def process_events(h, clusters):
"""
Takes clusters of node uuids and
builds, modifies, and deletes events out of them.
"""
now = datetime.utcnow()
# Get existing event clusters.
event_map = {}
existing = {}
for e in Event.all_active():
# Map event ids to their event, for lookup later.
event_map[e.id] = e
# Map event ids to a list of their member node ids.
existing[e.id] = [a.node_id for a in e.articles]
# Figure out which events to update, delete, and create.
to_update, to_create, to_delete, unchanged = triage(existing, clusters)
for a_ids in to_create:
articles = Article.query.filter(Article.node_id.in_([id.item() for id in a_ids])).order_by(Article.created_at.desc()).all()
e = Event(articles)
e.created_at = articles[0].created_at
e.updated_at = articles[-1].updated_at
rep_article = representative_article(h, a_ids, articles)
e.title = rep_article.title
e.image = rep_article.image
db.session.add(e)
for e_id, a_ids in to_update.items():
e = event_map[e_id]
articles = Article.query.filter(Article.node_id.in_([id.item() for id in a_ids])).all()
e.members = articles
rep_article = representative_article(h, a_ids, articles)
e.title = rep_article.title
e.image = rep_article.image
e.update()
# Freeze expiring events and clean up their articles from the hierarchy.
for e_id in unchanged:
e = event_map[e_id]
if (now - e.updated_at).days > 3:
e.active = False
nodes = [h.to_iid(a.node_id) for a in e.articles]
h.prune(nodes)
# Do this LAST so any of this event's associated articles
# have a chance to be moved to their new clusters (if any).
for e_id in to_delete:
db.session.delete(event_map[e_id])
# does this need to prune the articles as well?
# i think the assumption is that a deleted event's articles have all migrated elsewhere.
db.session.commit()
