import csv
from itertools import islice
import numpy as np
import math


def item_similarity(train_data, max_id):
    #与item_cf.py中的相似度计算的区别在于：本相似度算法中加入了对活跃用户的软性惩罚
    similarity = np.zeros([max_id + 1, max_id + 1])
    #item_like_num表示喜欢每个物品的人数
    item_like_num = np.zeros(max_id + 1)
    count = 0
    for user_like_list in train_data:
        iuf = 1 / math.log(1 + len(user_like_list) * 1.0)
        #iuf：inverse user frequence:用户活跃度对数的倒数
        for i in range(len(user_like_list) - 1):
            item_like_num[user_like_list[i]] += 1
            for j in range(i + 1, len(user_like_list)):
                similarity[user_like_list[i]][user_like_list[j]] += iuf
                similarity[user_like_list[j]][user_like_list[i]] += iuf
        item_like_num[user_like_list[len(user_like_list) - 1]] += 1
        count += 1
        if count % 100 == 0:
            print("物品相似度计算进度: {:.2f}%".format(count / len(train_data) * 100))
    #此时similarity[i][j]计算的是同时喜欢物品 i 和物品 j 的用户数

    for i in range(max_id + 1):
        for j in range(max_id + 1):
            similarity[i][j] /= math.sqrt(item_like_num[i] * item_like_num[j])
    #similarity此时归一化
    return similarity

def recommendation(test_user_id_list, train_user_id_list, users_like_list, similarity, K):
    user_recommendation_list = {}
    count = 0
    for user_id in test_user_id_list:
        index = train_user_id_list.index(user_id)
        recommendation_score = {}
        for liked_item_id in users_like_list[index]:
            #找出与liked_item_id的物品相似的K个物品
            topk_similar = similarity[liked_item_id].argsort()[-K:][::-1]
            for similar_item_id in topk_similar:
                if recommendation_score.get(similar_item_id) is None:
                    recommendation_score[similar_item_id] = similarity[liked_item_id][similar_item_id]
                else:
                    recommendation_score[similar_item_id] += similarity[liked_item_id][similar_item_id]
        recommendation_score = sorted(recommendation_score.items(), key=lambda x: x[1], reverse=True)
        top10_recommendation = []
        for id, score in recommendation_score[:10]:
            top10_recommendation.append(id)
        user_recommendation_list[user_id] = top10_recommendation
        if count % 100 == 0:
            print("当前推荐进度: {:.2f}%".format(count / len(test_user_id_list) * 100))
        count += 1
    return user_recommendation_list


def load_test_data(data_path):
    file = csv.reader(open(data_path, "r", encoding="utf-8"))
    user_id_list = []
    for line in islice(file, 1, None):
        user_id_list.append(int(line[0]))
    return user_id_list

def load_train_data(data_path):
    #给每个用户生成一个喜欢的物品的列表
    #max_id记录物品id的最大值
    max_id = 0
    file = csv.reader(open(data_path, "r", encoding="utf-8"))
    users_like_list = []
    train_user_id_list = [0]
    user_id = 0
    user_like_list = []
    for line in islice(file, 1, None):
        if int(line[0]) == user_id:
            user_like_list.append(int(line[1]))
        else:
            user_id = int(line[0])
            train_user_id_list.append(user_id)
            if max(user_like_list) > max_id:
                max_id = max(user_like_list)
            users_like_list.append(user_like_list)
            user_like_list = []
            user_like_list.append(int(line[1]))
    users_like_list.append(user_like_list)
    return users_like_list, train_user_id_list, max_id

def write_res_to_csv(user_recommendation_res):
    file = open("submission_v1.csv", "w", encoding="utf-8", newline="")
    csv_writer = csv.writer(file)
    csv_writer.writerow(["user_id", "item_id"])

    for user_id, recommendation_list in user_recommendation_res.items():
        for item_id in recommendation_list:
            csv_writer.writerow([str(user_id), str(item_id)])
    file.close()

if __name__ == "__main__":
    train_data_path = "./dataset/book_train_dataset.csv"
    test_data_path = "./dataset/book_test_dataset.csv"
    # K = 10即用户喜欢物品A，则寻找与物品最相似的 K种物品
    K = 10
    users_like_list, train_user_id_list, max_id = load_train_data(train_data_path)
    print("训练数据读取完成...")
    similarity = item_similarity(users_like_list, max_id)
    print("物品相似度计算完成...")
    test_user_id_list = load_test_data(test_data_path)
    print("测试user_id读取完成...")
    user_recommendation_res = recommendation(test_user_id_list, train_user_id_list, users_like_list, similarity, K)
    print("完成推荐，正在写入文件...")
    write_res_to_csv(user_recommendation_res)
    print("submission文件生成.")