feat(route): 北极星光伏 (#9542)

* 新增北极星光伏网路由

* 规范化北极星光伏网路由

* 修改一些文本错误

* refactor: migrate to v2

docs/traditional-media.md

@@ -469,9 +469,21 @@ Provides all of the articles by the specified Yahoo! author.
 
 ## 北极星电力网
 
-### 北极星环保
+### 环保要闻
 
-<Route author="zsimple"  example="/bjx/huanbao" path="/bjx/huanbao" />
+<Route author="zsimple" example="/bjx/huanbao" path="/bjx/huanbao" radar="1" rssbud="1"/>
+
+### 光伏
+
+<Route author="Sxuet" example="/bjx/gf/sc" path="/bjx/gf/:type" :paramsDesc="['分类，北极星光伏最后的`type`字段']" radar="1" rssbud="1"/>
+
+`:type` 类型可选如下
+
+| 要闻 | 政策 | 市场行情 | 企业动态 | 独家观点 | 项目工程 | 招标采购 | 财经 | 国际行情 | 价格趋势 | 技术跟踪 |
+| -- | -- | ---- | ---- | ---- | ---- | ---- | -- | ---- | ---- | ---- |
+| yw | zc | sc   | mq   | dj   | xm   | zb   | cj | gj   | sj   | js   |
+
+</Route>
 
 ## 财新网
 

lib/router.js

@@ -1445,8 +1445,8 @@ router.get('/banyuetan/:name', lazyloadRouteHandler('./routes/banyuetan'));
 // router.get('/people/xjpjh/:keyword?/:year?', lazyloadRouteHandler('./routes/people/xjpjh'));
 // router.get('/people/cpc/24h', lazyloadRouteHandler('./routes/people/cpc/24h'));
 
-// 北极星电力网
-router.get('/bjx/huanbao', lazyloadRouteHandler('./routes/bjx/huanbao'));
+// 北极星电力网 migrated to v2
+// router.get('/bjx/huanbao', lazyloadRouteHandler('./routes/bjx/huanbao'));
 
 // gamersky
 router.get('/gamersky/news', lazyloadRouteHandler('./routes/gamersky/news'));

lib/routes/bjx/huanbao.js

@@ -1,78 +0,0 @@
-const got = require('@/utils/got');
-const date = require('@/utils/date');
-const cheerio = require('cheerio');
-const url = require('url');
-const iconv = require('iconv-lite');
-
-module.exports = async (ctx) => {
-    const listURL = 'http://huanbao.bjx.com.cn/NewsList';
-    const response = await got({
-        method: 'get',
-        url: listURL,
-    });
-
-    const $ = cheerio.load(response.data);
-    const list = $('.list_main .list_left_ul a')
-        .map((_, a) => $(a).attr('href'))
-        .get();
-
-    const out = await Promise.all(
-        // 服务器禁止单个IP大并发访问，只能少返回几条
-        list.slice(0, 3).map((link) => fetchPage(ctx, link))
-    );
-
-    ctx.state.data = {
-        title: '北极星环保 - 环保行业垂直门户网站',
-        link: listURL,
-        item: out,
-    };
-};
-
-async function fetchPage(ctx, link) {
-    const cache = await ctx.cache.get(link);
-    if (cache) {
-        return JSON.parse(cache);
-    }
-
-    // 可能一篇文章过长会分成多页
-    const pages = [];
-
-    const result = await got.get(link, { responseType: 'buffer' });
-    const $page = cheerio.load(iconv.decode(result.data, 'gbk'));
-    pages.push($page);
-
-    // 如果是有分页链接，则使用顺序加载以保证顺序
-    const pagelinks = $page('.list_detail div.page a');
-    if (pagelinks.length > 0) {
-        for (let i = 0; i < pagelinks.length; i++) {
-            const $a = $page(pagelinks[i]);
-            if (!/^\d+$/.test($a.text().trim())) {
-                continue;
-            }
-            const sublink = url.resolve(link, $a.attr('href'));
-            /* eslint-disable no-await-in-loop */
-            const result = await got.get(sublink, { responseType: 'buffer' });
-            pages.push(cheerio.load(iconv.decode(result.data, 'gbk')));
-        }
-    }
-
-    // 将懒加载的loading图片转换为真实图片
-    pages.forEach(($p) => {
-        $p('.list_detail')
-            .find('img[data-echo]')
-            .each((_, img) => {
-                const $img = $p(img);
-                $img.attr('src', $img.data('echo')).removeAttr('data-echo');
-            });
-    });
-
-    const item = {
-        title: $page('.list_detail > h1').text(),
-        description: pages.reduce((desc, $p) => desc + $p('.list_detail .newsrand').html(), ''),
-        pubDate: date($page('.list_detail .list_copy b').last().text()),
-        link,
-        author: $page('.list_detail .list_copy b').first().text(),
-    };
-    ctx.cache.set(link, JSON.stringify(item));
-    return item;
-}

lib/v2/bjx/huanbao.js

@@ -0,0 +1,78 @@
+const got = require('@/utils/got');
+const { parseDate } = require('@/utils/parse-date');
+const cheerio = require('cheerio');
+const timezone = require('@/utils/timezone');
+const asyncPool = require('tiny-async-pool');
+
+const asyncPoolAll = async (...args) => {
+    const results = [];
+    for await (const result of asyncPool(...args)) {
+        results.push(result);
+    }
+    return results;
+};
+
+module.exports = async (ctx) => {
+    const listURL = 'https://huanbao.bjx.com.cn/yw/';
+    const response = await got(listURL);
+
+    const $ = cheerio.load(response.data);
+    let items = $('.cc-layout-3 .cc-list-content li')
+        .toArray()
+        .map((e) => {
+            e = $(e);
+            return {
+                title: e.find('a').attr('title'),
+                link: e.find('a').attr('href'),
+                pubDate: parseDate(e.find('span').text()),
+            };
+        });
+
+    items = await asyncPoolAll(
+        // 服务器禁止单个IP大并发访问，只能少返回几条
+        3,
+        items,
+        (items) => fetchPage(ctx, items.link)
+    );
+
+    ctx.state.data = {
+        title: '北极星环保 - 环保行业垂直门户网站',
+        link: listURL,
+        item: items,
+    };
+};
+
+const fetchPage = (ctx, link) =>
+    ctx.cache.tryGet(link, async () => {
+        // 可能一篇文章过长会分成多页
+        const pages = [];
+
+        const result = await got(link);
+        const $page = cheerio.load(result.data);
+        pages.push($page);
+
+        // 如果是有分页链接，则使用顺序加载以保证顺序
+        const pagelinks = $page('#article_cont .cc-paging a');
+
+        if (pagelinks.length > 0) {
+            for (let i = 0; i < pagelinks.length; i++) {
+                const $a = $page(pagelinks[i]);
+                if (!/^\d+$/.test($a.text().trim())) {
+                    continue;
+                }
+                const sublink = new URL($a.attr('href'), link).href;
+                /* eslint-disable no-await-in-loop */
+                const result = await got(sublink);
+                pages.push(cheerio.load(result.data));
+            }
+        }
+
+        const item = {
+            title: $page('title').text(),
+            description: pages.reduce((desc, $p) => desc + $p('.cc-article').html(), ''),
+            pubDate: timezone(parseDate($page('.cc-headline .box p span').eq(0).text()), +8),
+            link,
+            author: $page('.cc-headline .box p span').eq(1).text(),
+        };
+        return item;
+    });

lib/v2/bjx/maintainer.js

@@ -0,0 +1,4 @@
+module.exports = {
+    '/gf/:type': ['Sxuet'],
+    '/huanbao': ['zsimple'],
+};

lib/v2/bjx/radar.js

@@ -0,0 +1,21 @@
+module.exports = {
+    'bjx.com.cn': {
+        _name: '北极星电力网',
+        guangfu: [
+            {
+                title: '光伏 - 分类',
+                docs: 'https://docs.rsshub.app/traditional-media.html#bei-ji-xing-dian-li-wang',
+                source: ['/:type', '/'],
+                target: '/bjx/gf/:type?',
+            },
+        ],
+        huanbao: [
+            {
+                title: '环保要闻',
+                docs: 'https://docs.rsshub.app/traditional-media.html#bei-ji-xing-dian-li-wang',
+                source: ['/yw', '/'],
+                target: '/bjx/huanbao',
+            },
+        ],
+    },
+};

lib/v2/bjx/router.js

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

lib/v2/bjx/types.js

@@ -0,0 +1,33 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const type = ctx.params.type;
+    const response = await got({
+        method: 'get',
+        url: `https://guangfu.bjx.com.cn/${type}/`,
+    });
+    const data = response.data;
+    const $ = cheerio.load(data);
+    const typeName = $('div.box2 em:last').text();
+    const list = $('div.cc-list-content ul li');
+    ctx.state.data = {
+        title: `北极星太阳能光大网${typeName}`,
+        description: $('meta[name="Description"]').attr('content'),
+        link: `https://guangfu.bjx.com.cn/${type}/`,
+        item:
+            list &&
+            list
+                .map((index, item) => {
+                    item = $(item);
+                    return {
+                        title: item.find('a').attr('title'),
+                        description: item.html(),
+                        link: item.find('a').attr('href'),
+                        pubDate: parseDate(item.find('span').text()),
+                    };
+                })
+                .get(),
+    };
+};