extract.py

   1 #!/usr/bin/env python3
   2 # Tool to extract famous phrases from wikipedia
   3 from bs4 import BeautifulSoup
   4 import requests
   5
   6 url = 'https://en.wikipedia.org/wiki/List_of_Latin_phrases_(full)'
   7 response = requests.get(url)
   8 html = response.content
   9
  10 soup = BeautifulSoup(html, "html.parser")
  11 list_table = soup.find_all("table", attrs={"class":"wikitable"})
  12
  13 output = []
  14
  15 for table in list_table:
  16     for row in table.find_all("tr")[1:]:
  17         cell = row.find_all("td")
  18         for content in cell:
  19             text = content.get_text()
  20             output.append(text)
  21
  22 print(output)
  23