fix(route): deeplearning (#16868)

* fix(route): deeplearning

* fix: url cleaning

lib/routes/deeplearning/thebatch.ts

@@ -1,6 +1,8 @@
 import { Route } from '@/types';
 import cache from '@/utils/cache';
-import got from '@/utils/got';
+import ofetch from '@/utils/ofetch';
+import * as cheerio from 'cheerio';
+import { parseDate } from '@/utils/parse-date';
 
 export const route: Route = {
     path: '/thebatch',
@@ -27,33 +29,43 @@ export const route: Route = {
 };
 
 async function handler() {
-    const page = await got({
-        method: 'get',
-        url: `https://www.deeplearning.ai/the-batch/`,
-    });
-    const nextJs = page.data.match(/<script id="__NEXT_DATA__" type="application\/json">(.*)<\/script>/)[1];
+    const baseUrl = 'https://www.deeplearning.ai';
+    const link = `${baseUrl}/the-batch/`;
+    const page = await ofetch(link);
+    const $ = cheerio.load(page);
+    const nextJs = $('script#__NEXT_DATA__').text();
     const nextBuildId = JSON.parse(nextJs).buildId;
 
-    const listing = await got({
-        method: 'get',
-        url: `https://www.deeplearning.ai/_next/data/${nextBuildId}/the-batch.json`,
-    });
+    const listing = await ofetch(`${baseUrl}/_next/data/${nextBuildId}/the-batch.json`);
 
-    const items = listing.data.pageProps.posts.map((item) => ({
+    const items = listing.pageProps.posts.map((item) => ({
         title: item.title,
-        link: `https://www.deeplearning.ai/the-batch/${item.slug}`,
-        jsonUrl: `https://www.deeplearning.ai/_next/data/${nextBuildId}/the-batch/${item.slug}.json`,
-        pubDate: new Date(item.published_at).toUTCString(),
+        link: `${link}${item.slug}`,
+        jsonUrl: `${baseUrl}/_next/data/${nextBuildId}/the-batch/${item.slug}.json`,
+        pubDate: parseDate(item.published_at),
     }));
 
     return {
-        title: `The Batch - a new weekly newsletter from deeplearning.ai`,
-        link: `https://www.deeplearning.ai/the-batch/`,
+        title: 'The Batch - a new weekly newsletter from deeplearning.ai',
+        link,
         item: await Promise.all(
             items.map((item) =>
                 cache.tryGet(item.link, async () => {
-                    const resp = await got({ method: 'get', url: item.jsonUrl });
-                    item.description = resp.data.pageProps.cmsData.post.html;
+                    const resp = await ofetch(item.jsonUrl);
+                    const $ = cheerio.load(resp.pageProps.cmsData.post.html);
+
+                    $('a').each((_, ele) => {
+                        if (ele.attribs.href?.includes('utm_campaign')) {
+                            const url = new URL(ele.attribs.href);
+                            url.searchParams.delete('utm_campaign');
+                            url.searchParams.delete('utm_source');
+                            url.searchParams.delete('utm_medium');
+                            url.searchParams.delete('_hsenc');
+                            ele.attribs.href = url.href;
+                        }
+                    });
+
+                    item.description = $.html();
                     return item;
                 })
             )