from pathlib import Path

from gui_batch_paths import (
    build_batch_output_dir,
    iter_pseudonymized_texts,
    list_supported_documents,
)


def test_list_supported_documents_excludes_gui_output_tree(tmp_path: Path):
    root = tmp_path / "cases"
    source_case = root / "001_patient_header_and_birth"
    source_case.mkdir(parents=True)
    nested_case = root / "002_contact_bundle"
    nested_case.mkdir(parents=True)
    output_case = root / "anonymise" / "001_patient_header_and_birth"
    output_case.mkdir(parents=True)

    source_txt = source_case / "test.txt"
    source_pdf = nested_case / "source.pdf"
    output_txt = output_case / "test.pseudonymise.txt"

    source_txt.write_text("source", encoding="utf-8")
    source_pdf.write_text("pdf", encoding="utf-8")
    output_txt.write_text("output", encoding="utf-8")

    documents = list_supported_documents(root, {".txt", ".pdf"})

    assert documents == [source_txt, source_pdf]


def test_build_batch_output_dir_preserves_relative_parent(tmp_path: Path):
    root = tmp_path / "cases"
    output_root = root / "anonymise"
    source = root / "010_spaced_establishment_header" / "test.txt"
    source.parent.mkdir(parents=True)
    source.write_text("test", encoding="utf-8")

    output_dir = build_batch_output_dir(root, output_root, source)

    assert output_dir == output_root / "010_spaced_establishment_header"


def test_build_batch_output_dir_keeps_root_files_at_output_root(tmp_path: Path):
    root = tmp_path / "cases"
    output_root = root / "anonymise"
    source = root / "test.txt"
    root.mkdir(parents=True)
    source.write_text("test", encoding="utf-8")

    output_dir = build_batch_output_dir(root, output_root, source)

    assert output_dir == output_root


def test_iter_pseudonymized_texts_is_recursive(tmp_path: Path):
    output_root = tmp_path / "anonymise"
    nested = output_root / "001_patient_header_and_birth"
    nested.mkdir(parents=True)
    top_level = output_root / "summary.pseudonymise.txt"
    nested_txt = nested / "test.pseudonymise.txt"
    other_file = nested / "audit.jsonl"

    top_level.write_text("top", encoding="utf-8")
    nested_txt.write_text("nested", encoding="utf-8")
    other_file.write_text("{}", encoding="utf-8")

    found = sorted(iter_pseudonymized_texts(output_root))

    assert found == [nested_txt, top_level]