tape_recorder/core/audio_player.py

"""
Audio Player — Play audio files through the EQ engine to the sound card output.
"""

import sounddevice as sd
import numpy as np
from typing import Optional, Callable
import threading

from .file_loader import load_audio, AudioData, normalize_audio
from .eq_engine import EQEngine
from .tape_processor import TapeProcessor
from .compressor import Compressor
from .audio_device import DeviceManager

class AudioPlayer:
    """
    Handles playback of audio files through a TapeProcessor and EQEngine.
    """
    def __init__(self, eq_engine: EQEngine, tape_processor: TapeProcessor, device_manager: DeviceManager):
        self.eq_engine = eq_engine
        self.tape_processor = tape_processor
        self.compressor = Compressor()
        self.device_manager = device_manager
        
        self.audio_data: Optional[AudioData] = None
        self.stream: Optional[sd.OutputStream] = None
        
        self.current_frame: float = 0.0
        self.is_playing: bool = False
        self.is_looping: bool = False
        
        # Callbacks for UI updates
        self.on_position_update: Optional[Callable[[float], None]] = None
        self.on_playback_stopped: Optional[Callable[[], None]] = None
        
        # VU meter levels (left, right RMS in dB)
        self.vu_levels = (-100.0, -100.0)
        
        self.playback_speed: float = 1.0
        self.output_volume: float = 1.0
        self._is_restarting: bool = False
        
        self._last_raw_chunk: Optional[np.ndarray] = None
        self._last_processed_chunk: Optional[np.ndarray] = None

    def load_file(self, file_path: str, normalize: bool = False):
        """Load an audio file and prepare for playback."""
        self.stop()
        audio = load_audio(file_path)
        
        if normalize:
            audio = normalize_audio(audio)
        

            
        # Convert mono to stereo for output if needed
        if audio.channels == 1 and self.device_manager.channels == 2:
            audio.samples = np.repeat(audio.samples, 2, axis=1)
            audio.channels = 2
            
        self.audio_data = audio
        self.current_frame = 0.0

    def play(self):
        """Start or resume playback."""
        if self.audio_data is None:
            return
            
        if self.is_playing:
            return
            
        out_sr = self.device_manager.sample_rate
        
        self.eq_engine.reset_state()
        self.eq_engine.set_sample_rate(out_sr)
        
        self.tape_processor.reset_state()
        self.tape_processor.set_sample_rate(out_sr)
        
        kwargs = self.device_manager.get_sd_output_kwargs()
        
        self.stream = sd.OutputStream(
            callback=self._audio_callback,
            finished_callback=self._on_finished,
            **kwargs
        )
        
        self.is_playing = True
        self.stream.start()

    def pause(self):
        """Pause playback."""
        if self.stream:
            try:
                self.stream.stop()
                self.stream.close()
            except Exception as e:
                print(f"Error closing stream: {e}")
            self.stream = None
        self.is_playing = False

    def stop(self):
        """Stop playback and reset position."""
        self.pause()
        self.current_frame = 0.0
        self.vu_levels = (-100.0, -100.0)
        if self.on_position_update:
            self.on_position_update(0.0)

    def set_speed(self, speed: float):
        """Set playback speed. Stream adjusts seamlessly via interpolation."""
        self.playback_speed = max(0.1, speed)

    def seek(self, position_sec: float):
        """Seek to a specific position in seconds."""
        if self.audio_data is None:
            return
        
        frame = position_sec * self.audio_data.sample_rate
        self.current_frame = max(0.0, min(float(frame), float(self.audio_data.num_samples - 1)))
        
        if self.is_playing:
            self.eq_engine.reset_state()
            self.tape_processor.reset_state()
            
    def _audio_callback(self, outdata: np.ndarray, frames: int, time, status: sd.CallbackFlags):
        """sounddevice callback for generating output audio."""
        if status:
            print(f"Playback status: {status}")
            
        if self.audio_data is None or not self.is_playing:
            outdata.fill(0)
            return

        # Ratio of input samples to advance per output sample
        ratio = (self.audio_data.sample_rate / self.device_manager.sample_rate) * self.playback_speed
        
        # We need to produce `frames` samples
        out_indices = np.arange(frames)
        in_indices = self.current_frame + out_indices * ratio
        
        # Find which indices are valid
        valid_mask = in_indices < (self.audio_data.num_samples - 1)
        valid_count = int(np.sum(valid_mask))
        
        if valid_count == 0:
            if self.is_looping:
                self.current_frame = 0.0
                self.eq_engine.reset_state()
                self.tape_processor.reset_state()
                in_indices = self.current_frame + out_indices * ratio
                valid_mask = in_indices < (self.audio_data.num_samples - 1)
                valid_count = int(np.sum(valid_mask))
            else:
                outdata.fill(0)
                raise sd.CallbackStop()

        # Get integer part and fractional part
        in_idx_int = in_indices[:valid_count].astype(np.int32)
        in_idx_frac = in_indices[:valid_count] - in_idx_int
        
        # Broadcast for channels (e.g. 2 channels)
        in_idx_frac = in_idx_frac[:, np.newaxis].astype(np.float32)
        
        c0 = self.audio_data.samples[in_idx_int]
        c1 = self.audio_data.samples[in_idx_int + 1]
        
        raw_chunk = c0 + (c1 - c0) * in_idx_frac
        
        # Tape Processing & EQ Chain
        chunk = self.tape_processor.process_pre_eq(raw_chunk)
        chunk = self.eq_engine.process(chunk)
        chunk = self.tape_processor.process_post_eq(chunk)
        
        # Dynamic Range Compression
        chunk = self.compressor.process(chunk)
        
        # Apply output volume
        if self.output_volume != 1.0:
            chunk = chunk * self.output_volume
        
        # Write to output buffer
        outdata[:valid_count] = chunk
        
        self._last_raw_chunk = np.copy(raw_chunk)
        self._last_processed_chunk = np.copy(chunk)
        
        # Pad with zeros if we reached the end
        if valid_count < frames:
            outdata[valid_count:] = 0
            
        # Update VU meters
        self._update_vu_levels(chunk)
            
        self.current_frame = float(in_indices[valid_count - 1] + ratio)
        
        # Notify UI (safely, could be high rate, UI should throttle)
        if self.on_position_update:
            pos_sec = self.current_frame / self.audio_data.sample_rate
            self.on_position_update(pos_sec)

        if valid_count < frames and not self.is_looping:
            raise sd.CallbackStop()

    def _update_vu_levels(self, chunk: np.ndarray):
        """Calculate RMS levels for VU meters."""
        if chunk.size == 0:
            return
            
        # Calculate RMS per channel
        rms = np.sqrt(np.mean(chunk**2, axis=0))
        # Convert to dB, avoid log(0)
        db = 20 * np.log10(np.maximum(rms, 1e-5))
        
        if chunk.shape[1] >= 2:
            self.vu_levels = (float(db[0]), float(db[1]))
        else:
            self.vu_levels = (float(db[0]), float(db[0]))

    def _on_finished(self):
        """Called when stream stops."""
        if self._is_restarting:
            return
        self.is_playing = False
        if self.on_playback_stopped:
            self.on_playback_stopped()