from __future__ import annotations

from datetime import UTC, date, datetime

import numpy as np
import polars as pl
import pytest

from countryatlas.pipeline.changes import Series, compute_changes_and_events, detect_changes, detect_events
from countryatlas.registry import indicators_by_id


def _series(ind: str, values: list[float], start: int = 2000, country: str = "CAN") -> Series:
    years = np.arange(start, start + len(values), dtype=np.int32)
    return Series(country, ind, [date(int(y), 1, 1) for y in years], years, np.array(values, dtype=float))


def test_inflation_drop_headline_and_since() -> None:
    ind = indicators_by_id()["inflation"]
    vals = [2.0, 2.1, 1.9, 2.2, 2.0, 2.3, 8.1, 5.9, 5.5, 3.4]  # 2015: 5.5 → 2016: 3.4 (−2.1 pts) hmm: last drop 2.1
    vals = [2.0, 2.1, 1.9, 2.2, 2.0, 2.3, 2.1, 2.0, 6.8, 3.4]
    out = detect_changes(_series("inflation", vals, 2016), ind)
    kinds = {r["kind"]: r for r in out}
    assert "yoy_drop" in kinds
    h = kinds["yoy_drop"]["headline"]
    assert h.startswith("Inflation fell 3.4 points to 3.4 % in 2025")
    assert "largest drop" in h
    assert 0 < kinds["yoy_drop"]["severity"] <= 1


def test_floor_blocks_small_moves() -> None:
    ind = indicators_by_id()["inflation"]  # change_floor 2 points
    vals = [2.0, 2.0, 2.0, 2.0, 2.0, 2.0, 2.0, 2.0, 2.0, 3.0]  # +1 pt: huge z but below the floor
    out = detect_changes(_series("inflation", vals), ind)
    assert not any(r["kind"] in ("yoy_jump", "yoy_drop") for r in out)


def test_record_high_and_relative_headline() -> None:
    ind = indicators_by_id()["gdp-per-capita"]
    vals = [1000 * 1.03**k for k in range(15)]
    vals[5] *= 0.9  # a dip: the series must not be monotone, otherwise records are (rightly) not newsworthy
    vals[-1] = vals[-2] * 1.25
    out = detect_changes(_series("gdp-per-capita", vals), ind)
    kinds = {r["kind"] for r in out}
    assert "record_high" in kinds and "yoy_jump" in kinds
    jump = next(r for r in out if r["kind"] == "yoy_jump")
    assert "25.0 %" in jump["headline"] and "US$" in jump["headline"]


def test_sign_flip_and_events_history() -> None:
    ind = indicators_by_id()["gdp-growth"]
    vals = [3.0, 2.5, 3.1, 2.8, 3.0, 2.9, -4.5, 5.0, 2.0, -1.0]
    ch = detect_changes(_series("gdp-growth", vals), ind)
    assert any(r["kind"] == "sign_flip" for r in ch)
    ev = detect_events(_series("gdp-growth", vals), ind)
    flips = [r for r in ev if r["kind"] == "sign_flip"]
    assert len(flips) == 3  # 2006, 2007, 2009
    assert any(r["kind"] == "yoy_drop" and r["year"] == 2006 for r in ev)


def test_monotone_and_cumulative_series_are_silent() -> None:
    inds = indicators_by_id()
    vals = [100.0 * 1.02**k for k in range(30)]  # strictly increasing: every point is a "record"
    ch = detect_changes(_series("population", vals), inds["population"])
    assert not any(r["kind"] in ("record_high", "n_year_high") for r in ch)
    rows = [{"country_id": "CAN", "indicator_id": "cumulative-co2", "period": date(1990 + k, 1, 1), "year": 1990 + k, "value": v}
            for k, v in enumerate(vals)]
    ch2, ev2 = compute_changes_and_events(pl.DataFrame(rows), inds)
    assert ch2.height == 0 and ev2.height == 0


def test_changes_are_recent_and_weighted() -> None:
    inds = indicators_by_id()
    old = [2.0, 2.1, 1.9, 2.2, 2.0, 2.3, 2.1, 2.0, 6.8, 3.4]  # series ends in 2008
    rows = [{"country_id": "CAN", "indicator_id": "inflation", "period": date(1999 + k, 1, 1), "year": 1999 + k, "value": v}
            for k, v in enumerate(old)]
    rows += [{"country_id": "FRA", "indicator_id": "inflation", "period": date(2016 + k, 1, 1), "year": 2016 + k, "value": v}
             for k, v in enumerate(old)]
    from datetime import UTC, datetime

    ch, ev = compute_changes_and_events(pl.DataFrame(rows), inds, headline_ids={"inflation"}, now=datetime(2026, 9, 1, tzinfo=UTC))
    assert set(ch["country_id"].to_list()) == {"FRA"}  # Canada's 2008 drop is an event, not a change
    assert "CAN" in set(ev["country_id"].to_list())
    import json

    d = json.loads(ch["detail"][0])
    assert d["weight"] == 1.0 and "raw_severity" in d
    ch_w, _ = compute_changes_and_events(pl.DataFrame(rows), inds, headline_ids=set(), now=datetime(2026, 9, 1, tzinfo=UTC))
    assert json.loads(ch_w["detail"][0])["weight"] == 0.9  # inflation is featured but not headline here
    assert ch_w["severity"][0] == pytest.approx(min(1.0, d["raw_severity"] * 0.9), rel=1e-3)


def test_index_series_need_10pct_and_quarterly_is_annualised() -> None:
    inds = indicators_by_id()
    ind = inds["real-house-price-index"]  # format index, ranking_eligible false
    assert ind.format == "index" and not ind.ranking_eligible
    smooth = [100 + k for k in range(12)]
    smooth[-1] = smooth[-2] * 1.05  # +5 %: big z, but below the 10 % rule
    assert not any(r["kind"] in ("yoy_jump", "yoy_drop") for r in detect_changes(_series("real-house-price-index", smooth), ind))
    big = list(smooth)
    big[-1] = big[-2] * 1.2  # +20 %
    assert any(r["kind"] == "yoy_jump" for r in detect_changes(_series("real-house-price-index", big), ind))
    # quarterly rows collapse to one value per year (last value for an index) → at most one detection per year
    rows = []
    for y in range(2015, 2026):
        for q, month in enumerate((1, 4, 7, 10)):
            v = 100 + (y - 2015) * 2 + q * 0.3
            if y == 2025:
                v = 150.0 + q  # +~25 % jump in 2025
            rows.append({"country_id": "CAN", "indicator_id": "real-house-price-index", "period": date(y, month, 1), "year": y,
                         "frequency": "Q", "value": v})
    ch, ev = compute_changes_and_events(pl.DataFrame(rows), inds, now=datetime(2026, 1, 1, tzinfo=UTC))
    assert ch.filter(pl.col("kind") == "yoy_jump").height == 1
    assert ev.group_by(["indicator_id", "year"]).len()["len"].max() <= 3
    assert ev["severity"].min() >= 0.35
    assert ch.filter(pl.col("kind") == "yoy_jump")["value"][0] == 153.0  # last quarter of 2025


def test_driver_over_frame() -> None:
    inds = indicators_by_id()
    rows = []
    for c in ("CAN", "FRA"):
        for k, v in enumerate([2.0, 2.1, 1.9, 2.2, 2.0, 2.3, 2.1, 2.0, 6.8, 3.4]):
            rows.append({"country_id": c, "indicator_id": "inflation", "period": date(2016 + k, 1, 1), "year": 2016 + k, "value": v})
    df = pl.DataFrame(rows)
    ch, ev = compute_changes_and_events(df, inds)
    assert ch.filter(pl.col("kind") == "yoy_drop").height == 2
    assert set(ch.columns) >= {"id", "country_id", "indicator_id", "kind", "headline", "severity", "detected_at"}
    assert ev.height >= 2


# ------------------------------------------------------------------------------------------------ change detection 2.0
def test_structural_break_in_changes_and_events() -> None:
    ind = indicators_by_id()["unemployment-rate"]
    # 12 years around 5 % (with small noise so the series is not monotone), then 8 years around 12 % → break in 2012
    vals = [5.0, 5.2, 4.9, 5.1, 5.0, 5.3, 4.8, 5.1, 5.0, 5.2, 4.9, 5.1, 12.0, 12.2, 11.9, 12.1, 12.0, 12.3, 11.8, 12.1]
    ch = detect_changes(_series("unemployment-rate", vals, 2000), ind)  # last year 2019, break 2012 → within 10 years
    brk = [r for r in ch if r["kind"] == "structural_break"]
    assert len(brk) == 1
    b = brk[0]
    assert b["year"] == 2012 and "shifted to a higher level around 2012" in b["headline"]
    assert 0.45 <= b["severity"] <= 0.7
    import json as _json

    d = b["detail"] if isinstance(b["detail"], dict) else _json.loads(b["detail"])
    assert d["break_year"] == 2012 and d["gain"] >= 0.5 and d["shift_ratio"] >= 1.5 and d["min_segment"] == 5
    ev = detect_events(_series("unemployment-rate", vals, 2000), ind)
    assert len([r for r in ev if r["kind"] == "structural_break"]) == 1
    # an old break (25 years ago) stays an event but is not a recent change
    old = vals + [12.0, 12.2, 11.9, 12.1, 12.0, 12.3, 11.8, 12.1, 12.0, 12.2, 11.9, 12.1, 12.0, 12.3, 11.8]
    ch_old = detect_changes(_series("unemployment-rate", old, 1988), ind)  # last year 2022, break 2000 → 22 years ago
    assert not any(r["kind"] == "structural_break" for r in ch_old)
    assert any(r["kind"] == "structural_break" for r in detect_events(_series("unemployment-rate", old, 1988), ind))
    # a smooth trend is not a level shift
    trend = [5.0 + 0.3 * k for k in range(20)]
    assert not any(r["kind"] == "structural_break" for r in detect_changes(_series("unemployment-rate", trend, 2000), ind))


def test_trend_reversal() -> None:
    ind = indicators_by_id()["unemployment-rate"]  # floor 1 point
    vals = [8.0, 8.0, 8.0, 9.0, 8.0, 7.0, 6.0, 7.0, 8.0, 9.0]  # 3 falls (9→6) then 3 rises (6→9): +3 points ≥ floor
    ch = detect_changes(_series("unemployment-rate", vals, 2010), ind)
    rev = [r for r in ch if r["kind"] == "trend_reversal"]
    assert len(rev) == 1
    assert rev[0]["headline"].startswith("Unemployment has risen for three consecutive years after falling for three")
    assert rev[0]["severity"] == pytest.approx(0.55) and rev[0]["ref_value"] == 6.0
    down = [3.0, 3.0, 3.0, 2.0, 3.0, 4.0, 5.0, 4.0, 3.0, 2.0]
    r2 = [r for r in detect_changes(_series("unemployment-rate", down, 2010), ind) if r["kind"] == "trend_reversal"]
    assert len(r2) == 1 and "has fallen for three consecutive years after rising for three" in r2[0]["headline"]
    # too small a swing (0.3 points) is silent
    tiny = [8.0, 8.0, 8.0, 8.3, 8.2, 8.1, 8.0, 8.1, 8.2, 8.3]
    assert not any(r["kind"] == "trend_reversal" for r in detect_changes(_series("unemployment-rate", tiny, 2010), ind))


def test_volatility_spike() -> None:
    ind = indicators_by_id()["inflation"]  # floor 2 points
    calm = [2.0 + 0.1 * ((-1) ** k) for k in range(16)]  # 15 tiny diffs (±0.2)
    wild = [2.0, 9.0, 1.0, 8.0, 0.0]  # 5 large swings
    ch = detect_changes(_series("inflation", calm + wild, 2005), ind)
    vol = [r for r in ch if r["kind"] == "volatility_spike"]
    assert len(vol) == 1
    assert "unusually volatile" in vol[0]["headline"] and 0.45 <= vol[0]["severity"] <= 0.7
    import json as _json

    d = vol[0]["detail"] if isinstance(vol[0]["detail"], dict) else _json.loads(vol[0]["detail"])
    assert d["ratio"] >= 3.0 and d["recent_years"] == 5 and d["baseline_years"] == 15
    # same pattern but the recent swings are below the 2-point floor → silent
    small = calm + [2.0, 2.5, 2.0, 2.6, 2.1]
    assert not any(r["kind"] == "volatility_spike" for r in detect_changes(_series("inflation", small, 2005), ind))
    # too short a history → silent
    assert not any(r["kind"] == "volatility_spike" for r in detect_changes(_series("inflation", wild * 2, 2015), ind))


def test_new_kinds_flow_through_driver_with_weights() -> None:
    inds = indicators_by_id()
    vals = [5.0, 5.2, 4.9, 5.1, 5.0, 5.3, 4.8, 5.1, 5.0, 5.2, 4.9, 5.1, 12.0, 12.2, 11.9, 12.1, 12.0, 12.3, 11.8, 12.1]
    rows = [{"country_id": "CAN", "indicator_id": "unemployment-rate", "period": date(2006 + k, 1, 1), "year": 2006 + k, "value": v}
            for k, v in enumerate(vals)]
    ch, ev = compute_changes_and_events(pl.DataFrame(rows), inds, headline_ids={"unemployment-rate"}, now=datetime(2026, 9, 1, tzinfo=UTC))
    kinds = set(ch["kind"].to_list())
    assert "structural_break" in kinds
    assert ev.filter(pl.col("kind") == "structural_break").height == 1