stu

#_____________________________________________
import requests
from bs4 import BeautifulSoup
import pprint as ppr
import re
import pymongo
from openpyxl import load_workbook
#_____________________________________________
class Web_crawling:
    def __init__(self, url):
        self.req_data = requests.get(url) # insert url data
        self.html_data = self.req_data.text
        self.soup_data = BeautifulSoup(self.html_data, "html.parser")
        self.re_sorting_1 = re.compile("http*")
        self.re_sorting_2 = re.compile(".jpg")
        self.Image_url = list()
        self.wb_excel = load_workbook('C:\\Users\\sleep\\Desktop\\stu.xlsx')

    def HTML_INFOMATION(self):
        #print (self.soup_data)
        pass

    def Want_data_sorting(self):
        ws = self.wb_excel.create_sheet(title="Test_url")
        tmp_value = self.soup_data.findAll("img")
        for Link in tmp_value:
            Link = str(Link)
            LeftIndex = self.re_sorting_1.search(Link)
            RigthIndex = self.re_sorting_2.search(Link)
            try:
                indx_L= LeftIndex.span()
                indx_R= RigthIndex.span()
                self.Image_url.append(Link[indx_L[0]:indx_R[1]])
            except:
                pass
        ppr.pprint (self.Image_url)

        Indx = 1
        for v in self.Image_url:  # e_cp
            In_alphabet = 66
            # print (k, type(v)) # , dictionary
            s = chr(65) + str(Indx)
            print(s)
            ws[s] = v
            s = chr(In_alphabet) + str(Indx)
            print(s)
            ws[s] = v
            In_alphabet += 1
            Indx += 1

    def __del__(self):
         self.wb_excel.save('C:\\Users\\sleep\\Desktop\\stu.xlsx')

def main():
    me_craw = Web_crawling("http://www.goodhousekeeping.com/home/decorating-ideas/tips/g3902/interior-paint-colors")
    me_craw.HTML_INFOMATION()
    me_craw.Want_data_sorting()
if __name__ == "__main__":
    main()