pdf-ocr-hotfolder/tests/test_config_warnings.py

"""Tests für Legacy-Warnungen und unbekannte Config-Keys.

Zwei stille Fallen:
- [ocr].timeout bedeutet seit 0.4.0 Sekunden pro SEITE (vorher Gesamtlauf)
- unbekannte Keys (Tippfehler!) wurden beim Laden stumm verworfen
"""
from __future__ import annotations

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

from pdf_ocr_hotfolder.config import (
    config_warnings,
    legacy_warnings,
    load_config,
    unknown_key_warnings,
)

_PATHS = """
[paths]
incoming = "/tmp/in"
outgoing = "/tmp/out"
working  = "/tmp/work"
error    = "/tmp/err"
"""


def _write(tmp_path: Path, extra: str = "") -> Path:
    cfg = tmp_path / "config.toml"
    cfg.write_text(_PATHS + extra)
    return cfg


# ---------------- Legacy: [ocr].timeout ----------------

def test_legacy_timeout_warns(tmp_path: Path) -> None:
    """Ein Altwert (Gesamt-Timeout 1800) muss deutlich benannt werden."""
    cfg = load_config(_write(tmp_path, "\n[ocr]\ntimeout = 1800\n"))
    warnings = legacy_warnings(cfg)
    assert len(warnings) == 1
    assert "timeout" in warnings[0]
    assert "PRO SEITE" in warnings[0]
    assert "300" in warnings[0]


def test_timeout_at_threshold_warns(tmp_path: Path) -> None:
    cfg = load_config(_write(tmp_path, "\n[ocr]\ntimeout = 900\n"))
    assert legacy_warnings(cfg)


def test_sane_timeout_does_not_warn(tmp_path: Path) -> None:
    cfg = load_config(_write(tmp_path, "\n[ocr]\ntimeout = 300\n"))
    assert legacy_warnings(cfg) == []


def test_default_config_has_no_warnings(tmp_path: Path) -> None:
    cfg = load_config(_write(tmp_path))
    assert config_warnings(cfg) == []


# ---------------- Legacy: [ocr].pdfa_level ----------------

def test_pdfa_level_warns_about_ghostscript(tmp_path: Path) -> None:
    cfg = load_config(_write(tmp_path, '\n[ocr]\npdfa_level = "2"\n'))
    warnings = legacy_warnings(cfg)
    assert len(warnings) == 1
    assert "pdfa_level" in warnings[0]
    assert "Ghostscript" in warnings[0]
    assert "10.02.0" in warnings[0]


def test_empty_pdfa_level_does_not_warn(tmp_path: Path) -> None:
    cfg = load_config(_write(tmp_path, '\n[ocr]\npdfa_level = ""\n'))
    assert legacy_warnings(cfg) == []


def test_both_legacy_warnings_together(tmp_path: Path) -> None:
    cfg = load_config(_write(tmp_path, '\n[ocr]\ntimeout = 1800\npdfa_level = "1"\n'))
    assert len(legacy_warnings(cfg)) == 2


# ---------------- Unbekannte Keys ----------------

def test_typo_key_is_collected(tmp_path: Path) -> None:
    """`langauges` statt `languages` darf nicht mehr stumm verschwinden."""
    cfg = load_config(_write(tmp_path, '\n[ocr]\nlangauges = "deu"\n'))
    assert cfg.unknown_keys == ["[ocr].langauges"]
    assert "[ocr].langauges" in unknown_key_warnings(cfg)[0]
    # Der Rest wird weiterhin normal geladen
    assert cfg.ocr.languages == "deu+eng"


def test_known_keys_are_not_reported(tmp_path: Path) -> None:
    cfg = load_config(_write(tmp_path, '\n[ocr]\nlanguages = "deu"\njobs = 2\n'))
    assert cfg.unknown_keys == []
    assert cfg.ocr.languages == "deu"


def test_unknown_keys_in_all_sections(tmp_path: Path) -> None:
    cfg = load_config(_write(tmp_path, """
[ocr]
foo = 1

[output]
bar = "x"

[verapdf]
baz = true

[upload.folder]
qux = ""

[upload.nextcloud]
quux = ""

[upload.sftp]
corge = 0

[notify.email]
grault = ""

[logging]
level = "INFO"
garply = 1
"""))
    assert cfg.unknown_keys == [
        "[ocr].foo", "[output].bar", "[verapdf].baz",
        "[upload.folder].qux", "[upload.nextcloud].quux", "[upload.sftp].corge",
        "[notify.email].grault", "[logging].garply",
    ]


def test_unknown_top_level_key_is_reported(tmp_path: Path) -> None:
    """Ein Key ausserhalb jeder Sektion (z.B. vergessene Sektionszeile)."""
    cfg_file = tmp_path / "config.toml"
    cfg_file.write_text('log_level = "DEBUG"\n' + _PATHS)
    cfg = load_config(cfg_file)
    assert cfg.unknown_keys == ["log_level"]


def test_unknown_section_is_reported(tmp_path: Path) -> None:
    cfg = load_config(_write(tmp_path, '\n[ocrr]\nlanguages = "deu"\n\n[upload.ftp]\nhost = "x"\n'))
    assert "[ocrr]" in cfg.unknown_keys
    assert "[upload.ftp]" in cfg.unknown_keys


def test_unknown_key_inside_paths(tmp_path: Path) -> None:
    """Ein zusätzlicher Key direkt in [paths] wird ebenfalls gemeldet."""
    cfg_file = tmp_path / "config.toml"
    cfg_file.write_text(_PATHS + 'archive  = "/tmp/a"\n')
    cfg = load_config(cfg_file)
    assert cfg.unknown_keys == ["[paths].archive"]


def test_load_config_works_without_logging(tmp_path: Path) -> None:
    """load_config() darf nichts loggen müssen — Tests rufen sie direkt auf."""
    cfg_file = _write(tmp_path, '\n[ocr]\nlangauges = "deu"\n')
    with patch("logging.Logger.warning") as warn:
        cfg = load_config(cfg_file)
    warn.assert_not_called()
    assert cfg.unknown_keys


# ---------------- Warnungen beim Dienststart ----------------

def _argv(monkeypatch, cfg_file: Path, *extra: str) -> None:
    monkeypatch.setattr(sys, "argv",
                        ["pdf-ocr-hotfolder", "--config", str(cfg_file), *extra])


def test_warnings_are_logged_on_service_start(tmp_path: Path, tmp_config,
                                              monkeypatch, caplog) -> None:
    """Beim normalen Start landen die Warnungen im Log (nicht nur im Check)."""
    cfg_file = tmp_path / "cfg.toml"
    cfg_file.write_text(f"""
[paths]
incoming = "{tmp_config.paths.incoming}"
outgoing = "{tmp_config.paths.outgoing}"
working  = "{tmp_config.paths.working}"
error    = "{tmp_config.paths.error}"

[ocr]
timeout = 1800
langauges = "deu"
""")
    _argv(monkeypatch, cfg_file, "--once")

    from pdf_ocr_hotfolder.__main__ import main

    with caplog.at_level(logging.WARNING, logger="pdf_ocr_hotfolder.__main__"), \
         patch("pdf_ocr_hotfolder.service.shutil.which", return_value="/usr/bin/fake"):
        assert main() == 0

    text = caplog.text
    assert "PRO SEITE" in text
    assert "[ocr].langauges" in text