#!/usr/bin/env python3
"""Validate the frozen paper evaluation files without requiring source sets."""

from __future__ import annotations

import argparse
import hashlib
import json
from pathlib import Path


EXPECTED = {
    "data/evaluation/metric_qa/requests_paper_um2000.jsonl": {
        "rows": 2000,
        "sha256": "5c6fa80d160c587cc0a2f7da5c92fd8aa1472aaac40ac45be487fcb10f1093df",
    },
    "data/evaluation/caption/requests_paper_um2000.jsonl": {
        "rows": 2000,
        "sha256": "c6e27dae5461da60e74962e4ec45597534687303530eb82ea60c999895019e2d",
    },
    "data/evaluation/caption/hidden_gt_paper_um2000.jsonl": {
        "rows": 2000,
        "sha256": "1fc8ffa3966a7d8d14284cb5b79c286ab8469db3527a6fff94578d45045c8647",
    },
    "data/evaluation/tsqa/series_overlap_audit.json": {
        "sha256": "e568e2b01799ec83f48f499ca36cf748ab44b7baba65c4c01e085c653ec18ccb",
    },
    "data/evaluation/tsqa/requests_clean3264.jsonl": {
        "rows": 3264,
        "sha256": "fc48b3447fb8d52b1f6d90a7770ba9d0f8e57ae61225f99a8c43cf9795f259b7",
    },
}


def sha256(path: Path) -> str:
    digest = hashlib.sha256()
    with path.open("rb") as handle:
        for chunk in iter(lambda: handle.read(1024 * 1024), b""):
            digest.update(chunk)
    return digest.hexdigest()


def read_ids(path: Path) -> list[str]:
    return [
        str(json.loads(line)["request_id"])
        for line in path.read_text(encoding="utf-8").splitlines()
        if line.strip()
    ]


def main() -> None:
    parser = argparse.ArgumentParser(description=__doc__)
    parser.add_argument("--repo-root", type=Path, required=True)
    args = parser.parse_args()
    repo = args.repo_root.resolve()

    for relative, expected in EXPECTED.items():
        path = repo / relative
        if not path.is_file():
            raise FileNotFoundError(path)
        actual_hash = sha256(path)
        if actual_hash != expected["sha256"]:
            raise RuntimeError(
                f"{relative}: expected SHA256 {expected['sha256']}, found {actual_hash}"
            )
        if "rows" in expected:
            with path.open("rb") as handle:
                rows = sum(1 for line in handle if line.strip())
            if rows != expected["rows"]:
                raise RuntimeError(
                    f"{relative}: expected {expected['rows']} rows, found {rows}"
                )

    caption_requests = read_ids(
        repo / "data/evaluation/caption/requests_paper_um2000.jsonl"
    )
    caption_hidden = read_ids(
        repo / "data/evaluation/caption/hidden_gt_paper_um2000.jsonl"
    )
    if caption_requests != caption_hidden:
        raise RuntimeError("Caption request and hidden-GT ID order differs")

    audit = json.loads(
        (repo / "data/evaluation/tsqa/series_overlap_audit.json").read_text(
            encoding="utf-8"
        )
    )
    excluded = {str(item) for item in audit["affected_test_request_ids"]}
    clean_ids = read_ids(repo / "data/evaluation/tsqa/requests_clean3264.jsonl")
    if len(excluded) != 255 or len(clean_ids) != len(set(clean_ids)):
        raise RuntimeError("Invalid TSQA exclusion audit or clean request IDs")
    if set(clean_ids) & excluded:
        raise RuntimeError("TSQA clean requests contain an excluded request ID")

    print("Frozen paper evaluation data verified.")


if __name__ == "__main__":
    main()