Scraping Nick Saban's seasons as Alabama head coach

saban-seasons.py

"""
Scraping Nick Saban's seasons as Alabama head coach

I was curious what % of his time Alabama has spent at #1
"""

from collections import Counter
from bs4 import BeautifulSoup
import requests


def to_numeric(s):
    try:
        return int(s)
    except:
        return None

def parse_season(season: int):
    url = f"https://www.sports-reference.com/cfb/schools/alabama/{season}-schedule.html"
    response = requests.get(url)
    soup = BeautifulSoup(response.text, 'lxml')
    rows = soup.find('table', id='polls').find_all('tr')
    headers = [th.get('data-stat') + f'/{season}'
               for th in rows[0].find_all('th', recursive=False)]
    ranks = [to_numeric(td.get_text()) for td in rows[1].children]
    return dict(zip(headers, ranks))

def main():
    career = {}
    for season in range(2007, 2019):
        career.update(parse_season(season))
    
    total = len(career)
    counts = Counter(career.values())
    for rank in range(1, 25):
        weeks_at_rank = counts.get(rank, 0)
        print(rank, weeks_at_rank / total)

if __name__ == "__main__":
    main()