feat: add openai blog (#6301)

docs/en/new-media.md

@@ -147,6 +147,18 @@ This route provides a flexible plan with full text content to subscribe specific
 
 </RouteEn>
 
+## OpenAI
+
+### Blog
+
+<RouteEm author="ncziztk" example="/openai/blog" path="/openai/blog/:tag" :paramsDesc="['Tag, see below, All by default']">
+
+| All | Research | Announcements | Events | Milestones |
+| --- | -------- | ------------- | ------ | ---------- |
+|     | research | announcements | events | milestones |
+
+</RouteEn>
+
 ## Quanta Magazine
 
 ### Archive

docs/new-media.md

@@ -387,6 +387,18 @@ IPFS 网关有可能失效，那时候换成其他网关。
 
 <Route author="emdoe" example="/nautilus/topic/Art" path="/nautilus/topic/:tid" :paramsDesc="['话题 id, 可在页面上方 TOPICS 栏目处找到']"/>
 
+## OpenAI
+
+### Blog
+
+<Route author="ncziztk" example="/openai/blog" path="/openai/blog/:tag" :paramsDesc="['标签，见下表，默认为 All']">
+
+| All | Research | Announcements | Events | Milestones |
+| --- | -------- | ------------- | ------ | ---------- |
+|     | research | announcements | events | milestones |
+
+</Route>
+
 ## PMCAFF
 
 ### 今日推荐 / 精选

lib/router.js

@@ -3516,6 +3516,9 @@ router.get('/uisdc/news', require('./routes/uisdc/news'));
 router.get('/uisdc/zt/:title?', require('./routes/uisdc/zt'));
 router.get('/uisdc/topic/:title?/:sort?', require('./routes/uisdc/topic'));
 
+// OpenAI
+router.get('/openai/blog/:tag?', require('./routes/openai/blog'));
+
 // 小木虫
 router.get('/muchong/journal/:type?', require('./routes/muchong/journal'));
 router.get('/muchong/:id/:type?/:sort?', require('./routes/muchong/index'));

lib/routes/openai/blog.js

@@ -0,0 +1,53 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+
+module.exports = async (ctx) => {
+    const tag = ctx.params.tag || '';
+
+    const rootUrl = 'https://openai.com';
+    const currentUrl = `${rootUrl}/blog${tag === '' ? '' : `/tags/${tag}`}`;
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = cheerio.load(response.data);
+
+    const list = $('.col-12 a')
+        .slice(0, 5)
+        .map((_, item) => {
+            item = $(item);
+            return {
+                link: `${rootUrl}${item.attr('href')}`,
+            };
+        })
+        .get();
+
+    const items = await Promise.all(
+        list.map(
+            async (item) =>
+                await ctx.cache.tryGet(item.link, async () => {
+                    const detailResponse = await got({
+                        method: 'get',
+                        url: item.link,
+                    });
+                    const content = cheerio.load(detailResponse.data);
+
+                    item.title = content('title').text();
+
+                    content('.aside').remove();
+
+                    item.description = content('.content').html();
+                    item.pubDate = new Date(content('time').attr('datetime')).toUTCString();
+
+                    return item;
+                })
+        )
+    );
+
+    ctx.state.data = {
+        title: $('title').text(),
+        link: currentUrl,
+        item: items,
+    };
+};