Files
FabledCurator/tests/test_importer_attachments.py
T

84 lines
2.9 KiB
Python

"""FC-2d-iii: loose non-media → PostAttachment (no skip)."""
import pytest
from sqlalchemy import func, select
from backend.app.models import Artist, ImportSettings, PostAttachment
from backend.app.services.importer import Importer
from backend.app.services.thumbnailer import Thumbnailer
pytestmark = pytest.mark.integration
@pytest.fixture
def import_layout(tmp_path):
import_root = tmp_path / "import"
images_root = tmp_path / "images"
import_root.mkdir()
images_root.mkdir()
return import_root, images_root
@pytest.fixture
def importer(db_sync, import_layout):
import_root, images_root = import_layout
settings = db_sync.execute(
select(ImportSettings).where(ImportSettings.id == 1)
).scalar_one()
return Importer(
session=db_sync, images_root=images_root, import_root=import_root,
thumbnailer=Thumbnailer(images_root=images_root), settings=settings,
)
def test_loose_non_media_becomes_attachment(importer, import_layout):
import_root, _ = import_layout
f = import_root / "Alice" / "manifesto.pdf"
f.parent.mkdir(parents=True, exist_ok=True)
f.write_bytes(b"%PDF-1.4 stuff")
r = importer.import_one(f)
assert r.status == "attached"
att = importer.session.execute(select(PostAttachment)).scalar_one()
artist = importer.session.execute(
select(Artist).where(Artist.slug == "alice")
).scalar_one()
assert att.artist_id == artist.id
assert att.post_id is None
assert att.original_filename == "manifesto.pdf"
assert att.ext == ".pdf"
def test_json_sidecar_is_not_attached(importer, import_layout):
import_root, _ = import_layout
j = import_root / "Alice" / "x.json"
j.parent.mkdir(parents=True, exist_ok=True)
j.write_text("{}")
r = importer.import_one(j)
assert r.status == "skipped"
n = importer.session.execute(
select(func.count()).select_from(PostAttachment)
).scalar_one()
assert n == 0
def test_mangled_filename_extension_is_sanitized(importer, import_layout):
"""gallery-dl sometimes URL-encodes a query string into the basename
(`...https___www.patreon.com_media-u_Z0F...`). Python's Path.suffix
returns 50+ chars of base64-ish junk for those, which blows the
PostAttachment.ext varchar(32) column. Operator-flagged 2026-05-25.
The importer should record an empty ext rather than crash."""
import_root, _ = import_layout
f = (
import_root / "Alice"
/ "79507046_media_https___www.patreon.com_media-u_Z0FBQUFBQm5q"
)
f.parent.mkdir(parents=True, exist_ok=True)
f.write_bytes(b"binary blob")
r = importer.import_one(f)
assert r.status == "attached"
att = importer.session.execute(select(PostAttachment)).scalar_one()
# Junk "extension" -> stored as empty string (not the 50-char garbage).
assert att.ext == ""
# original_filename is Text-typed so the full name survives intact.
assert att.original_filename.endswith("_Z0FBQUFBQm5q")