From d53db732485df4112b0eb2eb126d3fc189c58154 Mon Sep 17 00:00:00 2001
From: H Lohaus <hlohaus@users.noreply.github.com>
Date: Fri, 15 Nov 2024 18:21:22 +0100
Subject: Update internet.py

---
 g4f/gui/server/internet.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'g4f')

diff --git a/g4f/gui/server/internet.py b/g4f/gui/server/internet.py
index a2d0165b..6426f608 100644
--- a/g4f/gui/server/internet.py
+++ b/g4f/gui/server/internet.py
@@ -67,7 +67,7 @@ def scrape_text(html: str, max_words: int = None) -> str:
         if select:
             select.extract()
     clean_text = ""
-    for paragraph in soup.select("p"):
+    for paragraph in soup.select("p, h1, h2, h3, h4, h5, h6"):
         text = paragraph.get_text()
         for line in text.splitlines():
             words = []
-- 
cgit v1.2.3