GallerySubscriber/backend/tests/services/test_gallery_dl_validation.py

"""Tests for GalleryDLService's post-download validation hook.

Focus: a successful gallery-dl run that wrote a truncated JPEG must
result in DownloadResult.success=False with VALIDATION_FAILED, the bad
file moved to the quarantine tree, and a sidecar JSON written next to
it. A run with no validation issues must behave exactly as before.
"""
from pathlib import Path
from types import SimpleNamespace
from unittest.mock import patch

import pytest

from app.services.file_validator import JPEG_HEAD, JPEG_TAIL
from app.services.gallery_dl import (
    DownloadResult,
    ErrorType,
    GalleryDLService,
    _summarize_validation_failures,
)


def _make_service(tmp_path: Path, validate: bool = True) -> GalleryDLService:
    """Service wired to a tmp download root, no real filesystem config."""
    fake_settings = SimpleNamespace(
        download_path=str(tmp_path / "downloads"),
        config_path=str(tmp_path / "config"),
        download_rate_limit=1.0,
    )
    Path(fake_settings.download_path).mkdir(parents=True, exist_ok=True)
    Path(fake_settings.config_path).mkdir(parents=True, exist_ok=True)
    with patch("app.services.gallery_dl.get_settings", return_value=fake_settings), \
         patch.object(GalleryDLService, "_load_base_config", return_value={}):
        return GalleryDLService(rate_limit=1.0, validate_files=validate)


def _write_jpeg(path: Path, *, truncated: bool) -> None:
    path.parent.mkdir(parents=True, exist_ok=True)
    body = JPEG_HEAD + b"\x00" * 64
    if not truncated:
        body += JPEG_TAIL
    path.write_bytes(body)


# --- _validate_and_quarantine ---


def test_quarantines_truncated_jpeg(tmp_path):
    service = _make_service(tmp_path)
    bad = Path(service.settings.download_path) / "Artist" / "patreon" / "01.jpg"
    _write_jpeg(bad, truncated=True)

    paths, failures = service._validate_and_quarantine(
        written_paths=[bad],
        subscription_name="Artist",
        platform="patreon",
        url="https://patreon.com/post/1",
    )

    assert len(paths) == 1
    assert len(failures) == 1
    assert failures[0]["format"] == "jpeg"
    # Original is gone, moved to _quarantine root
    assert not bad.exists()
    quarantine_root = (
        Path(service.settings.download_path) / "_quarantine" / "Artist" / "patreon"
    )
    moved = quarantine_root / "01.jpg"
    assert moved.exists()
    sidecar = quarantine_root / "01.jpg.quarantine.json"
    assert sidecar.exists()
    assert "patreon.com/post/1" in sidecar.read_text()


def test_passes_through_well_formed_files(tmp_path):
    service = _make_service(tmp_path)
    good = Path(service.settings.download_path) / "Artist" / "patreon" / "01.jpg"
    _write_jpeg(good, truncated=False)

    paths, failures = service._validate_and_quarantine(
        written_paths=[good],
        subscription_name="Artist",
        platform="patreon",
        url="https://patreon.com/post/1",
    )

    assert paths == []
    assert failures == []
    assert good.exists()  # unchanged


def test_skips_unvalidatable_extensions(tmp_path):
    service = _make_service(tmp_path)
    sidecar = Path(service.settings.download_path) / "Artist" / "patreon" / "01.json"
    sidecar.parent.mkdir(parents=True, exist_ok=True)
    sidecar.write_bytes(b'{"junk": true}')

    paths, _ = service._validate_and_quarantine(
        written_paths=[sidecar],
        subscription_name="Artist",
        platform="patreon",
        url="https://example/1",
    )
    assert paths == []
    assert sidecar.exists()


def test_quarantine_filename_collision_does_not_overwrite(tmp_path):
    service = _make_service(tmp_path)
    quarantine_root = (
        Path(service.settings.download_path) / "_quarantine" / "Artist" / "patreon"
    )
    quarantine_root.mkdir(parents=True, exist_ok=True)
    (quarantine_root / "01.jpg").write_bytes(b"PRIOR_QUARANTINE")

    bad = Path(service.settings.download_path) / "Artist" / "patreon" / "01.jpg"
    _write_jpeg(bad, truncated=True)

    paths, _ = service._validate_and_quarantine(
        written_paths=[bad],
        subscription_name="Artist",
        platform="patreon",
        url="https://patreon.com/2",
    )

    # Original quarantine file untouched, new one suffixed.
    assert (quarantine_root / "01.jpg").read_bytes() == b"PRIOR_QUARANTINE"
    assert (quarantine_root / "01.1.jpg").exists()
    assert paths and paths[0].endswith("01.1.jpg")


# --- _written_paths ---


def test_written_paths_extracts_only_absolute_paths(tmp_path):
    service = _make_service(tmp_path)
    stdout = (
        "/data/downloads/Artist/patreon/01.jpg\n"
        "# /data/downloads/Artist/patreon/02.jpg\n"  # archive skip — must be excluded
        "/data/downloads/Artist/patreon/03.jpg\n"
        "[patreon][info] some log line\n"
        "  /data/downloads/Artist/patreon/04.jpg  \n"  # surrounding whitespace
    )
    paths = service._written_paths(stdout)
    names = [p.name for p in paths]
    assert names == ["01.jpg", "03.jpg", "04.jpg"]


# --- _summarize_validation_failures ---


def test_summary_with_uniform_reason():
    failures = [
        {"reason": "missing JPEG EOI marker (FF D9)"} for _ in range(3)
    ]
    msg = _summarize_validation_failures(failures)
    assert "3 files" in msg
    assert "EOI" in msg


def test_summary_with_mixed_reasons():
    failures = [
        {"reason": "missing JPEG EOI marker (FF D9)"},
        {"reason": "missing JPEG EOI marker (FF D9)"},
        {"reason": "zero-byte file"},
    ]
    msg = _summarize_validation_failures(failures)
    assert "3 files" in msg
    assert "mixed" in msg


# --- validate_files=False respects opt-out ---


def test_validate_files_false_skips_quarantine(tmp_path):
    service = _make_service(tmp_path, validate=False)
    bad = Path(service.settings.download_path) / "Artist" / "patreon" / "01.jpg"
    _write_jpeg(bad, truncated=True)

    # download() is async + spawns subprocess. Cheaper test: confirm the
    # _validate_files flag is wired and the helper isn't called when false.
    # We don't run download() here; the constructor flag is the contract.
    assert service._validate_files is False