feat: add google sites (#4119)

assets/radar-rules.js

@@ -1124,6 +1124,14 @@
                 script: "({id: document.querySelector('html').innerHTML.match(/photos.app.goo.gl\\/(.*?)\"/)[1]})",
             },
         ],
+        sites: [
+            {
+                title: 'Sites',
+                docs: 'https://docs.rsshub.app/blog.html#google-sites',
+                source: ['/site/:id/*', '/site/:id'],
+                target: '/google/sites/:id',
+            },
+        ],
     },
     'javlibrary.com': {
         _name: 'javlibrary',

docs/blog.md

@@ -10,6 +10,12 @@ pageClass: routes
 
 <Route author="kt286" example="/archdaily" path="/archdaily"/>
 
+## Google Sites
+
+### 文章更新
+
+<Route author="hoilc" example="/google/sites/outlierseconomics" path="/google/sites/:id" :paramsDesc="['Site ID, 可在 URL 中找到']" radar="1" />
+
 ## Hexo
 
 ### Next 主题博客

docs/en/blog.md

@@ -10,6 +10,12 @@ pageClass: routes
 
 <RouteEn author="kt286" example="/archdaily" path="/archdaily"/>
 
+## Google Sites
+
+### Articles
+
+<RouteEn author="hoilc" example="/google/sites/outlierseconomics" path="/google/sites/:id" :paramsDesc="['Site ID, can be found in URL']" />
+
 ## Hexo Blog
 
 ### Blog using Next theme

lib/router.js

@@ -476,6 +476,7 @@ router.get('/google/citations/:id', require('./routes/google/citations'));
 router.get('/google/scholar/:query', require('./routes/google/scholar'));
 router.get('/google/doodles/:language?', require('./routes/google/doodles'));
 router.get('/google/album/:id', require('./routes/google/album'));
+router.get('/google/sites/:id', require('./routes/google/sites'));
 
 // Awesome Pigtals
 router.get('/pigtails', require('./routes/pigtails'));

lib/routes/google/sites.js

@@ -0,0 +1,75 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+
+const base = 'https://sites.google.com/';
+
+module.exports = async (ctx) => {
+    const id = ctx.params.id;
+
+    const init_url = `https://sites.google.com/site/${id}/system/app/pages/sitemap/list?offset=0`;
+    const init_response = await got.get(init_url);
+
+    const test_pages_count = init_response.data.match(/'sites-pagination-next-link-top', (\d+),/);
+    if (!test_pages_count) {
+        throw 'Site Not Found';
+    }
+    const pages_count = parseInt(test_pages_count[1]);
+
+    const url = pages_count > 20 ? `https://sites.google.com/site/${id}/system/app/pages/sitemap/list?offset=${pages_count - 20}` : init_url;
+    const response = url === init_url ? init_response : await got.get(url);
+
+    const $ = cheerio.load(response.data);
+
+    const site_name = $('a#sites-chrome-userheader-title').text();
+    const list = $('.sites-table > tbody > tr').get();
+
+    const parseContent = async (htmlString) => {
+        const $ = cheerio.load(htmlString);
+
+        const content = $('#sites-canvas-main-content');
+
+        return {
+            description: content.html(),
+        };
+    };
+
+    const out = await Promise.all(
+        list.map(async (item, index) => {
+            const $ = cheerio.load(item);
+
+            const title = $('a');
+            const path = title.attr('href');
+            const link = base + path;
+            const time = new Date();
+            time.setMinutes(time.getMinutes() - pages_count + index);
+
+            const cache = await ctx.cache.get(link);
+            if (cache) {
+                return Promise.resolve(JSON.parse(cache));
+            }
+
+            const rssitem = {
+                title: title.text().trim(),
+                link: link,
+                pubDate: time,
+                author: site_name,
+            };
+
+            try {
+                const response = await got.get(link);
+                const result = await parseContent(response.data);
+                rssitem.description = result.description;
+            } catch (err) {
+                return Promise.resolve('');
+            }
+            ctx.cache.set(link, JSON.stringify(rssitem));
+            return Promise.resolve(rssitem);
+        })
+    );
+
+    ctx.state.data = {
+        title: `${site_name} - Google Sites`,
+        link: url,
+        item: out.filter((item) => item !== ''),
+    };
+};