import csv
import json
import re
import xlwt
class IO_rw(object):
def __init__(self):
#excel
self.f = xlwt.Workbook(encoding='utf-8')
self.sheet1 = self.f.add_sheet(u'sheet1', cell_overwrite_ok=True)
#csv
self.file_json = open("post_weibo_simple.json", "r+", encoding="utf-8")
self.csvfile = open("test.csv", "w")
self.writer = csv.writer(self.csvfile)
b = ['id','name','location',"url","commentCount","pDate","fansCount","description","likeCount","shareCount","friendCount",
"postCount"]
self.writer.writerow(b)
def process_item(self):
dict_json = json.loads(self.file_json.read())
for i_from in dict_json['data']:
shareCount = i_from['shareCount']
id = i_from['id']
commentCount = i_from['commentCount']
pDate = i_from['pDate']
dict_i = i_from['from']
friendCount = dict_i['friendCount']
name = dict_i['name']
likeCount = dict_i['likeCount']
description = dict_i['description']
url = dict_i['url']
postCount = dict_i['postCount']
fansCount = dict_i['fansCount']
loca = dict_i['extend']
location = loca['location']
a = [id,name,location,url,commentCount,pDate,fansCount,description,likeCount,shareCount,friendCount,postCount]
self.writer.writerow(a)
def close_spider(self):
self.file_json.close()
self.csvFile.close()
if __name__ == '__main__':
r = IO_rw()
r.process_item()