1
0
mirror of https://github.com/pirate/ArchiveBox.git synced 2025-09-02 19:02:37 +02:00

Handle list of tags in JSON, and be more clever about comma vs. space

This commit is contained in:
jim winstead
2024-02-28 17:38:49 -08:00
parent 178e676e0f
commit ccabda4c7d
3 changed files with 21 additions and 4 deletions

View File

@@ -66,9 +66,14 @@ def parse_generic_json_export(json_file: IO[str], **_kwargs) -> Iterable[Link]:
elif link.get('name'):
title = link['name'].strip()
tags = ''
if link.get('tags'):
tags = link.get('tags').replace(' ',',')
# if we have a list, join it with commas
tags = link.get('tags')
if type(tags) == list:
tags = ','.join(tags)
elif type(tags) == str:
# if there's no comma, assume it was space-separated
if ',' not in tags:
tags = tags.replace(' ', ',')
yield Link(
url=htmldecode(url),