Prevent inserting duplicate comments in one transaction

twitch-cache-api/sync.py

@@ -68,11 +68,19 @@ class Sync(object):
                 db.session.add(video)
             db.session.commit()
             app.logger.info('Updated videos: %s', ', '.join(updated) if updated else 'NONE')
+            comments = {}
             for com in twitch.fetch_comments(updated):
-                q = db.session.query(Video).filter(Video.id == cls._get(com, 'content_id'))
+                id = cls._get(com, '_id')
-                video = q.first()
+                assoc = dict(
-                if not video:
+                    video_id=cls._get(com, 'content_id'),
-                    continue
+                    offset=cls._get(com, 'content_offset_seconds'))
+                if id in comments:
+                    comments[id].update(com)
+                    comments[id]['assocs'].append(assoc)
+                else:
+                    comments[id] = com
+                    comments[id]['assocs'] = [assoc]
+            for com in comments.values():
                 id = cls._get(com, '_id')
                 q = db.session.query(Comment).filter(Comment.id == id)
                 comment = q.first()
@@ -91,6 +99,11 @@ class Sync(object):
                 comment.message_user_badges = badges
                 comment.created_at = cls._to_datetime(cls._get(com, 'created_at'))
                 comment.updated_at = cls._to_datetime(cls._get(com, 'updated_at'))
+                for assc in com['assocs']:
+                    q = db.session.query(Video).filter(Video.id == assc['video_id'])
+                    video = q.first()
+                    if not video:
+                        continue
                     q = db.session.query(Association).filter(
                             Association.video_id == video.id,
                             Association.comment_id == comment.id)
@@ -98,7 +111,7 @@ class Sync(object):
                     if not assoc:
                         assoc = Association()
                     assoc.comment = comment
-                assoc.offset = cls._get(com, 'content_offset_seconds')
+                    assoc.offset = assc['offset']
                     video.associations.append(assoc)
                     db.session.add(video)
             db.session.commit()