"""The point-in-time table, and the arithmetic it is allowed to do."""

from __future__ import annotations

from datetime import UTC, date, datetime

import polars as pl

from recipe.pipeline import with_rate_changes
from recipe.pit import build_pit, horizon_label, weighted_median
from recipe.schema import EVENT_SCHEMA, align


def test_horizon_is_relative_to_the_meeting_that_projected_it():
    assert horizon_label("2026", 2026) == "Y0"
    assert horizon_label("2028", 2026) == "Y2"
    assert horizon_label("longer_run", 2026) == "LR"
    # A projection for a year already past is not a horizon; it is a misread
    # column, and it is dropped rather than filed under a negative offset.
    assert horizon_label("2019", 2026) is None


def test_the_median_dot_counts_each_participant_once():
    assert weighted_median([3.0, 3.5, 4.0], [1, 1, 1]) == 3.5
    assert weighted_median([3.0, 4.0], [2, 2]) == 3.5
    assert weighted_median([3.0, 3.5, 4.0], [5, 1, 1]) == 3.0
    assert weighted_median([], []) is None


def _events() -> pl.DataFrame:
    rows = [
        {
            "event_id": "US_FOMC_2026-01-28_STATEMENT", "meeting_id": "US_FOMC_2026-01-28",
            "event_type": "STATEMENT",
            "knowledge_at": datetime(2026, 1, 28, 19, 0, tzinfo=UTC),
            "target_rate_lower": 3.75, "target_rate_upper": 4.0, "target_rate_mid": 3.875,
            "decision": "hold", "vote_for": 12, "vote_against": 0,
            "knowledge_time_precision": "exact", "source_url": "https://example.invalid/a",
            "parser_version": "1.0.0",
        },
        {
            "event_id": "US_FOMC_2026-03-18_STATEMENT", "meeting_id": "US_FOMC_2026-03-18",
            "event_type": "STATEMENT",
            "knowledge_at": datetime(2026, 3, 18, 18, 0, tzinfo=UTC),
            "target_rate_lower": 3.5, "target_rate_upper": 3.75, "target_rate_mid": 3.625,
            "decision": "cut", "vote_for": 11, "vote_against": 1,
            "knowledge_time_precision": "exact", "source_url": "https://example.invalid/b",
            "parser_version": "1.0.0",
        },
    ]
    return align(pl.DataFrame(rows, infer_schema_length=None), EVENT_SCHEMA)


def _meetings() -> pl.DataFrame:
    return pl.DataFrame(
        {
            "meeting_id": ["US_FOMC_2026-01-28", "US_FOMC_2026-03-18"],
            "meeting_end_date": [date(2026, 1, 28), date(2026, 3, 18)],
        }
    )


def test_rate_change_is_measured_against_the_previous_statement():
    events = with_rate_changes(_events())
    changes = dict(zip(events["event_id"], events["rate_change_bp"], strict=True))
    assert changes["US_FOMC_2026-01-28_STATEMENT"] is None
    assert changes["US_FOMC_2026-03-18_STATEMENT"] == -25.0


def test_pit_rows_carry_the_instant_the_event_did():
    frames = {
        "events": _events(),
        "meetings": _meetings(),
        "sep": pl.DataFrame(schema={"meeting_id": pl.String, "median": pl.Float64}),
        "dot_plot": pl.DataFrame(
            schema={
                "meeting_id": pl.String, "projection_year": pl.String,
                "rate_level": pl.Float64, "dot_count": pl.Int32,
                "knowledge_at": pl.Datetime(time_unit="us", time_zone="UTC"),
                "source_url": pl.String,
            }
        ),
    }
    pit = build_pit(frames)
    assert set(pit["entity_id"].unique()) == {
        "US_FOMC_TARGET_RATE_LOWER", "US_FOMC_TARGET_RATE_UPPER", "US_FOMC_TARGET_RATE_MID",
    }
    mid = pit.filter(pl.col("entity_id") == "US_FOMC_TARGET_RATE_MID").sort("knowledge_date")
    assert mid["actual"].to_list() == [3.875, 3.625]
    assert mid["previous"].to_list() == [None, 3.875]
    assert mid["change"].to_list() == [None, -0.25]
    # Nothing is ever knowable before it was published.
    assert (mid["knowledge_date"] >= mid["event_date"].cast(pl.Datetime("us", "UTC"))).all()


def test_forecast_is_always_null():
    """Consensus estimates are licensed data; this dataset carries none."""

    frames = {
        "events": _events(), "meetings": _meetings(),
        "sep": pl.DataFrame(schema={"meeting_id": pl.String, "median": pl.Float64}),
        "dot_plot": pl.DataFrame(
            schema={
                "meeting_id": pl.String, "projection_year": pl.String,
                "rate_level": pl.Float64, "dot_count": pl.Int32,
                "knowledge_at": pl.Datetime(time_unit="us", time_zone="UTC"),
                "source_url": pl.String,
            }
        ),
    }
    assert build_pit(frames)["forecast"].null_count() == build_pit(frames).height