Tweak/fix opengraph parsing

This commit is contained in:
Thomas Sileo 2022-12-11 18:15:30 +01:00
parent 7621a19489
commit 22410862f3

View file

@ -62,6 +62,13 @@ def _scrap_og_meta(url: str, html: str) -> OpenGraphMeta | None:
if u := raw.get(maybe_rel): if u := raw.get(maybe_rel):
raw[maybe_rel] = make_abs(u, url) raw[maybe_rel] = make_abs(u, url)
if not is_url_valid(raw[maybe_rel]):
logger.info(f"Invalid url {raw[maybe_rel]}")
if maybe_rel == "url":
raw["url"] = url
elif maybe_rel == "image":
raw["image"] = None
return OpenGraphMeta.parse_obj(raw) return OpenGraphMeta.parse_obj(raw)