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.
126 lines
5.6 KiB
Python
126 lines
5.6 KiB
Python
"""Verify EIA units, calendar exceptions, archives and reporting boundaries."""
|
|
import asyncio
|
|
from datetime import UTC, date, datetime, timedelta
|
|
from pathlib import Path
|
|
from types import SimpleNamespace
|
|
from unittest.mock import AsyncMock
|
|
from urllib.parse import urlparse
|
|
|
|
import httpx
|
|
import pytest
|
|
|
|
from alphapulse.calendar import eia
|
|
|
|
FIXTURES = Path(__file__).parent / "fixtures/calendar"
|
|
CSV = (FIXTURES / "eia-2026-10-23.csv").read_text()
|
|
SCHEDULE = (FIXTURES / "eia-schedule.html").read_text()
|
|
|
|
|
|
def test_official_csv_preserves_precision_units_and_excludes_spr():
|
|
report = eia.report(CSV, SCHEDULE)
|
|
assert report.at == datetime(2026, 10, 28, 14, 30, tzinfo=UTC)
|
|
assert report.period == "2026-10-23"
|
|
assert report.values["eia_crude"]["actual"] == "-189.200"
|
|
assert report.values["eia_gasoline"]["actual"] == "387.000"
|
|
assert report.values["eia_distillate"]["actual"] == "-146.200"
|
|
assert all(v["previous"] is None for v in report.values.values())
|
|
|
|
|
|
def test_schedule_uses_holiday_times_and_eastern_dst():
|
|
events = eia.schedule(SCHEDULE, datetime(2026, 10, 30, tzinfo=UTC))
|
|
by_period = {item.period: item.at for item in events}
|
|
assert by_period["2026-10-23"] == datetime(2026, 10, 28, 14, 30, tzinfo=UTC)
|
|
assert by_period["2026-10-30"] == datetime(2026, 11, 4, 15, 30, tzinfo=UTC)
|
|
assert by_period["2026-11-06"] == datetime(2026, 11, 12, 17, tzinfo=UTC)
|
|
assert eia.release_at(date(2025, 12, 19), eia.exceptions(SCHEDULE)) == datetime(2025, 12, 29, 22, tzinfo=UTC)
|
|
|
|
|
|
@pytest.mark.parametrize("body", [
|
|
CSV.replace('"Difference"', '"Changed heading"', 1),
|
|
CSV.replace('"10/16/26"', '"10/15/26"', 1),
|
|
CSV.replace('"Commercial (Excluding SPR)"', '"Other stock"'),
|
|
CSV.replace('"-1.892"', '"NaN"', 1),
|
|
])
|
|
def test_unrecognized_csv_fails_instead_of_filling_other_data(body):
|
|
with pytest.raises(ValueError):
|
|
eia.report(body, SCHEDULE)
|
|
|
|
|
|
def test_following_supply_table_does_not_override_inventory_data():
|
|
body = CSV + '\n"Commercial (Excluding SPR)",1,2,3,4,5,6,7'
|
|
assert eia.report(body, SCHEDULE).values == eia.report(CSV, SCHEDULE).values
|
|
|
|
|
|
@pytest.mark.parametrize("body", ["Access denied", SCHEDULE.replace("10:30 a.m. eastern time on Wednesday", "Changed schedule"),
|
|
SCHEDULE.replace("12:00 p.m.", "Time pending", 1)])
|
|
def test_unrecognized_schedule_fails(body):
|
|
with pytest.raises(ValueError):
|
|
eia.exceptions(body)
|
|
|
|
|
|
def test_archive_discovery_follows_published_report_and_csv_links():
|
|
url = "https://www.eia.gov/petroleum/supply/weekly/archive/2026/2026_10_28/wpsr_2026_10_28.php"
|
|
body = f'<a href="{url}">28</a><a href="https://example.com{urlparse(url).path}">Other host</a>'
|
|
assert eia.archive_links(body, date(2026, 10, 1), date(2026, 10, 30)) == [url]
|
|
assert eia.csv_link('<a href="download/table1.csv">CSV</a>', url) == url.rsplit("/", 1)[0] + "/download/table1.csv"
|
|
with pytest.raises(ValueError):
|
|
eia.archive_links(body, date(2025, 1, 1), date(2025, 12, 31))
|
|
with pytest.raises(ValueError):
|
|
eia.csv_link('<a href="other/table1.csv">Other file</a>', url)
|
|
|
|
|
|
@pytest.mark.parametrize("offset,received,expected", [(0, False, "page"), (7, False, "pending"), (-7, True, "pending"), (-7, False, "error")])
|
|
def test_worker_does_not_apply_latest_values_to_another_week(trading, offset, received, expected):
|
|
from alphapulse.calendar.worker import CalendarWorker
|
|
|
|
async def run():
|
|
worker = CalendarWorker()
|
|
try:
|
|
worker.eia_schedule_body = SCHEDULE
|
|
response = httpx.Response(200, text=CSV, request=httpx.Request("GET", eia.LATEST))
|
|
worker.fetch = AsyncMock(return_value=(response, 1))
|
|
at = datetime(2026, 10, 28, 14, 30, tzinfo=UTC) + timedelta(days=offset)
|
|
row = SimpleNamespace(source="eia", family="eia_stocks", period="2026-10-23", release_url=None,
|
|
scheduled_at=at, first_received_at=at if received else None)
|
|
if expected == "error":
|
|
with pytest.raises(ValueError, match="所选日程不符"):
|
|
await worker.result_page(row)
|
|
else:
|
|
result = await worker.result_page(row)
|
|
assert (result is not None) == (expected == "page")
|
|
finally:
|
|
await worker.http.aclose()
|
|
await worker.bls.aclose()
|
|
asyncio.run(run())
|
|
|
|
|
|
@pytest.mark.parametrize("archive_date,valid", [("2026_10_28", True), ("2026_10_21", False)])
|
|
def test_history_preview_requires_matching_archive_date_without_writes(trading, monkeypatch, archive_date, valid):
|
|
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 []
|
|
|
|
now = datetime(2026, 10, 30, tzinfo=UTC)
|
|
docs = [backfill.Document("eia", eia.SCHEDULE, SCHEDULE.encode(), now),
|
|
backfill.Document("eia", f"https://www.eia.gov/petroleum/supply/weekly/archive/2026/{archive_date}/download/table1.csv", CSV.encode(), now)]
|
|
monkeypatch.setattr(backfill, "SessionLocal", Session)
|
|
monkeypatch.setattr(backfill, "fetch_documents", AsyncMock(return_value=docs))
|
|
save = AsyncMock()
|
|
monkeypatch.setattr(backfill, "save_announcement", save)
|
|
result = asyncio.run(backfill.run(date(2026, 10, 28), date(2026, 10, 28), sources={"eia"}))
|
|
assert result["mode"] == "dry-run"
|
|
assert result["announcements"].get("eia_stocks", 0) == int(valid)
|
|
assert bool(result["errors"]) != valid
|
|
save.assert_not_awaited()
|