GallerySubscriber/backend/tests/services/test_file_validator.py

"""Unit tests for file_validator.

Covers the actual production failure mode (JPEG missing FF D9 EOI) and
the symmetric checks for the other formats we validate. Unknown formats
must pass through — we don't want to police gallery-dl's own JSON
sidecars or zip ugoira archives.
"""
from pathlib import Path

import pytest

from app.services.file_validator import (
    GIF_HEAD_89A,
    JPEG_HEAD,
    JPEG_TAIL,
    PNG_HEAD,
    PNG_TAIL,
    WEBP_FORMAT,
    WEBP_RIFF,
    is_validatable,
    validate_file,
)


def _write(tmp_path: Path, name: str, data: bytes) -> Path:
    p = tmp_path / name
    p.write_bytes(data)
    return p


# --- JPEG ---


def test_jpeg_well_formed_passes(tmp_path):
    body = JPEG_HEAD + b"\x00" * 64 + JPEG_TAIL
    p = _write(tmp_path, "ok.jpg", body)
    result = validate_file(p)
    assert result.ok
    assert result.format == "jpeg"


def test_jpeg_missing_eoi_fails(tmp_path):
    """The exact production failure: file ends 6 bytes short, no FF D9.

    Downstream PIL.ImageFile.load raised "image file is truncated (6 bytes
    not processed)" — we must catch this at write time.
    """
    body = JPEG_HEAD + b"\x00" * 64  # No JPEG_TAIL
    p = _write(tmp_path, "truncated.jpg", body)
    result = validate_file(p)
    assert not result.ok
    assert result.format == "jpeg"
    assert "EOI" in result.reason


def test_jpeg_missing_soi_fails(tmp_path):
    p = _write(tmp_path, "bogus.jpg", b"NOTAJPEG" * 8 + JPEG_TAIL)
    result = validate_file(p)
    assert not result.ok
    assert "SOI" in result.reason


def test_jpeg_extension_jpeg_also_works(tmp_path):
    body = JPEG_HEAD + b"\x00" * 32 + JPEG_TAIL
    p = _write(tmp_path, "ok.jpeg", body)
    assert validate_file(p).ok


# --- PNG ---


def test_png_well_formed_passes(tmp_path):
    body = PNG_HEAD + b"\x00" * 32 + PNG_TAIL
    p = _write(tmp_path, "ok.png", body)
    result = validate_file(p)
    assert result.ok
    assert result.format == "png"


def test_png_missing_iend_fails(tmp_path):
    body = PNG_HEAD + b"\x00" * 32  # No IEND chunk
    p = _write(tmp_path, "truncated.png", body)
    result = validate_file(p)
    assert not result.ok
    assert "IEND" in result.reason


def test_png_missing_signature_fails(tmp_path):
    p = _write(tmp_path, "bogus.png", b"\x00" * 32 + PNG_TAIL)
    result = validate_file(p)
    assert not result.ok
    assert "signature" in result.reason


# --- GIF ---


def test_gif_well_formed_passes(tmp_path):
    body = GIF_HEAD_89A + b"\x00" * 32 + b"\x3b"
    p = _write(tmp_path, "ok.gif", body)
    result = validate_file(p)
    assert result.ok
    assert result.format == "gif"


def test_gif_missing_trailer_fails(tmp_path):
    body = GIF_HEAD_89A + b"\x00" * 32  # No 0x3B
    p = _write(tmp_path, "truncated.gif", body)
    result = validate_file(p)
    assert not result.ok
    assert "trailer" in result.reason


# --- WEBP ---


def test_webp_well_formed_passes(tmp_path):
    payload = b"\x00" * 32
    declared = (4 + len(payload)).to_bytes(4, "little")  # WEBP marker + payload
    body = WEBP_RIFF + declared + WEBP_FORMAT + payload
    p = _write(tmp_path, "ok.webp", body)
    result = validate_file(p)
    assert result.ok, f"unexpected failure: {result.reason}"
    assert result.format == "webp"


def test_webp_size_mismatch_fails(tmp_path):
    """RIFF declares more bytes than the file actually has — classic truncation."""
    payload = b"\x00" * 32
    lying_size = (4 + len(payload) + 100).to_bytes(4, "little")
    body = WEBP_RIFF + lying_size + WEBP_FORMAT + payload
    p = _write(tmp_path, "truncated.webp", body)
    result = validate_file(p)
    assert not result.ok
    assert "size mismatch" in result.reason


# --- Unknown formats / edge cases ---


def test_zero_byte_file_fails(tmp_path):
    p = _write(tmp_path, "empty.jpg", b"")
    result = validate_file(p)
    assert not result.ok
    assert "zero-byte" in result.reason


def test_missing_file_fails(tmp_path):
    p = tmp_path / "ghost.jpg"
    result = validate_file(p)
    assert not result.ok
    assert "not found" in result.reason


def test_unknown_extension_passes_through(tmp_path):
    """JSON sidecars from gallery-dl's metadata postprocessor and other
    non-image artifacts must not be quarantined."""
    p = _write(tmp_path, "metadata.json", b'{"ok": true}')
    result = validate_file(p)
    assert result.ok
    assert result.format is None


def test_zip_extension_passes_through(tmp_path):
    """Ugoira archives (Pixiv) — gallery-dl writes .zip, not our concern."""
    p = _write(tmp_path, "anim.zip", b"PK\x03\x04random")
    result = validate_file(p)
    assert result.ok


def test_is_validatable_flags_known_formats():
    assert is_validatable(Path("a.jpg"))
    assert is_validatable(Path("a.JPG"))  # case-insensitive
    assert is_validatable(Path("a.png"))
    assert is_validatable(Path("a.gif"))
    assert is_validatable(Path("a.webp"))
    assert not is_validatable(Path("a.json"))
    assert not is_validatable(Path("a.mp4"))