From 22410862f3d2cab24a5888f0c7245d63f107c3d7 Mon Sep 17 00:00:00 2001
From: Thomas Sileo <t@a4.io>
Date: Sun, 11 Dec 2022 18:15:30 +0100
Subject: [PATCH] Tweak/fix opengraph parsing

---
 app/utils/opengraph.py | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/app/utils/opengraph.py b/app/utils/opengraph.py
index 24eb873..05e7752 100644
--- a/app/utils/opengraph.py
+++ b/app/utils/opengraph.py
@@ -62,6 +62,13 @@ def _scrap_og_meta(url: str, html: str) -> OpenGraphMeta | None:
         if u := raw.get(maybe_rel):
             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)