Musiksammlung/tests/test_ocr.py

"""Tests für OCR-Funktionen (subprocess via Mock)."""

from __future__ import annotations

from pathlib import Path
from unittest.mock import MagicMock, patch

from PIL import Image

from musiksammlung.ocr import _detect_and_fix_rotation, ocr_images, run_ocr


def _fake_run(stdout: str = "", returncode: int = 0) -> MagicMock:
    """Erstellt ein Mock-subprocess.CompletedProcess."""
    result = MagicMock()
    result.returncode = returncode
    result.stdout = stdout
    result.stderr = ""
    return result


def _make_image(path: Path, size: tuple[int, int] = (100, 100)) -> Path:
    Image.new("RGB", size, color=(200, 100, 50)).save(str(path), "JPEG")
    return path


# ---------------------------------------------------------------------------
# run_ocr
# ---------------------------------------------------------------------------


class TestRunOcr:
    def test_returns_stdout_on_success(self, tmp_path: Path) -> None:
        img = _make_image(tmp_path / "test.jpg")
        with patch("musiksammlung.ocr.subprocess.run", return_value=_fake_run("Trackliste\n")):
            result = run_ocr(img)
        assert result == "Trackliste"

    def test_calls_tesseract_with_correct_args(self, tmp_path: Path) -> None:
        img = _make_image(tmp_path / "test.jpg")
        with patch("musiksammlung.ocr.subprocess.run", return_value=_fake_run()) as mock_run:
            run_ocr(img, languages="deu+eng")
        args = mock_run.call_args[0][0]
        assert args[0] == "tesseract"
        assert str(img) in args
        assert "deu+eng" in args
        assert "--psm" in args

    def test_raises_on_nonzero_returncode(self, tmp_path: Path) -> None:
        img = _make_image(tmp_path / "test.jpg")
        with patch(
            "musiksammlung.ocr.subprocess.run",
            return_value=_fake_run("", returncode=1),
        ):
            try:
                run_ocr(img)
                assert False, "RuntimeError expected"
            except RuntimeError as e:
                assert "Tesseract" in str(e)

    def test_custom_language(self, tmp_path: Path) -> None:
        img = _make_image(tmp_path / "test.jpg")
        with patch("musiksammlung.ocr.subprocess.run", return_value=_fake_run()) as mock_run:
            run_ocr(img, languages="fra")
        args = mock_run.call_args[0][0]
        assert "fra" in args


# ---------------------------------------------------------------------------
# _detect_and_fix_rotation
# ---------------------------------------------------------------------------


class TestDetectAndFixRotation:
    def test_no_rotation_needed(self) -> None:
        img = Image.new("L", (200, 100))
        osd_output = "Rotate: 0\nOrientation in degrees: 0\n"
        with patch("musiksammlung.ocr.subprocess.run", return_value=_fake_run(osd_output)):
            result = _detect_and_fix_rotation(img)
        assert result.size == (200, 100)

    def test_rotates_90_degrees(self) -> None:
        img = Image.new("L", (200, 100))  # breit
        osd_output = "Rotate: 90\nOrientation in degrees: 90\n"
        with patch("musiksammlung.ocr.subprocess.run", return_value=_fake_run(osd_output)):
            result = _detect_and_fix_rotation(img)
        # 200×100 rotiert um 90° → 100×200
        assert result.size == (100, 200)

    def test_rotates_180_degrees(self) -> None:
        img = Image.new("L", (200, 100))
        osd_output = "Rotate: 180\n"
        with patch("musiksammlung.ocr.subprocess.run", return_value=_fake_run(osd_output)):
            result = _detect_and_fix_rotation(img)
        # 180° Rotation ändert Größe nicht
        assert result.size == (200, 100)

    def test_fallback_brute_force_when_osd_fails(self) -> None:
        img = Image.new("L", (200, 100))
        call_count = [0]

        def side_effect(cmd, **kwargs):
            call_count[0] += 1
            if call_count[0] == 1:
                # Erster Aufruf (OSD) schlägt fehl
                return _fake_run("", returncode=1)
            # Weitere Aufrufe (brute-force): 90° liefert die meisten Buchstaben
            if "--psm" in cmd and "6" in cmd:
                return _fake_run("Allegro Andante Beethoven" if call_count[0] == 3 else "a")
            return _fake_run("")

        with patch("musiksammlung.ocr.subprocess.run", side_effect=side_effect):
            _detect_and_fix_rotation(img)
        # Brute-force wurde verwendet (mind. 4 Subprocess-Aufrufe)
        assert call_count[0] >= 2


# ---------------------------------------------------------------------------
# ocr_images
# ---------------------------------------------------------------------------


class TestOcrImages:
    def test_returns_text_for_single_image(self, tmp_path: Path) -> None:
        img = _make_image(tmp_path / "img.jpg")

        with (
            patch("musiksammlung.ocr.preprocess_image", return_value=img),
            patch("musiksammlung.ocr.run_ocr", return_value="Track 1\nTrack 2"),
        ):
            result = ocr_images([img])

        assert result == "Track 1\nTrack 2"

    def test_concatenates_multiple_images(self, tmp_path: Path) -> None:
        img1 = _make_image(tmp_path / "img1.jpg")
        img2 = _make_image(tmp_path / "img2.jpg")
        preprocessed = tmp_path / "pre.png"
        preprocessed.write_bytes(b"fake")

        with (
            patch("musiksammlung.ocr.preprocess_image", return_value=preprocessed),
            patch("musiksammlung.ocr.run_ocr", side_effect=["Text A", "Text B"]),
        ):
            result = ocr_images([img1, img2])

        assert result == "Text A\n\nText B"

    def test_skips_empty_ocr_result(self, tmp_path: Path) -> None:
        img = _make_image(tmp_path / "img.jpg")
        preprocessed = tmp_path / "pre.png"
        preprocessed.write_bytes(b"fake")

        with (
            patch("musiksammlung.ocr.preprocess_image", return_value=preprocessed),
            patch("musiksammlung.ocr.run_ocr", return_value=""),
        ):
            result = ocr_images([img])

        assert result == ""

    def test_cleans_up_temp_file(self, tmp_path: Path) -> None:
        img = _make_image(tmp_path / "img.jpg")
        preprocessed = tmp_path / "pre.png"
        preprocessed.write_bytes(b"fake")

        with (
            patch("musiksammlung.ocr.preprocess_image", return_value=preprocessed),
            patch("musiksammlung.ocr.run_ocr", return_value="text"),
        ):
            ocr_images([img])

        # Temporäre Datei wurde gelöscht
        assert not preprocessed.exists()

    def test_cleans_up_even_on_error(self, tmp_path: Path) -> None:
        img = _make_image(tmp_path / "img.jpg")
        preprocessed = tmp_path / "pre.png"
        preprocessed.write_bytes(b"fake")

        err = RuntimeError("Tesseract fehlgeschlagen")
        with (
            patch("musiksammlung.ocr.preprocess_image", return_value=preprocessed),
            patch("musiksammlung.ocr.run_ocr", side_effect=err),
        ):
            try:
                ocr_images([img])
            except RuntimeError:
                pass

        assert not preprocessed.exists()