def record_heatmaps(mongodb):
"""
Record heatmap bins for each video, based on segments
for a single video?
"""
start_time = time.time()
# TODO: handle cut segments (i.e., start event exists but end event missing)
# TODO: only remove the corresponding entries in the database: (video, user)
collection = mongodb['video_segments']
segments = list(collection.find())
collection = mongodb['video_heatmaps']
collection.remove()
print len(segments), "segments found"
results = defaultdict(dict)
for segment in segments:
if not segment["user_id"] in results[segment["video_id"]]:
results[segment["video_id"]][segment["user_id"]] = []
results[segment["video_id"]][segment["user_id"]].append(segment)
vid_col = mongodb['videos']
for video_id in results:
result = list(vid_col.find({"video_id": video_id}))
if len(result):
process_heatmaps(mongodb, results[video_id], video_id, result[0]["duration"])
else:
print "ERROR in video information retrieval"
# Make sure the collection is indexed.
from pymongo import ASCENDING
collection.ensure_index([("video_id", ASCENDING)])
# [("video_id", ASCENDING), ("time", ASCENDING)])
print sys._getframe().f_code.co_name, "COMPLETED", (time.time() - start_time), "seconds"
def record_heatmaps_ajax(mongodb, index):
"""
Record heatmap bins for each video, based on segments
for a single video?
"""
bin_size = 100000
start_time = time.time()
collection = mongodb[HEATMAPS_COL]
collection.remove()
# TODO: handle cut segments (i.e., start event exists but end event missing)
# TODO: only remove the corresponding entries in the database: (video, user)
vid_col = mongodb['videos']
video_list = list(vid_col.find())
num_videos = len(video_list)
for index, video in enumerate(video_list):
video_id = video["video_id"]
loop_start_time = time.time()
collection = mongodb[SEGMENTS_COL]
segments = list(collection.find({"video_id": video_id}))
#segments = collection.find().limit(bin_size).skip(index*bin_size) #.batch_size(1000)
print index, "/", num_videos, video_id, ":", len(segments), "segments", (time.time() - loop_start_time), "seconds"
if len(segments):
loop_start_time2 = time.time()
results = defaultdict(dict)
for segment in segments:
if not segment["user_id"] in results[segment["video_id"]]:
results[segment["video_id"]][segment["user_id"]] = []
results[segment["video_id"]][segment["user_id"]].append(segment)
process_heatmaps(mongodb, results[video_id], video_id, video["duration"])
print (time.time() - loop_start_time2), "seconds"
# Make sure the collection is indexed.
from pymongo import ASCENDING
collection.ensure_index([("video_id", ASCENDING)])
# [("video_id", ASCENDING), ("time", ASCENDING)])
print sys._getframe().f_code.co_name, "COMPLETED", (time.time() - start_time), "seconds"