From df6572b659bf25cabc6a71108b8cd9b70056262c Mon Sep 17 00:00:00 2001
From: LinxHex <92703641+LinxHex@users.noreply.github.com>
Date: Fri, 13 Mar 2026 23:57:36 +0800
Subject: [PATCH] fix(routes/miyuki): format news titles with bracketed
 categories and keep full-text body-only (#21377)

* fix(routes/miyuki): format news titles with bracketed categories and keep full-text body-only

* fix(routes/miyuki): clean news body extraction and normalize images

- Updated lib/routes/miyuki/news.ts to keep item titles as plain article titles and store category information only in the `category` field.
- Switched the route to use the canonical Miyuki domain and added the route `url` metadata.
- Refined full-text extraction so `description` keeps only the main article body and excludes page header metadata such as section title, article title, date, and category.
- Fixed malformed article-body handling for pages like `10217`, where the source HTML structure breaks the original body selector.
- Normalized article images by deduplicating `for_pc` / `for_sp` duplicates, converting relative image URLs to absolute URLs, and replacing list-based photo blocks with plain containers so RSS readers do not render black bullets.
- Resolved the related CodeFactor/lint issues and verified the file with `oxlint` and `eslint`.

* feat(apple): add full-text Apple Newsroom route

Add /apple/newsroom for the official Apple Newsroom China feed.
Fetch article pages to provide full-text content with cached detail requests, and normalize article metadata and media URLs.
Remove gallery navigation, download buttons, video playback controls, empty wrapper nodes, and other non-content UI elements from articles.
Format image and video captions as distinct descriptions, and drop duplicate captions when their text matches the article body.

* fix(routes/miyuki): simplify image handling in news descriptions

* Remove unrelated changes from PR
---
 lib/routes/miyuki/news.ts | 72 ++++++++++++++++++++++++++++++++++-----
 1 file changed, 63 insertions(+), 9 deletions(-)

diff --git a/lib/routes/miyuki/news.ts b/lib/routes/miyuki/news.ts
index 7ffb135f8..43a933a74 100644
--- a/lib/routes/miyuki/news.ts
+++ b/lib/routes/miyuki/news.ts
@@ -6,13 +6,21 @@ import ofetch from '@/utils/ofetch';
 import { parseDate } from '@/utils/parse-date';
 import timezone from '@/utils/timezone';
 
-const ORIGIN = 'https://miyuki.jp';
+const ORIGIN = 'https://www.miyuki.jp';
 const NEWS_LINK = `${ORIGIN}/s/y10/news/list`;
+const DETAIL_HEADER_SELECTOR = [
+    '.pc__news_detail__title',
+    '.pc__news_detail__title__japanese',
+    '.news_detail__date',
+    '.news_detail__title',
+    '.news_detail__ganre',
+].join(', ');
 
 export const route: Route = {
     path: '/news',
     example: '/miyuki/news',
     name: 'News',
+    url: 'www.miyuki.jp/s/y10/news/list',
     categories: ['new-media'],
     maintainers: ['KarasuShin'],
     features: {
@@ -33,21 +41,18 @@ async function handler() {
     const items = await Promise.all(
         $('.list__side_border li')
             .toArray()
-            .map(async (item) => {
+            .map((item) => {
                 const $item = $(item);
                 const link = `${ORIGIN}${$item.find('a').attr('href')!}`;
-                return await cache.tryGet(link, async () => {
+                return cache.tryGet(link, async () => {
                     const category = $item.find('p span').last().text();
+                    const title = $item.find('a').text();
                     return {
-                        title: `${category} - ${$item.find('a').text()}`,
+                        title,
                         link,
                         pubDate: timezone(parseDate($item.find('p span').first().text()), +9),
                         category: [category],
-                        description: await cache.tryGet(link, async () => {
-                            const detailHtml = await ofetch(link);
-                            const $detail = load(detailHtml);
-                            return $detail('.contents_area__inner').html()!;
-                        }),
+                        description: await getDescription(link),
                     } as DataItem;
                 });
             })
@@ -59,3 +64,52 @@ async function handler() {
         item: items,
     };
 }
+
+async function getDescription(link: string) {
+    const detailHtml = await ofetch(link);
+    const $ = load(detailHtml);
+    const content = $('.contents_area__inner');
+    content.children(DETAIL_HEADER_SELECTOR).remove();
+    normalizePhotoLists($, content);
+    content.children().each((_, element) => {
+        const child = $(element);
+        if (!hasMeaningfulHtml(child.html())) {
+            child.remove();
+        }
+    });
+
+    return content.html()?.trim() ?? '';
+}
+
+function hasMeaningfulHtml(html?: string | null) {
+    return Boolean(html?.replaceAll(/<br\s*\/?>/g, '').replaceAll('&nbsp;', '').trim());
+}
+
+function normalizePhotoLists($, content) {
+    content.find('.news_detail__photo_list').each((_, element) => {
+        const list = $(element);
+        const items = list
+            .children('li')
+            .toArray()
+            .flatMap((item) => {
+                const photoItem = $(item);
+                photoItem.find('img.for_sp').remove();
+                photoItem.find('img').each((__, image) => {
+                    const img = $(image);
+                    const src = img.attr('src');
+                    if (!src) {
+                        img.remove();
+                        return;
+                    }
+
+                    img.attr('src', new URL(src, ORIGIN).href);
+                    img.removeAttr('class');
+                });
+
+                const html = photoItem.html();
+                return hasMeaningfulHtml(html) ? [`<div>${html!.trim()}</div>`] : [];
+            });
+
+        list.replaceWith(items.join('<br /><br />'));
+    });
+}