mirror of
https://github.com/bellingcat/cisticola.git
synced 2026-06-11 04:48:33 +03:00
Fix issues with new datetime baed 'media_archived' column
This commit is contained in:
@@ -3,6 +3,7 @@ import os
|
||||
from io import BytesIO
|
||||
from urllib.parse import urlparse
|
||||
import tempfile
|
||||
from datetime import datetime, timezone
|
||||
|
||||
import boto3
|
||||
from loguru import logger
|
||||
@@ -417,7 +418,7 @@ class ScraperController:
|
||||
|
||||
session = self.session()
|
||||
|
||||
posts = session.query(ScraperResult).where(ScraperResult.media_archived == False).order_by(func.random()).all()
|
||||
posts = session.query(ScraperResult).where(ScraperResult.media_archived == None).order_by(func.random()).all()
|
||||
|
||||
logger.info(f"Found {len(posts)} posts without media. Archiving now")
|
||||
|
||||
@@ -431,7 +432,7 @@ class ScraperController:
|
||||
post = scraper.archive_files(post)
|
||||
|
||||
if post:
|
||||
session.query(ScraperResult).where(ScraperResult.id == post.id).update({'archived_urls': post.archived_urls, 'media_archived': True})
|
||||
session.query(ScraperResult).where(ScraperResult.id == post.id).update({'archived_urls': post.archived_urls, 'media_archived': datetime.now(timezone.utc)})
|
||||
session.commit()
|
||||
|
||||
break
|
||||
|
||||
Reference in New Issue
Block a user