from urllib import request
from time import sleep
import bs4
import json

BASE_URL="https://docs.aws.amazon.com/AWSCloudFormation/latest/UserGuide"
service_model = {}


def get_cf_service_resource_types(url: str):
    with request.urlopen(url) as fd:
        html = fd.read().decode('utf-8')
    child_soup = bs4.BeautifulSoup(html, 'html.parser')
    resource_types = [rt.p.a.contents[0] for rt in child_soup.find_all('li', attrs={'class': 'listitem'})]
    return resource_types


with request.urlopen(f"{BASE_URL}/aws-template-resource-type-ref.html") as fd:
    aws_cf_services = fd.read().decode('utf-8')

services_soup = bs4.BeautifulSoup(aws_cf_services, 'html.parser')
services_ul = services_soup.find(name='div', attrs={'class': 'highlights'}).ul
cf_services_list = [services.a.contents[0] for services in services_ul.children]
for service in services_ul.children:
    sobj = {'name': service.a.contents[0]}
    child_url = f"{BASE_URL}/{service.a.attrs['href'].lstrip('./')}"
    # print(f"Fetching {child_url}")
    sleep(0.1)
    for resource_identifier in get_cf_service_resource_types(child_url):
        try:
            service_provider, service_name, resource_type = resource_identifier.split('::')
            if service_provider not in service_model.keys():
                service_model[service_provider] = {'services': {}}
            if service_name not in service_model[service_provider]['services'].keys():
                service_model[service_provider]['services'][service_name] = []
            service_model[service_provider]['services'][service_name].append(resource_type)
        except Exception as exc:
            # print(exc)
            pass
print(json.dumps(service_model))