Deep_match_rank_emb/emb_faiss/faiss-tzld-videos-users-item2vec/embedding_manager_user.py

# encoding:utf-8
import pandas as pd
import json
import numpy as np
# import faiss
import time


class EmbeddingManagerUser(object):

    def __init__(self, fpath, key_name, value_name):

        begin_time = time.time()
        # pandas.dataframe
        self.df = pd.read_csv(fpath)
        read_time = time.time()
        print("read csv embedding file cost time is:  " + str(read_time - begin_time))

        # 将文件中的embedding加载到内存
        self.dict_embedding = self.load_embedding_to_dict(key_name, value_name)
        emb_time = time.time()
        print("load embedding to dict cost time is:  " + str(emb_time - read_time))

    def get_embedding(self, key):
        if str(key) in self.dict_embedding.keys():
            return self.dict_embedding[str(key)]
        else:
            return ""

    def load_embedding_to_dict(self, key_name, value_name):

        return {
            str(row[key_name]): row[value_name]
            for index, row in self.df.iterrows()
        }