From d449cb8babd3ec3f7263a9a1dc039d4876ca1d4c Mon Sep 17 00:00:00 2001
From: sigoden <sigoden@gmail.com>
Date: Wed, 17 Jul 2024 10:25:38 +0000
Subject: [PATCH] refactor: improve fetch_url_via_curl

---
 tools/fetch_url_via_curl.sh | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/tools/fetch_url_via_curl.sh b/tools/fetch_url_via_curl.sh
index 2f2efdc..540984e 100755
--- a/tools/fetch_url_via_curl.sh
+++ b/tools/fetch_url_via_curl.sh
@@ -8,8 +8,8 @@ set -e
 main() {
     # span and div tags are dropped from the HTML https://pandoc.org/MANUAL.html#raw-htmltex and sed removes any inline SVG images in image tags from the Markdown content.
     curl -fsSL "$argc_url" | \
-        pandoc -f html-native_divs-native_spans -t gfm-raw_html | \
-        sed -E 's/!\[.*?\]\((data:image\/svg\+xml[^)]+)\)//g' \
+        pandoc -f html-native_divs-native_spans -t gfm-raw_html --wrap=none | \
+        sed -E 's/!\[[^]]*\]\([^)]*\)//g' \
         >> "$LLM_OUTPUT"
 }