Python DB.get_instance方法代碼示例

# 需要導入模塊: from db import DB [as 別名]
# 或者: from db.DB import get_instance [as 別名]
	def load(user_id=None, user_name=None):
		assert user_id is not None or user_name is not None

		u = None
		if user_id is not None:
			u = DB.get_instance().query(Author).filter_by(id=user_id).first()
		elif user_name is not None:
			u = DB.get_instance().query(Author).filter_by(name=user_name).first()

		return u

# 需要導入模塊: from db import DB [as 別名]
# 或者: from db.DB import get_instance [as 別名]
	def __init__(self, user):
		self.user = user.id
		self.user_vec = ProfileOneHotEncoder.add_info_to_vec(self.user.vector, self.user.gender, self.user.location,
		                                                     self.user.emotion)
		self.graph = Author.load_graph()
		self.graph.add_node(user.id)
		favorite_tweets = DB.get_instance().query(Favorite.tweet_id).filter(Favorite.user_id == user.id)
		self.authors_liked = DB.get_instance().query(Tweet.user_name).filter(
			Tweet.id.in_(favorite_tweets.subquery())).all()
		for a in self.authors_liked:
			self.graph.add_edge(user.id, a[0])

# 需要導入模塊: from db import DB [as 別名]
# 或者: from db.DB import get_instance [as 別名]
	def create_authors(limit=None):
		from models.tweet import Tweet
		tpc = TopicsClassifier(limit=limit)
		pp = PredictionProfile(limit=limit)

		query = DB.get_instance().query(Tweet.user_id, Tweet.vector, Tweet.user_name)

		if limit is not None:
			query = query.limit(limit)

		for t in query.all():
			a = Author.load(user_name=t.user_name)
			if a is None:
				a = Author()
				a.id = t.user_id
				a.name = t.user_name
				a.nb_click = 0
				a.vector = np.zeros(300)
				DB.get_instance().add(a)
			a.update_profile(np.array(t.vector))
			DB.get_instance().commit()

		print('Authors created')

		# graph = Author.load_graph()
		# centralities = nx.eigenvector_centrality(graph)
		for author in DB.get_instance().query(Author).all():
			# author.centrality = centralities[author.name] if author.name in centralities else 0.
			author.predict_profile(tpc, pp)
			DB.get_instance().commit()

# 需要導入模塊: from db import DB [as 別名]
# 或者: from db.DB import get_instance [as 別名]
	def _load_enc():
		encoder = OneHotEncoder()
		X = []
		query = DB.get_instance().query(Tweet.gender, Tweet.country, Tweet.sentiment).distinct(Tweet.country).group_by(
			Tweet.country, Tweet.gender, Tweet.sentiment)
		for gender, country, sentiment in query.all():
			X.append([gender, country, sentiment])
		encoder.fit(X)
		return encoder

# 需要導入模塊: from db import DB [as 別名]
# 或者: from db.DB import get_instance [as 別名]
	def update_profile(self, vec):
		"""
		Update the profile of the user with the new vec param
		:param vec: (np.array) vector of the tweet to add
		:return:
		"""

		assert type(vec) == np.array

		self.nb_click += 1
		for i in range(len(self.vector)):
			self.vector[i] = (self.vector[i] * (self.nb_click - 1)) / self.nb_click + (vec[i] / self.nb_click)

		tpc = TopicsClassifier()
		pp = PredictionProfile()
		self.predict_profile(tpc, pp)

		DB.get_instance().commit()

# 需要導入模塊: from db import DB [as 別名]
# 或者: from db.DB import get_instance [as 別名]
	def remove_favorite(self, tweet):

		from models.favorite import Favorite
		from models.tweet import Tweet

		DB.get_instance().query(Favorite).filter_by(user_id=self.id).filter_by(tweet_id=tweet.id).delete(
			synchronize_session='fetch')

		favs = DB.get_instance().query(Favorite.tweet_id, Tweet.vector).join(Tweet,
																			 Tweet.id == Favorite.tweet_id).filter(
			'favorites.user_id=' + str(self.id)).all()
		vects = [f[1] for f in favs]
		self.vector = np.mean(vects, axis=0)

		tpc = TopicsClassifier()
		pp = PredictionProfile()
		self.predict_profile(tpc, pp)

		DB.get_instance().commit()

# 需要導入模塊: from db import DB [as 別名]
# 或者: from db.DB import get_instance [as 別名]
 def __init__(self, user, parent):
     QAbstractTableModel.__init__(self, parent)
     self.current_user_name = user
     self.db = DB.get_instance()
     self.words = self.get_words()
     self.columns_keys = self.get_columns_keys()
     self.columns_names = self.get_columns_names()
     self.rows_count = len(self.words)
     self.columns_count = len(self.columns_names)
     self.portuguese_index = self.get_portuguese_index()

# 需要導入模塊: from db import DB [as 別名]
# 或者: from db.DB import get_instance [as 別名]
	def rerank_authors(self, authors_prio):
		"""
		rerank results from authors_prio based on their similarity with the user
		:param authors_prio: the jaccard coefficient for the link prediction between the user and each kept author
		:return:
		"""
		reranked_reco = []
		for a, p in authors_prio:
			author = DB.get_instance().query(Author).filter(Author.name == a).first()
			author_vec = ProfileOneHotEncoder.add_info_to_vec(author.vector, author.gender, author.localisation,
			                                                  author.emotion).reshape(1, -1)
			sim = cosine_similarity(self.user_vec, author_vec)
			reranked_reco.append({'author_name': a, 'sim': sim[0]})
		return sorted(reranked_reco, key=lambda k: k['sim'], reverse=True)

# 需要導入模塊: from db import DB [as 別名]
# 或者: from db.DB import get_instance [as 別名]
	def __init__(self):
		self.corpus = None
		# self.load_corpus()
		self.parser = Parser()
		self.tweets = DB.get_instance().query(Tweet).limit(100)
		print('tweets loaded')