Update pelican_import.py

Use a set to avoid downloading duplicate attachments on a post more than once.
This commit is contained in:
Stuart Axon 2018-11-16 13:54:18 +00:00 committed by Stuart Axon
commit 4d1869002e
2 changed files with 11 additions and 9 deletions

View file

@ -699,7 +699,7 @@ def get_attachments(xml):
else:
filename = get_filename(filename, post_id)
names[post_id] = filename
attachedposts = defaultdict(list)
attachedposts = defaultdict(set)
for parent, url in attachments:
try:
parent_name = names[parent]
@ -707,7 +707,7 @@ def get_attachments(xml):
# attachment's parent is not a valid post
parent_name = None
attachedposts[parent_name].append(url)
attachedposts[parent_name].add(url)
return attachedposts