build_parquet.py

"""Parquet build for the PyAra (Russian audio deepfake) HF dataset repo.

PyAra (https://www.kaggle.com/datasets/alep079/pyara) — a Russian-language
anti-spoofing set. Source layout:
  Real/<stem>.wav   (fake=0) -> bonafide  (73,583)
  Fake/<stem>.wav   (fake=1) -> spoof     (128,195, 5 TTS/VC algorithms alg_1..alg_5)
  final_dataset.tsv : path, sentence, age, gender, fake, algorithm, length

A full decode probe of all clips runs first; 0 failures -> CLEAN raw-byte embed
(no decode/re-encode). All source audio is already 16 kHz mono PCM_16 WAV, every
clip >= 3.0 s. Records are processed sorted by utterance_id for stable sharding
and near-sequential HDD reads.

Sample mode (--limit N): first N rows into a single shard, skipping full-count
asserts -- used for the fast offline validate-dataset pass.
"""

import argparse
import csv
import io
import json
import os
import tempfile
from concurrent.futures import ProcessPoolExecutor
from pathlib import Path

import datasets  # noqa: E402
import pyarrow.parquet as pq  # noqa: E402
import soundfile as sf  # noqa: E402
from datasets import Audio, ClassLabel, Dataset, Features, Value  # noqa: E402
from tqdm.auto import tqdm  # noqa: E402

try:
    datasets.disable_progress_bars()
except AttributeError:
    from datasets.utils.logging import disable_progress_bar

    disable_progress_bar()

REPO_ROOT = Path(__file__).resolve().parent
SRC_ROOT = Path("/home/kirill/mnt/users_4tb/datasets/final_dataset")
META_PATH = SRC_ROOT / "final_dataset.tsv"
PARQUET_DIR = REPO_ROOT / "data"
NUM_SHARDS = 100
EXPECTED_ROWS = 201778
EXPECTED_BONAFIDE = 73583
EXPECTED_SPOOF = 128195
TARGET_SR = 16000
WORKERS = int(os.environ.get("PYARA_BUILD_WORKERS", "64"))

FEATURES = Features(
    {
        "path": Value("string"),
        "audio": Audio(sampling_rate=16000),
        "label": ClassLabel(names=["bonafide", "spoof"]),
        "notes": Value("string"),
    }
)


def parse_metadata():
    """Parse final_dataset.tsv -> records. Columns:
    path, sentence, age, gender, fake, algorithm, length."""
    records = []
    seen = set()
    with open(META_PATH, encoding="utf-8") as fh:
        reader = csv.DictReader(fh, delimiter="\t")
        for row in reader:
            rel = row["path"].strip()
            if not rel:
                continue
            stem = Path(rel).stem
            if stem in seen:
                raise ValueError(f"Duplicate utterance_id stem: {stem}")
            seen.add(stem)
            is_spoof = row["fake"].strip() == "1"
            records.append(
                {
                    "uid": stem,
                    "relpath": rel,
                    "abspath": str(SRC_ROOT / rel),
                    "label": "spoof" if is_spoof else "bonafide",
                    "algorithm": (row.get("algorithm") or "").strip() or ("bonafide" if not is_spoof else ""),
                    "gender": (row.get("gender") or "").strip(),
                    "age": (row.get("age") or "").strip(),
                    "length": (row.get("length") or "").strip(),
                }
            )
    return records


def build_notes(rec):
    return json.dumps(
        {
            "utterance_id": rec["uid"],
            "algorithm": rec["algorithm"],
            "gender": rec["gender"],
            "age": rec["age"],
            "length": rec["length"],
        }
    )


def _probe_one(abspath):
    try:
        data, _ = sf.read(abspath)
        if data.shape[0] == 0:
            return f"{abspath}: empty"
        return None
    except Exception as e:  # noqa: BLE001
        return f"{abspath}: {str(e).splitlines()[0][:100]}"


def probe_decodability(records):
    paths = [r["abspath"] for r in records]
    failures = []
    with ProcessPoolExecutor(max_workers=WORKERS) as ex:
        for err in ex.map(_probe_one, paths, chunksize=32):
            if err:
                failures.append(err)
    print(f"Probe: {len(failures)}/{len(paths)} clips failed soundfile decode")
    if failures:
        for f in failures[:10]:
            print(f"  {f}")
        raise RuntimeError(
            "Source audio no longer cleanly decodable; the CLEAN raw-embed path "
            "is unsafe. Re-introduce a re-encode stage (see ASVspoof2021_LA)."
        )


def _clip_duration(abspath):
    info = sf.info(abspath)
    return info.frames / info.samplerate


def _ensure_long_first_row(records):
    for i in range(len(records)):
        if _clip_duration(records[i]["abspath"]) >= 1.0:
            if i != 0:
                records[0], records[i] = records[i], records[0]
            return
    raise RuntimeError("No clip with duration >= 1.0s found")


def _build_shard(task):
    shard_index, rows, num_shards = task
    shard_name = f"test-{shard_index:05d}-of-{num_shards:05d}.parquet"
    final = PARQUET_DIR / shard_name
    if final.exists() and final.stat().st_size > 0:
        return (shard_index, len(rows), "skipped")

    def row_gen():
        for rec in rows:
            yield {
                "path": rec["relpath"],
                "audio": {
                    "bytes": Path(rec["abspath"]).read_bytes(),
                    "path": rec["relpath"],
                },
                "label": rec["label"],
                "notes": build_notes(rec),
            }

    with tempfile.TemporaryDirectory() as cache:
        ds = Dataset.from_generator(row_gen, features=FEATURES, cache_dir=cache)
        tmp = PARQUET_DIR / f".{shard_name}.tmp"
        ds.to_parquet(str(tmp))
        os.replace(tmp, final)
    return (shard_index, len(rows), "built")


def _partition(records, num_shards):
    n = len(records)
    per = (n + num_shards - 1) // num_shards
    out = []
    for i in range(num_shards):
        chunk = records[i * per : (i + 1) * per]
        if chunk:
            out.append(chunk)
    return out


def build():
    parser = argparse.ArgumentParser()
    parser.add_argument("--limit", type=int, default=None)
    args = parser.parse_args()
    limit = args.limit
    sample_mode = limit is not None

    print(f"Reading metadata from {META_PATH}")
    records = parse_metadata()
    print(f"Parsed {len(records)} rows")
    if not sample_mode:
        assert len(records) == EXPECTED_ROWS, f"Expected {EXPECTED_ROWS}, got {len(records)}"

    check = records if not sample_mode else records[: max(limit * 4, limit)]
    missing = [r["uid"] for r in check if not Path(r["abspath"]).exists()]
    assert not missing, f"{len(missing)} wav files missing, e.g. {missing[:5]}"

    records.sort(key=lambda r: r["uid"])
    _ensure_long_first_row(records)
    probe_decodability(records)

    if sample_mode:
        records = records[:limit]
        num_shards = 1
        print(f"SAMPLE MODE: {len(records)} rows -> 1 shard")
    else:
        num_shards = NUM_SHARDS

    bona = sum(1 for r in records if r["label"] == "bonafide")
    spoof = sum(1 for r in records if r["label"] == "spoof")
    print(f"  bonafide={bona} spoof={spoof} total={len(records)}")

    PARQUET_DIR.mkdir(parents=True, exist_ok=True)
    keep_suffix = f"-of-{num_shards:05d}.parquet"
    for stale in PARQUET_DIR.glob("test-*.parquet"):
        if not stale.name.endswith(keep_suffix):
            print(f"Removing stale shard {stale.name}")
            stale.unlink()
    shards = _partition(records, num_shards)
    tasks = [(i, rows, num_shards) for i, rows in enumerate(shards)]
    stage_workers = min(WORKERS, len(tasks))
    print(f"Building {len(tasks)} shard(s) with {stage_workers} workers...")
    built = skipped = 0
    with ProcessPoolExecutor(max_workers=stage_workers) as ex:
        for idx, n, status in tqdm(
            ex.map(_build_shard, tasks), total=len(tasks), desc="shards", unit="shard"
        ):
            if status == "built":
                built += 1
            elif status == "skipped":
                skipped += 1
    print(f"Done: {built} built, {skipped} skipped")

    _verify(num_shards, sample_mode)
    print("All verifications passed!")

    if not sample_mode:
        from speech_spoof_bench import labels

        out = labels.emit_labels(REPO_ROOT)
        print(f"Wrote {out}")


def _verify(num_shards, sample_mode):
    shards = sorted(PARQUET_DIR.glob("test-*.parquet"))
    total = sum(pq.read_metadata(str(f)).num_rows for f in shards)
    uid_set, path_set, bona, spoof = set(), set(), 0, 0
    for f in shards:
        t = pq.read_table(str(f), columns=["path", "label", "notes"])
        for p, lab, n in zip(
            t.column("path").to_pylist(),
            t.column("label").to_pylist(),
            t.column("notes").to_pylist(),
        ):
            path_set.add(p)
            uid_set.add(json.loads(n)["utterance_id"])
            if lab == 0:
                bona += 1
            elif lab == 1:
                spoof += 1
    assert len(uid_set) == total, "Duplicate utterance_ids"
    assert len(path_set) == total, "Duplicate paths"
    if not sample_mode:
        assert total == EXPECTED_ROWS, f"{total} != {EXPECTED_ROWS}"
        assert bona == EXPECTED_BONAFIDE, f"bonafide {bona} != {EXPECTED_BONAFIDE}"
        assert spoof == EXPECTED_SPOOF, f"spoof {spoof} != {EXPECTED_SPOOF}"
    t0 = pq.read_table(str(shards[0]))
    assert set(t0.column_names) == {"path", "audio", "label", "notes"}, t0.column_names
    audio0 = t0.column("audio")[0].as_py()
    data, sr = sf.read(io.BytesIO(audio0["bytes"]))
    dur = len(data) / sr
    assert sr == 16000, f"row0 sr {sr} != 16000"
    assert dur >= 1.0, f"row0 dur {dur:.2f}s < 1.0s"
    print(f"  verify: {total} rows, row0 {sr}Hz {dur:.2f}s decodable OK")


if __name__ == "__main__":
    build()