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" }