diff options
| author | sigoden <sigoden@gmail.com> | 2024-07-17 18:29:41 +0800 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-07-17 18:29:41 +0800 |
| commit | 4d18d6346d4aea16cc9dac1bcfc3ccc4c15a7bd0 (patch) | |
| tree | 7375770b97ab1ee2daec1c812d4ccca6ad419f03 | |
| parent | 2ee24731dc39985c5ed2014dc38edf44033af51d (diff) | |
| download | llm-functions-docker-4d18d6346d4aea16cc9dac1bcfc3ccc4c15a7bd0.tar.gz | |
refactor: improve fetch_url_via_curl (#77)
| -rwxr-xr-x | tools/fetch_url_via_curl.sh | 4 |
1 files 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" } |