json的格式化与文件输入输出
with open("datas.json",'w',encoding="utf-8") as f:
values = json.dumps(databasetables,indent=4,sort_keys=True,separators=(",",":"),ensure_ascii=False)
f.write(values)
爬虫
urls = "https://url?page={}&size={}"
def online(page=1,size=40):
query_url =urls.format(page,size)
session = requests.session()
res = session.get(url=query_url,headers=header)
data = json.loads(res.content)['data']
"do something by yourself"
"""
"""
excel
import xlrd
import xlwt
import time
import json
import os
filename = os.getcwd()+"/5月.xls"
xld = xlrd.open_workbook(filename)
target = xld.sheet_names()[0]
xl = xld.sheet_by_name(target)
databasetables = {
}
def initDatabase(index,name):
global databasetables
flag = 0
if databasetables.get(index) is None:
databasetables[index] = {}
else:
flag = 1
if databasetables.get(index).get(name) is None:
databasetables[index][name] = {
"value":[],
"lens":0
}
else:
flag = 2
return flag
for row in range(1,xl.nrows):
line = xl.row_values(row)
time = int(line[0]) if isinstance(line[0],float) else row
name = line[1].strip() if isinstance(line[1],str) else row
if time > row:
initDatabase(time,name)
channel = line[2].strip() if isinstance(line[2],str) else "400"
method = line[3].strip() if isinstance(line[3],str) else "其他"
province = line[4].strip() if isinstance(line[4],str) else "未知"
product = line[8].strip() if isinstance(line[8],str) else "未知"
data = {
"channel":channel,
"method":method,
"province":province,
"product":product,
}
databasetables[time][name]["lens"] += 1
databasetables[time][name]["value"].append(data)
nohup jupyter notebook > jupyter.log 2>&1 &