import pytest

from giant.data.loader import find_parquet_files


def _touch(path):
    path.parent.mkdir(parents=True, exist_ok=True)
    path.touch()
    return path


def test_find_parquet_files_single_file(tmp_path):
    f = _touch(tmp_path / "shard-000.parquet")
    assert find_parquet_files(f) == [f]


def test_find_parquet_files_directory_glob(tmp_path):
    a = _touch(tmp_path / "shard-000.parquet")
    b = _touch(tmp_path / "shard-001.parquet")
    _touch(tmp_path / "not_a_parquet.root")
    assert find_parquet_files(tmp_path) == sorted([a, b])


def test_find_parquet_files_empty_directory_raises(tmp_path):
    with pytest.raises(FileNotFoundError):
        find_parquet_files(tmp_path)


def test_manifest_resolves_relative_to_its_own_directory(tmp_path):
    target = _touch(tmp_path / "processed" / "pbwo4" / "shard-000.parquet")
    manifest_dir = tmp_path / "pools" / "pbwo4"
    manifest_dir.mkdir(parents=True)
    manifest = manifest_dir / "full.manifest"
    manifest.write_text("../../processed/pbwo4/shard-000.parquet\n")

    assert find_parquet_files(manifest) == [target.resolve()]


def test_manifest_skips_blank_lines_and_comments(tmp_path):
    target = _touch(tmp_path / "shard-000.parquet")
    manifest = tmp_path / "full.manifest"
    manifest.write_text("\n# a comment\nshard-000.parquet\n\n")

    assert find_parquet_files(manifest) == [target.resolve()]


def test_manifest_missing_file_raises(tmp_path):
    manifest = tmp_path / "full.manifest"
    manifest.write_text("does_not_exist.parquet\n")

    with pytest.raises(FileNotFoundError):
        find_parquet_files(manifest)


def test_manifest_with_no_entries_raises(tmp_path):
    manifest = tmp_path / "full.manifest"
    manifest.write_text("# only comments\n")

    with pytest.raises(FileNotFoundError):
        find_parquet_files(manifest)