codeforces.py

#!/usr/bin/python3
import re
from pathlib import Path
import datetime
import requests
import os
import time
from sys import argv
from bs4 import BeautifulSoup

extensions = {
    "GNU C11": ".c",
    "Clang++17 Diagnostics": ".cpp",
    "GNU C++11": ".cpp",
    "GNU C++14": ".cpp",
    "GNU C++17": ".cpp",
    "MS C++": ".cpp",
    "MS C++ 2017": ".cpp",
    "GNU C++17 (64)": ".cpp",
    "Mono C#": ".cs",
    "D": ".d",
    "Go": ".go",
    "Haskell": ".hs",
    "Java 11": ".java",
    "Java 8": ".java",
    "Kotlin": ".kt",
    "Ocaml": ".ml",
    "Delphi": ".dpr",
    "FPC": ".pp",
    "PascalABC.NET": ".pas",
    "Perl": ".pl",
    "PHP": ".php",
    "Python 2": ".py",
    "Python 3": ".py",
    "PyPy 2": ".py",
    "PyPy 3": ".py",
    "Ruby": ".rb",
    "Rust": ".rs",
    "Scala": ".scala",
    "JavaScript": ".js",
    "Node.js": ".js",
}

s = requests.session()

for row in [
    row.find_all("td")
    for page in [
        BeautifulSoup(
            s.get(f"https://codeforces.com/submissions/{argv[1]}/page/{pageNo}").text,
            "html.parser",
        ).find_all("tr", attrs={"data-submission-id": True})
        for pageNo in range(
            1,
            int(
                (lambda lastPage: 1 if lastPage is None else lastPage.text)(
                    BeautifulSoup(
                        s.get(f"https://codeforces.com/submissions/{argv[1]}/").text,
                        "html.parser",
                    ).select_one("div.pagination > ul > li:nth-last-child(2)")
                )
            )
            + 1,
        )
    ]
    for row in page
    if row.find("a", attrs={"href": re.compile("^/contest/")}) is not None
]:
    submission, when, who, problem, lang, verdict, duration, mem = [
        cell.get_text(strip=True) for cell in row
    ]
    contest = (
        row[3].find("a", attrs={"href": re.compile("^/contest/")})["href"].split("/")[2]
    )
    epoch = time.mktime(datetime.datetime.strptime(when, "%b/%d/%Y %H:%M").timetuple())

    (
        filename := Path(
            f"./{argv[1]}/{contest}/{problem}/{submission} - {verdict}{extensions.get(lang, ' ' + lang)}"
        )
    ).parent.mkdir(parents=True, exist_ok=True)
    print(filename)

    with open(filename, "w+") as f:
        url = f"https://codeforces.com/contest/{contest}/submission/{submission}"

        while (
            not (resp := s.get(url)).ok
            or (
                source := BeautifulSoup(resp.text, "html.parser").find(
                    "pre", id="program-source-text"
                )
            )
            is None
        ):
            print(f"Retrying {url}...")
            time.sleep(30)

        f.write(source.text)

    os.utime(filename, (epoch, epoch))
    # os.sleep(1)