include more output file locations when considering whether snapshot.is_archived

2025-08-24 07:03:03 +02:00 · 2024-01-19 03:43:36 -08:00
parent 0a25495520
commit 1773146833
1 changed files with 6 additions and 2 deletions
--- a/archivebox/index/schema.py
+++ b/archivebox/index/schema.py
@@ -379,11 +379,15 @@ class Link:

        output_paths = (
            domain(self.url),
+            'output.html',
            'output.pdf',
            'screenshot.png',
-            'output.html',
+            'singlefile.html',
+            'readability/content.html',
+            'mercury/content.html',
+            'htmltotext.txt',
            'media',
-            'singlefile.html'
+            'git',
        )

        return any(