load_codeforces_submissions.py

#!/usr/bin/env python3
from itertools import islice

from bs4 import BeautifulSoup
from codeforces import CodeforcesAPI, VerdictType
from urllib import request
import sys


def load_source_code_from_submission(submission):
    url = generate_submission_url(submission)
    return load_submission_source_by_url(url)


def generate_submission_url(submission):
    return 'http://codeforces.com/contest/{}/submission/{}'.format(submission.contest_id, submission.id)


def load_submission_source_by_url(url):
    with request.urlopen(url) as page:
        soup = BeautifulSoup(page.readall())
        return parse_submission_source_from_tag(soup.find('pre', class_='prettyprint'))


def parse_submission_source_from_tag(soup_tag):
    return soup_tag.string


def filter_successful_submissions(submissions):
    return filter(lambda s: s.verdict == VerdictType.ok, submissions)


def filter_submissions_by_problem_index(problem_index, submissions):
    return filter(lambda s: s.problem.index == problem_index, submissions)


def load_successful_submissions(contest_id, problem_index, start):
    api = CodeforcesAPI()
    submissions = api.contest_status(contest_id)
    submissions = filter_submissions_by_problem_index(problem_index, filter_successful_submissions(submissions))
    return islice(submissions, start, None)


def load_src_from_file(path):
    with open(path, 'r') as f:
        return f.read()


def main(contest_id, problem_id, start):
    submissions = load_successful_submissions(contest_id, problem_id, start)

    for i, s in enumerate(submissions):
        print('Processing submission #{} ({})'.format(s.id, i + start))
        submission_src = load_source_code_from_submission(s)

        with open('submissions_{}/{}'.format(contest_id, s.id), 'w') as f:
            f.write(submission_src)


if __name__ == '__main__':
    main(int(sys.argv[1]), sys.argv[2], int(sys.argv[3]))