from __future__ import annotations

import gzip
import json
import sys
from pathlib import Path

import pandas as pd


ROOT = Path(__file__).resolve().parent
WORKSPACE = Path(__file__).resolve().parents[3]
sys.path.insert(0, str(WORKSPACE / "src"))

from oa_tl import multiscale_pseudosnp_linear_autoencoder as model  # noqa: E402


EXPECTED_FEATURE_HASH = "a9ce199b9a6e7c6185cd3945ef468e0051f89fc813277e86ab4c8ce23cdbc858"
DEVELOPMENT = WORKSPACE / "outputs" / "runs" / "20260807T1305Z_expanded_discovery_dataset" / "development.tsv"


def read_jsonl_gz(path: Path):
    with gzip.open(path, "rt", encoding="utf-8") as handle:
        for line in handle:
            yield json.loads(line)


def main() -> None:
    if model._sha256_file(DEVELOPMENT) != model.GOVERNED_DEVELOPMENT_SHA256:
        raise RuntimeError("governed development table hash changed")
    full = pd.read_csv(DEVELOPMENT, sep="\t", low_memory=False)
    cohort = model.select_oa2025_pip_triplets(full, threshold=model.PIP_THRESHOLD)
    fold_map, fold_receipt = model.formal_full_graph_fold_map(full)
    cohort["outer_fold"] = cohort["split_group_id"].astype(str).map(fold_map).astype(int)

    exported = list(read_jsonl_gz(ROOT / "data" / "supervised_model_input.jsonl.gz"))
    if len(exported) != len(cohort):
        raise RuntimeError("exported supervised cohort row count differs")
    for expected, actual in zip(cohort.itertuples(index=False), exported):
        if (
            str(expected.example_id) != actual["example_id"]
            or int(expected.label) != actual["label"]
            or str(expected.sequence_ref) != actual["sequence_ref"]
            or str(expected.sequence_alt) != actual["sequence_alt"]
            or int(expected.outer_fold) != actual["outer_fold"]
        ):
            raise RuntimeError(f"supervised source alignment failed: {actual['example_id']}")

    pseudo_paths = sorted((ROOT / "data").glob("pretraining-*.jsonl.gz"))
    pairs = (
        (record["sequence_ref"], record["sequence_alt"])
        for path in pseudo_paths
        for record in read_jsonl_gz(path)
    )
    features = model.multiscale_feature_matrix(pairs)
    feature_hash = model._sha256_csr(features)
    if feature_hash != EXPECTED_FEATURE_HASH:
        raise RuntimeError(f"pseudo-SNP feature hash mismatch: {feature_hash}")
    print(
        "PASS: governed 168-row cohort and fold map align; "
        f"20,000 x 9,216 feature CSR SHA256={feature_hash}; "
        f"fold_assignment_sha256={fold_receipt['assignment_sha256']}"
    )


if __name__ == "__main__":
    main()