Get n_solved / n_submitted by scraping the competitive coding websites

codechef_spoj_hackerearth_userstats.py

import requests
import bs4

def get_codechef_userstats(user_id):
    soup = bs4.BeautifulSoup(
    requests.get('https://www.codechef.com/users/%s' % user_id).content)
    # Parse the data in the first td tag of the 2nd tr tag of the table with id "problem_stats"
    stats_row = soup.find('table', {'id': 'problem_stats'}).findAll('tr')[1].findAll('td')
    n_solved = int(stats_row[0].text)
    n_submitted = int(stats_row[2].text)
    
    return (n_solved, n_submitted)

def get_spoj_userstats(user_id):
    soup = bs4.BeautifulSoup(
    requests.get('https://www.spoj.com/users/%s' % user_id).content)
    stats_row = soup.find('dl').findAll('dd')
    n_solved = int(stats_row[0].text)
    n_submitted = int(stats_row[1].text)
    return n_solved, n_submitted

def get_hackerearth_userstats(user_id):
    soup = bs4.BeautifulSoup(requests.get(
        'https://www.hackerearth.com/users/pagelets/%s/hackerearth-profile-overview/'
        % user_id).content)
    stats_row = soup.find('tbody').findAll('td')
    n_solved = int(stats_row[0].text)
    n_submitted = int(stats_row[2].text)
    return n_solved, n_submitted

def get_userstats(site, user_id):        
    """Get the (n_solved, n_submitted) for the given user_id"""
    user_id = user_id.lower().strip()
    site = site.lower().strip()
    try:
        return globals()["get_%s_userstats" % site](user_id)
    except:
        return (-1, -1)