"""Tests for LocalStorage — fs-backed Storage protocol for the training pipeline.

Storage is the swap-point between local-dev (LocalStorage rooted at ./data/) and the
eventual S3-backed impl. Downstream stages (bulk_fetch, write_parquet) talk to the
Storage protocol only, not Path.
"""

from pathlib import Path

import pytest

from ml_training_data.storage import LocalStorage


def test_write_bytes_then_read_bytes_returns_same_data(tmp_path: Path) -> None:
    # Arrange
    storage = LocalStorage(root=tmp_path)
    payload = b"hello world"

    # Act
    storage.write_bytes("greetings/hello.txt", payload)
    out = storage.read_bytes("greetings/hello.txt")

    # Assert
    assert out == payload


def test_exists_is_false_before_write_and_true_after(tmp_path: Path) -> None:
    # Arrange
    storage = LocalStorage(root=tmp_path)

    # Act
    before = storage.exists("a/b.bin")
    storage.write_bytes("a/b.bin", b"x")
    after = storage.exists("a/b.bin")

    # Assert
    assert before is False
    assert after is True


def test_iter_keys_yields_every_written_key(tmp_path: Path) -> None:
    # Arrange
    storage = LocalStorage(root=tmp_path)
    storage.write_bytes("certs/a.json", b"1")
    storage.write_bytes("certs/b.json", b"2")
    storage.write_bytes("manifest.json", b"3")

    # Act
    keys = sorted(storage.iter_keys())

    # Assert
    assert keys == ["certs/a.json", "certs/b.json", "manifest.json"]


def test_iter_keys_filters_by_prefix(tmp_path: Path) -> None:
    # Arrange
    storage = LocalStorage(root=tmp_path)
    storage.write_bytes("certs/a.json", b"1")
    storage.write_bytes("certs/b.json", b"2")
    storage.write_bytes("manifest.json", b"3")

    # Act
    keys = sorted(storage.iter_keys(prefix="certs/"))

    # Assert
    assert keys == ["certs/a.json", "certs/b.json"]


def test_read_bytes_raises_filenotfound_for_missing_key(tmp_path: Path) -> None:
    # Arrange
    storage = LocalStorage(root=tmp_path)

    # Act / Assert
    with pytest.raises(FileNotFoundError):
        storage.read_bytes("nope.bin")


def test_open_read_returns_seekable_binary_stream(tmp_path: Path) -> None:
    # Arrange
    storage = LocalStorage(root=tmp_path)
    storage.write_bytes("big.bin", b"abcdefghij")

    # Act
    with storage.open_read("big.bin") as f:
        f.seek(4)
        chunk = f.read(3)

    # Assert
    assert chunk == b"efg"