feat(route): Update route OpenAI (#12113)

* Updated route OpenAI

* style: auto format

* Improved route OpenAI

* Use cheerio to remove comments

* style: auto format

* Update lib/v2/openai/radar.js

---------

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

docs/en/new-media.md

@@ -554,11 +554,11 @@ This route provides a flexible plan with full text content to subscribe specific
 
 ### Blog
 
-<RouteEn author="ncziztk" example="/openai/blog" path="/openai/blog/:tag" :paramsDesc="['Tag, see below, All by default']">
+<RouteEn author="ncziztk StevenRCE0" example="/openai/blog" path="/openai/blog/:tag?" :paramsDesc="['Tag, see below, All by default']">
 
-| All | Research | Announcements | Events | Milestones |
-| --- | -------- | ------------- | ------ | ---------- |
-|     | research | announcements | events | milestones |
+| All | Announcements | Events | Safety & Alignment | Community | Product | Culture & Careers   | Milestones | Research |
+| --- | ------------- | ------ | ------------------ | --------- | ------- | ------------------- | ---------- | -------- |
+|     | announcements | events | safety-alignment   | community | product | culture-and-careers | milestones | research |
 
 </RouteEn>
 

docs/new-media.md

@@ -1024,11 +1024,11 @@ IPFS 网关有可能失效，那时候换成其他网关。
 
 ### Blog
 
-<Route author="ncziztk" example="/openai/blog" path="/openai/blog/:tag" :paramsDesc="['标签，见下表，默认为 All']">
+<Route author="ncziztk StevenRCE0" 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 |
+| All | Announcements | Events | Safety & Alignment | Community | Product | Culture & Careers   | Milestones | Research |
+| --- | ------------- | ------ | ------------------ | --------- | ------- | ------------------- | ---------- | -------- |
+|     | announcements | events | safety-alignment   | community | product | culture-and-careers | milestones | research |
 
 </Route>
 

lib/router.js

@@ -3419,9 +3419,6 @@ router.get('/blizzard/news/:language?/:category?', lazyloadRouteHandler('./route
 // DeepL
 router.get('/deepl/blog/:lang?', lazyloadRouteHandler('./routes/deepl/blog'));
 
-// OpenAI
-router.get('/openai/blog/:tag?', lazyloadRouteHandler('./routes/openai/blog'));
-
 // 小木虫
 router.get('/muchong/journal/:type?', lazyloadRouteHandler('./routes/muchong/journal'));
 router.get('/muchong/:id/:type?/:sort?', lazyloadRouteHandler('./routes/muchong/index'));

lib/routes/openai/blog.js

@@ -1,52 +0,0 @@
-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((item) =>
-            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,
-    };
-};

lib/v2/openai/blog.js

@@ -0,0 +1,97 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { art } = require('@/utils/render');
+const { toTitleCase } = require('@/utils/common-utils');
+const path = require('path');
+
+module.exports = async (ctx) => {
+    const tag = ctx.params.tag || '';
+
+    const rootUrl = 'https://openai.com';
+    const blogRootUrl = 'https://openai.com/blog';
+    const blogOriginUrl = `${rootUrl}/blog${tag === '' ? '' : `?topics=${tag}`}`;
+
+    // Find API base URL
+    const initResponse = await got({
+        method: 'get',
+        url: blogRootUrl,
+    });
+
+    let apiBaseUrl = initResponse.data
+        .toString()
+        .match(/(?<=TWILL_API_BASE:").+?(?=")/)[0]
+        .replaceAll('\\u002F', '/');
+    apiBaseUrl = apiBaseUrl + '/api/v1/blog-details';
+    const apiUrl = new URL(apiBaseUrl);
+
+    // Construct API query
+    apiUrl.searchParams.append('sort', '-publicationDate,-createdAt');
+    apiUrl.searchParams.append('page[size]', '20');
+    apiUrl.searchParams.append('page[number]', '1');
+    apiUrl.searchParams.append('include', 'media,topics,authors');
+    if (tag) {
+        apiUrl.searchParams.append('filter[topics][slugs][0]', tag);
+    }
+
+    const response = await got({
+        method: 'get',
+        url: apiUrl,
+    });
+
+    const list = response.data.data.filter((entry) => entry.type === 'blog-details');
+
+    const items = await Promise.all(
+        list.map((item) => {
+            const attributes = item.attributes;
+            const textUrl = `${blogRootUrl}/${attributes.slug}`;
+            return ctx.cache.tryGet(attributes.slug, async () => {
+                const detailResponse = await got({
+                    method: 'get',
+                    url: textUrl,
+                });
+                let content = cheerio.load(detailResponse.data);
+
+                const authors = content('[aria-labelledby="metaAuthorsHeading"] > li > a > span > span')
+                    .toArray()
+                    .map((entry) => content(entry).text())
+                    .join(', ');
+
+                // Leave out comments
+                const comments = content('*')
+                    .contents()
+                    .filter(function () {
+                        return this.nodeType === 8;
+                    });
+                comments.remove();
+
+                content = content('#content');
+
+                const imageSrc = attributes.seo.ogImageSrc;
+                const imageAlt = attributes.seo.ogImageAlt;
+
+                const article = art(path.join(__dirname, 'templates/article.art'), {
+                    content,
+                    imageSrc,
+                    imageAlt,
+                });
+
+                return {
+                    title: attributes.title,
+                    author: authors,
+                    description: article,
+                    pubDate: attributes.createdAt,
+                    category: attributes.tags.map((tag) => tag.title),
+                    link: textUrl,
+                };
+            });
+        })
+    );
+
+    const title = `OpenAI Blog${tag ? ` - ${toTitleCase(tag)}` : ''}`;
+
+    ctx.state.data = {
+        title,
+        link: blogOriginUrl,
+        item: items,
+    };
+};

lib/v2/openai/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/blog/:tag?': ['StevenRCE0', 'nczitzk'],
+};

lib/v2/openai/radar.js

@@ -0,0 +1,19 @@
+module.exports = {
+    'openai.com': {
+        _name: 'OpenAI',
+        '.': [
+            {
+                title: 'Blog',
+                docs: 'https://docs.rsshub.app/en/new-media.html#openai',
+                source: '/blog',
+                target: (_, url) => {
+                    const topics = new URL(url).searchParams.get('topics');
+                    if (topics) {
+                        return `/openai/blog/${topics}`;
+                    }
+                    return '/openai/blog';
+                },
+            },
+        ],
+    },
+};

lib/v2/openai/router.js

@@ -0,0 +1,3 @@
+module.exports = function (router) {
+    router.get('/blog/:tag?', require('./blog'));
+};

lib/v2/openai/templates/article.art

@@ -0,0 +1,2 @@
+<img src="{{ imageSrc }}" alt="{{ imageAlt }}"/>
+{{@ content }}