import time

start = time.time()
print("hello")


# Code to create a DF with the right shape (NxN) and initializing each cell to 0
similarity_matrix = cosine_similarity(df_ratings_dummy, df_ratings_dummy)
similarity_matrix_df = pd.DataFrame(similarity_matrix, index=df_ratings.index, columns=df_ratings.index)
similarity_matrix_df[:] = 0



sim_score = 0
for user1_id, row in similarity_matrix_df.iterrows(): #iterate over rows
    print("Calculating sim column for user_id: " + str(user1_id))
    for user2_id, value in row.items():


        if(user1_id == user2_id):
            similarity_matrix_df[user1_id][user2_id] = 1
        else:
            #Step 1: Get co-rated items between user1 and user 2
            user1_user2_df = df_ratings.loc[[user1_id, user2_id]]
            all_rated_items = user1_user2_df[user1_user2_df.columns[~user1_user2_df.isnull().all()]]
            co_rated_items = all_rated_items.dropna(axis=1)

            #Step 2: For each co-rated item, calc the sim using PIP
            #Step 3: Loop over each co-rated item/movie, and calculate the sim using user1_rating and user2_rating
            for movie_id in co_rated_items.columns:
                mu_k = df_ratings[movie_id].mean()
                user1_rating = co_rated_items[movie_id][user1_id]
                user2_rating = co_rated_items[movie_id][user2_id]

                pip = proximity(user1_rating, user2_rating) * impact(user1_rating, user2_rating) * popularity(user1_rating, user2_rating, mu_k)
                sim_score = sim_score + pip

            similarity_matrix_df[user1_id][user2_id] = sim_score
            #Step 4: Reset the sim_score
            sim_score = 0


end = time.time()
print(end - start)