eezd
8d6cbeeb40
Collect weekly claims and precise commercial petroleum inventory changes with official schedule exceptions, distinct reporting periods, raw evidence, and revisions. Extend explicit history backfills without changing their default source scope. Report archive gaps and reject mismatched weeks instead of filling current values into history.
181 lines
8.7 KiB
Python
181 lines
8.7 KiB
Python
"""DOL regression coverage uses a public release extract and isolated I/O."""
|
|
import asyncio
|
|
from datetime import UTC, date, datetime, timedelta
|
|
from pathlib import Path
|
|
from types import SimpleNamespace
|
|
from unittest.mock import AsyncMock
|
|
|
|
import httpx
|
|
import pytest
|
|
|
|
from alphapulse.calendar import dol
|
|
|
|
REPORT = (Path(__file__).parent / "fixtures/calendar/dol-current.txt").read_text()
|
|
SCHEDULE = """<html><body>
|
|
Publication Schedule: The UI Weekly Claims News Release is published each week on Thursday morning at 8:30am EST.
|
|
<table><thead><tr><th>Release Date</th><th>Release Time</th></tr></thead><tbody>
|
|
<tr><td>Wednesday, November 25, 2026</td><td>8:30 AM EST</td></tr>
|
|
</tbody></table></body></html>"""
|
|
|
|
|
|
def test_report_separates_statistical_weeks_and_revised_previous():
|
|
initial, continued = (dol.report(REPORT)[family] for family in dol.FAMILIES)
|
|
assert initial.at == continued.at == datetime(2026, 10, 29, 12, 30, tzinfo=UTC)
|
|
assert initial.period == "2026-10-24" and continued.period == "2026-10-17"
|
|
assert initial.values["initial_claims"]["actual"] == "19.4"
|
|
assert initial.values["initial_claims"]["previous"] == "19.5"
|
|
assert continued.values["continued_claims"]["actual"] == "170.1"
|
|
assert continued.values["continued_claims"]["previous"] == "172.2"
|
|
assert "修订" in continued.values["continued_claims"]["previous_basis"]
|
|
|
|
|
|
def test_report_resolves_new_year_from_publication_date():
|
|
text = REPORT.replace("October 29, 2026", "January 2, 2027").replace("October 24", "December 26").replace("October 17", "December 19")
|
|
result = dol.report(text)
|
|
assert result["initial_claims"].period == "2026-12-26"
|
|
assert result["continued_claims"].period == "2026-12-19"
|
|
assert result["initial_claims"].at == datetime(2027, 1, 2, 13, 30, tzinfo=UTC)
|
|
|
|
|
|
def test_historical_report_keeps_explicit_unrevised_previous():
|
|
# Source: https://oui.doleta.gov/press/2025/010825.pdf (first two pages).
|
|
text = (Path(__file__).parent / "fixtures/calendar/dol-2025-01-08.txt").read_text()
|
|
result = dol.report(text)
|
|
assert result["initial_claims"].at == datetime(2025, 1, 8, 13, 30, tzinfo=UTC)
|
|
assert result["initial_claims"].values["initial_claims"]["actual"] == "20.1"
|
|
assert result["initial_claims"].values["initial_claims"]["previous"] == "21.1"
|
|
assert result["continued_claims"].period == "2024-12-28"
|
|
assert result["continued_claims"].values["continued_claims"]["previous"] == "183.4"
|
|
|
|
|
|
@pytest.mark.parametrize("text", [
|
|
"Access denied",
|
|
REPORT.replace("SEASONALLY ADJUSTED DATA", "CHANGED HEADER"),
|
|
REPORT.replace("initial claims was 194,000", "initial claims pending"),
|
|
REPORT.replace("October 24", "September 1"),
|
|
REPORT.replace("195,000.", "unavailable."),
|
|
])
|
|
def test_unrecognized_report_does_not_fabricate_values(text):
|
|
with pytest.raises(ValueError):
|
|
dol.report(text)
|
|
|
|
|
|
def test_schedule_respects_dst_and_thanksgiving_exception():
|
|
events = dol.schedule(SCHEDULE, datetime(2026, 10, 30, tzinfo=UTC))
|
|
initial = {item.at.date(): item for item in events if item.family == "initial_claims"}
|
|
assert initial[date(2026, 10, 29)].at.hour == 12
|
|
assert initial[date(2026, 11, 5)].at.hour == 13
|
|
assert initial[date(2026, 11, 25)].at.hour == 13
|
|
assert date(2026, 11, 26) not in initial
|
|
assert all(item.period is None for item in events)
|
|
assert len({item.external_id for item in events}) == len(events)
|
|
|
|
|
|
@pytest.mark.parametrize("body", [
|
|
SCHEDULE.replace("Thursday morning at 8:30am EST", "Schedule changed"),
|
|
SCHEDULE.replace("8:30 AM EST", "9:00 AM EST"),
|
|
SCHEDULE.replace("November 25, 2026", "11/25/2026"),
|
|
SCHEDULE.replace("Release Date", "Changed heading"),
|
|
])
|
|
def test_changed_schedule_or_exception_format_fails_explicitly(body):
|
|
with pytest.raises(ValueError):
|
|
dol.schedule(body, datetime(2026, 10, 30, tzinfo=UTC))
|
|
|
|
|
|
def test_archives_only_follow_official_links_and_requested_range():
|
|
body = '''<a href="/press/2025/010825.pdf">8</a><a href="/press/2025/010825.pdf">Duplicate</a>
|
|
<a href="https://example.com/press/2025/011625.pdf">Other host</a>
|
|
<a href="/press/2024/122624.pdf">Old</a>'''
|
|
assert dol.archive_links(body, date(2025, 1, 1), date(2025, 12, 31)) == ["https://oui.doleta.gov/press/2025/010825.pdf"]
|
|
with pytest.raises(ValueError, match="历史覆盖尚未确认"):
|
|
dol.archive_links(body, date(2026, 1, 1), date(2026, 12, 31))
|
|
with pytest.raises(ValueError, match="年份"):
|
|
dol.archive_links('<a href="/press/2025/010826.pdf">Bad year</a>', date(2025, 1, 1), date(2026, 12, 31))
|
|
|
|
|
|
@pytest.mark.parametrize("offset,received,expected", [(0, False, "page"), (7, False, "pending"), (-7, True, "pending"), (-7, False, "error")])
|
|
def test_live_result_never_applies_another_week(trading, monkeypatch, offset, received, expected):
|
|
from alphapulse.calendar import worker as module
|
|
|
|
async def run():
|
|
worker = module.CalendarWorker()
|
|
try:
|
|
response = httpx.Response(200, content=b"synthetic PDF", request=httpx.Request("GET", dol.LATEST))
|
|
worker.fetch = AsyncMock(return_value=(response, 7))
|
|
monkeypatch.setattr(module, "pdf_text", lambda body: REPORT)
|
|
at = datetime(2026, 10, 29, 12, 30, tzinfo=UTC) + timedelta(days=offset)
|
|
row = SimpleNamespace(source="dol", family="initial_claims", release_url=None, scheduled_at=at,
|
|
first_received_at=at if received else None)
|
|
if expected == "error":
|
|
with pytest.raises(ValueError, match="公告时间"):
|
|
await worker.result_page(row)
|
|
else:
|
|
page = await worker.result_page(row)
|
|
assert (page is not None) == (expected == "page")
|
|
finally:
|
|
await worker.http.aclose()
|
|
await worker.bls.aclose()
|
|
asyncio.run(run())
|
|
|
|
|
|
def test_weekly_rows_show_their_own_week_and_uncertain_crypto_impact(trading):
|
|
from alphapulse.calendar.service import release_rows
|
|
announcement = dol.report(REPORT)["continued_claims"]
|
|
row = SimpleNamespace(id=1, family="continued_claims", source="dol", title="Weekly claims",
|
|
period=announcement.period, stage=None, scheduled_at=announcement.at, schedule_status="scheduled",
|
|
first_received_at=announcement.at, last_checked_at=announcement.at, values=announcement.values,
|
|
error=None, release_url=dol.LATEST, schedule_url=dol.SCHEDULE, revision_no=2)
|
|
[item] = release_rows(row, announcement.at)
|
|
assert item["actual"] == "170.1" and item["unit"] == "万人"
|
|
assert "2026-10-17" in item["event_name"]
|
|
assert item["impact"] == "uncertain" and item["forecast"] is None
|
|
|
|
|
|
def test_history_preview_includes_both_series_without_writing(trading, monkeypatch):
|
|
from alphapulse.calendar import backfill
|
|
|
|
class Session:
|
|
async def __aenter__(self):
|
|
return self
|
|
|
|
async def __aexit__(self, *_):
|
|
pass
|
|
|
|
async def execute(self, _):
|
|
return SimpleNamespace(all=lambda: [])
|
|
|
|
async def scalars(self, _):
|
|
return []
|
|
|
|
doc = backfill.Document("dol", dol.LATEST, b"%PDF synthetic report", datetime(2026, 10, 30, tzinfo=UTC))
|
|
monkeypatch.setattr(backfill, "SessionLocal", Session)
|
|
monkeypatch.setattr(backfill, "fetch_documents", AsyncMock(return_value=[doc]))
|
|
monkeypatch.setattr(backfill, "pdf_text", lambda body: REPORT)
|
|
save = AsyncMock()
|
|
monkeypatch.setattr(backfill, "save_announcement", save)
|
|
result = asyncio.run(backfill.run(date(2026, 10, 29), date(2026, 10, 29), sources={"dol"}))
|
|
assert result["mode"] == "dry-run"
|
|
assert result["announcements"] == {"initial_claims": 1, "continued_claims": 1}
|
|
assert not result["errors"] and not result["missing_announcements"]
|
|
save.assert_not_awaited()
|
|
|
|
|
|
def test_archive_year_gaps_are_explicit_and_current_report_is_still_collected(trading, monkeypatch):
|
|
from alphapulse.calendar import backfill
|
|
requests = []
|
|
|
|
def respond(request):
|
|
requests.append(str(request.url))
|
|
if request.url.path.endswith("archive.asp"):
|
|
return httpx.Response(200, text='<a href="/press/2025/010825.pdf">8</a>')
|
|
return httpx.Response(200, content=b"%PDF synthetic report")
|
|
|
|
client = httpx.AsyncClient
|
|
monkeypatch.setattr(httpx, "AsyncClient", lambda **kwargs: client(**kwargs, transport=httpx.MockTransport(respond)))
|
|
monkeypatch.setattr(backfill, "ensure_quiet_window", AsyncMock())
|
|
monkeypatch.setattr(backfill.asyncio, "sleep", AsyncMock())
|
|
errors = []
|
|
documents = asyncio.run(backfill.fetch_documents(date(2025, 1, 1), date(2026, 10, 30), set(), errors, sources={"dol"}))
|
|
assert len(documents) == 2 and dol.LATEST in requests
|
|
assert len(errors) == 1 and "2026" in errors[0]["error"] and "不完整" in errors[0]["error"]
|