Files
F-r-ldraledigplannerare/backend/tests/unit/test_fk_scraper.py
T

735 lines
27 KiB
Python

"""Tests for the FK scraper: extractor, validation, and session state machine.
Three independent test concerns:
1. _extract_parental_leave — parses the FK #/valjdagar HTML fixture and
produces a correct ParentalLeaveFetch. No network, no BankID.
2. Validation helpers — out-of-bounds numbers raise rather than persist.
3. Session state machine — start/poll/cancel transitions, TTL auto-cancel,
double-cancel safety. Browser is mocked; no headless Chromium needed.
"""
from __future__ import annotations
import re
import uuid
from datetime import datetime
from pathlib import Path
from typing import Any
from unittest.mock import AsyncMock, MagicMock, patch
import pytest
from app.entities.parental_leave import ParentalLeaveFetch, TierData
from app.scrapers.fk_scraper import (
FKScraperService,
_SessionState,
_do_extract,
_set_status,
_sessions,
)
# ---------------------------------------------------------------------------
# Helpers for the HTML-fixture tests
# ---------------------------------------------------------------------------
CAPTURES_DIR = Path(__file__).parents[3] / "captures"
def _load_fixture(name: str) -> str:
"""Read a page capture from captures/, skipping if not present."""
path = CAPTURES_DIR / name
if not path.exists():
pytest.skip(f"Fixture {name} not found — run the captures step first")
return path.read_text(encoding="utf-8")
# ---------------------------------------------------------------------------
# Minimal mock of a Playwright Page that answers our CSS queries from HTML
# ---------------------------------------------------------------------------
def _make_mock_page(html: str):
"""
Build a minimal async mock page that serves inner_text / all_inner_texts
for the selectors used by _extract_parental_leave.
Uses basic regex over the fixture HTML rather than a real browser — fast
and dependency-free.
"""
NIVA_SEL = "[data-test='forDig'] .visa-dagar__desktop .niva-statistik__dagar span"
def _for_dig_kvar() -> str:
m = re.search(
r'data-test="forDig".*?'
r'class="visa-dagar__dagar-kvar">(\d+)<',
html,
re.DOTALL,
)
if m:
return m.group(1)
raise ValueError("No forDig dagar-kvar found in fixture")
def _for_dig_discrete() -> str:
m = re.search(
r'data-test="forDig".*?'
r'class="visa-dagar__discrete">(.*?)<',
html,
re.DOTALL,
)
if m:
return m.group(1)
raise ValueError("No forDig discrete found in fixture")
def _dubbeldag_kvar() -> str:
m = re.search(
r'data-test="dubbeldag".*?'
r'class="visa-dagar__dagar-kvar">(\d+)<',
html,
re.DOTALL,
)
if m:
return m.group(1)
raise ValueError("No dubbeldag dagar-kvar found in fixture")
def _niva_all_texts() -> list[str]:
m_block = re.search(
r'data-test="forDig"(.*?)data-test="footer"',
html,
re.DOTALL,
)
block = m_block.group(1) if m_block else ""
desktop_m = re.search(r"visa-dagar__desktop(.*?)visa-dagar__mobile", block, re.DOTALL)
target = desktop_m.group(1) if desktop_m else block
return re.findall(r"\(\d+ dagar\)", target)
page = MagicMock()
def locator_factory(sel: str) -> MagicMock:
loc = MagicMock()
if sel == NIVA_SEL:
loc.first.inner_text = AsyncMock(return_value="")
loc.first.wait_for = AsyncMock()
loc.all_inner_texts = AsyncMock(return_value=_niva_all_texts())
elif "visa-dagar__dagar-kvar" in sel and "forDig" in sel:
loc.first.inner_text = AsyncMock(return_value=_for_dig_kvar())
loc.first.wait_for = AsyncMock()
loc.all_inner_texts = AsyncMock(return_value=[])
elif "visa-dagar__discrete" in sel and "forDig" in sel:
loc.first.inner_text = AsyncMock(return_value=_for_dig_discrete())
loc.first.wait_for = AsyncMock()
loc.all_inner_texts = AsyncMock(return_value=[])
elif "dubbeldag" in sel and "visa-dagar__dagar-kvar" in sel:
loc.first.inner_text = AsyncMock(return_value=_dubbeldag_kvar())
loc.first.wait_for = AsyncMock()
loc.all_inner_texts = AsyncMock(return_value=[])
else:
loc.first.inner_text = AsyncMock(return_value="")
loc.first.wait_for = AsyncMock()
loc.all_inner_texts = AsyncMock(return_value=[])
return loc
page.locator.side_effect = locator_factory
return page
# ---------------------------------------------------------------------------
# Extract from real HTML fixture
# ---------------------------------------------------------------------------
# Import the extractor directly (no browser needed)
from app.scrapers.fk_scraper import _extract_parental_leave # noqa: E402
class TestExtractFromFixture:
"""Parse the captured overview HTML and assert correct field values."""
async def test_days_left_and_total_parsed(self) -> None:
html = _load_fixture("05_foraldrapenning_overview.html")
state = _SessionState(
session_id="test-1",
workspace_id=uuid.uuid4(),
personal_data_uuid=uuid.uuid4(),
child_id=uuid.uuid4(),
)
page = _make_mock_page(html)
result = await _extract_parental_leave(page, state)
# From the capture: 164 av 240 dagar
assert result.days_left == 164
assert result.total_days == 240
assert result.days_used == 76
async def test_tier_values_parsed(self) -> None:
html = _load_fixture("05_foraldrapenning_overview.html")
state = _SessionState(
session_id="test-2",
workspace_id=uuid.uuid4(),
personal_data_uuid=uuid.uuid4(),
child_id=uuid.uuid4(),
)
page = _make_mock_page(html)
result = await _extract_parental_leave(page, state)
# From the capture: sjukpenning 119, lägstanivå 45
assert result.tier_data.sjukpenning_left == 119
assert result.tier_data.lagsta_niva_left == 45
async def test_dubbeldagar_parsed(self) -> None:
html = _load_fixture("05_foraldrapenning_overview.html")
state = _SessionState(
session_id="test-3",
workspace_id=uuid.uuid4(),
personal_data_uuid=uuid.uuid4(),
child_id=uuid.uuid4(),
)
page = _make_mock_page(html)
result = await _extract_parental_leave(page, state)
# From the capture: 39 av 60 dubbeldagar
assert result.dubbeldagar_left == 39
async def test_source_is_scraped(self) -> None:
html = _load_fixture("05_foraldrapenning_overview.html")
state = _SessionState(
session_id="test-4",
workspace_id=uuid.uuid4(),
personal_data_uuid=uuid.uuid4(),
child_id=uuid.uuid4(),
)
page = _make_mock_page(html)
result = await _extract_parental_leave(page, state)
assert result.source == "scraped"
async def test_reserved_days_derived(self) -> None:
html = _load_fixture("05_foraldrapenning_overview.html")
state = _SessionState(
session_id="test-5",
workspace_id=uuid.uuid4(),
personal_data_uuid=uuid.uuid4(),
child_id=uuid.uuid4(),
)
page = _make_mock_page(html)
result = await _extract_parental_leave(page, state)
# 76 days used < 150 transferable → all reserved days remain
assert result.reserved_days_total == 90
assert result.reserved_days_left == 90
assert result.transferable_days_left == 240 - 90 - 76 # 74
async def test_tier_totals_sum_to_total_days(self) -> None:
html = _load_fixture("05_foraldrapenning_overview.html")
state = _SessionState(
session_id="test-6",
workspace_id=uuid.uuid4(),
personal_data_uuid=uuid.uuid4(),
child_id=uuid.uuid4(),
)
page = _make_mock_page(html)
result = await _extract_parental_leave(page, state)
assert (
result.tier_data.sjukpenning_total + result.tier_data.lagsta_niva_total
== result.total_days
)
async def test_tier_used_nonnegative(self) -> None:
html = _load_fixture("05_foraldrapenning_overview.html")
state = _SessionState(
session_id="test-7",
workspace_id=uuid.uuid4(),
personal_data_uuid=uuid.uuid4(),
child_id=uuid.uuid4(),
)
page = _make_mock_page(html)
result = await _extract_parental_leave(page, state)
assert result.tier_data.sjukpenning_used >= 0
assert result.tier_data.lagsta_niva_used >= 0
# ---------------------------------------------------------------------------
# Validation: bad numbers raise rather than persist
# ---------------------------------------------------------------------------
class TestValidation:
"""Feed impossible numbers and assert the extractor raises."""
def _make_bad_page(
self, days_left: int, total_days: int, sjukpenning: int, lagsta: int, dubbel: int
) -> Any:
NIVA_SEL = "[data-test='forDig'] .visa-dagar__desktop .niva-statistik__dagar span"
page = MagicMock()
def locator_factory(sel: str) -> MagicMock:
loc = MagicMock()
if sel == NIVA_SEL:
loc.first.inner_text = AsyncMock(return_value="")
loc.first.wait_for = AsyncMock()
loc.all_inner_texts = AsyncMock(
return_value=[f"({sjukpenning} dagar)", f"({lagsta} dagar)"]
)
elif "visa-dagar__dagar-kvar" in sel and "forDig" in sel:
loc.first.inner_text = AsyncMock(return_value=str(days_left))
loc.first.wait_for = AsyncMock()
loc.all_inner_texts = AsyncMock(return_value=[])
elif "visa-dagar__discrete" in sel and "forDig" in sel:
loc.first.inner_text = AsyncMock(
return_value=f"av {total_days} dagar kvar efter ansökan"
)
loc.first.wait_for = AsyncMock()
loc.all_inner_texts = AsyncMock(return_value=[])
elif "dubbeldag" in sel:
loc.first.inner_text = AsyncMock(return_value=str(dubbel))
loc.first.wait_for = AsyncMock()
loc.all_inner_texts = AsyncMock(return_value=[])
else:
loc.first.inner_text = AsyncMock(return_value="")
loc.first.wait_for = AsyncMock()
loc.all_inner_texts = AsyncMock(return_value=[])
return loc
page.locator.side_effect = locator_factory
return page
def _state(self) -> _SessionState:
return _SessionState(
session_id="val-test",
workspace_id=uuid.uuid4(),
personal_data_uuid=uuid.uuid4(),
child_id=uuid.uuid4(),
)
async def test_negative_days_left_raises(self) -> None:
page = self._make_bad_page(-1, 240, 100, 45, 39)
with pytest.raises(ValueError, match="negative"):
await _extract_parental_leave(page, self._state())
async def test_total_days_parse_failure_raises(self) -> None:
NIVA_SEL = "[data-test='forDig'] .visa-dagar__desktop .niva-statistik__dagar span"
page = MagicMock()
def locator_factory(sel: str) -> MagicMock:
loc = MagicMock()
if sel == NIVA_SEL:
loc.first.inner_text = AsyncMock(return_value="")
loc.first.wait_for = AsyncMock()
loc.all_inner_texts = AsyncMock(return_value=["(80 dagar)", "(20 dagar)"])
elif "dagar-kvar" in sel and "forDig" in sel:
loc.first.inner_text = AsyncMock(return_value="100")
loc.first.wait_for = AsyncMock()
loc.all_inner_texts = AsyncMock(return_value=[])
elif "discrete" in sel and "forDig" in sel:
loc.first.inner_text = AsyncMock(return_value="INVALID TEXT")
loc.first.wait_for = AsyncMock()
loc.all_inner_texts = AsyncMock(return_value=[])
elif "dubbeldag" in sel:
loc.first.inner_text = AsyncMock(return_value="39")
loc.first.wait_for = AsyncMock()
loc.all_inner_texts = AsyncMock(return_value=[])
else:
loc.first.inner_text = AsyncMock(return_value="0")
loc.first.wait_for = AsyncMock()
loc.all_inner_texts = AsyncMock(return_value=[])
return loc
page.locator.side_effect = locator_factory
with pytest.raises(ValueError, match="parse total days"):
await _extract_parental_leave(page, self._state())
async def test_fewer_than_two_niva_values_raises(self) -> None:
NIVA_SEL = "[data-test='forDig'] .visa-dagar__desktop .niva-statistik__dagar span"
page = MagicMock()
def locator_factory(sel: str) -> MagicMock:
loc = MagicMock()
if sel == NIVA_SEL:
loc.first.inner_text = AsyncMock(return_value="")
loc.first.wait_for = AsyncMock()
loc.all_inner_texts = AsyncMock(return_value=["(119 dagar)"]) # only one
elif "dagar-kvar" in sel and "forDig" in sel:
loc.first.inner_text = AsyncMock(return_value="164")
loc.first.wait_for = AsyncMock()
loc.all_inner_texts = AsyncMock(return_value=[])
elif "discrete" in sel and "forDig" in sel:
loc.first.inner_text = AsyncMock(return_value="av 240 dagar kvar efter ansökan")
loc.first.wait_for = AsyncMock()
loc.all_inner_texts = AsyncMock(return_value=[])
elif "dubbeldag" in sel:
loc.first.inner_text = AsyncMock(return_value="39")
loc.first.wait_for = AsyncMock()
loc.all_inner_texts = AsyncMock(return_value=[])
else:
loc.first.inner_text = AsyncMock(return_value="0")
loc.first.wait_for = AsyncMock()
loc.all_inner_texts = AsyncMock(return_value=[])
return loc
page.locator.side_effect = locator_factory
with pytest.raises(ValueError, match="Expected 2 tier values"):
await _extract_parental_leave(page, self._state())
async def test_zero_total_days_raises(self) -> None:
"""Guard against placeholder zero data: total_days=0 is impossible and indicates hydration failure."""
page = self._make_bad_page(164, 0, 119, 45, 39)
with pytest.raises(ValueError, match="total_days=0.*had not hydrated"):
await _extract_parental_leave(page, self._state())
# ---------------------------------------------------------------------------
# Session state machine
# ---------------------------------------------------------------------------
class TestSessionStateMachine:
"""Session transitions, TTL watchdog, and double-cancel safety."""
def setup_method(self) -> None:
_sessions.clear()
def teardown_method(self) -> None:
_sessions.clear()
async def test_start_session_creates_entry(self) -> None:
service = FKScraperService()
with patch("app.scrapers.fk_scraper._run_session", new_callable=AsyncMock) as mock_run:
mock_run.return_value = None
sid = await service.start_session(
workspace_id=uuid.uuid4(),
personal_data_uuid=uuid.uuid4(),
child_id=uuid.uuid4(),
)
assert sid in _sessions
assert _sessions[sid].status == "launching"
async def test_poll_unknown_session_returns_error(self) -> None:
service = FKScraperService()
result = await service.poll_session("nonexistent")
assert result["status"] == "error"
assert "Unknown" in result["message"]
async def test_poll_awaiting_scan_returns_qr_data(self) -> None:
service = FKScraperService()
sid = str(uuid.uuid4())
state = _SessionState(
session_id=sid,
workspace_id=uuid.uuid4(),
personal_data_uuid=uuid.uuid4(),
child_id=uuid.uuid4(),
status="awaiting_scan",
qr_data=b"PNG_BYTES",
)
_sessions[sid] = state
result = await service.poll_session(sid)
assert result["status"] == "awaiting_scan"
assert "qr_data" in result
import base64
assert base64.b64decode(result["qr_data"]) == b"PNG_BYTES"
async def test_cancel_sets_status(self) -> None:
service = FKScraperService()
sid = str(uuid.uuid4())
state = _SessionState(
session_id=sid,
workspace_id=uuid.uuid4(),
personal_data_uuid=uuid.uuid4(),
child_id=uuid.uuid4(),
status="awaiting_scan",
)
_sessions[sid] = state
await service.cancel_session(sid)
assert _sessions[sid].status == "cancelled"
async def test_cancel_nonexistent_session_is_safe(self) -> None:
service = FKScraperService()
await service.cancel_session("does-not-exist") # should not raise
async def test_ttl_watchdog_cancels_stale_session(self) -> None:
from app.scrapers.fk_scraper import _ttl_watchdog
with patch("app.scrapers.fk_scraper.settings") as mock_settings:
mock_settings.scraper_session_ttl_seconds = 0 # immediate
sid = str(uuid.uuid4())
state = _SessionState(
session_id=sid,
workspace_id=uuid.uuid4(),
personal_data_uuid=uuid.uuid4(),
child_id=uuid.uuid4(),
status="awaiting_scan",
)
_sessions[sid] = state
await _ttl_watchdog(sid)
assert _sessions[sid].status == "error"
assert "timed out" in (_sessions[sid].message or "")
async def test_set_status_updates_message(self) -> None:
sid = str(uuid.uuid4())
state = _SessionState(
session_id=sid,
workspace_id=uuid.uuid4(),
personal_data_uuid=uuid.uuid4(),
child_id=uuid.uuid4(),
)
_set_status(state, "error", "Something went wrong")
assert state.status == "error"
assert state.message == "Something went wrong"
async def test_do_extract_calls_persist_callback_on_success(self) -> None:
"""_do_extract invokes the persist_callback with the extracted result."""
persisted: list[ParentalLeaveFetch] = []
async def mock_persist(fetch: ParentalLeaveFetch) -> None:
persisted.append(fetch)
now = datetime.now()
expected = ParentalLeaveFetch(
id=uuid.uuid4(),
personal_data_uuid=uuid.uuid4(),
workspace_id=uuid.uuid4(),
child_id=uuid.uuid4(),
source="scraped",
fetched_at=now,
total_days=240,
days_used=76,
days_left=164,
reserved_days_total=90,
reserved_days_left=90,
transferable_days_left=74,
dubbeldagar_left=39,
tier_data=TierData(
sjukpenning_total=195,
sjukpenning_used=76,
sjukpenning_left=119,
lagsta_niva_total=45,
lagsta_niva_used=0,
lagsta_niva_left=45,
),
created_at=now,
updated_at=now,
)
sid = str(uuid.uuid4())
state = _SessionState(
session_id=sid,
workspace_id=uuid.uuid4(),
personal_data_uuid=uuid.uuid4(),
child_id=uuid.uuid4(),
persist_callback=mock_persist,
)
_sessions[sid] = state
page = MagicMock()
page.wait_for_selector = AsyncMock()
page.wait_for_function = AsyncMock()
def locator_factory(sel: str) -> MagicMock:
loc = MagicMock()
loc.first.click = AsyncMock()
return loc
page.locator.side_effect = locator_factory
with patch(
"app.scrapers.fk_scraper._extract_parental_leave", new_callable=AsyncMock
) as mock_extract:
mock_extract.return_value = expected
await _do_extract(page, state)
assert state.status == "done"
assert state.result is expected
assert len(persisted) == 1
assert persisted[0] is expected
async def test_do_extract_sets_error_when_persist_callback_fails(self) -> None:
"""_do_extract sets status=error if the persist_callback raises."""
async def failing_persist(fetch: ParentalLeaveFetch) -> None:
raise RuntimeError("DB unavailable")
now = datetime.now()
dummy = ParentalLeaveFetch(
id=uuid.uuid4(),
personal_data_uuid=uuid.uuid4(),
workspace_id=uuid.uuid4(),
child_id=uuid.uuid4(),
source="scraped",
fetched_at=now,
total_days=240,
days_used=0,
days_left=240,
reserved_days_total=90,
reserved_days_left=90,
transferable_days_left=150,
dubbeldagar_left=60,
tier_data=TierData(195, 0, 195, 45, 0, 45),
created_at=now,
updated_at=now,
)
sid = str(uuid.uuid4())
state = _SessionState(
session_id=sid,
workspace_id=uuid.uuid4(),
personal_data_uuid=uuid.uuid4(),
child_id=uuid.uuid4(),
persist_callback=failing_persist,
)
_sessions[sid] = state
page = MagicMock()
page.wait_for_selector = AsyncMock()
page.wait_for_function = AsyncMock()
def locator_factory(sel: str) -> MagicMock:
loc = MagicMock()
loc.first.click = AsyncMock()
return loc
page.locator.side_effect = locator_factory
with patch(
"app.scrapers.fk_scraper._extract_parental_leave", new_callable=AsyncMock
) as mock_extract:
mock_extract.return_value = dummy
await _do_extract(page, state)
assert state.status == "error"
assert "persist" in (state.message or "").lower()
async def test_do_extract_waits_for_hydration_before_extracting(self) -> None:
"""_do_extract calls wait_for_function to ensure data hydrates before extraction."""
now = datetime.now()
expected = ParentalLeaveFetch(
id=uuid.uuid4(),
personal_data_uuid=uuid.uuid4(),
workspace_id=uuid.uuid4(),
child_id=uuid.uuid4(),
source="scraped",
fetched_at=now,
total_days=240,
days_used=0,
days_left=240,
reserved_days_total=90,
reserved_days_left=90,
transferable_days_left=150,
dubbeldagar_left=60,
tier_data=TierData(195, 0, 195, 45, 0, 45),
created_at=now,
updated_at=now,
)
sid = str(uuid.uuid4())
state = _SessionState(
session_id=sid,
workspace_id=uuid.uuid4(),
personal_data_uuid=uuid.uuid4(),
child_id=uuid.uuid4(),
)
_sessions[sid] = state
page = MagicMock()
page.wait_for_selector = AsyncMock()
page.wait_for_function = AsyncMock()
def locator_factory(sel: str) -> MagicMock:
loc = MagicMock()
loc.first.click = AsyncMock()
return loc
page.locator.side_effect = locator_factory
with patch(
"app.scrapers.fk_scraper._extract_parental_leave", new_callable=AsyncMock
) as mock_extract:
mock_extract.return_value = expected
await _do_extract(page, state)
# Verify wait_for_function was called with the correct selector
page.wait_for_function.assert_called_once()
call_args = page.wait_for_function.call_args
assert call_args[1]["arg"] == "[data-test='forDig'] .visa-dagar__discrete"
assert call_args[1]["timeout"] == 30_000
# ---------------------------------------------------------------------------
# Persistence integration: real SQLite repo round-trip
# ---------------------------------------------------------------------------
class TestPersistenceIntegration:
"""Run _do_extract with a real SQLiteParentalLeaveRepository and verify the row."""
def setup_method(self) -> None:
_sessions.clear()
def teardown_method(self) -> None:
_sessions.clear()
async def test_scraped_row_is_readable_after_session_done(
self, parental_leave_repo, seeded_ids
) -> None:
"""After _do_extract completes, repo.get_latest_by_parent returns the scraped row."""
async def persist_callback(fetch: ParentalLeaveFetch) -> None:
await parental_leave_repo.create(fetch)
now = datetime.now()
scraped = ParentalLeaveFetch(
id=uuid.uuid4(),
personal_data_uuid=seeded_ids.personal_data_uuid,
workspace_id=seeded_ids.workspace_id,
child_id=seeded_ids.child_id,
source="scraped",
fetched_at=now,
total_days=240,
days_used=76,
days_left=164,
reserved_days_total=90,
reserved_days_left=90,
transferable_days_left=74,
dubbeldagar_left=39,
tier_data=TierData(195, 76, 119, 45, 0, 45),
created_at=now,
updated_at=now,
)
sid = str(uuid.uuid4())
state = _SessionState(
session_id=sid,
workspace_id=seeded_ids.workspace_id,
personal_data_uuid=seeded_ids.personal_data_uuid,
child_id=seeded_ids.child_id,
persist_callback=persist_callback,
)
_sessions[sid] = state
page = MagicMock()
page.wait_for_selector = AsyncMock()
page.wait_for_function = AsyncMock()
def locator_factory(sel: str) -> MagicMock:
loc = MagicMock()
loc.first.click = AsyncMock()
return loc
page.locator.side_effect = locator_factory
with patch(
"app.scrapers.fk_scraper._extract_parental_leave", new_callable=AsyncMock
) as mock_extract:
mock_extract.return_value = scraped
await _do_extract(page, state)
assert state.status == "done"
result = await parental_leave_repo.get_latest_by_parent(
seeded_ids.personal_data_uuid, seeded_ids.child_id
)
assert result is not None
assert result.source == "scraped"
assert result.days_left == 164
assert result.total_days == 240