Merge branch 'iss2'

recommendation_engine.py

@@ -58,6 +58,10 @@ class Engine:
             self.state['userKeywordsWeights'] = {
                 'themeTags': {'locked': False, 'value': 0.3},
                 'characterTags': {'locked': False, 'value': 0.7}
+<<<<<<< HEAD
+=======
+
+>>>>>>> iss2
             }
         self.update_keywords()
 
@@ -89,7 +93,10 @@ class Engine:
                     clips[inpoint['index']]['out'] = self.pandora.get(video_id, ['duration'])['duration']
         return clips
 
+
     def get_videos(self, user):
+        # Update self_playlists first to reflect changes
+        update_user_playlists(self.playlists, user)
 
         if user.get('events', [{}])[0].get("event")=="login":
             return {
@@ -106,30 +113,26 @@ class Engine:
         # check if there were grid events for all indexes.
         grid_events = {}
         (nc, np, ns) = (grid_change.get("nextClip"), grid_change.get("nextPlaylist"), grid_change.get("staySame"))
-        # this assumes np + nc + ns = total number of videos in the grid view (16). 
+        video_num = nc + np + ns
-        # Make sure sanity check exists in front-end (error if it does not add up to 16).
-        video_num = nc + np + ns 
 
-        # for event in user.get('events', []):
+        # # The version where the loop also extract play_index; requires "index" in play event data (previously unavailable):
-        #     if event.get('event') == "grid" and event.get('data').get('index') not in grid_events:
-        #         grid_events[event.get('data').get('index')] = event.get('data')
-        #     if len(grid_events) == video_num:
-        #         break
-
-        # # The version where the loop also extract play_index (requires "index" in play event data):
         play_index = None
         for event in user.get('events', []):
             if event.get('event') == "grid" and event.get('data').get('index') not in grid_events:
                 grid_events[event.get('data').get('index')] = event.get('data')
-            if event.get('event') == "play" and not play_index:
+            if event.get('event') == "play" and event["data"].get("type") == "video" and not play_index:
-                play_index = event.get('data').get('index') 
+                play_index = event.get('data').get('index')
             if len(grid_events) == video_num and play_index:
-                break 
+                break
 
         prev_grid_list = sorted([v for v in grid_events.values()], key=lambda k:k['index'])
 
         # if there were no grid events for all, initialize all grids.
+<<<<<<< HEAD
         if len(prev_grid_list) < video_num: 
+=======
+        if len(prev_grid_list) < video_num:
+>>>>>>> iss2
             return {
                 'user': {
                     'keywords': user.get('keywords', {})
@@ -160,7 +163,6 @@ class Engine:
                     next_playlist_index.append(next_clip_index)
                     break
                 else:
-                # if "playlist" and "playlistPostion" (if not, default to 0) exists in grid event
                     for playlist in self.playlists:
                         if playlist.get('name')== prev_grid_list[i].get('playlist'):
                             if len(playlist["clips"]) == 1:
@@ -199,7 +201,10 @@ class Engine:
                 },
                 'videos': videos_
             }
+<<<<<<< HEAD
 
+=======
+>>>>>>> iss2
 
 
     def get_recommendations(self, user, vids_exclude = []):
@@ -226,7 +231,11 @@ class Engine:
         character_tags["FEDOR MIKHAILOVICH SOFRONOV"] = character_tags.get("FYODOR MIKHAILOVICH SOFRONOV",0)
         character_tags["SHKABARNYA OLGA SERGEEVNA"] = character_tags.get("OLGA SERGEEVNA SHKABARNYA",0)
         character_tags["VICTORIA OLEGOVNA SKITSKAYA"] = character_tags.get("VIKTORIA OLEGOVNA SKITSKAYA",0)
+<<<<<<< HEAD
         
+=======
+
+>>>>>>> iss2
         score = {}
         for playlist in playlists:
             score[playlist['name']] = random.random() * 0.001
@@ -264,7 +273,68 @@ class Engine:
         } for video in videos]
 
 
+    def update_user_playlists(playlists, user, watch_cutoff = 0.9):
+    # Output: playlists with updated in/out time of clips that have been watched.
+    # Watched is defined as a video being played in full screen.
+    # "watch_cutoff" parameter: the portion of the clip duration to be determined as watched the whole clip. should be [0,1]
+    # + check (play, pause) pairs and eliminate unusual cases most likely due to a bug.
+    # + If (play, pause) pairs exceed XX(80-90?) percent of the clip length, eliminate the clip from the playlist.
+    # + Otherwise, find the last pause position of a clip and record it as "in" position of the clip.
+    # + If the clips are all eliminated from a playlist, eliminate the playlist.
+        play = {}
+        watched = []
+        clip_max_dur = 10800 # = 3 hours; arbitrary max duration allowed for (pause time - play time) to detect outlier/bugs
+        # The current max time of a clip duration is 10379.383333377269 from "DDLaunch: Erik Verlinde, Gravity as an emergent force (1956)"
+        # A user could potentially spend more than 3 hours if they keep watching after the clip enters into the subsequent "scene"
+        for event in user.get('events', [])[::-1]:
+            if event["event"] == "play" and event["data"].get("type") == "video":
+                play = event
+            elif event["event"] == "pause" and play!={} and event["data"].get("type") == "video":
+                if "position" not in play["data"]:
+                    play = {}
+                    break
+                if play["data"].get("playlist") == event["data"].get("playlist"):
+                    if event["data"]["position"] - play["data"]["position"] > 0 and event["data"]["position"] - play["data"]["position"] < clip_max_dur and event["data"].get("playlistPosition") == play["data"].get("playlistPosition") and event["data"].get("playlistPosition") is not None:
+                        i = event["data"]["playlistPosition"]
+                        for playlist in playlists:
+                            if playlist["name"] == event["data"]["playlist"] and i < len(playlist["clips"]):
+                                if play["data"]["position"] >= max(playlist["clips"][i]["in"] - 15, 0) and event["data"]["position"] <= playlist["clips"][i]["out"] + 15:
+                                    # This assumes the (play, pause) fits inside the clip's (in, out) segment with +/- 15secs buffer. There were newer edits of clip positions with 12 seconds difference.
+                                    # instances where this might not be the case: clip in/out may be largely edited (before after edit inconsistency); skip may trigger jump to a wrong clip (bug)
+                                    if "orig_in" not in playlist["clips"][i]:
+                                        cutoff_pos = (playlist["clips"][i]["out"]-playlist["clips"][i]["in"])*watch_cutoff + playlist["clips"][i]["in"]
+                                    else:
+                                        cutoff_pos = (playlist["clips"][i]["out"]-playlist["clips"][i]["orig_in"])*watch_cutoff + playlist["clips"][i]["orig_in"]
+
+                                    if event["data"]["position"] >= cutoff_pos:
+                                        watched.append((playlist["name"],i))
+                                    else:
+                                        if "orig_in" not in playlist["clips"][i]:
+                                            # record the original "in" position to calculate cutoff position in the future
+                                            playlist["clips"][i]["orig_in"] = playlist["clips"][i]["in"]
+                                        # update "in" position of the clip in the playlist
+                                        playlist["clips"][i]["in"] = event["data"]["position"]
+                                break
+                play = {}
+
+        d_watched = defaultdict(set)
+        for k, v in watched:
+            d_watched[k].add(v)
+        for k, v in d_watched.items():
+            for playlist in playlists:
+                if playlist["name"] == k:
+                    if len(v) == len(playlist["clips"]):
+                        playlists.remove(playlist)
+                    else:
+                        playlist["clips"] = [playlist["clips"][i] for i in range(len(playlist["clips"])) if i not in v]
+                    break
+        return(playlists)
+
+
     def get_next(self, user, position):
+        # Update self_playlists first to reflect changes
+        update_user_playlists(self.playlists, user)
+        
         grid_events = {}
         video_num = 16