打包的代码如下：
import requests
import re
import openpyxl
url='http://www.tensorfly.cn/tfdoc/get_started/basic_usage.html'
saveas='Tensorflow.xlsx'
def get_html(url):
r=requests.get(url)
print(r.status_code)
#r.encodeing='utf-8'
print(r.encoding)
html=r.text.encode(r.encoding).decode('utf-8')
#print(html)
return html
def get_string(html):
string=r'(.*?)[\W^\u4e00-\u9fa5]*?(.*?)[\W^\u4e00-\u9fa5]*?'
string2=r'(.*?)([\w\W]*?)'
s=re.compile(string2)
codelist=s.findall(html)
print(codelist)
return codelist
def clean_string(codelist):
exlist=[]
for i in range(0,len(codelist)):
#print(list(codelist[i]))
exlist.append(list(codelist[i]))
exlist[i][1]=exlist[i][1].replace('\n','')
exlist[i][1]=exlist[i][1].replace(' ','')
print(exlist)
return exlist
def get_excel_list(codelist,saveas):
wk=openpyxl.load_workbook(saveas)
st=wk.worksheets[0]
print(len(codelist))
for i in range(1,len(codelist)+1):
st.cell(i,1).value=codelist[i-1][0]
st.cell(i,2).value=codelist[i-1][1]
wk.save(saveas)
html=get_html(url)
codelist=get_string(html)
exlist=clean_string(codelist)
get_excel_list(exlist,saveas)
求大神帮助分析问题原因
