From 94aead998809ad40dc1d50963cefb15af16d35bf Mon Sep 17 00:00:00 2001
From: Hongbo Wu <hongbo@omnivore.app>
Date: Tue, 16 May 2023 14:46:44 +0800
Subject: [PATCH] Add content handler for weixin.qq.com

---
 packages/content-handler/src/index.ts         |  2 ++
 .../src/websites/weixin-qq-handler.ts         | 32 +++++++++++++++++++
 2 files changed, 34 insertions(+)
 create mode 100644 packages/content-handler/src/websites/weixin-qq-handler.ts

diff --git a/packages/content-handler/src/index.ts b/packages/content-handler/src/index.ts
index 2745ee8ba..c8f5d7baf 100644
--- a/packages/content-handler/src/index.ts
+++ b/packages/content-handler/src/index.ts
@@ -33,6 +33,7 @@ import { ScrapingBeeHandler } from './websites/scrapingBee-handler'
 import { StackOverflowHandler } from './websites/stack-overflow-handler'
 import { TDotCoHandler } from './websites/t-dot-co-handler'
 import { TwitterHandler } from './websites/twitter-handler'
+import { WeixinQqHandler } from './websites/weixin-qq-handler'
 import { WikipediaHandler } from './websites/wikipedia-handler'
 import { YoutubeHandler } from './websites/youtube-handler'
 
@@ -75,6 +76,7 @@ const contentHandlers: ContentHandler[] = [
   new StackOverflowHandler(),
   new EnergyWorldHandler(),
   new PipedVideoHandler(),
+  new WeixinQqHandler(),
 ]
 
 const newsletterHandlers: ContentHandler[] = [
diff --git a/packages/content-handler/src/websites/weixin-qq-handler.ts b/packages/content-handler/src/websites/weixin-qq-handler.ts
new file mode 100644
index 000000000..feff90955
--- /dev/null
+++ b/packages/content-handler/src/websites/weixin-qq-handler.ts
@@ -0,0 +1,32 @@
+import { ContentHandler } from '../content-handler'
+
+export class WeixinQqHandler extends ContentHandler {
+  constructor() {
+    super()
+    this.name = 'Weixin QQ'
+  }
+
+  shouldPreParse(url: string, dom: Document): boolean {
+    return new URL(url).hostname.endsWith('weixin.qq.com')
+  }
+
+  async preParse(url: string, dom: Document): Promise<Document> {
+    // This replace the class name of the article info to preserve the block
+    dom
+      .querySelector('.rich_media_meta_list')
+      ?.setAttribute('class', '_omnivore_rich_media_meta_list')
+
+    // This removes the title
+    dom.querySelector('.rich_media_title')?.remove()
+
+    // This removes the profile info
+    dom.querySelector('.profile_container')?.remove()
+
+    //  This removes the footer
+    dom.querySelector('#content_bottom_area')?.remove()
+    dom.querySelector('.rich_media_area_extra')?.remove()
+    dom.querySelector('#js_pc_qr_code')?.remove()
+
+    return Promise.resolve(dom)
+  }
+}