from pathlib import Path

from manual_masking import (
    DEFAULT_MASK_OUTPUT_DIRNAME,
    DEFAULT_MASK_PREVIEW_DIRNAME,
    append_jsonl_file,
    ensure_mask_templates_dir,
    list_mask_templates,
    mask_templates_dir,
    mask_template_label,
    resolve_manual_mask_pdf,
)


def test_mask_templates_dir_is_under_config():
    base = Path("/tmp/anonymisation")
    assert mask_templates_dir(base) == base / "config" / "mask_templates"


def test_ensure_mask_templates_dir_creates_folder(tmp_path: Path):
    created = ensure_mask_templates_dir(tmp_path)
    assert created == tmp_path / "config" / "mask_templates"
    assert created.is_dir()


def test_resolve_manual_mask_pdf_accepts_only_pdf():
    assert resolve_manual_mask_pdf(Path("/tmp/test.pdf")) == Path("/tmp/test.pdf")
    assert resolve_manual_mask_pdf(Path("/tmp/test.PDF")) == Path("/tmp/test.PDF")
    assert resolve_manual_mask_pdf(Path("/tmp/test.docx")) is None
    assert resolve_manual_mask_pdf(None) is None


def test_manual_mask_outputs_follow_project_convention():
    assert DEFAULT_MASK_OUTPUT_DIRNAME == "anonymise"
    assert DEFAULT_MASK_PREVIEW_DIRNAME == "anonymise_preview"


def test_list_mask_templates_filters_supported_extensions(tmp_path: Path):
    templates_dir = ensure_mask_templates_dir(tmp_path)
    kept = templates_dir / "alpha.yml"
    other = templates_dir / "beta.txt"
    nested = templates_dir / "nested" / "gamma.json"
    nested.parent.mkdir(parents=True)
    kept.write_text("x", encoding="utf-8")
    other.write_text("x", encoding="utf-8")
    nested.write_text("x", encoding="utf-8")

    assert list_mask_templates(tmp_path) == [kept, nested]
    assert mask_template_label(nested, tmp_path) == "nested/gamma.json"


def test_append_jsonl_file_appends_non_empty_content(tmp_path: Path):
    target = tmp_path / "target.jsonl"
    extra = tmp_path / "extra.jsonl"
    target.write_text('{"kind":"A"}\n', encoding="utf-8")
    extra.write_text('{"kind":"B"}\n', encoding="utf-8")

    append_jsonl_file(target, extra)

    assert target.read_text(encoding="utf-8") == '{"kind":"A"}\n{"kind":"B"}\n'