I just did a version guessing using the most common patterns. It has most of the failures as the other solutions but with comments explicitly pointing out the issues.

You should be able to click 'View Solution' under my post to see it. If not then this link should work: https://www.codewars.com/kata/reviews/553a8bb91e0399d6f70001b9/groups/5ff1ed99f0bb620001b79f96

I suggest that the url secure.ftp.g.cn be added to the test suite.

g is the domain name. secure and ftp are two subdomains and the TLD is cn.

The url is fictitious but g.cn is a real website while multiple sub-domains are a possibility.

Fair point. I deleted the comment without reading to start from a clean slate, before trying to find the schema names.

There's two bugs with the random tests when using Python. Here's the code I used to get around the bugs.

  item_count = ...
  
  # the test script seems to have several bugs in it..... !?
  # Bug A - when the index is invalid and negative return page_size
  if page_index < 0:
      return self.page_size
      
  # Bug B - when page requested is last page, return 0 if page is full
  if page_index+1 == len(self.pages) and item_count == self.page_size:
     return 0
     
  return item_count

Here's also a fix for the test script:

  item_count = b-a
  page_number = x
  max_page_index = int(ceil(item_count / float(pagelength))) - 1

  if page_number >= 0 and page_number < max_page_index:
      solution = pagelength
  elif page_number == max_page_index:
      solution = (item_count % pagelength) or pagelength
  else:
      solution = -1