url_check

"""
regex based to search if a page is inner page or home page or category page.
"""
import re


def url_check(url):
    url = url.split("/")
    url = list(filter(None, url))
    if "http" in url[0]:
        url.pop(0)
    if len(url) > 1:
        category_check = re.search(r'.*\d+', url[-1])
        if category_check is not None or ".html" in url[-1]:
            url_type = "subpage"
        else:
            url_type = "categorypage"
    else:
        url_type = "homepage"
    return url_type


print(url_check("https://edition.cnn.com/"))
print(url_check("https://edition.cnn.com/health"))
print(url_check("https://edition.cnn.com/2019/08/27/health/heart-disease-stroke-diabetes-death-rates-study/index.html"))
"""
RESULT: 
homepage
categorypage
subpage
"""