"""tests/test_analyst.py — unit tests for ingestion/analyst.py.

Mocks yfinance so tests are hermetic and fast. Verifies parsing of
earnings_estimate / eps_trend and the 30-day revision computation.
"""
from __future__ import annotations

from datetime import datetime, timedelta
from unittest.mock import MagicMock, patch

import pandas as pd
import pytest

from ingestion import analyst


# Disable cache for these tests (we don't want stale DB state interfering)
@pytest.fixture(autouse=True)
def _no_cache(monkeypatch):
    monkeypatch.setattr("storage.earnings_cache.get", lambda *a, **k: None)
    monkeypatch.setattr("storage.earnings_cache.set", lambda *a, **k: None)


# ── fetch_analyst_estimates ───────────────────────────────────────────────────

def _build_estimate_df(avg_eps=2.35):
    return pd.DataFrame(
        {"avg": [avg_eps, 2.50], "low": [2.20, 2.40], "high": [2.50, 2.60]},
        index=["0q", "+1q"],
    )


def _build_revenue_df(avg_rev=91_400_000_000):
    return pd.DataFrame(
        {"avg": [avg_rev, 95_000_000_000]},
        index=["0q", "+1q"],
    )


def _build_trend_df(current=2.35, ago30=2.31):
    return pd.DataFrame(
        {"current": [current, 2.50], "30daysAgo": [ago30, 2.45]},
        index=["0q", "+1q"],
    )


def test_fetch_analyst_estimates_success():
    fake = MagicMock()
    fake.earnings_estimate = _build_estimate_df()
    fake.revenue_estimate = _build_revenue_df()
    fake.eps_trend = _build_trend_df(current=2.40, ago30=2.30)

    with patch("yfinance.Ticker", return_value=fake):
        data, err = analyst.fetch_analyst_estimates("AAPL")

    assert err is None
    assert data["consensus_eps_est"] == 2.35
    assert data["consensus_rev_est"] == 91_400_000_000
    # (2.40 - 2.30) / 2.30 * 100 ≈ 4.35
    assert abs(data["estimate_revision_30d_pct"] - 4.35) < 0.01


def test_fetch_analyst_estimates_handles_empty_dataframes():
    fake = MagicMock()
    fake.earnings_estimate = pd.DataFrame()
    fake.revenue_estimate = pd.DataFrame()
    fake.eps_trend = pd.DataFrame()

    with patch("yfinance.Ticker", return_value=fake):
        data, err = analyst.fetch_analyst_estimates("XYZ")

    assert err is None
    assert data["consensus_eps_est"] is None
    assert data["consensus_rev_est"] is None
    assert data["estimate_revision_30d_pct"] is None
    assert data["period_aligned"] is False
    assert data["comparison_allowed"] is False


def test_fetch_analyst_estimates_handles_zero_prior_estimate():
    """Avoid division by zero when 30daysAgo == 0."""
    fake = MagicMock()
    fake.earnings_estimate = _build_estimate_df()
    fake.revenue_estimate = _build_revenue_df()
    fake.eps_trend = _build_trend_df(current=2.30, ago30=0.0)

    with patch("yfinance.Ticker", return_value=fake):
        data, err = analyst.fetch_analyst_estimates("AAPL")

    assert err is None
    assert data["estimate_revision_30d_pct"] is None


def test_fetch_analyst_estimates_propagates_exception():
    with patch("yfinance.Ticker", side_effect=RuntimeError("rate limit")):
        data, err = analyst.fetch_analyst_estimates("AAPL")
    assert data is None
    assert "rate limit" in err


# ── fetch_price_reaction ──────────────────────────────────────────────────────

def _make_close_df(start_date: str, prices: list[float]):
    """Return a price-history DataFrame indexed by trading days starting at start_date."""
    idx = pd.bdate_range(start=start_date, periods=len(prices))
    return pd.DataFrame({"Close": prices, "Open": prices, "High": prices, "Low": prices}, index=idx)


def _event_provenance(timing: str, event_date: str = "2026-04-01"):
    published_at = {
        "before_open": f"{event_date}T12:00:00+00:00",  # 08:00 New York (EDT)
        "after_close": f"{event_date}T20:05:00+00:00",  # 16:05 New York (EDT)
    }[timing]
    return analyst.make_earnings_event_provenance(
        ticker="AAPL",
        published_at=published_at,
        source_type="company_ir",
        source_url="https://investor.example.com/earnings-release",
        source_excerpt="Apple announced its quarterly financial results after market close.",
    )


def test_fetch_price_reaction_fails_closed_without_verified_provenance():
    hist = _make_close_df("2026-04-01", [
        100.0,  # 2026-04-01 Wed
        101.0,  # 2026-04-02 Thu
        102.0,  # 2026-04-03 Fri
        103.0,  # 2026-04-06 Mon
        104.0,  # 2026-04-07 Tue
        105.0,  # 2026-04-08 Wed
        106.0,  # 2026-04-09 Thu
        107.0,  # 2026-04-10 Fri
    ])
    with patch("yfinance.download", return_value=hist) as mock_download:
        data, err = analyst.fetch_price_reaction("AAPL", "2026-04-01")

    assert err is None
    assert data["d1_pct"] is None
    assert data["d5_pct"] is None
    assert data["since_release_pct"] is None
    assert data["event_aligned"] is False
    assert data["comparison_allowed"] is False
    assert data["alignment_status"] == "UNVERIFIED_EVENT_PROVENANCE"
    mock_download.assert_not_called()


def test_caller_strings_alone_never_verify_an_earnings_event():
    hist = _make_close_df("2026-04-01", [100.0 + i for i in range(8)])
    with patch("yfinance.download", return_value=hist) as mock_download:
        data, err = analyst.fetch_price_reaction(
            "AAPL", "2026-04-01",
            event_kind="earnings_release", event_timing="after_close",
        )
    assert err is None
    assert data["event_aligned"] is False
    assert data["comparison_allowed"] is False
    assert data["alignment_reason"] == "missing_event_provenance"
    mock_download.assert_not_called()


def test_verified_after_close_uses_event_close_then_next_and_fifth_sessions():
    hist = _make_close_df("2026-04-01", [100.0 + i for i in range(8)])
    provenance = _event_provenance("after_close")
    with patch("yfinance.download", return_value=hist):
        data, err = analyst.fetch_price_reaction(
            "AAPL", "2026-04-01",
            event_kind="earnings_release", event_timing="after_close",
            event_provenance=provenance,
        )
    assert err is None
    assert data["event_aligned"] is True
    assert data["comparison_allowed"] is True
    assert data["alignment_status"] == "VERIFIED_EARNINGS_EVENT"
    assert data["baseline_trading_date"] == "2026-04-01"
    assert data["d1_trading_date"] == "2026-04-02"
    assert data["d5_trading_date"] == "2026-04-08"
    assert data["d1_pct"] == pytest.approx(1.0)
    assert data["d5_pct"] == pytest.approx(5.0)


def test_verified_before_open_uses_prior_close_and_event_session_as_d1():
    hist = _make_close_df("2026-03-31", [100.0 + i for i in range(9)])
    provenance = _event_provenance("before_open")
    with patch("yfinance.download", return_value=hist):
        data, err = analyst.fetch_price_reaction(
            "AAPL", "2026-04-01",
            event_kind="earnings_release", event_timing="before_open",
            event_provenance=provenance,
        )
    assert err is None
    assert data["event_aligned"] is True
    assert data["baseline_trading_date"] == "2026-03-31"
    assert data["d1_trading_date"] == "2026-04-01"
    assert data["d5_trading_date"] == "2026-04-07"
    assert data["d1_pct"] == pytest.approx(1.0)
    assert data["d5_pct"] == pytest.approx(5.0)


def test_tampered_event_provenance_fails_closed_without_price_fetch():
    provenance = _event_provenance("after_close")
    provenance["content_hash"] = "0" * 64
    with patch("yfinance.download") as mock_download:
        data, err = analyst.fetch_price_reaction(
            "AAPL", "2026-04-01",
            event_kind="earnings_release", event_timing="after_close",
            event_provenance=provenance,
        )
    assert err is None
    assert data["event_aligned"] is False
    assert data["alignment_reason"] == "event_content_hash_mismatch"
    mock_download.assert_not_called()


def test_caller_timing_must_match_timestamp_derived_provenance():
    provenance = _event_provenance("after_close")
    with patch("yfinance.download") as mock_download:
        data, err = analyst.fetch_price_reaction(
            "AAPL", "2026-04-01",
            event_kind="earnings_release", event_timing="before_open",
            event_provenance=provenance,
        )
    assert err is None
    assert data["event_aligned"] is False
    assert data["alignment_reason"] == "event_timing_mismatch"
    mock_download.assert_not_called()


def test_fetch_price_reaction_handles_invalid_date():
    data, err = analyst.fetch_price_reaction("AAPL", "not-a-date")
    assert data is None
    assert "invalid filing_date" in err


def test_fetch_price_reaction_handles_empty_history():
    provenance = _event_provenance("after_close")
    with patch("yfinance.download", return_value=pd.DataFrame()):
        data, err = analyst.fetch_price_reaction(
            "AAPL", "2026-04-01",
            event_kind="earnings_release", event_timing="after_close",
            event_provenance=provenance,
        )
    assert data is None
    assert "no price history" in err


# ── post_earnings_returns_batch ───────────────────────────────────────────────

def test_post_earnings_returns_batch_fails_closed_without_provenance():
    hist = _make_close_df("2026-04-01", [100.0 + i for i in range(20)])
    with patch("yfinance.download", return_value=hist) as mock_download:
        out = analyst.post_earnings_returns_batch("AAPL", ["2026-04-01", "2026-04-08"])
    assert out == {}
    mock_download.assert_not_called()


def test_post_earnings_returns_batch_uses_verified_event_windows():
    hist = _make_close_df("2026-04-01", [100.0 + i for i in range(20)])
    provenance = {
        "2026-04-01": _event_provenance("after_close", "2026-04-01"),
        "2026-04-08": _event_provenance("after_close", "2026-04-08"),
    }
    with patch("yfinance.download", return_value=hist):
        out = analyst.post_earnings_returns_batch(
            "AAPL",
            ["2026-04-01", "2026-04-08"],
            event_provenance_by_date=provenance,
        )

    assert "2026-04-01" in out
    assert "2026-04-08" in out
    # Returns are decimals, not percents
    assert 0 < out["2026-04-01"]["d1"] < 1
    assert out["2026-04-01"]["event_aligned"] is True