Add Google Scholar closes #363 (#463)

README.md

@@ -227,6 +227,8 @@ RSSHub 是一个轻量、易于扩展的 RSS 生成器，可以给任何奇奇
     -   资讯
 -   All the Flight Deals
     -   特价机票 Flight deals
+-   Google
+    -   谷歌学术关键词更新
 
 </details>
 

docs/README.md

@@ -1826,3 +1826,16 @@ locations: 始发地，由「国家，参见 ISO 3166-1 国家代码」和「城
 ISO 3166-1 国家代码列表请参见 [https://en.wikipedia.org/wiki/ISO_3166-1](https://en.wikipedia.org/wiki/ISO_3166-1)
 
 nearby: 可选 0 或 1，默认 0 为不包括，是否包括临近机场
+
+## Google
+
+### 谷歌学术关键词更新 <Author uid="HenryQW"/>
+
+举例: [https://rsshub.app/google/scholar/data+visualizaton」](https://rsshub.app/google/scholar/data+visualizaton)
+
+路由: `/google/scholar/:query`
+
+参数：query: 查询语句，支持「简单」和「高级」两种模式：
+
+1. 简单模式，例如「data visualizaton」，[https://rsshub.app/google/scholar/data+visualizaton](https://rsshub.app/google/scholar/data+visualizaton)。
+2. 高级模式，前往 [Google Scholar](https://scholar.google.com/schhp?hl=zh-cn&as_sdt=0,5)，点击左上角，选择高级搜索并提交查询。此时 URL 应为：[https://scholar.google.com/scholar?as_q=data+visualizaton&as_epq=&as_oq=&as_eq=&as_occt=any&as_sauthors=&as_publication=&as_ylo=2018&as_yhi=&hl=zh-CN&as_sdt=0%2C5](https://scholar.google.com/scholar?as_q=data+visualizaton&as_epq=&as_oq=&as_eq=&as_occt=any&as_sauthors=&as_publication=&as_ylo=2018&as_yhi=&hl=zh-CN&as_sdt=0%2C5)，复制`https://scholar.google.com/scholar?`后的所有语句作为本路由的查询参数。例子所对应的完整路由为[https://rsshub.app/google/scholar/as_q=data+visualizaton&as_epq=&as_oq=&as_eq=&as_occt=any&as_sauthors=&as_publication=&as_ylo=2018&as_yhi=&hl=zh-CN&as_sdt=0%2C5](https://rsshub.app/google/scholar/as_q=data+visualizaton&as_epq=&as_oq=&as_eq=&as_occt=any&as_sauthors=&as_publication=&as_ylo=2018&as_yhi=&hl=zh-CN&as_sdt=0%2C5)。

router.js

@@ -415,4 +415,7 @@ router.get('/atfd/:locations/:nearby?', require('./routes/atfd/index'));
 // Fir
 router.get('/fir/update/:id', require('./routes/fir/update'));
 
+// Google
+router.get('/google/scholar/:query', require('./routes/google/scholar'));
+
 module.exports = router;

routes/google/scholar.js

@@ -0,0 +1,58 @@
+const axios = require('../../utils/axios');
+const config = require('../../config');
+const cheerio = require('cheerio');
+
+module.exports = async (ctx) => {
+    let params = ctx.params.query;
+    let query = params;
+    let description = `Google Scholar Monitor Query: ${query}`;
+
+    if (params.indexOf('as_q=', 0) !== -1) {
+        const reg = /as_q=(.*?)&/g;
+        query = reg.exec(params)[1];
+        description = `Google Scholar Monitor Advanced Query: ${query}`;
+    } else {
+        params = 'q=' + params;
+    }
+
+    const url = `https://scholar.google.com/scholar?${params}`;
+
+    const response = await axios({
+        method: 'get',
+        url,
+        headers: {
+            'User-Agent': config.ua,
+        },
+    });
+
+    const $ = cheerio.load(response.data);
+    const list = $('#gs_res_ccl_mid .gs_r.gs_or.gs_scl .gs_ri');
+
+    const out = [];
+
+    for (let i = 0; i < list.length; i++) {
+        const $ = cheerio.load(list[i]);
+        const itemUrl = $('h3 a').attr('href');
+        const cache = await ctx.cache.get(itemUrl);
+        if (cache) {
+            out.push(JSON.parse(cache));
+            continue;
+        }
+        const single = {
+            title: $('h3 a').text(),
+            author: $('.gs_a').text(),
+            description: $('.gs_rs').text(),
+            link: itemUrl,
+            guid: itemUrl,
+        };
+
+        out.push(single);
+    }
+
+    ctx.state.data = {
+        title: `Google Scholar Monitor: ${query}`,
+        link: url,
+        description,
+        item: out,
+    };
+};