Fallback on first body image

cache.py

@@ -161,7 +161,7 @@ def fetch_metadata(title, url, description):
     """Fetch additional metadata."""
     parsed_url = urlparse(url)
     root_url = f"{parsed_url.scheme}://{parsed_url.netloc}/"
-    data = lassie.fetch(url)
+    data = lassie.fetch(url, all_images=True)
     og_image = ""
     favicon = ""
     for image in data.get("images"):
@@ -182,6 +182,14 @@ def fetch_metadata(title, url, description):
             elif ".ico" in favicon and ".ico" not in image_src:
                 favicon = image_src
 
+    if not og_image:
+        for image in data.get("images"):
+            if image_type == "body_image":
+                image_src = image.get("src")
+                if "favicon" not in image_src:
+                    og_image = image_src
+                    break
+
     # Fallback on server's default.
     if not favicon:
         favico_url = f"{root_url}favicon.ico"
@@ -199,21 +207,24 @@ def fetch_metadata(title, url, description):
 def metadata():
     """Fetch additional metadata for existing archives."""
     for cache in Cache.all():
-        # That one is takin way too long.
+        # That one is taking way too long.
         if cache.url.startswith("https://tw5.immateriel.fr"):
-            print("Skipping", cache.url)
+            print("Skipping (too long)", cache.url)
+            continue
+        if cache.og_image and cache.description and cache.favicon:
+            print("Skipping (all good)", cache.url)
+            continue
+        if cache.url.startswith(
+            (
+                "https://www.la-grange.net",
+                "https://gilest.org",
+                "https://vasilis.nl",
+                "https://www.danmcquillan.org",
+                "https://public-inbox.org",
+            )
+        ) and (cache.og_image or cache.description or cache.favicon):
+            print("Skipping (known missing infos)", cache.url)
             continue
-        # if cache.description or cache.url.startswith(
-        #     (
-        #         "https://www.la-grange.net",
-        #         "https://gilest.org",
-        #         "https://vasilis.nl",
-        #         "https://www.danmcquillan.org",
-        #         "https://public-inbox.org",
-        #     )
-        # ):
-        #     print("Skipping", cache.url)
-        #     continue
         print("Fetching metadata for", cache.url, cache.title)
         og_image, description, favicon = fetch_metadata(
             cache.title, cache.url, cache.description

cache/2024/1d60fc5548a6fe61da80a4e16892fa0c/index.md

@@ -2,7 +2,7 @@ title: Deep Democracy - IAPOP
 url: https://iapop.com/deep-democracy/
 hash_url: 1d60fc5548a6fe61da80a4e16892fa0c
 archive_date: 2024-01-31
-og_image: 
+og_image: https://iapop.com/wp-content/uploads/2018/03/iapop_world_icon-blue-150px-tiny.png
 description: Definition of Deep Democracy The concept of Deep Democracy was developed by Arnold Mindell. It is defined as an attitude and a principle. Attitude: Deep Democracy is an attitude that […]
 favicon: https://iapop.com/wp-content/uploads/2018/03/iapop_world_icon-blue-150px-tiny.png
 

cache/2024/1fe484434058e9c44d21bfebb0ddba31/index.md

@@ -2,7 +2,7 @@ title: ce que nous laissons
 url: https://www.la-grange.net/2024/01/23/legacy
 hash_url: 1fe484434058e9c44d21bfebb0ddba31
 archive_date: 2024-01-28
-og_image: 
+og_image: https://www.la-grange.net/2024/01/23/3152-ciel.jpg
 description: 
 favicon: https://www.la-grange.net/favicon.ico
 

cache/2024/790f724c45b26de460f9eeac04d48884/index.md

@@ -2,7 +2,7 @@ title: carnet d'archéologue
 url: https://www.la-grange.net/2024/01/22/carnet
 hash_url: 790f724c45b26de460f9eeac04d48884
 archive_date: 2024-01-28
-og_image: 
+og_image: https://www.la-grange.net/2024/01/20/3134-carnets.jpg
 description: 
 favicon: https://www.la-grange.net/favicon.ico
 

cache/2024/87c468a4eddabe5d2c28e902d7f17504/index.md

@@ -2,7 +2,7 @@ title: je ne sais pas pourquoi
 url: https://www.la-grange.net/2024/01/11/pourquoi
 hash_url: 87c468a4eddabe5d2c28e902d7f17504
 archive_date: 2024-01-11
-og_image: 
+og_image: https://www.la-grange.net/2024/01/09/3045-glace.jpg
 description: 
 favicon: https://www.la-grange.net/favicon.ico
 

cache/2024/956819385548bba6e768563b12edc2d6/index.md

@@ -2,7 +2,7 @@ title: herbe
 url: https://www.la-grange.net/2024/01/24/herbe
 hash_url: 956819385548bba6e768563b12edc2d6
 archive_date: 2024-01-28
-og_image: 
+og_image: https://www.la-grange.net/2024/01/24/3154-herbe.jpg
 description: 
 favicon: https://www.la-grange.net/favicon.ico
 

cache/2024/9bc04d41d25fc73391116d99b7259a3d/index.md

@@ -2,7 +2,7 @@ title: notes
 url: https://www.la-grange.net/2023/07/10/notes-train
 hash_url: 9bc04d41d25fc73391116d99b7259a3d
 archive_date: 2024-01-07
-og_image: 
+og_image: https://www.la-grange.net/2023/07/10/0797-batiment-visage.jpg
 description: 
 favicon: https://www.la-grange.net/favicon.ico
 

cache/2024/b1da1249f2db388d7e84d6ad23c2fc5d/index.md

@@ -2,7 +2,7 @@ title: Data Luddism
 url: https://www.danmcquillan.org/dataluddism.html
 hash_url: b1da1249f2db388d7e84d6ad23c2fc5d
 archive_date: 2024-01-09
-og_image: 
+og_image: https://www.danmcquillan.org/images/burnmill.jpg
 description: 
 favicon: 
 

cache/2024/b692faaa55fd2775e957b20e833e9e5e/index.md

@@ -2,7 +2,7 @@ title: fraîchement
 url: https://www.la-grange.net/2024/01/26/fraichement
 hash_url: b692faaa55fd2775e957b20e833e9e5e
 archive_date: 2024-01-28
-og_image: 
+og_image: https://www.la-grange.net/2024/01/26/3159-chauffage.jpg
 description: 
 favicon: https://www.la-grange.net/favicon.ico
 

cache/2024/c4751e7c80b292e3533ee6b3e057b702/index.md

@@ -2,7 +2,7 @@ title: Manuel de survie de la femme dans la tech
 url: https://www.duchess-france.fr/dossier/women%20in%20tech/alli%C3%A9s/2023/01/15/manuel-survie-femme-tech.html
 hash_url: c4751e7c80b292e3533ee6b3e057b702
 archive_date: 2024-01-21
-og_image: 
+og_image: https://www.duchess-france.fr/assets/bandeau.jpeg
 description: Je vois de plus en plus de femmes rejoindre l’informatique, et c’est une très bonne chose. Je vois aussi trop de femmes patir de sexisme ordinaire, se remettre en question encore et encore… et quitter le milieu au bout de quelques années. J’ai mis du temps à apprendre certaines choses.
 favicon: https://www.duchess-france.fr/favicon.ico
 

cache/2024/ce5fdc61fd66cdb9ce548fb543eba986/index.md

@@ -2,7 +2,7 @@ title: Unsigned Commits
 url: https://blog.glyph.im/2024/01/unsigned-commits.html
 hash_url: ce5fdc61fd66cdb9ce548fb543eba986
 archive_date: 2024-01-25
-og_image: 
+og_image: https://blog.glyph.im/images/back9.png
 description: Deciphering Glyph, the blog of Glyph Lefkowitz.
 favicon: https://blog.glyph.im/images/favicon.ico
 

cache/2024/d236f33cf82727313d17cb23bf36a395/index.md

@@ -2,7 +2,7 @@ title: Reconsider your partnership with Brave
 url: https://kagifeedback.org/d/2808-reconsider-your-partnership-with-brave/6
 hash_url: d236f33cf82727313d17cb23bf36a395
 archive_date: 2024-01-07
-og_image: 
+og_image: https://kagifeedback.org/assets/favicon-bmwk4ltf.png
 description: Brave, as you know, is led by Brendan Eich. s homophobia is so disgusting that he was forced to resign as the leader...
 favicon: https://kagifeedback.org/assets/favicon-bmwk4ltf.png
 

cache/2024/d75afc90a9d3c3b5a56b69446795fbb5/index.md

@@ -2,7 +2,7 @@ title: plaisir d'ébauche
 url: https://www.la-grange.net/2024/01/06/ebauche
 hash_url: d75afc90a9d3c3b5a56b69446795fbb5
 archive_date: 2024-01-07
-og_image: 
+og_image: https://www.la-grange.net/2024/01/06/3008-furikake.jpg
 description: 
 favicon: https://www.la-grange.net/favicon.ico
 

cache/2024/faa1d8cae94da6838ff9351e5df791ca/index.md

@@ -2,7 +2,7 @@ title: Make the indie web easier
 url: https://gilest.org/indie-easy.html
 hash_url: faa1d8cae94da6838ff9351e5df791ca
 archive_date: 2024-01-09
-og_image: 
+og_image: https://gilest.org/2024/dangerously-muddy.jpg
 description: 
 favicon: https://gilest.org/favicon.ico
 

cache/2024/ff566a58892db07815a327802fea66d3/index.md

@@ -2,7 +2,7 @@ title: In Loving Memory of Square Checkbox
 url: https://tonsky.me/blog/checkbox/
 hash_url: ff566a58892db07815a327802fea66d3
 archive_date: 2024-01-28
-og_image: 
+og_image: https://tonsky.me/blog/checkbox/checkbox@2x.png?t=1706539628
 description: History of checkboxes and radio buttons in user interfaces
 favicon: https://tonsky.me/i/favicon.png