"""Macro Data Loading: FRED Economic Series — load and explore macroeconomic time series from FRED."""

import plotly.graph_objects as go
import polars as pl
from plotly.subplots import make_subplots

from data import load_macro, load_macro_metadata
from utils.style import COLORS

# Production defaults — Papermill injects overrides for CI

macro = load_macro()
date_col = "timestamp" if "timestamp" in macro.columns else "date"

print(f"Shape: {macro.shape}")
print(f"Date range: {macro[date_col].min()} to {macro[date_col].max()}")

# Load metadata
meta = load_macro_metadata()
category_col = "category" if "category" in meta.columns else "group"
series_col = "series_id" if "series_id" in meta.columns else "series"
print(f"Series count: {len(meta)}")
print(f"Metadata columns: {meta.columns}")

# Group by category
categories = meta.group_by(category_col).agg(
    [
        pl.len().alias("series_count"),
        pl.col(series_col).alias("series"),
    ]
)

for row in categories.iter_rows(named=True):
    print(f"\n{row[category_col].upper()} ({row['series_count']} series):")
    for s in row["series"]:
        print(f"  - {s}")

# Treasury yields - select available columns
yield_cols = ["dgs2", "dgs10"]
yields = macro.select([date_col] + yield_cols).drop_nulls()
yields = yields.sort(date_col)

# Note: "Latest" refers to latest in this offline dataset, not current market
latest_10y = yields.item(-1, "dgs10")
latest_date = yields.item(-1, date_col)

print(f"Observations: {len(yields):,}")
print(f"Latest in dataset ({latest_date}): 10Y = {latest_10y:.2f}%")

# Visualize Treasury yields (last 5 years)
yields_recent = yields.filter(pl.col(date_col) >= pl.lit("2020-01-01").str.to_date())
yields_pd = yields_recent.to_pandas()

# Build the whole chart in one cell. Splitting the traces and the layout across cells
# auto-displays a title-less, annotation-less intermediate before the finished figure
# (the last trace-adding statement returns the figure and renders it). 2Y in copper vs
# 10Y in blue stay distinguishable where the curve inverts and the lines cross.
fig = go.Figure()
fig.add_trace(
    go.Scatter(
        x=yields_pd[date_col],
        y=yields_pd["dgs2"],
        mode="lines",
        name="2-Year Treasury",
        line=dict(color=COLORS["copper"], width=1.5),
    )
)
fig.add_trace(
    go.Scatter(
        x=yields_pd[date_col],
        y=yields_pd["dgs10"],
        mode="lines",
        name="10-Year Treasury",
        line=dict(color=COLORS["blue"], width=1.5),
    )
)
for date, label, y_pos in [
    ("2022-03-16", "Fed hike cycle begins", 0.5),
    ("2023-07-26", "Peak rate 5.25-5.50%", 5.5),
]:
    fig.add_annotation(
        x=date, y=y_pos, text=label, showarrow=True, arrowhead=2, arrowsize=1, ax=30, ay=-30
    )
fig.update_layout(
    title="Both Treasury yields climbed from near zero to about 5% through the 2022-23 hikes",
    xaxis_title="Date",
    yaxis_title="Yield (%)",
    height=400,
    legend=dict(orientation="h", yanchor="bottom", y=1.02, xanchor="center", x=0.5),
)
fig.show()

print("\nKey insight: When 2Y > 10Y, the yield curve is 'inverted', which has")
print("historically preceded recessions by 6-18 months.")

# VIX (volatility) - lowercase column names
vix = macro.select([date_col, "vixcls"]).drop_nulls().sort(date_col)
latest_vix = vix.item(-1, "vixcls")
latest_vix_date = vix.item(-1, date_col)

print(f"Observations: {len(vix):,}")
print(f"Latest in dataset ({latest_vix_date}): {latest_vix:.1f}")
print(f"Historical mean: {vix['vixcls'].mean():.1f}")
print(f"Historical max: {vix['vixcls'].max():.1f} (crisis peak)")

# Whole figure in one cell so only the finished chart renders (the hline calls below
# return the figure and would otherwise auto-display a title-less intermediate). The
# right margin leaves room for the regime-threshold labels so they are not clipped.
vix_recent = vix.filter(pl.col(date_col) >= pl.lit("2020-01-01").str.to_date())
vix_pd = vix_recent.to_pandas()

fig = go.Figure()
fig.add_trace(
    go.Scatter(
        x=vix_pd[date_col],
        y=vix_pd["vixcls"],
        mode="lines",
        name="VIX",
        line=dict(color=COLORS["amber"], width=1.5),
        fill="tozeroy",
        fillcolor="rgba(212, 168, 75, 0.15)",  # translucent COLORS["amber"]
    )
)
# Regime thresholds
fig.add_hline(
    y=20,
    line_dash="dash",
    line_color=COLORS["neutral"],
    annotation_text="Normal (<20)",
    annotation_position="right",
)
fig.add_hline(
    y=30,
    line_dash="dash",
    line_color=COLORS["slate"],
    annotation_text="Elevated (>30)",
    annotation_position="right",
)
for date, label, y_pos in [
    ("2020-03-16", "COVID crash", 82),
    ("2022-01-24", "Fed taper tantrum", 38),
    ("2023-03-13", "SVB collapse", 29),
]:
    fig.add_annotation(x=date, y=y_pos, text=label, showarrow=True, arrowhead=2, ax=0, ay=-30)
fig.update_layout(
    title="VIX spiked near 82 in the 2020 COVID crash, then held mostly below 30",
    xaxis_title="Date",
    yaxis_title="VIX Index",
    height=400,
    showlegend=False,
    margin=dict(r=110),
)
fig.show()

print("\nRegime thresholds:")
print("  VIX < 15: Low volatility (risk-on)")
print("  VIX 15-20: Normal")
print("  VIX 20-30: Elevated")
print("  VIX > 30: High volatility (risk-off)")

slope = macro.select([date_col, "t10y2y"]).drop_nulls()
inversions = slope.filter(pl.col("t10y2y") < 0).height
total_days = slope.height

print("t10y2y = DGS10 - DGS2 (10-Year minus 2-Year Treasury Constant Maturity)")
print(
    f"Yield curve inversions (spread < 0): {inversions:,} days ({100 * inversions / total_days:.1f}%)"
)

# Yield curve spread visualization
slope_recent = slope.filter(pl.col(date_col) >= pl.lit("2020-01-01").str.to_date())
slope_pd = slope_recent.to_pandas()

# Whole figure in one cell so only the finished chart renders (the hline call returns
# the figure and would otherwise auto-display a title-less intermediate); the right
# margin keeps the inversion-line label on-plot.
fig = go.Figure()
fig.add_trace(
    go.Scatter(
        x=slope_pd[date_col],
        y=slope_pd["t10y2y"],
        mode="lines",
        name="10Y-2Y Spread",
        line=dict(color=COLORS["blue"], width=1.5),
    )
)
# Shade inversion periods (spread < 0)
inverted = slope_pd[slope_pd["t10y2y"] < 0]
if len(inverted) > 0:
    fig.add_trace(
        go.Scatter(
            x=inverted[date_col],
            y=inverted["t10y2y"],
            mode="none",
            fill="tozeroy",
            fillcolor="rgba(200, 117, 51, 0.3)",  # translucent COLORS["copper"]
            name="Inversion",
            showlegend=True,
        )
    )
fig.add_hline(
    y=0,
    line_dash="solid",
    line_color=COLORS["negative"],
    line_width=2,
    annotation_text="Inversion line",
    annotation_position="right",
)
fig.update_layout(
    title="The 10Y-2Y spread stayed inverted from mid-2022 into 2024, reaching -1.08 pts",
    xaxis_title="Date",
    yaxis_title="Spread (percentage points)",
    height=400,
    margin=dict(r=110),
    legend=dict(orientation="h", yanchor="bottom", y=1.02, xanchor="center", x=0.5),
)
fig.show()

print("Inversion statistics (since 2020):")
print(f"  Days inverted: {len(inverted)} ({100 * len(inverted) / len(slope_pd):.1f}%)")
print(f"  Deepest inversion: {slope_pd['t10y2y'].min():.2f}%")
print(f"  Most recent value: {slope_pd['t10y2y'].iloc[-1]:.2f}%")

for series in ["dgs10", "unrate", "icsa"]:
    non_null = macro.filter(pl.col(series).is_not_null())
    freq_info = meta.filter(pl.col(series_col) == series)
    freq = freq_info["native_frequency"][0]
    print(f"  {series.upper():6}  {freq:8}  {len(non_null):,} observations")

macro_filled = macro.with_columns(
    pl.col("unrate").forward_fill().alias("unrate_filled"),
)
(
    macro_filled.select([date_col, "unrate", "unrate_filled"])
    .filter(pl.col(date_col).dt.year() == 2024)
    .head(10)
)

# Build the three-panel macro dashboard in a single cell so the inline
# renderer captures the fully populated figure rather than a partial,
# first-panel-only intermediate.
fig = make_subplots(
    rows=3,
    cols=1,
    shared_xaxes=True,
    subplot_titles=(
        "Treasury Yields (2Y, 10Y)",
        "Yield Curve Spread (10Y-2Y)",
        "VIX Volatility Index",
    ),
    vertical_spacing=0.08,
)

# Filter to recent data
start_date = "2020-01-01"
recent = macro.filter(pl.col(date_col) >= pl.lit(start_date).str.to_date()).to_pandas()

# Panel 1: Yields
fig.add_trace(
    go.Scatter(
        x=recent[date_col],
        y=recent["dgs2"],
        mode="lines",
        name="2Y",
        line=dict(color=COLORS["copper"]),
    ),
    row=1,
    col=1,
)
fig.add_trace(
    go.Scatter(
        x=recent[date_col],
        y=recent["dgs10"],
        mode="lines",
        name="10Y",
        line=dict(color=COLORS["blue"]),
    ),
    row=1,
    col=1,
)

# Panel 2: Spread
fig.add_trace(
    go.Scatter(
        x=recent[date_col],
        y=recent["t10y2y"],
        mode="lines",
        name="10Y-2Y",
        line=dict(color=COLORS["blue"]),
        fill="tozeroy",
        fillcolor="rgba(26, 45, 74, 0.12)",  # translucent COLORS["slate"]
    ),
    row=2,
    col=1,
)
fig.add_hline(y=0, line_dash="dash", line_color=COLORS["negative"], row=2, col=1)

# Panel 3: VIX
fig.add_trace(
    go.Scatter(
        x=recent[date_col],
        y=recent["vixcls"],
        mode="lines",
        name="VIX",
        line=dict(color=COLORS["amber"]),
        fill="tozeroy",
        fillcolor="rgba(212, 168, 75, 0.15)",  # translucent COLORS["amber"]
    ),
    row=3,
    col=1,
)
fig.add_hline(y=20, line_dash="dash", line_color=COLORS["neutral"], row=3, col=1)

fig.update_layout(
    height=650,
    title_text="Yields, curve inversion, and VIX all track the 2022-24 tightening cycle",
    showlegend=True,
    legend=dict(orientation="h", yanchor="bottom", y=1.02, xanchor="center", x=0.5),
)

fig.update_yaxes(title_text="Yield (%)", row=1, col=1)
fig.update_yaxes(title_text="Spread (%)", row=2, col=1)
fig.update_yaxes(title_text="Index", row=3, col=1)

fig.show()

print("\nDashboard interpretation:")
print("  - Rising yields + inverted curve = Tightening cycle, recession risk")
print("  - VIX spikes often coincide with curve moves")
print("  - Use together for regime detection in Chapter 8")