fix(modb): cache article bodies with bounded detail requests

lib/routes/modb/topic.ts

@@ -1,5 +1,6 @@
 import { load } from 'cheerio';
 import { escapeUTF8 } from 'entities/escape';
+import pMap from 'p-map';
 
 import type { DataItem, Route } from '@/types';
 import cache from '@/utils/cache';
@@ -34,6 +35,20 @@ interface TopicDoc {
     tags?: string[];
 }
 
+class ArticleBodyUnavailableError extends Error {}
+
+const getArticleDescription = (link: string) =>
+    cache.tryGet(`modb:topic:article-body:v1:${link}`, async () => {
+        const response = await ofetch(link);
+        const $ = load(response);
+        const article = $('div.editor-content-styl.article-style').first();
+        article.find('script, style').remove();
+        if (!article.text().trim() && !article.find('img, video, audio, iframe, table').length) {
+            throw new ArticleBodyUnavailableError(`The MODB article does not expose a public body: ${link}`);
+        }
+        return { description: article.html()! };
+    });
+
 async function handler(ctx) {
     const baseUrl = 'https://www.modb.pro';
     const topicId = ctx.req.param('id');
@@ -44,7 +59,7 @@ async function handler(ctx) {
             columnId: topicId,
         },
     });
-    const list = response.list.flatMap((item): Array<DataItem & { link: string }> => {
+    const list: Array<DataItem & { link: string }> = response.list.flatMap((item): Array<DataItem & { link: string }> => {
         const doc: TopicDoc | undefined = item.type === 0 ? item.knowledge : item.type === 1 ? item.dbDoc : undefined;
         if (!doc?.title || item.rid === undefined || item.rid === null) {
             logger.error(`Unsupported or incomplete topic entry of type ${item.type}`);
@@ -64,17 +79,23 @@ async function handler(ctx) {
         ];
     });
 
-    const items = await Promise.all(
-        list.map((item) =>
-            item.description === undefined
-                ? cache.tryGet(item.link, async () => {
-                      const response = await ofetch(item.link);
-                      const $ = load(response);
-                      item.description = $('div.editor-content-styl.article-style').first().html();
-                      return item;
-                  })
-                : item
-        )
+    const items = await pMap(
+        list,
+        async (item) => {
+            if (item.description !== undefined) {
+                return item;
+            }
+            try {
+                const details = await getArticleDescription(item.link);
+                return { ...item, ...details };
+            } catch (error) {
+                if (error instanceof ArticleBodyUnavailableError) {
+                    return item;
+                }
+                throw error;
+            }
+        },
+        { concurrency: 3 }
     );
 
     return {