75 lines
2.7 KiB
Python
75 lines
2.7 KiB
Python
|
"""Tests fuer app/protokoll_parsers/__init__.py — Registry + Dispatch (#126)."""
|
||
|
|
from __future__ import annotations
|
||
|
|
|
||
|
|
import pytest
|
||
|
|
|
||
|
|
from app.protokoll_parsers import (
|
||
|
|
PROTOKOLL_PARSERS,
|
||
|
|
parse_protocol,
|
||
|
|
supported_bundeslaender,
|
||
|
|
)
|
||
|
|
|
||
|
|
|
||
|
|
class TestRegistry:
|
||
|
|
def test_nrw_registered(self):
|
||
|
|
"""NRW ist die Referenz-Implementierung — muss da sein."""
|
||
|
|
assert "NRW" in PROTOKOLL_PARSERS
|
||
|
|
|
||
|
|
def test_supported_includes_nrw(self):
|
||
|
|
assert "NRW" in supported_bundeslaender()
|
||
|
|
|
||
|
|
def test_supported_returns_sorted(self):
|
||
|
|
codes = supported_bundeslaender()
|
||
|
|
assert codes == sorted(codes)
|
||
|
|
|
||
|
|
def test_registry_values_are_callable(self):
|
||
|
|
for code, parser in PROTOKOLL_PARSERS.items():
|
||
|
|
assert callable(parser), f"Parser fuer {code} ist nicht callable"
|
||
|
|
|
||
|
|
|
||
|
|
class TestDispatch:
|
||
|
|
def test_unknown_bl_raises_not_implemented(self):
|
||
|
|
with pytest.raises(NotImplementedError) as exc:
|
||
|
|
parse_protocol("XX", "/dev/null")
|
||
|
|
msg = str(exc.value)
|
||
|
|
assert "XX" in msg
|
||
|
|
# Liste der unterstuetzten BL muss in der Message stehen
|
||
|
|
assert "NRW" in msg
|
||
|
|
# Issue-Referenz fuer Folge-Arbeit
|
||
|
|
assert "#126" in msg
|
||
|
|
|
||
|
|
def test_known_bl_delegates_to_registered_parser(self, tmp_path, monkeypatch):
|
||
|
|
"""parse_protocol delegiert an den BL-Parser aus der Registry."""
|
||
|
|
called_with: list[str] = []
|
||
|
|
|
||
|
|
def fake_parser(pdf_path: str) -> list[dict]:
|
||
|
|
called_with.append(pdf_path)
|
||
|
|
return [{"drucksache": "18/1", "ergebnis": "angenommen", "votes": {"ja": [], "nein": [], "enthaltung": []}}]
|
||
|
|
|
||
|
|
# Temporaer einen TEST-Parser registrieren, dann wieder entfernen
|
||
|
|
monkeypatch.setitem(PROTOKOLL_PARSERS, "TEST", fake_parser)
|
||
|
|
|
||
|
|
result = parse_protocol("TEST", str(tmp_path / "x.pdf"))
|
||
|
|
|
||
|
|
assert called_with == [str(tmp_path / "x.pdf")]
|
||
|
|
assert len(result) == 1
|
||
|
|
assert result[0]["drucksache"] == "18/1"
|
||
|
|
|
||
|
|
|
||
|
|
class TestParserSchema:
|
||
|
|
"""Vertrag: jeder registrierte Parser muss Result-Dicts mit minimalem
|
||
|
|
Schema liefern — drucksache (str|None), ergebnis (str), votes (dict)."""
|
||
|
|
|
||
|
|
def test_nrw_result_dict_has_expected_keys(self):
|
||
|
|
"""Smoke-Test mit handgemachtem Plenarprotokoll-Snippet — pruefen,
|
||
|
|
dass das Schema des Output-Dicts die in __init__.py dokumentierten
|
||
|
|
Keys enthaelt."""
|
||
|
|
from app.protokoll_parsers.nrw import find_results
|
||
|
|
|
||
|
|
text = "Damit ist der Antrag Drucksache 18/100 angenommen."
|
||
|
|
results = find_results(text)
|
||
|
|
assert results, "find_results sollte mindestens einen Treffer liefern"
|
||
|
|
for r in results:
|
||
|
|
for key in ("drucksache", "ergebnis", "kind", "einstimmig"):
|
||
|
|
assert key in r, f"Key '{key}' fehlt im Result"
|