feat: add 中国政府网生猪信息 & 商品价格 (#5305)

docs/government.md

@@ -131,6 +131,24 @@ pageClass: routes
 
 ## 中国政府
 
+### 数据
+
+<Route author="nczitzk" example="/gov/shuju/shengzhu/baitiaorou" path="/gov/shuju/:caty/:item" :paramsDesc="['分类，“生猪”对应`shengzhu`，“价格”对应`jiage`', '项目，见表']">
+
+生猪分类可选项目：
+
+| 白条猪价格 | 生猪屠宰与活体交易 | 仔猪价格 | 生猪出场价与玉米价  | 生猪存栏信息和生猪疫情 |
+| ---------- | ------------------ | -------- | ------------------- | ---------------------- |
+| baitiaorou | huotijiaoyi        | zizhu    | chuchangjia_yumijia | cunlan_yiqing          |
+
+价格分类可选项目：
+
+| 商品价格 | 农产品价格  | 油价   |
+| -------- | ----------- | ------ |
+| shangpin | nongchanpin | youjia |
+
+</Route>
+
 ### 最新政策
 
 <Route author="SettingDust" example="/gov/zhengce/zuixin" path="/gov/zhengce/zuixin"/>

lib/router.js

@@ -1208,6 +1208,7 @@ router.get('/gov/province/:name/:category', require('./routes/gov/province'));
 router.get('/gov/city/:name/:category', require('./routes/gov/city'));
 router.get('/gov/statecouncil/briefing', require('./routes/gov/statecouncil/briefing'));
 router.get('/gov/news/:uid', require('./routes/gov/news'));
+router.get('/gov/shuju/:caty/:item', require('./routes/gov/shuju'));
 
 // 苏州
 router.get('/gov/suzhou/news/:uid', require('./routes/gov/suzhou/news'));

lib/routes/gov/shuju/index.js

@@ -0,0 +1,51 @@
+const url = require('url');
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+
+const rootUrl = 'http://www.gov.cn';
+
+module.exports = async (ctx) => {
+    const currentUrl = `${rootUrl}/shuju/${ctx.params.caty}/${ctx.params.item}.htm`;
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+    const $ = cheerio.load(response.data);
+    const list = $('ul li')
+        .map((_, item) => {
+            item = $(item);
+            const a = item.find('a');
+            return {
+                title: a.text(),
+                link: url.resolve(rootUrl, a.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 content = cheerio.load(detailResponse.data);
+
+                    content('.font').remove();
+                    content('.pages_print').remove();
+
+                    item.pubDate = new Date(content('div.pages-date').text().trim()).toUTCString();
+                    item.description = content('#UCAP-CONTENT').html();
+
+                    return item;
+                })
+        )
+    );
+
+    ctx.state.data = {
+        title: $('title').text(),
+        link: currentUrl,
+        item: items,
+    };
+};