def cluster(similarity=pearson):
db = SQLite3().cursor()
pins = Pins()
words = Words()
maxcount = pins.size()
labels = [r[0] for r in words.set()]
offset = 0
n = 1
sum_num = sum([i for i in range(1, maxcount)])
while True:
pin_a = pins.find_by_offset(offset)
pin_a_words = [w[0] for w in words.find_by_pinid(pin_a[0])]
pin_a_wordcount = [pin_a_words.count(w) for w in list(set(labels))]
# calculate distance of two pins.
for i in range(offset + 1, maxcount):
pin_b = pins.find_by_offset(i)
pin_b_words = [w[0] for w in words.find_by_pinid(pin_b[0])]
pin_b_wordcount = [pin_b_words.count(w) for w in list(set(labels))]
# calculate distance of two pins
print '[%s] %s / %s calculate score of %s and %s' % (
datetime.today().strftime('%Y-%m-%d %H:%M:%S'), n, sum_num,
pin_a[0], pin_b[0])
sim = 1.0 - similarity(pin_a_wordcount, pin_b_wordcount)
# save distance to database for cache
clusters = Clusters()
clusters.data['pin_id_a'] = pin_a[0]
clusters.data['pin_id_b'] = pin_b[0]
clusters.data['score'] = sim
clusters.save()
n += 1
if offset >= maxcount - 1:
break
offset += 1
def save(self):
id = self.get_image_id()
if not id:
return 0
pins = Pins()
pins.data['pin_id'] = id
pins.data['image_url'] = self.get_image()
pins.data['description'] = self.get_description()
pins.data['refer_url'] = self.get_refer_url()
if not pins.save():
return 0
repins = self.get_repins()
wordcount = {}
for user in repins:
words = Words()
words.data['pin_id'] = id
words.data['user'] = user
# delete and insert in preperation for change of board name.
words.delete()
for w in repins[user]:
words.data['word'] = w
words.save()
return 1
def get_candidates(offset=0, num=20, threshold=0.5):
pins = Pins()
pins = pins.get_candidates(offset, num, threshold)
return [{'pin_id': str(pin[0]), 'uri': str(pin[1])} for pin in pins]