পেজিনেশন লিংক স্ক্রাপ করা

from bs4 import BeautifulSoup
import requests
root = 'https://subslikescript.com'
website = f'{root}/movies_letter-A'
result =requests.get(website)
soup = BeautifulSoup(result.content,'html.parser')
# print(soup)

# pagination
pagination = soup.find('ul', class_='pagination')
pages = pagination.find_all('li',class_='page-item')
lastpage = pages[-2].get_text()
print(lastpage)

for page in range(1,int(lastpage)+1 )[:2]:
   website = f'{root}/movies_letter-A?page={page}'
   result = requests.get(website)
   soup = BeautifulSoup(result.content,'html.parser')
   box = soup.find('article',class_="main-article")
   links=[] 
   for link in box.find_all('a',href=True):
       links.append(link['href']) 

   for link in links:
    try:
        website = f'{root}/{link}'
        result =requests.get(website)
        soup = BeautifulSoup(result.content,'html.parser')
        box = soup.find('article',class_="main-article")
        title = box.find('h1').get_text()
        transcript = box.find('div',class_='full-script').get_text()
        with open(f'{title}.txt','w',encoding='utf-8') as file:
            file.write(transcript)
        
    except:
        print('--------- Link Not Found---------')

import string

def sanitize_file_name(title):
    # Define a translation table to remove unwanted characters
    # We will keep all alphanumeric characters and spaces and replace others with underscores
    allowed_characters = string.ascii_letters + string.digits + ' '  # Alphanumeric characters and spaces
    translation_table = str.maketrans('', '', ''.join(c for c in string.printable if c not in allowed_characters))

    # Remove unwanted characters from the title
    sanitized_title = title.translate(translation_table)

    # Remove leading and trailing whitespaces and replace spaces with underscores
    sanitized_title = sanitized_title.strip().replace(' ', '_')

    # Limit the maximum length of the file name to 100 characters (you can adjust this value)
    max_length = 100
    sanitized_title = sanitized_title[:max_length]

    return sanitized_title
    
    sanitized_title = sanitize_file_name(title)
    
          with open(f"data/a/{sanitized_title}.txt",'w',encoding='utf-8') as file:
                file.write(transcript)