refactor: improve fetch_url_via_curl (#77)

This commit is contained in:
sigoden
2024-07-17 18:29:41 +08:00
committed by GitHub
parent 2ee24731dc
commit 4d18d6346d
+2 -2
View File
@@ -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"
}