pdf-ocr-hotfolder/tests/test_error_collision.py

"""Punkt 2: error/ darf nichts mehr still überschreiben.

Scheiterte dieselbe `scan.pdf` zweimal, ersetzte die zweite die erste in
error/ — dieselbe Datenverlust-Klasse, die für outgoing/ bereits geschlossen
ist. Betrifft `_move_to_error()` und damit auch `_rescue_to_error()`.
"""
from __future__ import annotations

import logging
from pathlib import Path
from unittest.mock import patch

from pdf_ocr_hotfolder.config import OcrConfig, OutputConfig, VeraPdfConfig
from pdf_ocr_hotfolder.processor import _move_to_error, process_pdf
from pdf_ocr_hotfolder.service import HotfolderService

ERSTE = b"%PDF-1.4 erste\n"
ZWEITE = b"%PDF-1.4 zweite\n"


# ---------------- _move_to_error direkt ----------------

def test_move_to_error_keeps_existing_file(tmp_path: Path) -> None:
    error_dir = tmp_path / "error"
    error_dir.mkdir()
    (error_dir / "scan.pdf").write_bytes(ERSTE)

    zweite = tmp_path / "scan.pdf"
    zweite.write_bytes(ZWEITE)
    _move_to_error(zweite, error_dir)

    assert (error_dir / "scan.pdf").read_bytes() == ERSTE
    ausweich = list(error_dir.glob("scan_*.pdf"))
    assert len(ausweich) == 1
    assert ausweich[0].read_bytes() == ZWEITE
    assert not zweite.exists()


def test_move_to_error_without_collision_keeps_name(tmp_path: Path) -> None:
    error_dir = tmp_path / "error"
    src = tmp_path / "scan.pdf"
    src.write_bytes(ERSTE)

    _move_to_error(src, error_dir)

    assert (error_dir / "scan.pdf").read_bytes() == ERSTE
    assert list(error_dir.iterdir()) == [error_dir / "scan.pdf"]


def test_move_to_error_logs_warning_on_collision(tmp_path: Path, caplog) -> None:
    error_dir = tmp_path / "error"
    error_dir.mkdir()
    (error_dir / "scan.pdf").write_bytes(ERSTE)
    src = tmp_path / "scan.pdf"
    src.write_bytes(ZWEITE)

    with caplog.at_level(logging.WARNING, logger="pdf_ocr_hotfolder.processor"):
        _move_to_error(src, error_dir)

    assert "scan.pdf" in caplog.text
    assert "überschrieben" in caplog.text


def test_move_to_error_creates_dir(tmp_path: Path) -> None:
    src = tmp_path / "scan.pdf"
    src.write_bytes(ERSTE)
    error_dir = tmp_path / "tief" / "error"
    _move_to_error(src, error_dir)
    assert (error_dir / "scan.pdf").exists()


def test_move_to_error_survives_oserror(tmp_path: Path, caplog) -> None:
    """Ein fehlgeschlagener Move darf weiterhin nur geloggt werden."""
    src = tmp_path / "scan.pdf"
    src.write_bytes(ERSTE)
    error_dir = tmp_path / "error"

    with patch("pdf_ocr_hotfolder.processor.shutil.move",
               side_effect=OSError("read-only")):
        _move_to_error(src, error_dir)  # darf nicht werfen

    assert src.exists()


# ---------------- über process_pdf: zweimal dieselbe Datei kaputt ----------------

def _prepare(tmp_path: Path) -> dict:
    dirs = {name: tmp_path / name
            for name in ("incoming", "working", "outgoing", "error")}
    for d in dirs.values():
        d.mkdir(parents=True, exist_ok=True)
    return dirs


def _run_failing_ocr(dirs: dict, src: Path):
    with patch("pdf_ocr_hotfolder.processor.run_ocr",
               side_effect=RuntimeError("ocr kaputt")):
        return process_pdf(
            src=src,
            working_dir=dirs["working"],
            outgoing_dir=dirs["outgoing"],
            error_dir=dirs["error"],
            ocr_cfg=OcrConfig(),
            vera_cfg=VeraPdfConfig(enabled=False),
            output_cfg=OutputConfig(),
        )


def test_same_name_failing_twice_keeps_both(tmp_path: Path) -> None:
    dirs = _prepare(tmp_path)

    for inhalt in (ERSTE, ZWEITE):
        src = dirs["incoming"] / "scan.pdf"
        src.write_bytes(inhalt)
        result = _run_failing_ocr(dirs, src)
        assert not result.success

    dateien = sorted(p.read_bytes() for p in dirs["error"].iterdir())
    assert len(dateien) == 2
    assert sorted([ERSTE, ZWEITE]) == dateien


# ---------------- _rescue_to_error erbt den Schutz ----------------

def test_rescue_to_error_keeps_existing_file(tmp_config) -> None:
    """Der Rettungspfad nach einer unerwarteten Exception ebenso."""
    (tmp_config.paths.error / "boom.pdf").write_bytes(ERSTE)
    src = tmp_config.paths.incoming / "boom.pdf"
    src.write_bytes(ZWEITE)

    service = HotfolderService(tmp_config)
    try:
        service._rescue_to_error(src)
    finally:
        service._executor.shutdown(wait=False)

    assert (tmp_config.paths.error / "boom.pdf").read_bytes() == ERSTE
    ausweich = list(tmp_config.paths.error.glob("boom_*.pdf"))
    assert len(ausweich) == 1
    assert ausweich[0].read_bytes() == ZWEITE