فتاوى نور على الدرب الصوتية

alandals_scraper.py

from pathlib import Path

import httpx
from parsel import Selector

root_url = "http://alandals.net"
response = httpx.get(f"{root_url}/Sections.php")
selector = Selector(response.text)
sections_links = selector.css("tr a::attr(href)").getall()

csv_file = Path("data.csv")
csv_file.write_text("section, question, url\n")

section_count = 1
for section in sections_links[1:]:
  response = httpx.get(f"{root_url}/{section}", follow_redirects=True)
  selector = Selector(response.text)
  section_title = selector.css(
      "tr:nth-child(1) > td > p > font > b::text").get()
  questions_links = selector.css("tr a::attr(href)").getall()

  question_count = 0
  for question_link in questions_links[1:]:
    response = httpx.get(f"{root_url}/{question_link}", follow_redirects=True)
    selector = Selector(response.text)
    question_text = selector.css("td[colspan='2'] center::text").get()
    mp3_link = selector.css('a[href$=".mp3"]::attr(href)').get()
    mp3_url = f"{root_url}/{mp3_link}"
    question_count += 1
    print(f"{section_count} - {question_count} - {mp3_url}")
    with csv_file.open(mode="a", encoding="utf-8") as file:
      file.write(f"{section_title},{question_text},{mp3_url}\n")
  section_count += 1