scripts/download_benchmarks.py

"""Download public JSP / FJSP benchmark instances.

Sources
-------
- Taillard:    canonical files on Eric Taillard's homepage and the JSPLIB GitHub
- Lawrence:    OR-Library (J. E. Beasley)
- Brandimarte: IDSIA mirror
- DMU:         JSPLIB GitHub mirror

We download from the JSPLIB GitHub mirror (https://github.com/tamy0612/JSPLIB)
because it provides a single canonical text representation, public and stable,
under permissive licensing.
"""

from __future__ import annotations

import argparse
import hashlib
import sys
from pathlib import Path
from typing import Iterable, List
from urllib.error import HTTPError, URLError
from urllib.request import urlopen


JSPLIB_RAW = (
    "https://raw.githubusercontent.com/tamy0612/JSPLIB/master/instances/"
)


def _download(url: str) -> bytes:
    try:
        with urlopen(url, timeout=30) as resp:
            return resp.read()
    except (HTTPError, URLError) as e:
        raise SystemExit(f"Failed to download {url}: {e}")


def _save(text: bytes, path: Path) -> str:
    path.parent.mkdir(parents=True, exist_ok=True)
    path.write_bytes(text)
    return hashlib.sha256(text).hexdigest()[:16]


def download_taillard(data_dir: Path, names: Iterable[str]) -> dict:
    """Download Taillard instances by name (e.g., 'ta01')."""
    out = {}
    for name in names:
        url = f"{JSPLIB_RAW}{name}"
        try:
            blob = _download(url)
            sha = _save(blob, data_dir / f"{name}.txt")
            out[name] = {"url": url, "sha256_16": sha, "size": len(blob)}
            print(f"  taillard/{name} ok ({len(blob)} bytes, sha {sha})")
        except SystemExit as e:
            print(f"  taillard/{name} FAILED: {e}", file=sys.stderr)
            out[name] = {"url": url, "error": str(e)}
    return out


def download_lawrence(data_dir: Path, names: Iterable[str]) -> dict:
    out = {}
    for name in names:
        url = f"{JSPLIB_RAW}{name}"
        try:
            blob = _download(url)
            sha = _save(blob, data_dir / f"{name}.txt")
            out[name] = {"url": url, "sha256_16": sha, "size": len(blob)}
            print(f"  lawrence/{name} ok ({len(blob)} bytes, sha {sha})")
        except SystemExit as e:
            print(f"  lawrence/{name} FAILED: {e}", file=sys.stderr)
            out[name] = {"url": url, "error": str(e)}
    return out


def download_dmu(data_dir: Path, names: Iterable[str]) -> dict:
    out = {}
    for name in names:
        url = f"{JSPLIB_RAW}{name}"
        try:
            blob = _download(url)
            sha = _save(blob, data_dir / f"{name}.txt")
            out[name] = {"url": url, "sha256_16": sha, "size": len(blob)}
            print(f"  dmu/{name} ok ({len(blob)} bytes, sha {sha})")
        except SystemExit as e:
            print(f"  dmu/{name} FAILED: {e}", file=sys.stderr)
            out[name] = {"url": url, "error": str(e)}
    return out


# Brandimarte canonical mirror
BRANDIMARTE_BASE = (
    "https://raw.githubusercontent.com/SchedulingLab/fjsp-instances/master/Brandimarte/"
)

def download_brandimarte(data_dir: Path, names: Iterable[str]) -> dict:
    out = {}
    for name in names:
        url = f"{BRANDIMARTE_BASE}{name}.fjs"
        try:
            blob = _download(url)
            sha = _save(blob, data_dir / f"{name}.txt")
            out[name] = {"url": url, "sha256_16": sha, "size": len(blob)}
            print(f"  brandimarte/{name} ok ({len(blob)} bytes, sha {sha})")
        except SystemExit as e:
            print(f"  brandimarte/{name} FAILED: {e}", file=sys.stderr)
            out[name] = {"url": url, "error": str(e)}
    return out


def main() -> None:
    parser = argparse.ArgumentParser()
    parser.add_argument(
        "--family",
        nargs="+",
        default=["taillard", "lawrence", "brandimarte", "dmu"],
        choices=["taillard", "lawrence", "brandimarte", "dmu", "all"],
    )
    parser.add_argument("--data-dir", default="data")
    args = parser.parse_args()

    if "all" in args.family:
        args.family = ["taillard", "lawrence", "brandimarte", "dmu"]

    root = Path(args.data_dir)
    manifest = {}

    if "taillard" in args.family:
        names = [f"ta{i:02d}" for i in range(1, 81)]
        manifest["taillard"] = download_taillard(root / "taillard", names)
    if "lawrence" in args.family:
        names = [f"la{i:02d}" for i in range(1, 41)]
        manifest["lawrence"] = download_lawrence(root / "lawrence", names)
    if "brandimarte" in args.family:
        names = [f"mk{i:02d}" for i in range(1, 11)]
        manifest["brandimarte"] = download_brandimarte(root / "brandimarte", names)
    if "dmu" in args.family:
        names = [f"dmu{i:02d}" for i in range(1, 81)]
        manifest["dmu"] = download_dmu(root / "dmu", names)

    import json
    (root / "MANIFEST.json").write_text(json.dumps(manifest, indent=2))
    print(f"\nManifest saved to {root / 'MANIFEST.json'}")


if __name__ == "__main__":
    main()