"""Push the gitignored run INPUTS to the Modal Volume.

Run this from a box that actually has the artifacts (the 96GB GPU box). The
queued jobs read these from out/ at train time; on Modal that out/ is the Volume.

What gets uploaded (all small -- KB to a few hundred MB):
  out/pairsets/        hand-authored persona contrastive pairs (prog_wide.json,
                       heldout_known_runtests.json, ...)
  data/pairs/          hand-authored Markdown pairsets
  out/vhack/           extracted hack-direction bases (v_hack_a5_runtests.safetensors,
                       v_hack_pairset_prog_wide_randomV.safetensors, ...)
  out/pools/           teacher rollout pools (substrate/, teacher_pool/)

NOT uploaded here (the Volume builds these itself, see app.py::warm):
  the Qwen3-4B weights  -> downloaded from HF into /cache/hf on first run
  svd_cache/            -> computed once on Modal and cached

Usage (on the GPU box, after `pip install modal` + `modal token new`):
  python modal/upload_inputs.py
"""
from __future__ import annotations

from pathlib import Path

import modal

# Volume paths are relative to the volume ROOT (which mounts at /cache in the
# container), so we upload to "out/..." NOT "/cache/out/...".
DIRS = [
    "out/pairsets",
    "data/pairs",
    "out/vhack",
    "out/pools",
    "external/rl-rewardhacking/results/data",  # the LeetCode problems (train/test/holdout jsonl)
]


def main():
    vol = modal.Volume.from_name("vgrout-cache", create_if_missing=True)
    repo = Path(__file__).resolve().parent.parent
    present = [(repo / d) for d in DIRS if (repo / d).exists()]
    missing = [d for d in DIRS if not (repo / d).exists()]
    if missing:
        print(f"[warn] absent locally, skipping: {missing}")
    assert present, f"none of {DIRS} exist under {repo} -- run from the box that has the artifacts"

    with vol.batch_upload(force=True) as batch:
        for local in present:
            remote = str(local.relative_to(repo))  # e.g. "out/pools"
            print(f"[upload] {local} -> {remote}")
            batch.put_directory(str(local), remote)
    print("[done] inputs on Volume. Verify: modal volume ls vgrout-cache out")


if __name__ == "__main__":
    main()