spb/hfmarketdata

1"""Bulk fundamentals extracts: listing, Parquet download, ETag / 304."""2from __future__ import annotations34import io56import pandas as pd7import pytest89pytestmark = pytest.mark.usefixtures("fundamentals_data")101112def test_bulk_list_and_download_with_etag(client):13    r = client.get("/v1/bulk/fundamentals")14    assert r.status_code == 20015    years = [y["year"] for y in r.json()["data"]]16    assert 2024 in years17    r = client.get("/v1/bulk/fundamentals/2024.parquet")18    assert r.status_code == 200 and r.headers["content-type"] == "application/vnd.apache.parquet"19    etag = r.headers["ETag"]20    assert etag.startswith('"') and int(r.headers["X-Row-Count"]) > 021    df = pd.read_parquet(io.BytesIO(r.content))22    assert set(df["ticker"]) == {"AAPL", "MSFT", "SHAK"} and set(df["fiscal_year"]) == {2024}23    assert {"income", "balance", "cashflow"} == set(df["statement"])24    aapl_q2 = df[(df.ticker == "AAPL") & (df.statement == "income") & (df.fiscal_quarter == 2)]25    assert len(aapl_q2) == 1 and float(aapl_q2["revenue"].iloc[0]) == 90_753_000_00026    # latest version only: one row per (company, statement, period)27    assert not df.duplicated(["cik", "statement", "fiscal_year", "fiscal_quarter"]).any()28    r304 = client.get("/v1/bulk/fundamentals/2024.parquet", headers={"If-None-Match": etag})29    assert r304.status_code == 304 and r304.headers["X-Row-Count"] == "0" and r304.headers["ETag"] == etag30    r = client.get("/v1/bulk/fundamentals/2024.parquet", headers={"If-None-Match": '"stale"'})31    assert r.status_code == 20032    listed = next(y for y in client.get("/v1/bulk/fundamentals").json()["data"] if y["year"] == 2024)33    assert listed["etag"] == etag and listed["stale"] is False343536def test_bulk_errors(client):37    r = client.get("/v1/bulk/fundamentals/1999.parquet")38    assert r.status_code == 400 and r.json()["error"]["code"] == "INVALID_PARAMETER"39    r = client.get("/v1/bulk/fundamentals/2099.parquet")40    assert r.status_code == 404 and r.json()["error"]["code"] == "NOT_FOUND" and "available_years" in r.json()["error"]["details"]41