fix(route): use McKinsey official insights RSS

Author: 1121eason

lib/routes/mckinsey/insights.ts

@@ -1,106 +1,53 @@
 import { load } from 'cheerio';
 
 import type { DataItem, Route } from '@/types';
-import cache from '@/utils/cache';
 import ofetch from '@/utils/ofetch';
 import { parseDate } from '@/utils/parse-date';
 
-const baseUrl = 'https://www.mckinsey.com';
-const link = `${baseUrl}/featured-insights`;
-
-const headers = {
-    'accept-language': 'en-US,en;q=0.9',
-    'user-agent': 'Mozilla/5.0 (compatible; RSSHub; +https://github.com/DIYgod/RSSHub)',
-};
-
-const datePattern = /(?:January|February|March|April|May|June|July|August|September|October|November|December)\s+\d{1,2},\s+\d{4}/;
-
-function normalizeText(value?: string | null): string {
-    return value?.replace(/\s+/g, ' ').trim() ?? '';
-}
-
-function isInsightUrl(url: URL): boolean {
-    if (url.hostname !== 'www.mckinsey.com') {
-        return false;
-    }
-
-    const pathname = url.pathname.toLowerCase();
-
-    if (pathname === '/featured-insights' || pathname === '/our-insights' || pathname === '/insights') {
-        return false;
-    }
-
-    return [
-        '/featured-insights/',
-        '/our-insights/',
-        '/industries/',
-        '/capabilities/',
-        '/quarterly/',
-        '/mgi/',
-        '/institute-for-economic-mobility/',
-    ].some((prefix) => pathname.startsWith(prefix) && !pathname.endsWith('/our-insights'));
-}
-
-function getContainerText($, element): string {
-    const containers = $(element)
-        .parents('article, li, div, section')
-        .toArray()
-        .map((container) => normalizeText($(container).text()))
-        .filter((text) => text.length >= 40 && text.length <= 1600);
-
-    return containers[0] ?? normalizeText($(element).parent().text());
-}
+const link = 'https://www.mckinsey.com/insights/rss';
+const siteUrl = 'https://www.mckinsey.com/insights';
 
 const handler: Route['handler'] = async () => {
-    const data = await ofetch(link, { headers });
-    const $ = load(data);
-    const seen = new Set<string>();
-
-    const item = $('a[href]')
+    const data = await ofetch(link, {
+        headers: {
+            accept: 'application/rss+xml, application/xml, text/xml;q=0.9, */*;q=0.8',
+            'accept-language': 'en-US,en;q=0.9',
+            'user-agent': 'Mozilla/5.0 (compatible; RSSHub; +https://github.com/DIYgod/RSSHub)',
+        },
+        retry: 1,
+        timeout: 15000,
+    });
+
+    const $ = load(data, { xmlMode: true });
+
+    const item = $('item')
         .toArray()
-        .map((element) => {
-            const $element = $(element);
-            const rawHref = $element.attr('href');
-            const title = normalizeText($element.attr('title') || $element.text());
-
-            if (!rawHref || title.length < 12 || /^more\b/i.test(title) || /^download\b/i.test(title)) {
-                return;
-            }
-
-            const itemUrl = new URL(rawHref, baseUrl);
-            itemUrl.hash = '';
-
-            if (!isInsightUrl(itemUrl) || seen.has(itemUrl.href)) {
-                return;
-            }
-
-            seen.add(itemUrl.href);
-
-            const containerText = getContainerText($, element);
-            const date = containerText.match(datePattern)?.[0];
-            const description = normalizeText(containerText.replace(title, '').replace(date ?? '', '')).slice(0, 800);
+        .map((entry) => {
+            const $entry = $(entry);
+            const categories = $entry
+                .find('category')
+                .toArray()
+                .map((category) => $(category).text().trim())
+                .filter(Boolean);
 
             return {
-                title,
-                link: itemUrl.href,
-                description: description || title,
-                pubDate: date ? parseDate(date) : undefined,
+                title: $entry.find('title').first().text().trim(),
+                link: $entry.find('link').first().text().trim(),
+                description: $entry.find('description').first().text().trim(),
+                pubDate: parseDate($entry.find('pubDate').first().text().trim()),
+                guid: $entry.find('guid').first().text().trim(),
+                category: categories,
             };
         })
-        .filter(Boolean)
-        .slice(0, 30) as DataItem[];
-
-    const cachedItems = await Promise.all(
-        item.map((entry) =>
-            cache.tryGet(`mckinsey:insights:${entry.link}`, async () => entry)
-        )
-    );
+        .filter((entry) => entry.title && entry.link)
+        .slice(0, 50) as DataItem[];
 
     return {
-        title: 'McKinsey Insights',
-        link,
+        title: $('channel > title').first().text().trim() || 'McKinsey Insights',
+        link: $('channel > link').first().text().trim() || siteUrl,
+        description: $('channel > description').first().text().trim(),
         language: 'en-US',
-        item: cachedItems,
+        item,
     };
 };