"""Before/after waveform and spectrum comparison plots."""

import numpy as np
import matplotlib
matplotlib.use("Agg")
import matplotlib.pyplot as plt


def _to_mono(audio):
    """Collapse to mono for plotting."""
    if audio.ndim > 1 and audio.shape[1] > 1:
        return audio.mean(axis=1)
    return audio.ravel()


def _downsample_for_plot(signal, time, max_points=500_000):
    """Reduce sample count so matplotlib stays responsive."""
    if len(signal) > max_points:
        step = len(signal) // max_points
        return signal[::step], time[::step]
    return signal, time


def plot_waveform_comparison(original, mastered, sample_rate):
    """Create a stacked before/after waveform plot.

    Returns a matplotlib Figure.
    """
    fig, axes = plt.subplots(2, 1, figsize=(8, 4), sharex=True)

    duration = len(original) / sample_rate
    time_o = np.linspace(0, duration, len(original))
    time_m = np.linspace(0, duration, len(mastered))

    orig_mono = _to_mono(original)
    mast_mono = _to_mono(mastered)

    orig_mono, time_o = _downsample_for_plot(orig_mono, time_o)
    mast_mono, time_m = _downsample_for_plot(mast_mono, time_m)

    axes[0].plot(time_o, orig_mono, color="#4a90d9", linewidth=0.3)
    axes[0].set_ylabel("Amplitude")
    axes[0].set_title("Original")
    axes[0].set_ylim(-1.05, 1.05)

    axes[1].plot(time_m, mast_mono, color="#d94a4a", linewidth=0.3)
    axes[1].set_ylabel("Amplitude")
    axes[1].set_title("Mastered")
    axes[1].set_xlabel("Time (seconds)")
    axes[1].set_ylim(-1.05, 1.05)

    plt.tight_layout()
    return fig


def plot_spectrum_comparison(original, mastered, sample_rate):
    """Create an overlaid frequency spectrum comparison.

    Returns a matplotlib Figure.
    """
    fig, ax = plt.subplots(1, 1, figsize=(8, 3))

    orig_mono = _to_mono(original)
    mast_mono = _to_mono(mastered)

    n_fft = 8192

    def avg_spectrum(signal, n_fft, sr):
        hop = n_fft // 2
        n_windows = max(1, (len(signal) - n_fft) // hop)
        spectra = []
        for i in range(min(n_windows, 100)):
            start = i * hop
            window = signal[start : start + n_fft] * np.hanning(n_fft)
            spectrum = np.abs(np.fft.rfft(window))
            spectra.append(spectrum)
        avg = np.mean(spectra, axis=0)
        freqs = np.fft.rfftfreq(n_fft, 1.0 / sr)
        avg_db = 20.0 * np.log10(avg + 1e-10)
        return freqs, avg_db

    freqs_o, spec_o = avg_spectrum(orig_mono, n_fft, sample_rate)
    freqs_m, spec_m = avg_spectrum(mast_mono, n_fft, sample_rate)

    ax.plot(freqs_o, spec_o, color="#4a90d9", alpha=0.7, linewidth=1, label="Original")
    ax.plot(freqs_m, spec_m, color="#d94a4a", alpha=0.7, linewidth=1, label="Mastered")
    ax.set_xscale("log")
    ax.set_xlim(20, sample_rate / 2)
    ax.set_xlabel("Frequency (Hz)")
    ax.set_ylabel("Magnitude (dB)")
    ax.set_title("Frequency Spectrum Comparison")
    ax.legend()
    ax.grid(True, alpha=0.3)

    plt.tight_layout()
    return fig