import numpy as np
from sklearn.metrics.pairwise import cosine_similarity

class MusicRecommendationSystem:
    def __init__(self, user_item_matrix):
        self.user_item_matrix = np.array(user_item_matrix)
        self.user_similarity_matrix = None

    def calculate_user_similarity(self):
        # 计算用户之间的余弦相似度
        self.user_similarity_matrix = cosine_similarity(self.user_item_matrix)
        np.fill_diagonal(self.user_similarity_matrix, 0)  # 将对角线上的相似度设为0

    def recommend(self, user_index, num_recommendations=5):
        # 为指定用户推荐音乐
        if self.user_similarity_matrix is None:
            self.calculate_user_similarity()

        # 获取指定用户的相似度向量
        user_similarity_vector = self.user_similarity_matrix[user_index]

        # 找到最相似的几个用户
        similar_users_indices = np.argsort(-user_similarity_vector)[1:num_recommendations+1]

        # 计算推荐分数
        user_ratings = self.user_item_matrix[user_index]
        similar_users_ratings = self.user_item_matrix[similar_users_indices]
        weighted_similar_users_ratings = similar_users_ratings * user_similarity_vector[similar_users_indices][:, np.newaxis]
        recommendation_scores = np.sum(weighted_similar_users_ratings, axis=0) / np.sum(np.abs(user_similarity_vector[similar_users_indices]))

        # 排序推荐分数并获取推荐的音乐索引
        recommended_music_indices = np.argsort(-recommendation_scores)

        # 过滤掉用户已经评分的音乐
        recommended_music_indices = [index for index in recommended_music_indices if user_ratings[index] == 0]

        return recommended_music_indices[:num_recommendations]

# 示例用户-音乐评分矩阵
user_item_matrix = [
    [5, 3, 0, 0],
    [4, 0, 0, 2],
    [1, 1, 0, 5],
    [1, 0, 0, 4],
    [0, 1, 5, 4],
]

# 创建推荐系统实例
recommendation_system = MusicRecommendationSystem(user_item_matrix)

# 计算用户相似度
recommendation_system.calculate_user_similarity()

# 为用户0推荐音乐
recommended_music_for_user0 = recommendation_system.recommend(user_index=0)

print("Recommended music indices for user 0:", recommended_music_for_user0)