From 3b6c1baa2869fe7d77553966e2a17b7d2eba5683 Mon Sep 17 00:00:00 2001
From: Martin Richtarsky <s@martinien.de>
Date: Tue, 13 Feb 2024 16:16:24 +0100
Subject: [PATCH] Look for class="article" for content

Quite a few sites use this, but fail the current
heuristic.
---
 src/utils/get-core-content-text.mjs | 15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)

diff --git a/src/utils/get-core-content-text.mjs b/src/utils/get-core-content-text.mjs
index f179bde8..bef4dc8f 100644
--- a/src/utils/get-core-content-text.mjs
+++ b/src/utils/get-core-content-text.mjs
@@ -43,23 +43,28 @@ function findLargestElement(e) {
 }
 
 export function getCoreContentText() {
+  function getTextFrom(e) {
+    return e.innerText || e.textContent
+  }
+
   for (const [siteName, selectors] of Object.entries(adapters)) {
     if (location.hostname.includes(siteName)) {
       const element = getPossibleElementByQuerySelector(selectors)
-      if (element) return element.innerText || element.textContent
+      if (element) return getTextFrom(element)
       break
     }
   }
 
+  const element = document.querySelector('article')
+  if (element) {
+    return getTextFrom(element)
+  }
+
   const largestElement = findLargestElement(document.body)
   const secondLargestElement = findLargestElement(largestElement)
   console.log(largestElement)
   console.log(secondLargestElement)
 
-  function getTextFrom(e) {
-    return e.innerText || e.textContent
-  }
-
   let ret
   if (!largestElement) {
     ret = getTextFrom(document.body)