new_tiku_structure_2021/utils/get_data.py

"""
从文件数据库中获取文本数据
"""
import json
import os
import re
import requests
import pymongo


class Mongo(object):
    def __init__(self):
        self.conn = pymongo.MongoClient("mongodb://192.168.1.140:27017/")  # 创建数据库连接
        # self.conn = pymongo.MongoClient('mongodb://localhost:27017/')  # 创建数据库连接
        self.db_name = 'zy'  # 数据库名称
        self.table_name = 'spider.files'  # 表名称
        self.spider_db = self.conn[self.db_name]  # 新建数据库名，
        self.meta_db = self.spider_db[self.table_name]  # meta data about files  文件题库


    def get_data_info(self, query):
        for data in self.meta_db.find(query):
            return data


if __name__ == "__main__":
    from bson.objectid import ObjectId  # 只需安装pymongo即可
    mongo = Mongo()
    aa = mongo.get_data_info({"_id" : ObjectId("5fc64c094994183dda7e7575")})
    print(aa["item_ocr"])