scripts/audit_codenet_metadata.py

#!/usr/bin/env python3
"""Create reproducible Project CodeNet language and problem-family censuses."""

from __future__ import annotations

import argparse
import re
from pathlib import Path

import duckdb
import pyarrow.csv as pacsv


def checked_sha256(value: str) -> str:
    if not re.fullmatch(r"[0-9a-f]{64}", value):
        raise argparse.ArgumentTypeError("expected a lowercase SHA-256 digest")
    return value


def write_query(connection: duckdb.DuckDBPyConnection, query: str, path: Path) -> None:
    path.parent.mkdir(parents=True, exist_ok=True)
    pacsv.write_csv(connection.execute(query).to_arrow_table(), path)


def main() -> None:
    parser = argparse.ArgumentParser()
    parser.add_argument("--metadata-root", type=Path, required=True)
    parser.add_argument("--language-output", type=Path, required=True)
    parser.add_argument("--problem-language-output", type=Path, required=True)
    parser.add_argument("--revision", required=True)
    parser.add_argument("--archive-sha256", type=checked_sha256, required=True)
    args = parser.parse_args()

    files = sorted(args.metadata_root.glob("p[0-9][0-9][0-9][0-9][0-9].csv"))
    if len(files) != 4_053:
        raise RuntimeError(f"expected 4,053 per-problem CSVs, found {len(files):,}")

    connection = duckdb.connect()
    connection.read_csv(
        str(args.metadata_root / "p?????.csv"),
        header=True,
        all_varchar=True,
        union_by_name=True,
    ).create_view("submissions")

    revision = args.revision.replace("'", "''")
    archive_sha256 = args.archive_sha256
    write_query(
        connection,
        f"""
        SELECT
            language,
            count(*) AS submissions,
            count(DISTINCT problem_id) AS problems,
            sum(CASE WHEN status = 'Accepted' THEN 1 ELSE 0 END) AS accepted_submissions,
            count(DISTINCT CASE WHEN status = 'Accepted' THEN problem_id END) AS accepted_problems,
            sum(try_cast(code_size AS HUGEINT)) AS source_bytes,
            sum(CASE WHEN status = 'Accepted' THEN try_cast(code_size AS HUGEINT) ELSE 0 END)
                AS accepted_source_bytes,
            '{revision}' AS source_revision,
            '{archive_sha256}' AS archive_sha256
        FROM submissions
        GROUP BY language
        ORDER BY submissions DESC, language
        """,
        args.language_output,
    )
    write_query(
        connection,
        f"""
        SELECT
            problem_id,
            language,
            count(*) AS submissions,
            sum(CASE WHEN status = 'Accepted' THEN 1 ELSE 0 END) AS accepted_submissions,
            sum(try_cast(code_size AS HUGEINT)) AS source_bytes,
            sum(CASE WHEN status = 'Accepted' THEN try_cast(code_size AS HUGEINT) ELSE 0 END)
                AS accepted_source_bytes,
            '{revision}' AS source_revision,
            '{archive_sha256}' AS archive_sha256
        FROM submissions
        GROUP BY problem_id, language
        ORDER BY problem_id, language
        """,
        args.problem_language_output,
    )


if __name__ == "__main__":
    main()