tests/test_ingest.py

from unittest.mock import patch, MagicMock, call
from ingest import _period_to_av_quarter


def test_quarterly_period():
    assert _period_to_av_quarter("Q42024") == "2024Q4"
    assert _period_to_av_quarter("Q12025") == "2025Q1"
    assert _period_to_av_quarter("Q22024") == "2024Q2"
    assert _period_to_av_quarter("Q32024") == "2024Q3"


def test_annual_period_maps_to_q4():
    assert _period_to_av_quarter("FY2024") == "2024Q4"


def test_empty_period_returns_empty():
    assert _period_to_av_quarter("") == ""


def test_malformed_period_returns_empty():
    assert _period_to_av_quarter("UNKNOWN") == ""


def _make_edgar(period, form_type):
    """Build a minimal EdgarData-like object for ingest tests."""
    from ingestion.edgar import EdgarData
    return EdgarData(
        ticker="NVDA",
        company_name="NVIDIA",
        filing_date="2025-02-26",
        period=period,
        form_type=form_type,
        revenue=None, revenue_yoy_pct=None, eps=None,
        gross_margin=None, operating_margin=None, free_cash_flow=None,
        mda_text="", risk_factors_text="",
    )


def test_10k_fetches_q4_transcript():
    """ingest() must call fetch_transcript with the Q4 quarter code for 10-K filings."""
    from ingest import ingest

    edgar_10k = _make_edgar("FY2025", "10-K")

    with (
        patch("ingest.fetch_all_edgar_data", return_value=[edgar_10k]),
        patch("ingest.fill_missing_metrics", side_effect=lambda x: x),
        patch("ingest.clear_ticker_data"),
        patch("ingest._extract_guidance", return_value=None),
        patch("ingest.parse_guidance", return_value={}),
        patch("ingest.upsert_metrics"),
        patch("ingest.upsert_section"),
        patch("ingest.embed_and_store_filing"),
        patch("ingest.embed_and_store_transcript"),
        patch("ingest.prune_old_metrics"),
        patch("ingest.init_db"),
        patch("ingest.init_sections_db"),
        patch("ingest.get_all_metrics", return_value=[]),
        patch("ingest.get_section", return_value=None),
        patch("ingest.fetch_transcript", return_value="Q4 transcript text") as mock_transcript,
    ):
        ingest("NVDA")

    mock_transcript.assert_called_once_with("NVDA", "2025Q4")


def test_10q_fetches_quarterly_transcript():
    """ingest() must call fetch_transcript with the correct quarterly code for 10-Q filings."""
    from ingest import ingest

    edgar_10q = _make_edgar("Q12025", "10-Q")

    with (
        patch("ingest.fetch_all_edgar_data", return_value=[edgar_10q]),
        patch("ingest.fill_missing_metrics", side_effect=lambda x: x),
        patch("ingest.clear_ticker_data"),
        patch("ingest._extract_guidance", return_value=None),
        patch("ingest.parse_guidance", return_value={}),
        patch("ingest.upsert_metrics"),
        patch("ingest.upsert_section"),
        patch("ingest.embed_and_store_filing"),
        patch("ingest.embed_and_store_transcript"),
        patch("ingest.prune_old_metrics"),
        patch("ingest.init_db"),
        patch("ingest.init_sections_db"),
        patch("ingest.get_all_metrics", return_value=[]),
        patch("ingest.get_section", return_value=None),
        patch("ingest.fetch_transcript", return_value="Q1 transcript text") as mock_transcript,
    ):
        ingest("NVDA")

    mock_transcript.assert_called_once_with("NVDA", "2025Q1")


def test_delta_backfills_missing_company_profile_sections_without_refetching_transcript():
    """A previously ingested 10-K must be revisited when its profile sections are absent."""
    from ingest import ingest

    edgar_10k = _make_edgar("FY2025", "10-K")
    edgar_10k.business_text = "NVIDIA designs accelerated computing platforms."
    edgar_10k.segments_geography_text = "Revenue is disclosed by reportable segment and geography."

    def stored_section(_ticker, _period, section):
        return "cached transcript" if section == "transcript" else None

    with (
        patch("ingest.fetch_all_edgar_data", return_value=[edgar_10k]),
        patch("ingest.fill_missing_metrics", side_effect=lambda x: x),
        patch("ingest.get_all_metrics", return_value=[{"period": "FY2025"}]),
        patch("ingest.get_section", side_effect=stored_section),
        patch("ingest._extract_guidance", return_value=None),
        patch("ingest.parse_guidance", return_value={}),
        patch("ingest.upsert_metrics"),
        patch("ingest.upsert_section") as mock_upsert_section,
        patch("ingest.embed_and_store_filing") as mock_embed_filing,
        patch("ingest.embed_and_store_transcript"),
        patch("ingest.prune_old_metrics"),
        patch("ingest.init_db"),
        patch("ingest.init_sections_db"),
        patch("ingest.fetch_transcript") as mock_fetch_transcript,
    ):
        ingest("NVDA")

    mock_fetch_transcript.assert_not_called()
    mock_embed_filing.assert_called_once()
    assert mock_embed_filing.call_args.kwargs["business_text"] == edgar_10k.business_text
    assert (
        mock_embed_filing.call_args.kwargs["segments_geography_text"]
        == edgar_10k.segments_geography_text
    )
    assert call(
        "NVDA",
        "FY2025",
        "10-K",
        "business",
        edgar_10k.business_text,
    ) in mock_upsert_section.call_args_list
    assert call(
        "NVDA",
        "FY2025",
        "10-K",
        "segments_geography",
        edgar_10k.segments_geography_text,
    ) in mock_upsert_section.call_args_list