增加 深圳大学研究生招生通知 (#2019)

* 增加 深圳大学研究生招生通知

* 对全文内容使用缓存

* code prettified

* fix cache implementation

* remove duplicate logging output

* update university.md

docs/university.md

@@ -709,6 +709,16 @@ https://rsshub.app/**nuist**/`bulletin` 或 https://rsshub.app/**nuist**/`bullet
 
 <Route name="广东海洋大学" author="Xiaotouming" example="/gdoujwc" path="/gdoujwc"/>
 
+## 深圳大学
+
+<Route name="深圳大学研究生招生网" author="NagaruZ" example="/szu/yz/1" path="/universities/szu/yz/:type?" :paramsDesc="['默认为1']" >
+
+| 研究生 | 博士生 |
+| ------ | ------ |
+| 1      | 2      |
+
+</Route>
+
 ## MIT
 
 <Route name="MIT graduateadmissions's all blogs" author="LogicJake" example="/mit/graduateadmissions/index/all" path="/universities/mit/graduateadmissions/index/all"/>

lib/router.js

@@ -742,6 +742,9 @@ router.get('/shu/jwc/:type?', require('./routes/universities/shu/jwc'));
 // 北京科技大学天津学院
 router.get('/ustb/tj/news/:type?', require('./routes/universities/ustb/tj/news'));
 
+// 深圳大学
+router.get('/szu/yz/:type?', require('./routes/universities/szu/yz'));
+
 // ifanr
 router.get('/ifanr/:channel?', require('./routes/ifanr/index'));
 

lib/routes/universities/szu/yz/index.js

@@ -0,0 +1,50 @@
+const axios = require('../../../../utils/axios');
+const cheerio = require('cheerio');
+const util = require('./utils');
+
+const map = new Map([[1, { title: '硕士招生 - 深圳大学研究生招生网' }], [2, { title: '博士招生 - 深圳大学研究生招生网' }]]);
+
+module.exports = async (ctx) => {
+    let type = Number.parseInt(ctx.params.type);
+    const struct = {
+        1: {
+            selector: {
+                list: '.list',
+                item: 'li',
+                content: '#vsb_content',
+            },
+            url: 'https://yz.szu.edu.cn/sszs/gg.htm',
+        },
+        2: {
+            selector: {
+                list: '.list',
+                item: 'li',
+                content: '#vsb_content',
+            },
+            url: 'https://yz.szu.edu.cn/bszs/gg.htm',
+        },
+    };
+    if (type !== 1 && type !== 2) {
+        // fallback to default
+        type = 1;
+    }
+    const url = struct[type].url;
+
+    const response = await axios.get(url);
+    const data = response.data;
+
+    const $ = cheerio.load(data);
+    const list = $(struct[type].selector.list)
+        .find(struct[type].selector.item)
+        .get();
+
+    const name = $('title').text();
+    const result = await util.ProcessFeed(list, ctx.cache, struct[type]);
+
+    ctx.state.data = {
+        title: map.get(type).title,
+        link: `${url}`,
+        description: `${name}`,
+        item: result,
+    };
+};

lib/routes/universities/szu/yz/utils.js

@@ -0,0 +1,89 @@
+const cheerio = require('cheerio');
+const axios = require('../../../../utils/axios');
+const url = require('url');
+
+const ProcessFeed = async (list, cache, current) =>
+    await Promise.all(
+        list
+            .filter(function(item) {
+                // 如果不包含链接说明不是新闻item，如表头的tr
+                const $ = cheerio.load(item);
+                if ($('a').length > 0) {
+                    return true;
+                }
+                return false;
+
+                // return typeof ($('a').attr('href')) !== undefined;
+                // return false;
+            })
+            .map(async (item) => {
+                let $ = cheerio.load(item);
+                const $url = url.resolve(current.url, $('a').attr('href'));
+                const key = $url;
+                // 检查缓存中是否存在该页面
+
+                const value = await cache.get(key);
+                if (value) {
+                    // 查询返回未过期缓存
+                    return JSON.parse(value);
+                } else {
+                    // 加载新闻内容页面
+                    const response = await axios.get($url);
+
+                    const data = response.data;
+                    $ = cheerio.load(data); // 使用 cheerio 加载返回的 HTML
+
+                    // 还原图片地址
+                    $('img').each((index, elem) => {
+                        const $elem = $(elem);
+                        const src = $elem.attr('src');
+                        if (src && src !== '') {
+                            $elem.attr('src', url.resolve(current.url, src));
+                        }
+                    });
+
+                    // 还原链接地址
+                    $('a').each((index, elem) => {
+                        const $elem = $(elem);
+                        const src = $elem.attr('href');
+                        if (src && src !== '') {
+                            $elem.attr('href', url.resolve(current.url, src));
+                        }
+                    });
+
+                    // 去除样式
+                    $('img').removeAttr('style');
+                    $('div').removeAttr('style');
+                    $('span').removeAttr('style');
+                    $('p').removeAttr('style');
+                    $('table').removeAttr('style');
+                    $('td').removeAttr('style');
+                    $('tr').removeAttr('style');
+                    $('style').remove();
+                    $('script').remove();
+
+                    const title = $('h2').text();
+
+                    const single = {
+                        title: title,
+                        description: $(current.selector.content).html(),
+                        link: $url,
+                        pubDate: new Date(
+                            $('div.ny_fbt')
+                                .text()
+                                .substr(6, 16)
+                        ).toUTCString(), // 混有发表时间和点击量，取出时间
+                        author: '深圳大学研究生招生网',
+                        guid: $url, // 文章唯一标识
+                    };
+                    // 将内容写入缓存
+                    cache.set(key, JSON.stringify(single), 24 * 60 * 60); // 缓存时间为24h
+                    // 返回列表上提取到的信息
+                    return single;
+                }
+            })
+    );
+
+module.exports = {
+    ProcessFeed,
+};