| 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657 |
- # coding=utf-8
- '''
- Created on 2016年4月5日
- 使用【kind_uu.csv】文件得到uu的类目
- @author: ChenHao
- '''
- from analysis import detailAnalysis_kind
- from pymongo.mongo_client import MongoClient
- from util_common import Constant
- file_path = "../spider_download/Other/kind_uu.csv"
- '''
- 1、根据提供的分类数据,得到解析所需的数组
- PS:在wps里面将文件转成csv时,考虑到内容中","的影响,先手动将全文的","(English)替换为","(汉字);在存储时还原
- '''
- fin = open(file_path, "r")
- lines = fin.readlines()
- fin.close()
- kinds_list = list()
- for index, line in enumerate(lines):
- temp_str = line.replace("\n", "")
- temp_list = temp_str.split(",")
- kind_list = list()
- for tl in temp_list[2:]:
- if tl is not "":
- kind_list.append(tl.replace(",", ","))
- print (index, kind_list)
-
- kinds_list.append(kind_list)
- '''
- 2、解析得到分类并存入mongodb数据库
- '''
- kindAnalysis = detailAnalysis_kind.DetailAnalysisKind(None)
- for kind_list in kinds_list:
- kindAnalysis._add_into_out_list(kind_list)
- kindAnalysis._set_detno_and_uuidForCmp()
- kindAnalysis._set_isLeaf()
- kind_uu = kindAnalysis.out_list
- print (kind_uu)
- '''
- 3、存入数据库
- '''
- cli = MongoClient(Constant.MONGODB_URL)
- db = cli.spider
- db.kind_uu.insert_many(kind_uu)
- cli.close()
|