因为有很多没评分的和很久一起的测评,所以我就强行正则了
# !-*-coding:utf-8-*-
# ! 2019/3/5 20:35
# !@Author:Cy 2019 03
# !@File:gamer-sky.py
import requests
import re
import time
import csv
res = r'title=\\"(.*?)\\">'
res2 = r'num\\">(.*?)</div>'
res3 = '《(.*?)》'
res4=r'<div class=\\"time\\">(.*?)</div>'
f = open('a.csv', 'w', encoding='utf-8')
ff = csv.writer(f)
for i in range(1, 50):
url = "https://db2.gamersky.com/LabelJsonpAjax.aspx?callback=jQuery18305156710057641445_" \
"1551772185891&jsondata={%22type%22%3A%22updatenodelabel%22%2C%22isCache%22%3Atrue%2C" \
"%22cacheTime%22%3A60%2C%22nodeId%22%3A%2220465%22%2C%22isNodeId%22%3A%22true%22%2C%22" \
"page%22%3A" + str(i) + "}&_=1551772191095"
a = requests.get(url).text
b = re.findall(res, a, re.S | re.M)
# print(b)
c = re.findall(res2, a, re.S | re.M)
t = re.findall(res4, a, re.S | re.M)
mz = []
allp = []
#print(t)
for j in range(len(b)):
try:
m = re.findall(res3, b[j], re.S | re.M)
mz.append(m[0])
sb = b[j].split('》')[-1]
allp.append(sb.split(' ')[-1] if (len(sb.split(' ')) >= 2) else sb.split(':')[-1])
except:
del c[j]
del t[j]
continue;
# print(allp,len(allp))
print(c,len(c))
# print(mz,len(mz))
print(t,len(t))
chang = len(c)
for j in range(chang):
if (c[j] == '--'):
'''
del mz[j]
del c[j]
del allp[j]
'''
print("1")
else:
#work=(mz[j],c[j],allp[j])
#list(work) #tuple(seq)将列表转换为元组。
work = []
work.append(mz[j])
work.append(c[j])
work.append(allp[j])
work.append(t[j])
print(work)
ff.writerow(work)
time.sleep(0.1)