298 lines
14 KiB
Python
298 lines
14 KiB
Python
import re
|
|
from datetime import date
|
|
from decimal import Decimal
|
|
from pathlib import Path
|
|
|
|
import pytest
|
|
|
|
from app.parsers.base import ParsedStatement, ParsedTransaction
|
|
|
|
H = {"Authorization": "Bearer test-key"}
|
|
|
|
FAKE = ParsedStatement(
|
|
bank="dkb", iban="DE02120300000000202051",
|
|
period_start=date(2026, 6, 1), period_end=date(2026, 6, 30),
|
|
opening_balance=Decimal("100.00"), closing_balance=Decimal("40.00"),
|
|
transactions=[ParsedTransaction(date(2026, 6, 3), date(2026, 6, 3),
|
|
Decimal("-60.00"), "Miete Juni", "Vermieter")])
|
|
|
|
|
|
@pytest.fixture
|
|
def fake_parse(monkeypatch):
|
|
monkeypatch.setattr("app.services.importer.parse_pdf", lambda p: FAKE)
|
|
|
|
|
|
# Ein Auszug mit ZWEI identischen Buchungen (gleicher dedup_hash). Saldo bleibt
|
|
# konsistent (100 - 60 - 60 = -20). Beide muessen bestaetigt werden koennen.
|
|
FAKE_TWINS = ParsedStatement(
|
|
bank="dkb", iban="DE02120300000000202051",
|
|
period_start=date(2026, 6, 1), period_end=date(2026, 6, 30),
|
|
opening_balance=Decimal("100.00"), closing_balance=Decimal("-20.00"),
|
|
transactions=[
|
|
ParsedTransaction(date(2026, 6, 3), date(2026, 6, 3),
|
|
Decimal("-60.00"), "Miete Juni", "Vermieter"),
|
|
ParsedTransaction(date(2026, 6, 3), date(2026, 6, 3),
|
|
Decimal("-60.00"), "Miete Juni", "Vermieter"),
|
|
])
|
|
|
|
|
|
@pytest.fixture
|
|
def fake_parse_twins(monkeypatch):
|
|
monkeypatch.setattr("app.services.importer.parse_pdf", lambda p: FAKE_TWINS)
|
|
|
|
|
|
def _upload(client, name="auszug.pdf"):
|
|
return client.post("/api/imports/upload", headers=H,
|
|
files={"file": (name, b"%PDF-fake", "application/pdf")})
|
|
|
|
|
|
def test_upload_preview_confirm(client, fake_parse, tmp_path, monkeypatch):
|
|
monkeypatch.setenv("FB_INBOX_DIR", str(tmp_path / "inbox"))
|
|
monkeypatch.setenv("FB_UPLOADS_DIR", str(tmp_path / "uploads"))
|
|
from app.config import get_settings
|
|
get_settings.cache_clear()
|
|
r = _upload(client)
|
|
assert r.status_code == 201
|
|
sid = r.json()["id"]
|
|
prev = client.get(f"/api/imports/{sid}/preview", headers=H).json()
|
|
assert prev["balance_ok"] is True
|
|
assert len(prev["transactions"]) == 1
|
|
assert client.post(f"/api/imports/{sid}/confirm", headers=H).status_code == 200
|
|
txs = client.get("/api/transactions", headers=H).json()
|
|
assert len(txs) == 1 and txs[0]["status"] == "confirmed"
|
|
# Zweiter Import desselben Auszugs: alles Duplikate
|
|
sid2 = _upload(client, "auszug2.pdf").json()["id"]
|
|
prev2 = client.get(f"/api/imports/{sid2}/preview", headers=H).json()
|
|
assert prev2["duplicates"] == 1
|
|
client.post(f"/api/imports/{sid2}/confirm", headers=H)
|
|
assert len(client.get("/api/transactions", headers=H).json()) == 1
|
|
|
|
|
|
def test_confirm_rechecks_duplicates_across_two_drafts(client, fake_parse, tmp_path, monkeypatch):
|
|
# Beide Auszuege werden importiert, bevor einer bestaetigt ist -> zum
|
|
# Importzeitpunkt ist keine Buchung ein Duplikat (dedup prueft nur gegen
|
|
# bestaetigte). Erst beim zweiten Confirm faellt auf, dass die Buchung nun
|
|
# bereits bestaetigt existiert. Der Re-Check muss sie ueberspringen, sonst
|
|
# wird dieselbe Buchung doppelt bestaetigt.
|
|
monkeypatch.setenv("FB_INBOX_DIR", str(tmp_path / "inbox"))
|
|
monkeypatch.setenv("FB_UPLOADS_DIR", str(tmp_path / "uploads"))
|
|
from app.config import get_settings
|
|
get_settings.cache_clear()
|
|
sid1 = _upload(client, "a1.pdf").json()["id"]
|
|
sid2 = _upload(client, "a2.pdf").json()["id"]
|
|
assert client.post(f"/api/imports/{sid1}/confirm", headers=H).status_code == 200
|
|
assert client.post(f"/api/imports/{sid2}/confirm", headers=H).status_code == 200
|
|
# Zweites Confirm darf die Buchung nicht erneut bestaetigen.
|
|
assert len(client.get("/api/transactions", headers=H).json()) == 1
|
|
# Zweites Confirm eines bereits bestaetigten Imports -> 409.
|
|
assert client.post(f"/api/imports/{sid1}/confirm", headers=H).status_code == 409
|
|
|
|
|
|
def test_confirm_keeps_identical_twins_within_same_statement(client, fake_parse_twins, tmp_path, monkeypatch):
|
|
# Zwei identische Buchungen im SELBEN Auszug teilen sich den dedup_hash. Der
|
|
# Confirm-Re-Check darf nur statement-uebergreifende Duplikate abfangen,
|
|
# sonst wuerde die erste bestaetigte Buchung ihre Zwillingsbuchung als
|
|
# Duplikat markieren und den vom Preview geprueften Saldo zerstoeren.
|
|
monkeypatch.setenv("FB_INBOX_DIR", str(tmp_path / "inbox"))
|
|
monkeypatch.setenv("FB_UPLOADS_DIR", str(tmp_path / "uploads"))
|
|
from app.config import get_settings
|
|
get_settings.cache_clear()
|
|
sid = _upload(client).json()["id"]
|
|
prev = client.get(f"/api/imports/{sid}/preview", headers=H).json()
|
|
assert prev["balance_ok"] is True
|
|
assert prev["duplicates"] == 0
|
|
assert client.post(f"/api/imports/{sid}/confirm", headers=H).status_code == 200
|
|
txs = client.get("/api/transactions", headers=H).json()
|
|
assert len(txs) == 2
|
|
assert all(t["status"] == "confirmed" for t in txs)
|
|
|
|
|
|
def test_upload_sanitizes_path_traversal_filename(client, fake_parse, tmp_path, monkeypatch):
|
|
monkeypatch.setenv("FB_INBOX_DIR", str(tmp_path / "inbox"))
|
|
monkeypatch.setenv("FB_UPLOADS_DIR", str(tmp_path / "uploads"))
|
|
from app.config import get_settings
|
|
get_settings.cache_clear()
|
|
r = client.post("/api/imports/upload", headers=H,
|
|
files={"file": ("../evil.pdf", b"%PDF-fake", "application/pdf")})
|
|
assert r.status_code == 201
|
|
# Der Dateiname darf keine Pfad-Anteile aus dem Client uebernehmen: nur
|
|
# der Basisname landet im Uploads-Verzeichnis, nichts entkommt nach oben.
|
|
assert not (tmp_path / "evil.pdf").exists()
|
|
assert (tmp_path / "uploads" / "evil.pdf").exists()
|
|
assert not (tmp_path / "inbox" / "evil.pdf").exists()
|
|
|
|
|
|
def test_upload_rejects_non_pdf(client, tmp_path, monkeypatch):
|
|
monkeypatch.setenv("FB_INBOX_DIR", str(tmp_path / "inbox"))
|
|
monkeypatch.setenv("FB_UPLOADS_DIR", str(tmp_path / "uploads"))
|
|
from app.config import get_settings
|
|
get_settings.cache_clear()
|
|
r = client.post("/api/imports/upload", headers=H,
|
|
files={"file": ("x.txt", b"not a pdf", "text/plain")})
|
|
assert r.status_code == 400
|
|
|
|
|
|
def test_delete_error_import_returns_204(client, db):
|
|
# Ein fehlgeschlagener Import (status="error") muss verworfen werden
|
|
# koennen - nur bestaetigte Importe sind vor dem Loeschen geschuetzt.
|
|
from app.models.tables import Statement
|
|
db.add(Statement(filename="kaputt.pdf", bank="dkb", status="error",
|
|
error_message="PDF nicht lesbar"))
|
|
db.commit()
|
|
sid = db.query(Statement).one().id
|
|
assert client.delete(f"/api/imports/{sid}", headers=H).status_code == 204
|
|
|
|
|
|
def test_rollback_confirmed_import_deletes_transactions_and_statement(client, fake_parse, tmp_path, monkeypatch):
|
|
# Ein bestaetigter Import laesst sich als Ganzes zurueckrollen: alle seine
|
|
# Buchungen UND der Auszug selbst verschwinden, das Konto bleibt bestehen
|
|
# (Name/IBAN fuer einen Re-Import mit verbessertem Parser).
|
|
monkeypatch.setenv("FB_INBOX_DIR", str(tmp_path / "inbox"))
|
|
monkeypatch.setenv("FB_UPLOADS_DIR", str(tmp_path / "uploads"))
|
|
from app.config import get_settings
|
|
get_settings.cache_clear()
|
|
sid = _upload(client).json()["id"]
|
|
assert client.post(f"/api/imports/{sid}/confirm", headers=H).status_code == 200
|
|
account_id = client.get("/api/accounts", headers=H).json()[0]["id"]
|
|
|
|
r = client.post(f"/api/imports/{sid}/rollback", headers=H)
|
|
assert r.status_code == 200
|
|
body = r.json()
|
|
assert body == {"deleted_transactions": 1, "statement_id": sid}
|
|
|
|
txs = client.get("/api/transactions", headers=H).json()
|
|
assert len(txs) == 0
|
|
imports = client.get("/api/imports", headers=H).json()
|
|
assert all(s["id"] != sid for s in imports)
|
|
accounts = client.get("/api/accounts", headers=H).json()
|
|
assert any(a["id"] == account_id for a in accounts)
|
|
|
|
|
|
def test_rollback_then_reimport_restores_transactions(client, fake_parse, tmp_path, monkeypatch):
|
|
# Der eigentliche Zweck des Rollbacks: denselben Auszug erneut importieren
|
|
# koennen (z.B. nach einem verbesserten Parser), ohne dass alte Buchungen
|
|
# als Duplikate im Weg stehen.
|
|
monkeypatch.setenv("FB_INBOX_DIR", str(tmp_path / "inbox"))
|
|
monkeypatch.setenv("FB_UPLOADS_DIR", str(tmp_path / "uploads"))
|
|
from app.config import get_settings
|
|
get_settings.cache_clear()
|
|
sid = _upload(client).json()["id"]
|
|
client.post(f"/api/imports/{sid}/confirm", headers=H)
|
|
client.post(f"/api/imports/{sid}/rollback", headers=H)
|
|
|
|
sid2 = _upload(client, "auszug_reimport.pdf").json()["id"]
|
|
prev = client.get(f"/api/imports/{sid2}/preview", headers=H).json()
|
|
assert prev["balance_ok"] is True
|
|
assert prev["duplicates"] == 0
|
|
assert client.post(f"/api/imports/{sid2}/confirm", headers=H).status_code == 200
|
|
txs = client.get("/api/transactions", headers=H).json()
|
|
assert len(txs) == 1 and txs[0]["status"] == "confirmed"
|
|
|
|
|
|
def test_rollback_draft_import_returns_409(client, fake_parse, tmp_path, monkeypatch):
|
|
monkeypatch.setenv("FB_INBOX_DIR", str(tmp_path / "inbox"))
|
|
monkeypatch.setenv("FB_UPLOADS_DIR", str(tmp_path / "uploads"))
|
|
from app.config import get_settings
|
|
get_settings.cache_clear()
|
|
sid = _upload(client).json()["id"]
|
|
r = client.post(f"/api/imports/{sid}/rollback", headers=H)
|
|
assert r.status_code == 409
|
|
|
|
|
|
def test_rollback_unknown_import_returns_404(client):
|
|
r = client.post("/api/imports/999999/rollback", headers=H)
|
|
assert r.status_code == 404
|
|
|
|
|
|
def _row_for(html: str, filename: str) -> str:
|
|
"""Isoliert den <tr>-Block eines bestimmten Import-Statements anhand seines
|
|
Dateinamens, damit Button-Zustaende eindeutig dem richtigen Statement
|
|
zugeordnet werden (statt nur global auf Disabled-Buttons zu zaehlen)."""
|
|
for row in html.split("<tr>"):
|
|
if filename in row:
|
|
return row
|
|
raise AssertionError(f"keine Zeile fuer {filename!r} gefunden")
|
|
|
|
|
|
def _has_active_button(row: str, label: str) -> bool:
|
|
for tag in re.finditer(r"<button[^>]*>" + re.escape(label) + r"</button>", row):
|
|
if "disabled" not in tag.group(0):
|
|
return True
|
|
return False
|
|
|
|
|
|
def _has_disabled_button(row: str, label: str) -> bool:
|
|
return bool(re.search(r"<button[^>]*disabled[^>]*>" + re.escape(label) + r"</button>", row))
|
|
|
|
|
|
def test_import_list_button_matrix(client, tmp_path, monkeypatch):
|
|
# UX-Regel: Bedienelemente sind immer sichtbar, ggf. ausgegraut (disabled),
|
|
# statt je nach Status komplett zu fehlen. Drei Statements mit den drei
|
|
# moeglichen Status erzeugen, dann pro Zeile die Button-Matrix pruefen.
|
|
monkeypatch.setenv("FB_INBOX_DIR", str(tmp_path / "inbox"))
|
|
monkeypatch.setenv("FB_UPLOADS_DIR", str(tmp_path / "uploads"))
|
|
from app.config import get_settings
|
|
get_settings.cache_clear()
|
|
|
|
# parse_pdf direkt (nicht ueber die fake_parse-Fixture) patchen/zuruecksetzen,
|
|
# damit der dritte Upload weiter unten wieder den echten Parser durchlaeuft
|
|
# und an kaputten Bytes wie in test_upload_corrupt_pdf_produces_error_statement_not_500
|
|
# scheitert - monkeypatch.undo() wuerde hier auch die setenv-Aenderungen
|
|
# oben zuruecknehmen.
|
|
import app.services.importer as importer_mod
|
|
original_parse_pdf = importer_mod.parse_pdf
|
|
importer_mod.parse_pdf = lambda p: FAKE
|
|
try:
|
|
sid_draft = _upload(client, "entwurf.pdf").json()["id"]
|
|
sid_confirmed = _upload(client, "bestaetigt.pdf").json()["id"]
|
|
assert client.post(f"/api/imports/{sid_confirmed}/confirm", headers=H).status_code == 200
|
|
finally:
|
|
importer_mod.parse_pdf = original_parse_pdf
|
|
|
|
r_err = client.post("/api/imports/upload", headers=H,
|
|
files={"file": ("fehler.pdf", b"not a pdf", "application/pdf")})
|
|
assert r_err.status_code == 201
|
|
assert r_err.json()["status"] == "error"
|
|
|
|
client.post("/login", data={"username": "admin", "password": "geheim"})
|
|
r = client.get("/import/list")
|
|
assert r.status_code == 200
|
|
|
|
draft_row = _row_for(r.text, "entwurf.pdf")
|
|
assert _has_active_button(draft_row, "Übernehmen")
|
|
assert _has_active_button(draft_row, "Verwerfen")
|
|
assert _has_disabled_button(draft_row, "Import zurückrollen")
|
|
assert not _has_active_button(draft_row, "Import zurückrollen")
|
|
|
|
confirmed_row = _row_for(r.text, "bestaetigt.pdf")
|
|
assert _has_disabled_button(confirmed_row, "Übernehmen")
|
|
assert _has_disabled_button(confirmed_row, "Verwerfen")
|
|
assert _has_active_button(confirmed_row, "Import zurückrollen")
|
|
assert not _has_disabled_button(confirmed_row, "Import zurückrollen")
|
|
|
|
error_row = _row_for(r.text, "fehler.pdf")
|
|
assert _has_active_button(error_row, "Verwerfen")
|
|
assert _has_disabled_button(error_row, "Übernehmen")
|
|
assert _has_disabled_button(error_row, "Import zurückrollen")
|
|
|
|
|
|
def test_upload_corrupt_pdf_produces_error_statement_not_500(client, tmp_path, monkeypatch):
|
|
# Datei besteht die Endungspruefung (.pdf), ist aber strukturell kein
|
|
# gueltiges PDF -> pdfplumber wirft eine eigene Exception (keine
|
|
# ParserError). Das darf nicht als unbehandelter 500 durchschlagen,
|
|
# sondern muss wie "Bank nicht erkannt" als sauberer Fehler-Import
|
|
# sichtbar werden.
|
|
monkeypatch.setenv("FB_INBOX_DIR", str(tmp_path / "inbox"))
|
|
monkeypatch.setenv("FB_UPLOADS_DIR", str(tmp_path / "uploads"))
|
|
from app.config import get_settings
|
|
get_settings.cache_clear()
|
|
r = client.post("/api/imports/upload", headers=H,
|
|
files={"file": ("kaputt.pdf", b"not a pdf", "application/pdf")})
|
|
assert r.status_code == 201
|
|
body = r.json()
|
|
assert body["status"] == "error"
|
|
assert "PDF nicht lesbar" in body["error_message"]
|
|
# Datei bleibt im Posteingang liegen, wird nicht ins Uploads-Verzeichnis verschoben.
|
|
assert (tmp_path / "inbox" / "kaputt.pdf").exists()
|
|
assert not (tmp_path / "uploads" / "kaputt.pdf").exists()
|