Python र HTML फाईलहरूको प्रयोग गरी साइटमाप्स XML सिर्जना गर्ने

"मूसिसभविष्य बाणि

<XML संस्करण="1.0"घेरिएको="UTF-8"?>
<<<<<<मूसिस
      XMLNs: XSI="Http: // Wooho। I.org/2001/xml स्क्रियन-उदाहरण" "
	  XMLNs: XHTML="Http: // Woohoo। Hever.org199999999999/xhml"
      Xsi: स्क्वामेन्ट="Http: // Woohoo .Sitemaps.org / Schamap / 0.9 HTTP: // WHTP
      XMLNs="Http: // Heohoo .siemapss.org / schahaas / chirmap / 0" "" "" "" "" "" "" "" "" "" "" "" "" "" "" "" "" "" ""भविष्य बाणि	  	  
---- www.Check-डोमेन :.com Peeemap बिजुली ----भविष्य बाणि

  
    <स्पिन>Https://tour-2ebsite.com/example- अफ्रिका
    <चट्टान आमोड>2021-11-11--30T7: 19: zat + + +00: 00
    <परिवर्तनफ्रेक>साप्ताहिक
    <प्राथमिकता>0.64400
  

import os
import re
import random
import unidecode


def read_text_from_file(file_path):
    """
    Aceasta functie returneaza continutul unui fisier.
    file_path: calea catre fisierul din care vrei sa citesti
    """
    with open(file_path, encoding='utf8') as f:
        text = f.read()
        return text


def write_to_file(text, file_path):
    """
    Aceasta functie scrie un text intr-un fisier.
    text: textul pe care vrei sa il scrii
    file_path: calea catre fisierul in care vrei sa scrii
    """
    with open(file_path, 'wb') as f:
        f.write(text.encode('utf8', 'ignore'))

def creeaza_fisier_xml():
    lista_nume_fisiere = preia_nume_fisiere_html(
        'e:\\YOUR-FOLDER-WITH-HTML-FILES',
        # files to ignore
        ['404-1.html', '404-2.html', '404-3.html']
    )
    start_xml = read_text_from_file('start_xml.txt')
    final_xml = read_text_from_file('final_xml.txt')
    model_xml = read_text_from_file('model_xml.txt')

    xml_text = start_xml + '\n'
    link_pattern = re.compile('(.*?)')
    link = re.findall(link_pattern, model_xml)
    if len(link) != 0:
        link = link[0]

    for nume_fisier in lista_nume_fisiere:
        model = model_xml
        model = model.replace(link, 'https://YOUR-WEBSITE.com/' + nume_fisier)
        xml_text = xml_text + model + '\n'

    xml_text = xml_text + final_xml

    write_to_file(xml_text, 'rss.xml')
    print("Scriere efectuata cu succes.")

def preia_nume_fisiere_html(folder_fisiere_html, lista_fisiere_de_ignorat):
    lista_nume_fisiere_html = list()
    for f in os.listdir(folder_fisiere_html):
            if f.endswith('.html'):
                lista_nume_fisiere_html.append(f)
    lista_finala = list()
    for f in lista_nume_fisiere_html:
        if f not in lista_fisiere_de_ignorat:
            lista_finala.append(f)
    return lista_finala

def main():
    creeaza_fisier_xml()

if __name__ == '__main__':
    main()