def video_single_query(vid):
"""
Return heatmap information from the database for a single video.
Example: http://localhost:9999/query/video_single?vid=2deIoNhqDsg
"""
import numpy as np
mongodb = get_db()
start_time = time.time()
collection = mongodb['visual_transition']
vtran_temp_data = list(collection.find({"vid": vid}))
# print vtran_temp_data
vtran_data = vtran_temp_data[0]["visual_diff"]
# vtran_np_data = np.array(vtran_data)
# print vtran_data
# print vtran_data[:,1]
vtran_peaks_raw = detect_peaks(np.array(vtran_data)[:,1], 3, "vtran")
# print vtran_peaks
vtran_peaks = json.dumps(vtran_peaks_raw, default=json_util.default)
collection = mongodb[HEATMAPS_COL]
# entries = list(collection.find({"video_id": vid}, {"completion_counts": 0}))
entries = list(collection.find({"video_id": vid}, {"daily_view_counts": 0, "raw_counts": 0, "playrate_counts": 0, "pause_counts": 0, "unique_counts": 0, "replay_counts": 0, "skip_counts": 0, "completion_counts": 0}))
# print vid, entries
# L@S 2014 analysis
# collection = mongodb["video_heatmaps_mitx_fall2012"]
# entries = list(collection.find({"video_id": vid}, {"completion_counts": 0}))
# if len(entries) == 0:
# collection = mongodb["video_heatmaps_harvardx_ph207x_fall2012"]
# entries = list(collection.find({"video_id": vid}, {"completion_counts": 0}))
# if len(entries) == 0:
# collection = mongodb["video_heatmaps_berkeleyx_cs188x_fall2012"]
# entries = list(collection.find({"video_id": vid}, {"completion_counts": 0}))
if len(entries):
# First, smooth the points and run peak detection
play_points = entries[0]["play_counts"]
play_points[0] = 0
#play_points[0:int(len(play_points)*0.03)] = [0]*int(len(play_points)*0.03)
play_kde = get_kde(np.array(play_points), 0.02)
# mask nan values
#print "contains nan:", np.all(np.isnan(play_kde[:,1]), 0)
#if np.all(np.isnan(play_kde[:,1]), 0):
# play_kde = [[index, point] for index, point in enumerate(play_points)]
for index, point in enumerate(play_kde[:,1]):
if np.isnan(point):
play_kde[:,1][index] = play_points[index]
masked_play_kde = np.ma.array(play_kde, mask=np.isnan(play_kde))
#print masked_play_kde, "max", np.max(masked_play_kde[:,1])
play_kde = masked_play_kde
play_peaks_raw = detect_peaks(play_kde[:,1], 2.2, "interaction")
entries[0]["play_kde"] = play_kde[:,1].tolist()
interaction_peaks = json.dumps(play_peaks_raw, default=json_util.default)
result = json.dumps(entries[0], default=json_util.default)
else:
result = ""
print sys._getframe().f_code.co_name, "COMPLETED", (time.time() - start_time), "seconds"
return [result, interaction_peaks, vtran_data, vtran_peaks]
def video_multiple_query(course=""):
"""
Return heatmap information from the database for a single video.
Example: http://localhost:9999/query/video_single?vid=2deIoNhqDsg
"""
import numpy as np
mongodb = get_db()
start_time = time.time()
all_result = []
all_interaction_peaks = []
collection = mongodb[VIDEOS_COL]
#UIST 2014
if course == "6.00x":
course_name = "6.00x-Fall-2012"
elif course == "3.091x":
course_name = "3.091x-Fall-2012"
else:
course_name = "6.00x-Fall-2012"
entries = list(collection.find({"course_name":course_name})
.sort([("week_number", ASCENDING), ("sequence_number", ASCENDING), ("module_index", ASCENDING)]))
for entry in entries:
# vtran data might not be necessary for the multiplayer view
# collection = mongodb['visual_transition']
# vtran_temp_data = list(collection.find({"vid": vid}))
# vtran_data = vtran_temp_data[0]["visual_diff"]
# vtran_peaks_raw = detect_peaks(np.array(vtran_data)[:,1], 3, "vtran")
# vtran_peaks = json.dumps(vtran_peaks_raw, default=json_util.default)
collection = mongodb[HEATMAPS_COL]
vid = entry["video_id"]
# entries = list(collection.find({"video_id": vid}, {"completion_counts": 0}))
single_entries = list(collection.find({"video_id": vid}, {"total_watching_time": 0, "daily_view_counts": 0, "raw_counts": 0, "playrate_counts": 0, "pause_counts": 0, "unique_counts": 0, "replay_counts": 0, "skip_counts": 0, "completion_counts": 0}))
if len(single_entries):
# First, smooth the points and run peak detection
play_points = single_entries[0]["play_counts"]
play_points[0] = 0
#play_points[0:int(len(play_points)*0.03)] = [0]*int(len(play_points)*0.03)
play_kde = get_kde(np.array(play_points), 0.02)
# mask nan values
#print "contains nan:", np.all(np.isnan(play_kde[:,1]), 0)
#if np.all(np.isnan(play_kde[:,1]), 0):
# play_kde = [[index, point] for index, point in enumerate(play_points)]
for index, point in enumerate(play_kde[:,1]):
if np.isnan(point):
play_kde[:,1][index] = play_points[index]
masked_play_kde = np.ma.array(play_kde, mask=np.isnan(play_kde))
#print masked_play_kde, "max", np.max(masked_play_kde[:,1])
play_kde = masked_play_kde
play_peaks_raw = detect_peaks(play_kde[:,1], 3, "interaction")
single_entries[0]["play_kde"] = play_kde[:,1].tolist()
interaction_peaks = json.dumps(play_peaks_raw, default=json_util.default)
result = json.dumps(single_entries[0], default=json_util.default)
all_result.append(result)
all_interaction_peaks.append(interaction_peaks)
else:
result = ""
print sys._getframe().f_code.co_name, "COMPLETED", (time.time() - start_time), "seconds"
# return [result, interaction_peaks, vtran_data, vtran_peaks]
return [all_result, all_interaction_peaks]
