogicJake" example="/gov/mofcom/article/xwfb" path="/gov/mofcom/article/:suffix+" :paramsDesc="['支持形如 `http://www.mofcom.gov.cn/article/*` 的网站，传入 article 之后的后缀，支持多级']" radar="1" rssbud="1"/>
 
 ## 中华人民共和国生态环境部
 

lib/router.js

@@ -1945,8 +1945,8 @@ router.get('/dhl/:id', lazyloadRouteHandler('./routes/dhl/shipment-tracking'));
 // Japanpost
 router.get('/japanpost/track/:reqCode/:locale?', lazyloadRouteHandler('./routes/japanpost/track'));
 
-// 中华人民共和国商务部
-router.get('/mofcom/article/:suffix', lazyloadRouteHandler('./routes/mofcom/article'));
+// 中华人民共和国商务部 migrated to v2
+// router.get('/mofcom/article/:suffix', lazyloadRouteHandler('./routes/mofcom/article'));
 
 // 品玩
 router.get('/pingwest/status', lazyloadRouteHandler('./routes/pingwest/status'));

lib/routes/mofcom/article.js

@@ -1,54 +0,0 @@
-const cheerio = require('cheerio');
-const got = require('@/utils/got');
-const { resolve } = require('url');
-
-const host = 'http://www.mofcom.gov.cn';
-
-module.exports = async (ctx) => {
-    const suffix = ctx.params.suffix;
-    const url = resolve('http://www.mofcom.gov.cn/article/', suffix);
-    const res = await got.get(url);
-    const $ = cheerio.load(res.data);
-    const title = $('head > title').text();
-    const list = $('.txtList_01 li').get();
-
-    const out = await Promise.all(
-        list.map(async (item) => {
-            const $ = cheerio.load(item);
-            const title = $('a').attr('title');
-            const sub_url = $('a').attr('href');
-            const itemUrl = resolve(host, sub_url);
-            const cache = await ctx.cache.get(itemUrl);
-            if (cache) {
-                return Promise.resolve(JSON.parse(cache));
-            }
-
-            let responses = await got.get(itemUrl);
-            const redirect = responses.data.match(/window\.location\.href='(.*)'/);
-            if (redirect) {
-                responses = await got.get(redirect[1], {
-                    headers: {
-                        Referer: itemUrl,
-                        'Accept-Encoding': 'gzip, deflate',
-                        'Accept-Language': 'zh-CN,zh;q=0.9',
-                        Accept: 'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3',
-                    },
-                });
-            }
-            const $d = cheerio.load(responses.data);
-
-            const single = {
-                title,
-                link: itemUrl,
-                description: $d('.art-con').html() || $d('.textlive').html(),
-            };
-            ctx.cache.set(itemUrl, JSON.stringify(single));
-            return Promise.resolve(single);
-        })
-    );
-    ctx.state.data = {
-        title,
-        link: url,
-        item: out,
-    };
-};

lib/v2/gov/maintainer.js

@@ -22,6 +22,7 @@ module.exports = {
     '/miit/wjgs': ['Yoge-Code'],
     '/miit/zcjd': ['Yoge-Code'],
     '/moe/:type': ['Crawler995'],
+    '/mofcom/article/:suffix+': ['LogicJake'],
     '/moj/aac/news/:type?': ['TonyRL'],
     '/nifdc/:path+': ['nczitzk'],
     '/nmpa/:path+': ['TonyRL'],

lib/v2/gov/mofcom/article.js

@@ -0,0 +1,68 @@
+const cheerio = require('cheerio');
+const got = require('@/utils/got');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+const host = 'http://www.mofcom.gov.cn';
+
+module.exports = async (ctx) => {
+    const suffix = ctx.params.suffix;
+    const url = `http://www.mofcom.gov.cn/article/${suffix}/`;
+    const { data: res } = await got(url);
+    const $ = cheerio.load(res);
+    $('.listline').remove();
+    const list = $('.txtList_01 li')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+            const date =
+                item.find('span').length !== 0
+                    ? item
+                          .find('span')
+                          .text()
+                          .match(/((\d{4}-\d{2}-\d{2})(\s\d{2}:\d{2}:\d{2})?)/)[1]
+                    : item
+                          .find('a')
+                          .attr('title')
+                          .match(/(\d{4}年\d{1,2}月\d{1,2})/)[1];
+            return {
+                title: item.find('a').attr('title'),
+                link: host + item.find('a').attr('href'),
+                pubDate: timezone(parseDate(date, ['YYYY-MM-DD HH:mm:ss', 'YYYY-MM-DD', 'YYYY年M月D']), 8),
+            };
+        });
+
+    await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                let responses = await got(item.link);
+                // xwfb/xwlxfbh || xwfb/xwztfbh
+                const redirect = responses.data.match(/_cofing1=\{href:"(.*)",type/) || responses.data.match(/window\.location\.href='(.*)'/);
+                if (redirect) {
+                    responses = await got(redirect[1], {
+                        headers: {
+                            Referer: item.link,
+                        },
+                    });
+                }
+                const $ = cheerio.load(responses.data);
+                const iframe = $('.art-con iframe').attr('src');
+                if ($('.art-con iframe').attr('src')) {
+                    const { data: res } = await got(iframe);
+                    item.description = cheerio.load(res)('tbody').html();
+                } else {
+                    item.description = $('.art-con').html() || /* xwfb/xwztfbh */ $('.textlive').html();
+                }
+                item.pubDate = $('meta[name="PubDate"]').length ? timezone(parseDate($('meta[name="PubDate"]').attr('content'), 'YYYY-MM-DD HH:mm'), 8) : item.pubDate;
+
+                return item;
+            })
+        )
+    );
+    ctx.state.data = {
+        title: $('head > title').text(),
+        description: $('meta[name=description]').attr('content'),
+        link: url,
+        item: list,
+    };
+};

lib/v2/gov/radar.js

@@ -526,6 +526,17 @@ module.exports = {
             },
         ],
     },
+    'mofcom.gov.cn': {
+        _name: '中华人民共和国商务部',
+        '.': [
+            {
+                title: '新闻发布',
+                docs: 'https://docs.rsshub.app/government.html#zhong-hua-ren-min-gong-he-guo-shang-wu-bu',
+                source: ['/article/*'],
+                target: (_, url) => `/gov/mofcom/article/${new URL(url).pathname.replace('/article/', '')}`,
+            },
+        ],
+    },
     'moj.gov.tw': {
         _name: '台灣法務部廉政署',
         'www.aac': [

lib/v2/gov/router.js

@@ -14,6 +14,7 @@ module.exports = function (router) {
     router.get('/miit/wjgs', require('./miit/wjgs'));
     router.get('/miit/zcjd', require('./miit/zcjd'));
     router.get('/moe/:type', require('./moe/moe'));
+    router.get('/mofcom/article/:suffix+', require('./mofcom/article'));
     router.get('/moj/aac/news/:type?', require('./moj/aac/news'));
     router.get(/nifdc\/([\w/-]+)?/, require('./nifdc'));
     router.get(/\/nmpa\/([\w][\w/]+)?/, require('./nmpa/generic'));