include more output file locations when considering whether snapshot.…

…is_archived
ArchiveBox · Jan 19, 2024 · 1773146 · 1773146
1 parent 0a25495
commit 1773146
Showing 1 changed file with 6 additions and 2 deletions.
diff --git a/archivebox/index/schema.py b/archivebox/index/schema.py
@@ -379,11 +379,15 @@ def is_archived(self) -> bool:
 
  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(