itsulu/FediFetcher
2
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Skip fetching context of private posts

Browse source 
Context fetching is performed without authentication, so it is only
possible for public and unlisted posts.
This commit is contained in:
Andrew Kvalheim 2024-07-01 18:01:03 -07:00
parent e85384a5a6
commit 5ed751a8c6
1 changed files with 9 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

10
find_posts.py
View file

@ -479,6 +479,14 @@ def get_reply_toots(user_id, server, access_token, seen_urls, reply_since):
f"Error getting replies for user {user_id} on server {server}. Status code: {resp.status_code}" f"Error getting replies for user {user_id} on server {server}. Status code: {resp.status_code}"
) )
def toot_context_can_be_fetched(toot):
fetchable = toot["visibility"] in ["public", "unlisted"]
if not fetchable:
logger.debug(f"Cannot fetch context of private toot {toot['uri']}")
return fetchable
def toot_context_should_be_fetched(toot): def toot_context_should_be_fetched(toot):
if toot['uri'] not in recently_checked_context: if toot['uri'] not in recently_checked_context:
recently_checked_context[toot['uri']] = toot recently_checked_context[toot['uri']] = toot
@ -512,7 +520,7 @@ def get_all_known_context_urls(server, reply_toots, parsed_urls, seen_hosts):
if toot_has_parseable_url(toot, parsed_urls): if toot_has_parseable_url(toot, parsed_urls):
url = toot["url"] if toot["reblog"] is None else toot["reblog"]["url"] url = toot["url"] if toot["reblog"] is None else toot["reblog"]["url"]
parsed_url = parse_url(url, parsed_urls) parsed_url = parse_url(url, parsed_urls)
if(toot_context_should_be_fetched(toot)): if toot_context_can_be_fetched(toot) and toot_context_should_be_fetched(toot):
recently_checked_context[toot['uri']]['lastSeen'] = datetime.now(datetime.now().astimezone().tzinfo) recently_checked_context[toot['uri']]['lastSeen'] = datetime.now(datetime.now().astimezone().tzinfo)
context = get_toot_context(parsed_url[0], parsed_url[1], url, seen_hosts) context = get_toot_context(parsed_url[0], parsed_url[1], url, seen_hosts)
if context is not None: if context is not None: