"""
Edge Color Control
==================

This example demonstrates how to control edge color across all plot types,
either globally through ``pp.rcParams['edgecolor']`` or per-call via the
``edgecolor=`` argument.

By default, ``pp.rcParams['edgecolor']`` is ``None``, which means each plot
picks its own edge automatically (typically the face/palette color). Setting
it once applies uniform edges across every plot in the figure.
"""

import publiplots as pp
import pandas as pd
import numpy as np

# Use a higher alpha for this example so face colors are vivid — it makes
# edge colors easier to spot by contrast.
pp.rcParams['alpha'] = 0.8

# %%
# Default Behavior: Automatic Edges
# ---------------------------------
# With the default rcParam (``None``), edges match the face color. Each bar's
# outline follows its palette entry.

np.random.seed(7)
bar_data = pd.DataFrame({
    'group': np.repeat(['A', 'B', 'C', 'D'], 12),
    'value': np.concatenate([
        np.random.normal(45, 7, 12),
        np.random.normal(60, 8, 12),
        np.random.normal(75, 9, 12),
        np.random.normal(90, 10, 12),
    ])
})

fig, ax = pp.subplots(axes_size=(60, 40))
pp.barplot(
    data=bar_data,
    x='group',
    y='value',
    hue='group',
    palette='pastel',
    errorbar='se',
    title='Default: edges match palette',
    ax=ax,
)
pp.show()

# %%
# Global Override via rcParams
# ----------------------------
# Set ``pp.rcParams['edgecolor']`` to apply a single edge color across every
# plot in the script. This is the recommended way to enforce a consistent
# publication style without repeating ``edgecolor=`` on every call.

pp.rcParams['edgecolor'] = 'black'

fig, ax = pp.subplots(axes_size=(60, 40))
pp.barplot(
    data=bar_data,
    x='group',
    y='value',
    hue='group',
    palette='pastel',
    errorbar='se',
    title="rcParams['edgecolor'] = 'black'",
    ax=ax,
)
pp.show()

# %%
# Uniform Edges Across Plot Types
# -------------------------------
# The same rcParam applies to every publiplots function: bars, boxes, violins,
# scatter, strip/swarm, point, raincloud, and dot-heatmaps.

np.random.seed(11)
mixed_data = pd.DataFrame({
    'group': np.repeat(['A', 'B', 'C'], 30),
    'value': np.concatenate([
        np.random.normal(50, 8, 30),
        np.random.normal(65, 9, 30),
        np.random.normal(80, 10, 30),
    ]),
})

# `pp.subplots` declares axes dimensions in mm and extends the figure to
# accommodate decorations. `pp.legend_group` reserves a shared legend
# column on the right whose width is auto-measured from the rendered
# entries — no manual `legend_column=...` guess.
#
# `pp.legend(anchor=...)` attaches BEFORE the plot calls. Plotting
# functions that stash entries (scatterplot, stripplot, swarmplot,
# pointplot in this release) see the group and skip their own per-axis
# legend — the group claims and renders their entries instead.
#
# For plots that don't yet stash (bar, box), we pass `legend=False`; the
# scatterplot on the right anchors the shared "group" entry for the whole
# row since all three panels share the same hue variable.
fig, axes = pp.subplots(1, 3, axes_size=(45, 30))
pp.legend(anchor=axes[-1])

pp.barplot(
    data=mixed_data, x='group', y='value', hue='group',
    palette='pastel', errorbar='se', title='Bar', ax=axes[0], legend=False,
)
pp.boxplot(
    data=mixed_data, x='group', y='value', hue='group',
    palette='pastel', title='Box', ax=axes[1], legend=False,
)
pp.scatterplot(
    data=mixed_data, x='group', y='value', hue='group',
    palette='pastel', title='Scatter', ax=axes[2],
)
pp.show()

# %%
# Composite Plots: Raincloud
# --------------------------
# Raincloud plots layer a violin (KDE cloud), a box, and a strip of raw points
# in one figure. Before the rcParam, each layer had to be passed ``edgecolor``
# individually (or inherit palette edges per layer). Setting the rcParam once
# gives every sub-artist the same outline, which reads much cleaner for
# publications — especially across groups sharing a palette.
#
# Note the ``rain_kws=dict(linewidth=0.5)`` override: the raincloud default
# draws rain points with ``linewidth=0`` for a clean look, but that also hides
# any edge color. Pass a small positive linewidth to make the edge visible on
# the individual points.

np.random.seed(21)
rain_data = pd.DataFrame({
    'condition': np.repeat(['Control', 'Low Dose', 'High Dose'], 40),
    'response': np.concatenate([
        np.random.normal(50, 8, 40),
        np.random.normal(62, 10, 40),
        np.random.normal(78, 9, 40),
    ]),
})

fig, axes = pp.subplots(1, 2, axes_size=(55, 50), sharey=True)

# Left: palette-driven edges (rcParam temporarily off)
pp.rcParams['edgecolor'] = None
pp.raincloudplot(
    data=rain_data,
    x='condition',
    y='response',
    hue='condition',
    palette='pastel',
    title='Auto edges (palette)',
    ax=axes[0],
    legend=False,
)

# Right: uniform black edges via rcParam + visible rain-point strokes.
pp.rcParams['edgecolor'] = 'black'
pp.raincloudplot(
    data=rain_data,
    x='condition',
    y='response',
    hue='condition',
    palette='pastel',
    rain_kws=dict(alpha=0.5, linewidth=0.5),
    title="rcParams['edgecolor'] = 'black'",
    ax=axes[1],
    legend=False,
)

# Both panels share the same groups, so one legend to the right of the
# rightmost axes serves both.
group = pp.legend(anchor=axes[-1])
group.add_legend(
    handles=pp.create_legend_handles(
        labels=['Control', 'Low Dose', 'High Dose'],
        colors=list(pp.color_palette('pastel', 3)),
        alpha=pp.rcParams['alpha'],
        linewidth=pp.rcParams['lines.linewidth'],
        edgecolors='black',
    ),
    label='condition',
)
pp.show()

# %%
# Per-Call Override
# -----------------
# A per-call ``edgecolor=`` argument always wins over the rcParam. Use this
# to tweak a single plot without resetting the global default.

fig, axes = pp.subplots(1, 2, axes_size=(55, 35))
pp.barplot(
    data=bar_data,
    x='group',
    y='value',
    hue='group',
    palette='pastel',
    errorbar='se',
    title="rcParam (black) inherited",
    ax=axes[0],
)
pp.barplot(
    data=bar_data,
    x='group',
    y='value',
    hue='group',
    palette='pastel',
    errorbar='se',
    edgecolor='#c0392b',
    title="edgecolor='#c0392b' (per-call)",
    ax=axes[1],
)
pp.show()

# %%
# Restoring the Default
# ---------------------
# Set the rcParam back to ``None`` to return to per-plot automatic edges.

pp.rcParams['edgecolor'] = None

fig, ax = pp.subplots(axes_size=(60, 40))
pp.barplot(
    data=bar_data,
    x='group',
    y='value',
    hue='group',
    palette='pastel',
    errorbar='se',
    title="Back to default (None = auto)",
    ax=ax,
)
pp.show()

# Restore the global alpha default.
pp.rcParams['alpha'] = 0.1