def fetch_flickr_photos(mood_type, terms, num_photos):
# grab from flickr
photos = flickr.walk(tag_mode = 'all', tags = terms, extras='url_z')
for i in range(0, num_photos):
photo = photos.next()
url = photo.get('url_z')
pid = photo.get('id')
score = computePictureScore(pid, mood_type)
assert Picture.add(pid, url, mood_type, initialScore = score)
def fetch_flickr_photos(mood_type, terms, num_photos):
# grab from flickr
photos = flickr.walk(tag_mode='all', tags=terms, extras='url_z')
for i in range(0, num_photos):
photo = photos.next()
url = photo.get('url_z')
pid = photo.get('id')
score = computePictureScore(pid, mood_type)
assert Picture.add(pid, url, mood_type, initialScore=score)
def pullPictures(mood, terms, add_num = 0):
# get pics
pics = flickr.photos_search(tags=terms,\
safe_search=1,\
per_page=_SEARCH_NUM)
# calculate how many pictures to do
length = len(pics[0])
myLen = length if add_num > length else add_num
# keep track of which picture we are looking at
picture_map = range(0, length)
random.shuffle(picture_map)
for i in range(0, myLen):
first_attrib = None
photo_id = 0
while len(picture_map) != 0:
idx = picture_map.pop()
first_attrib = pics[0][idx].attrib
photo_id = int(first_attrib['id'])
initial_score = computePictureScore(photo_id, mood)
try:
# throw error if bad things happen
flickr.photos_getInfo(photo_id=photo_id)
except Exception:
continue
try:
# throw error
Picture.objects.get(flickr_id=photo_id)
# photo id already exist in database
if len(picture_map) == 0:
raise Exception('I FAILED!!')
except Exception:
break
url = _getURL(first_attrib)
assert Picture.add(photo_id, url, mood, initialScore=initial_score)
def pullPictures(mood, terms, add_num=0):
# get pics
pics = flickr.photos_search(tags=terms,\
safe_search=1,\
per_page=_SEARCH_NUM)
# calculate how many pictures to do
length = len(pics[0])
myLen = length if add_num > length else add_num
# keep track of which picture we are looking at
picture_map = range(0, length)
random.shuffle(picture_map)
for i in range(0, myLen):
first_attrib = None
photo_id = 0
while len(picture_map) != 0:
idx = picture_map.pop()
first_attrib = pics[0][idx].attrib
photo_id = int(first_attrib['id'])
initial_score = computePictureScore(photo_id, mood)
try:
# throw error if bad things happen
flickr.photos_getInfo(photo_id=photo_id)
except Exception:
continue
try:
# throw error
Picture.objects.get(flickr_id=photo_id)
# photo id already exist in database
if len(picture_map) == 0:
raise Exception('I FAILED!!')
except Exception:
break
url = _getURL(first_attrib)
assert Picture.add(photo_id, url, mood, initialScore=initial_score)
def pullAndFilter(mood, terms, add_num, partition_num):
# get pics
pics = flickr.photos_search(tags=terms,\
safe_search=1,\
per_page=_SEARCH_NUM)
# calculate how many pictures to do
length = len(pics[0])
# myLen = min(partition_num, length)
myLen = length if partition_num > length else partition_num
myLen = max(myLen, 0)
#generate a random index
startIndex = random.randint(0, max(length - myLen, 0))
#generate map for keeping track
picture_map = range(0, myLen)
random.shuffle(picture_map)
added_picture_list = []
#looping myLen times, get out n number of pictures
for i in range(0, myLen):
first_attrib = None
photo_id = 0
print 'picture progress: ' + str(i) + '/' + str(myLen)
while len(picture_map) != 0:
idx = (picture_map.pop() + startIndex) % length
first_attrib = pics[0][idx].attrib
photo_id = int(first_attrib['id'])
initial_score = computePictureScore(photo_id, mood)
del_list = Deleted.objects.filter(content_type=Media.PICTURE, content_id=photo_id)
if len(del_list) != 0:
continue
try:
# throw error if bad things happen
flickr.photos_getInfo(photo_id=photo_id)
except Exception:
continue
try:
# throw error
Picture.objects.get(flickr_id=photo_id)
# photo id already exist in database
if len(picture_map) == 0:
raise Exception('I FAILED!!')
except Exception:
break
try:
url = _getURL(first_attrib)
added_picture_list.append((photo_id, url, mood, initial_score))
except Exception:
print 'The None type url error again, die gracefully'
sorted_picture_list = sorted(added_picture_list, key=lambda tuple: tuple[3], reverse=True)
# add till database is full
preadd = min(add_num, len(sorted_picture_list))
for i in range(0, preadd):
tuple = sorted_picture_list.pop(0)
assert Picture.add(tuple[0], tuple[1], tuple[2], initialScore=tuple[3])
# get media data
medias = Media.objects.filter(moods=mood, content_type=Media.PICTURE)
ScoreFilter._calculateFinalScore(medias)
mediaData = Media.objects.filter(moods=mood, score__final_score__gt=-1).order_by('score__final_score')
# allData = Media.objects.filter(moods=mood, content_type=Media.PICTURE)
# dataDict = {}
# for m in allData:
# if m.score.final_score == -1:
# dataDict[m] = m.score.initial_score
# else:
# dataDict[m] = m.score.final_score
#
# for key, value in sorted(dataDict.iteritems(), key=lambda (k,v): (v,k)):
# mediaData.append(key)
if len(mediaData) == 0:
return preadd
#Add to database if score is higher than existing scores
lowestMediaIndex = 0
while len(sorted_picture_list) != 0:
tuple = sorted_picture_list.pop(0)
score = tuple[3]
if score > mediaData[lowestMediaIndex].score.final_score:
destory(mediaData[lowestMediaIndex].id, Media.PICTURE)
assert Picture.add(tuple[0], tuple[1], tuple[2], initialScore=tuple[3])
lowestMediaIndex += 1
if lowestMediaIndex >= len(mediaData):
break
else:
break
return preadd + lowestMediaIndex
def pullAndFilter(mood, terms, add_num, partition_num):
# get pics
pics = flickr.photos_search(tags=terms,\
safe_search=1,\
per_page=_SEARCH_NUM)
# calculate how many pictures to do
length = len(pics[0])
# myLen = min(partition_num, length)
myLen = length if partition_num > length else partition_num
myLen = max(myLen, 0)
#generate a random index
startIndex = random.randint(0, max(length - myLen, 0))
#generate map for keeping track
picture_map = range(0, myLen)
random.shuffle(picture_map)
added_picture_list = []
#looping myLen times, get out n number of pictures
for i in range(0, myLen):
first_attrib = None
photo_id = 0
print 'picture progress: ' + str(i) + '/' + str(myLen)
while len(picture_map) != 0:
idx = (picture_map.pop() + startIndex) % length
first_attrib = pics[0][idx].attrib
photo_id = int(first_attrib['id'])
initial_score = computePictureScore(photo_id, mood)
del_list = Deleted.objects.filter(content_type=Media.PICTURE,
content_id=photo_id)
if len(del_list) != 0:
continue
try:
# throw error if bad things happen
flickr.photos_getInfo(photo_id=photo_id)
except Exception:
continue
try:
# throw error
Picture.objects.get(flickr_id=photo_id)
# photo id already exist in database
if len(picture_map) == 0:
raise Exception('I FAILED!!')
except Exception:
break
try:
url = _getURL(first_attrib)
added_picture_list.append((photo_id, url, mood, initial_score))
except Exception:
print 'The None type url error again, die gracefully'
sorted_picture_list = sorted(added_picture_list,
key=lambda tuple: tuple[3],
reverse=True)
# add till database is full
preadd = min(add_num, len(sorted_picture_list))
for i in range(0, preadd):
tuple = sorted_picture_list.pop(0)
assert Picture.add(tuple[0], tuple[1], tuple[2], initialScore=tuple[3])
# get media data
medias = Media.objects.filter(moods=mood, content_type=Media.PICTURE)
ScoreFilter._calculateFinalScore(medias)
mediaData = Media.objects.filter(
moods=mood, score__final_score__gt=-1).order_by('score__final_score')
# allData = Media.objects.filter(moods=mood, content_type=Media.PICTURE)
# dataDict = {}
# for m in allData:
# if m.score.final_score == -1:
# dataDict[m] = m.score.initial_score
# else:
# dataDict[m] = m.score.final_score
#
# for key, value in sorted(dataDict.iteritems(), key=lambda (k,v): (v,k)):
# mediaData.append(key)
if len(mediaData) == 0:
return preadd
#Add to database if score is higher than existing scores
lowestMediaIndex = 0
while len(sorted_picture_list) != 0:
tuple = sorted_picture_list.pop(0)
score = tuple[3]
if score > mediaData[lowestMediaIndex].score.final_score:
destory(mediaData[lowestMediaIndex].id, Media.PICTURE)
assert Picture.add(tuple[0],
tuple[1],
tuple[2],
initialScore=tuple[3])
lowestMediaIndex += 1
if lowestMediaIndex >= len(mediaData):
break
else:
break
return preadd + lowestMediaIndex
