Identify article using only the first 400 chars of its ID

Some feeds, like Google News, have very long article IDs because they
contain base64 encoded information. 400 chars seems enough to avoid
duplicates while maintaining a reasonably sized index.

reader/tasks.py

@@ -126,7 +126,7 @@ def synchronize_parsed_feed(feed: models.Feed, parsed_feed: ParsedFeed):
     articles_to_uncache = list()
     existing_articles = (
         models.Article.objects.filter(feed=feed)
-        .filter(id_in_feed__in={a.id for a in parsed_feed.articles})
+        .filter(id_in_feed__in={a.id[:400] for a in parsed_feed.articles})
         .select_related('feed')
         .prefetch_related('attachment_set')
     )
@@ -136,7 +136,7 @@ def synchronize_parsed_feed(feed: models.Feed, parsed_feed: ParsedFeed):
         article, created, modified = create_or_update_if_needed(
             models.Article,
             existing_articles,
-            id_in_feed=parsed_article.id,
+            id_in_feed=parsed_article.id[:400],
             feed=feed,
             defaults={
                 'uri': parsed_article.link or '',