XML_eshop_feed_analyser

#edit line 3 & 6

feed = "https://feeds.mergado.com/radiators-shop-google-nakupy-fc682e7b905babe86f3892c3f7c2b67a.xml"

#Enter the file path to save the result
out_path = "G:\\Dropbox\\xml_feed_attribute.xlsx"


#request the feed and parsen it
from bs4 import BeautifulSoup
import requests
r = requests.get(feed)
soup = BeautifulSoup(r.content, 'xml')



#get all the items
item_list = soup.findAll("item")

#genrate list of atribute based on the 10 first items.
Atribute_list = list()
for item in item_list[0:10]:
    stemp = item.findChildren()
    [Atribute_list.append(x.name) for x in stemp]
    
Atribute_list = list(set(Atribute_list))


#generate the list of avialable atribute
Result_dic = {}

for atribute in Atribute_list:
    stemp = soup.findAll(atribute)
    option_list = list()
    [option_list.append(x.text) for x in stemp]
    option_list = list(set(option_list))
    Result_dic[atribute] = option_list

#generate export file    
import pandas as pd
export = pd.DataFrame({"init": ["This automation was sponsored by https://www.smartlook.com/"]})

for atribute in Atribute_list:
    stemp = pd.DataFrame({atribute: Result_dic[atribute]})
    export = pd.concat([export, stemp],  axis=1)


writer = pd.ExcelWriter(out_path, engine='xlsxwriter')            
export.to_excel(writer,   header=True, index=False ) 
writer.close()