extract.py

   1 #!/usr/bin/env python3
   2 # Extract Latin famous phrases from wikipedia
   3 # Armaan Bhojwani 2020
   4
   5 from bs4 import BeautifulSoup
   6 import requests
   7
   8 url = 'https://en.wikipedia.org/wiki/List_of_Latin_phrases_(full)'
   9 response = requests.get(url)
  10 html = response.content
  11
  12 soup = BeautifulSoup(html, "html.parser")
  13 list_table = soup.find_all("table", attrs={"class":"wikitable"})
  14 f = open("phrases", "w")
  15
  16 for table in list_table:
  17     for row in table.find_all("tr")[1:]:
  18         f.write("%" )
  19         cell = row.find_all("td")
  20         for content in cell:
  21             text = content.get_text()
  22             f.write("\n" + text)
  23 f.close()