brandizzi icon

Untitled

brandizzi | PRO | 08/01/16 05:32:30 PM UTC | 0 ⭐ | 608 👁️ | Never ⏰ | []
Python |

870 B

|

None

|

0 👍

/

0 👎

# -*- coding: utf-8 -*-
from bs4 import BeautifulSoup
import urllib2
 
ukr = "http://www.ukrstat.gov.ua/operativ/operativ2013/rp/zn_ed_reg/zn_ed_reg_e/zn_ed_reg_2014k_e.htm"
page = urllib2.urlopen(ukr)
soup = BeautifulSoup(page)
table = soup.find('table', id = "table2")
 
NAM=[]
TOT=[]
AGR=[]
IND=[]
CON=[]
TRA=[]
STO=[]
ACC=[]
INF=[]
FIN=[]
 
for row in table.findAll("tr"):
    cells = row.findAll("td")
    if len(cells) == 10:
        NAM.append(cells[0].text.strip())
        TOT.append(cells[1].text.strip())
        AGR.append(cells[2].text.strip())
        IND.append(cells[3].text.strip())
        CON.append(cells[4].text.strip())
        TRA.append(cells[5].text.strip())
        STO.append(cells[6].text.strip())
        ACC.append(cells[7].text.strip())
        INF.append(cells[8].text.strip())
        FIN.append(cells[9].text.strip())
 
print NAM

Comments