feat: add 南京理工大学电光学院; fix: 南京理工大学研究生院 (DIYgod#4298)

hharzer · Apr 2, 2020 · 12162d1 · 12162d1
1 parent 23720b7
commit 12162d1
Show file tree

Hide file tree

Showing 5 changed files with 120 additions and 21 deletions.
diff --git a/docs/university.md b/docs/university.md
@@ -774,6 +774,24 @@ category 列表：
 
 </Route>
 
+### 南京理工大学电光学院
+
+<Route author="jasongzy" example="/njust/eo/17/tz" path="/njust/eo/:grade?/:type?" :paramsDesc="['年级默认为 `17`', '类别默认为 `tz`']">
+
+grade 列表：
+
+| 本科 2016 级 | 本科 2017 级 | 本科 2018 级 | 本科 2019 级 |
+| ------------ | ------------ | ------------ | ------------ |
+| 16           | 17           | 18           | 19           |
+
+type 列表：
+
+| 年级通知（通知公告） | 每日动态（主任寄语） |
+| -------------------- | -------------------- |
+| tz                   | dt                   |
+
+</Route>
+
 ## 南京林业大学
 
 ### 教务处

diff --git a/lib/router.js b/lib/router.js
@@ -663,6 +663,7 @@ router.get('/cczu/news/:category?', require('./routes/universities/cczu/news'));
 router.get('/njust/jwc/:type', require('./routes/universities/njust/jwc'));
 router.get('/njust/cwc/:type', require('./routes/universities/njust/cwc'));
 router.get('/njust/gs/:type', require('./routes/universities/njust/gs'));
+router.get('/njust/eo/:grade?/:type?', require('./routes/universities/njust/eo'));
 
 // 四川旅游学院
 router.get('/sctu/xgxy', require('./routes/universities/sctu/information-engineer-faculty/index'));

diff --git a/lib/routes/universities/njust/eo/index.js b/lib/routes/universities/njust/eo/index.js
@@ -0,0 +1,50 @@
+const date = require('@/utils/date');
+const cheerio = require('cheerio');
+const { getContent } = require('@/routes/universities/njust/eo/util');
+
+const map = new Map([
+    ['16tz', { title: '南京理工大学电光16 -- 通知公告', id: '/_t217/tzgg/list.htm' }],
+    ['16dt', { title: '南京理工大学电光16 -- 主任寄语', id: '/_t217/zrjy/list.htm' }],
+    ['17tz', { title: '南京理工大学电光17 -- 年级通知', id: '/_t689/njtz/list.htm' }],
+    ['17dt', { title: '南京理工大学电光17 -- 每日动态', id: '/_t689/mrdt/list.htm' }],
+    ['18tz', { title: '南京理工大学电光18 -- 年级通知', id: '/_t900/njtz_10234/list.htm' }],
+    ['18dt', { title: '南京理工大学电光18 -- 主任寄语', id: '/_t900/zrjy_10251/list.htm' }],
+    ['19tz', { title: '南京理工大学电光19 -- 通知公告', id: '/_t1163/tzgg_11606/list.htm' }],
+    ['19dt', { title: '南京理工大学电光19 -- 每日动态', id: '/_t1163/mrdt_11608/list.htm' }],
+]);
+
+const baseUrl = 'http://dgxg.njust.edu.cn';
+
+module.exports = async (ctx) => {
+    const grade = ctx.params.grade || '17';
+    const type = ctx.params.type || 'tz';
+    const category = grade + type;
+    const id = map.get(category).id;
+
+    const html = await getContent(baseUrl + id);
+    const $ = cheerio.load(html);
+    const list = $('li.list_item');
+
+    ctx.state.data = {
+        title: map.get(category).title,
+        link: baseUrl + '/' + id.split('/')[1] + '/main.htm',
+        item:
+            list &&
+            list
+                .map((index, item) => ({
+                    title: $(item)
+                        .find('a')
+                        .text()
+                        .trim(),
+                    pubDate: date(
+                        $(item)
+                            .find('span.Article_PublishDate')
+                            .text()
+                    ),
+                    link: $(item)
+                        .find('a')
+                        .attr('href'),
+                }))
+                .get(),
+    };
+};
diff --git a/lib/routes/universities/njust/eo/util.js b/lib/routes/universities/njust/eo/util.js
@@ -0,0 +1,28 @@
+const delay = (timeout = 1000) => new Promise((resolve) => setTimeout(resolve, timeout));
+
+async function getContent(url) {
+    const browser = await require('@/utils/puppeteer')();
+    try {
+        const page = await browser.newPage();
+        page.waitForNavigation({
+            timeout: 20000,
+        });
+        // 更改 window.navigator.webdriver 值以避开反爬
+        await page.evaluateOnNewDocument(() => {
+            // eslint-disable-next-line no-undef
+            Object.defineProperty(navigator, 'webdriver', {
+                get: () => undefined,
+            });
+        });
+        await page.goto(url);
+        await delay(1000);
+        const content = await page.content();
+        return content;
+    } finally {
+        browser.close();
+    }
+}
+
+module.exports = {
+    getContent,
+};
diff --git a/lib/routes/universities/njust/gs/index.js b/lib/routes/universities/njust/gs/index.js
@@ -1,40 +1,42 @@
-const got = require('@/utils/got');
+const date = require('@/utils/date');
 const cheerio = require('cheerio');
+const { getContent } = require('@/routes/universities/njust/eo/util');
 
 const map = new Map([
     [1, { title: '南京理工大学研究生院 -- 通知公告', id: '/sytzgg_4568' }],
     [2, { title: '南京理工大学研究生院 -- 学术公告', id: '/xshdggl' }],
 ]);
 
+const baseUrl = 'http://gs.njust.edu.cn';
+
 module.exports = async (ctx) => {
     const type = Number.parseInt(ctx.params.type);
     const id = map.get(type).id;
-    const baseUrl = 'http://gs.njust.edu.cn';
-    const res = await got({
-        method: 'get',
-        url: baseUrl + id + '/list.htm',
-        headers: {
-            Referer: 'https://gs.njust.edu.cn/',
-        },
-    });
-    const $ = cheerio.load(res.data);
-    const list = $('.wp_article_list li').slice(0, 10);
+
+    const html = await getContent(baseUrl + id + '/list.htm');
+    const $ = cheerio.load(html);
+    const list = $('li.list_item');
 
     ctx.state.data = {
         title: map.get(type).title,
-        link: 'https://gs.njust.edu.cn',
+        link: 'http://gs.njust.edu.cn',
         item:
             list &&
             list
-                .map((index, item) => {
-                    item = $(item);
-
-                    return {
-                        title: item.find('a').text(),
-                        link: `${baseUrl + $(item.find('a')).attr('href')}`,
-                        pubDate: new Date(item.find('.Article_PublishDate').text()).toUTCString(),
-                    };
-                })
+                .map((index, item) => ({
+                    title: $(item)
+                        .find('a')
+                        .text()
+                        .trim(),
+                    pubDate: date(
+                        $(item)
+                            .find('span.Article_PublishDate')
+                            .text()
+                    ),
+                    link: $(item)
+                        .find('a')
+                        .attr('href'),
+                }))
                 .get(),
     };
 };