import requests, base64
from bs4 import BeautifulSoup
import re
import gradio as gr

def get_blocked_urls():
  """
    Get a list of blocked URLs.

    Returns:
      list: A list of blocked URLs.

    Raises:
      None.
  """
  url = 'https://colab.research.google.com/'
  r = requests.get(url)
  if r.status_code == 200:
    result = []
    soup = BeautifulSoup(r.text, 'html.parser')
    # search for script that contains "external_polymer_binary" in attr
    for script in soup.find_all('script'):
      if "external_polymer_binary" in str(script):
        
        r_js = requests.get(script['src'])
        # print(r_js.text)

        pattern = r"'(.*?)webui(.*?)'"
        match = re.search(pattern, r_js.text)
        raw_string = match.group(0)

        # trim 1 char front and back, split the text with ';' into array
        raw_string = raw_string[1:-1].split(';')
        result = raw_string
        for i in range(len(result)):
          try:
            decodedurl = base64.b64decode(result[i]).decode('utf-8')
        
          except:
            decodedurl = None
          if decodedurl:
            result[i] = f"{result[i]} < {decodedurl} >[thisisb64]"

    if len(result) > 0:
      return (result)
    else:
      return (["failed :<"])

  else:
    return (["res code: "+r.status_code])


def handle_refresh():
  """
  Generates an HTML ordered list of blocked URLs.

  Returns:
    str: The HTML string containing the ordered list of blocked URLs.
  """
  xs = "<ol>"
  for url in get_blocked_urls():
    if "[thisisb64]" in url:
      url = url.replace("[thisisb64]", "")
      # primary color [primary_100]
      nondecoded = url.split('<')[0]
      decodedurl = url.split('<')[1]
      decodedurl = f"<{decodedurl}>"
      xs += "<li><code>"+nondecoded+"</code> <font color="#4EACEF">&lt;"+decodedurl+"&gt;</font></li>"
    else:
      xs += "<li><code>"+url+"</code></li>"
  xs += "</ol>"
  return xs
    
  

with gr.Blocks(
    analytics_enabled=False, title="GGL Checks", theme="NoCrypt/miku"
) as demo:
    gr.HTML("""<center><h1>GGL Checks</h1></center>""")
    refresh = gr.Button("Refresh", variant="primary")
    html = gr.HTML()
    refresh.click(handle_refresh, outputs=[html])


demo.launch(debug=True)