1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950 |
- import requests
- from bs4 import BeautifulSoup
- import sqlite3
- from progress.bar import IncrementalBar
- from tqdm import tqdm
- Database = sqlite3.connect("Film_list.db")
- cur = Database.cursor()
- cur.execute("""CREATE TABLE IF NOT EXISTS Film_list (
- ID INTEGER PRIMARY KEY,
- NAME TEXT,
- YEAR TEXT,
- LINK TEXT)""")
- Database.commit()
- url = f"https://hd-38.lordfilm-s.me/films/"
- r = requests.get(url)
- soup = BeautifulSoup(r.text, "lxml")
- number_of_films = int(soup.find("div", {"class": "navigation"}).text.split()[-1])
- name_year = [""]*24*number_of_films
- link = [""]*24*number_of_films
- counter = 0
- number = 1
- bar = IncrementalBar('Download links ', max = number_of_films)
- while number <= number_of_films:
- r = requests.get(url+f"page/{number}/")
- soup = BeautifulSoup(r.text, "lxml")
- for i, j, k in zip(soup.find_all("div", class_ = "th-title"),soup.find_all("div", class_ = "th-year"),soup.find_all("div", class_ = "th-item")):
- name_year[counter] = i.text + ' ('+ j.text+')'
- link[counter] = k.find("a", class_ = "th-in with-mask").get('href')
- counter += 1
- number += 1
- bar.next()
- bar.finish()
- for i in tqdm(range(24*number_of_films), desc="Upload in Database"):
- cur.execute("""INSERT INTO Film_list (NAME, LINK) VALUES(?, ?);""", (name_year[i], link[i]))
- Database.commit()
|