def get_title(title):
title_url = urllib2.unquote(title).replace(' ', '-').lower()
title = urllib2.quote(title_url.replace('-', ' '))
key = title_url
movie = memcache.get(key)
if movie is None:
logging.info('DB QUERY')
q = db.Query(Movie)
q.filter('title =', key)
result = q.fetch(limit=1)
if len(result) > 0:
logging.info('MOVIE FOUND IN DB')
row = result[0]
movie = row.data
memcache.set(key, movie)
if movie is None:
logging.info('FETCHING MOVIE FROM API')
movie = query.query_movie(title)
movie = set_star_ratings(movie)
movie = set_covers(movie)
m_dict = json.loads(movie)
if m_dict['total'] != 0:
m = Movie(title=key, data=movie)
m.put()
memcache.set(key, movie)
else:
movie = '{}'
return json.loads(movie)
def get(self, category):
if category == 'year':
return self.year
elif category == 'hosts':
if not self.hosts:
c = 0
count = Counter()
for tweet in self.corpus:
text = tweet['text']
if 'host' in text:
c += 1
tags = nltk.pos_tag(nltk.word_tokenize(text))
for i in xrange(len(tags)-1):
first = tags[i][1] == "NNP" and tags[i][0] not in stopwords
second = tags[i+1][1] == "NNP" and tags[i+1][0] not in stopwords
if first and second:
count["%s %s" % (tags[i][0], tags[i+1][0])] += 1
if c > 5:
break
most_common = count.most_common(2)
self.hosts = [most_common[0][0]]+[most_common[1][0]]
return self.hosts
elif category == 'awards':
return self.awards.keys()
elif category == 'winner':
winner_dict = {}
# count = Counter()
# win_pattern = re.compile('w[io]n', re.IGNORECASE)
for award in self.awards:
award_pattern = re.compile(self.regexes[award])
award_type = self.award_type[award]
if not self.awards[award]['winner']: # Fill the dict
name_query = []
if award_type:
name_query = query.query_name(self.corpus, award_pattern, g=2, n=1)
else:
name_query = query.query_movie(self.corpus, award_pattern, g=2, n=1)
if name_query:
name_tuple = name_query[0][0]
name_count = name_query[0][1]
print award, name_tuple, name_count
name_string = "%s %s" % name_tuple
winner_dict[award] = name_string
self.awards[award]['winner'] = name_string
else:
winner_dict[award] = self.awards[award]['winner']
return winner_dict
elif category == 'nominees':
nominees_dict = {}
# count = Counter()
# nomin_pattern = re.compile('(nomin|w[io]n)', re.IGNORECASE)
for award in self.awards:
award_pattern = re.compile(self.regexes[award])
award_type = self.award_type[award]
if not self.awards[award]['nominees']: #Fill the dict
name_query = []
if award_type:
name_query = query.query_name(self.corpus, award_pattern, g=2, n=5)
else:
name_query = query.query_movie(self.corpus, award_pattern, g=2, n=5)
nomin_list = []
for name in name_query:
if name != []:
name_tuple = name[0]
name_count = name[1]
print award, name_tuple, name_count
name_string = "%s %s" % name_tuple
nomin_list.append(name_string)
nominees_dict[award] = nomin_list
self.awards[award]['nominees'] = nomin_list
else:
nominees_dict[award] = self.awards[award]['nominees']
return nominees_dict
elif category == "presenters":
presenters_dict = {}
for award in self.awards:
award_pattern = re.compile(self.regexes[award])
# award_type = self.award_type[award] #No need to check award type - presenters only people
if not self.awards[award]['presenters']: #fill the dict
name_query = query.query_name(self.corpus, award_pattern, g=2, n=7)
presenter_list = []
for name in name_query[5:]:
if name != []:
name_tuple = name[0]
name_count = name[1]
print award, name_tuple, name_count
name_string = "%s %s" % name_tuple
presenter_list.append(name_string)
# print name_string, award
presenters_dict[award] = presenter_list
self.awards[award]['presenters'] = presenter_list
else:
presenters_dict[award] = self.awards[award]['presenters']
return presenters_dict
else:
raise KeyError("GET called with improper argument")
