feat(route/apnews): Support rss parsing. (#15440)

* feat(route/apnews): Support rss parsing. * Update lib/routes/apnews/topics.ts Co-authored-by: Tony <TonyRL@users.noreply.github.com> * Update lib/routes/apnews/rss.ts Co-authored-by: Tony <TonyRL@users.noreply.github.com> * Update rss.ts ---------
DIYgod · May 3, 2024 · 1756242 · 1756242
1 parent 98ceba2
commit 1756242
Show file tree

Hide file tree

Showing 3 changed files with 63 additions and 16 deletions.
diff --git a/lib/routes/apnews/rss.ts b/lib/routes/apnews/rss.ts
@@ -0,0 +1,41 @@
+import { Route } from '@/types';
+import parser from '@/utils/rss-parser';
+import { fetchArticle } from './utils';
+const HOME_PAGE = 'https://apnews.com';
+
+export const route: Route = {
+    path: '/rss/:rss?',
+    categories: ['traditional-media'],
+    example: '/apnews/rss/business',
+    parameters: { rss: 'Route name from the first segment of the corresponding site, or `index` for the front page(default).' },
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['apnews.com/:rss'],
+            target: '/rss/:rss',
+        },
+    ],
+    name: 'RSS',
+    maintainers: ['zoenglinghou', 'mjysci', 'TonyRL'],
+    handler,
+};
+
+async function handler(ctx) {
+    const { rss = 'index' } = ctx.req.param();
+    const url = `${HOME_PAGE}/${rss}.rss`;
+    const res = await parser.parseURL(url);
+
+    const items = await Promise.all(res.items.map((item) => fetchArticle(item)));
+
+    return {
+        ...rss,
+        item: items,
+    };
+}
diff --git a/lib/routes/apnews/topics.ts b/lib/routes/apnews/topics.ts
@@ -1,8 +1,7 @@
 import { Route } from '@/types';
-import cache from '@/utils/cache';
 import got from '@/utils/got';
 import { load } from 'cheerio';
-import timezone from '@/utils/timezone';
+import { fetchArticle } from './utils';
 const HOME_PAGE = 'https://apnews.com';
 
 export const route: Route = {
@@ -44,20 +43,7 @@ async function handler(ctx) {
                 link: $(e).find('a').attr('href'),
             }))
             .filter((e) => typeof e.link === 'string')
-            .map((item) =>
-                cache.tryGet(item.link, async () => {
-                    const { data: response } = await got(item.link);
-                    const $ = load(response);
-                    $('div.Enhancement').remove();
-                    return Object.assign(item, {
-                        pubDate: timezone(new Date($("meta[property='article:published_time']").attr('content')), 0),
-                        updated: timezone(new Date($("meta[property='article:modified_time']").attr('content')), 0),
-                        description: $('div.RichTextStoryBody').html(),
-                        category: $("meta[property='article:section']").attr('content'),
-                        guid: $("meta[name='brightspot.contentId']").attr('content'),
-                    });
-                })
-            )
+            .map((item) => fetchArticle(item))
     );
 
     return {

diff --git a/lib/routes/apnews/utils.ts b/lib/routes/apnews/utils.ts
@@ -0,0 +1,20 @@
+import cache from '@/utils/cache';
+import ofetch from '@/utils/ofetch';
+import { parseDate } from '@/utils/parse-date';
+import timezone from '@/utils/timezone';
+import { load } from 'cheerio';
+
+export function fetchArticle(item) {
+    return cache.tryGet(item.link, async () => {
+        const data = await ofetch(item.link);
+        const $ = load(data);
+        $('div.Enhancement').remove();
+        return Object.assign(item, {
+            pubDate: timezone(parseDate($("meta[property='article:published_time']").attr('content')), 0),
+            updated: timezone(parseDate($("meta[property='article:modified_time']").attr('content')), 0),
+            description: $('div.RichTextStoryBody').html(),
+            category: $("meta[property='article:section']").attr('content'),
+            guid: $("meta[name='brightspot.contentId']").attr('content'),
+        });
+    });
+}