From 3bd83f04623e4f8daf67e2bcbbdf7bb693b9ad8f Mon Sep 17 00:00:00 2001 From: rlx Date: Wed, 1 Nov 2017 18:14:15 +0000 Subject: [PATCH] implement get_videos --- recommendation_engine.py | 44 +++++++++++++++++++++++++++++++--------- 1 file changed, 34 insertions(+), 10 deletions(-) diff --git a/recommendation_engine.py b/recommendation_engine.py index a822f52..bec7138 100644 --- a/recommendation_engine.py +++ b/recommendation_engine.py @@ -58,10 +58,29 @@ class Engine: return clips[index:] + clips[:index - 1] def get_videos(self, user): - products = [] + # Count tags for the user + count = {} for event in user['events']: if 'product' in event['data']: - products.append(event['data']['product']) + count[event['data']['product']] = count.get( + event['data']['product'], 0 + ) + 1 + score = {} + # For each tag in playlist, increment score by count + for playlist in self.playlists: + score[playlist['name']] = random.random() + for tag in playlist['tags']: + score[playlist['name']] += count.get(tag, 0) + playlists = sorted( + self.playlists, + key=lambda playlist: -score[playlist['name']] + )[:16] + # Shuffle playlists (randomize layout) and shift clips (randomize start) + random.shuffle(playlists) + return [{ + 'clips': self._shift_clips(playlist['clips']), + 'name': playlist['name'] + } for playlist in playlists] def update(self): # Get all storylines with tags @@ -85,7 +104,7 @@ class Engine: }, ['id', 'in', 'out', 'value']) if clip['value'] in names] # Get list of ids for videos with clips ids = list(set([clip['id'].split('/')[0] for clip in clips])) - # Get (and cache) order (and code + name) for each video + # Get and cache video data filename = os.path.join(self.path, 'videos.json') if os.path.exists(filename): with open(filename) as f: @@ -99,21 +118,27 @@ class Engine: ], key=lambda video: video['order']) with open(filename, 'w') as f: f.write(json.dumps(videos, indent=4, sort_keys=True)) + # Get video order order = {video['id']: video['order'] for video in videos} # Sort clips clips = sorted( clips, key=lambda clip: order[clip['id'].split('/')[0]] * 1000000 + clip['in'] ) - # Get playlists + # Get and cache playlists self.playlists = [playlist for playlist in [{ 'name': storyline['name'], 'tags': storyline['tags'], - 'clips': [ - '{}_{:.3f}-{:.3f}'.format( - clip['id'].split('/')[0], clip['in'], clip['out'] - ) for clip in clips if clip['value'] == storyline['name'] - ] + 'clips': [{ + 'id': clip['id'], + 'in': clip['in'], + 'out': clip['out'] + } for clip in clips if clip['value'] == storyline['name']] + #'clips': [ + # '{}_{:.3f}-{:.3f}'.format( + # clip['id'].split('/')[0], clip['in'], clip['out'] + # ) for clip in clips if clip['value'] == storyline['name'] + #] } for storyline in storylines] if playlist['clips']] with open(os.path.join(self.path, 'playlists.json'), 'w') as f: f.write(json.dumps(self.playlists, indent=4, sort_keys=True)) @@ -121,4 +146,3 @@ class Engine: if __name__ == '__main__': engine = Engine('json') engine.update() -