Fase 3: import del subsistema de audio desde AEEA

Reemplaza el audio antiguo de orni_attack (singleton con new/delete
raw, sin efectos, sin crossfade) por el subsistema moderno de AEEA
(unique_ptr, RAII, crossfade nativo, echo/reverb, pitch-shift,
callbacks de fin de pista, getMusicDurationMs para timelines
deterministas).

Eliminados:
- source/core/audio/audio_cache.{hpp,cpp} (1 cache por subsistema)
- source/core/audio/jail_audio.hpp viejo (motor inline globals)
- source/external/stb_vorbis.h (v1.20)

Añadidos (copiados de AEEA, traducidos comentarios al castellano):
- source/core/audio/audio.{hpp,cpp} — singleton con Audio::Config inyectada
- source/core/audio/audio_adapter.{hpp,cpp} — adapter para getMusic/getSound
- source/core/audio/audio_effects.{hpp,cpp} — Schroeder reverb + echo DSP
- source/core/audio/jail_audio.{hpp,cpp} — Ja::Engine class-based, streaming
- source/core/audio/sound_effects_config.{hpp,cpp} — presets YAML (opcional)
- source/external/stb_vorbis.c (v1.22) — OGG decoder, versión más reciente
- source/external/stb_vorbis_impl.cpp — TU aislada para evitar clang-tidy

Adaptaciones:
- audio_adapter.cpp implementado a medida para orni: usa
  Resource::Helper::loadFile (no Resource::Cache de AEEA que orni no
  tiene). Cache local con unique_ptr<Ja::Music> / unique_ptr<Ja::Sound>.
- Includes: utils/defaults.hpp -> core/defaults.hpp, utils/log.hpp
  reemplazado por iostream con std::cerr/std::cout.

API breaking changes (callsites migrados):
- Audio::init() -> Audio::init(Config); el Director construye la Config
  desde Defaults::Audio::* (ENABLED, VOLUME, MUSIC_*, SOUND_*).
- Audio::get()->getMusicState() -> Audio::getMusicState() (ahora static).
- AudioCache::getMusic/getSound -> AudioResource::getMusic/getSound.

Defaults::Audio consolidado: ahora aglutina las constantes que antes
estaban repartidas entre namespace Audio (VOLUME, ENABLED), namespace
Music (VOLUME, ENABLED), namespace Sound (VOLUME, ENABLED). Las pistas
y rutas de efectos siguen en Music::* / Sound::*. Añadidas FREQUENCY,
FORMAT, CHANNELS, CROSSFADE_MS, VOLUME_STEP para el motor.

Beneficios para fases siguientes:
- Crossfade en transiciones de escena (uso: playMusic(name, -1, 1500)).
- Pitch-shift para variaciones de SFX (Audio::playSound(name, group, 0.95)).
- Echo/reverb DSP via playSoundWithEcho/Reverb (sounds.yaml presets).
- Callbacks setOnMusicEnded para sincronizar eventos con el fin de pista.

Compila y enlaza. Pendiente: test runtime del usuario.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-05-19 12:43:01 +02:00
parent a4f6a5514f
commit ed98ef612e
19 changed files with 1893 additions and 916 deletions
+645
View File
@@ -0,0 +1,645 @@
#include "core/audio/jail_audio.hpp"
#include <algorithm>
#include <cassert>
#include <cstdint>
#include <cstdio>
#include <memory>
#include <optional>
#include <vector>
#include "core/audio/audio_effects.hpp"
// Solo declaracions de stb_vorbis: STB_VORBIS_HEADER_ONLY omet el bloc
// d'implementació. Les definicions las aporta source/external/stb_vorbis_impl.cpp
// (TU aïllat porque clang-analyzer no dispari fals positius al nostre codi).
#define STB_VORBIS_HEADER_ONLY
// clang-format off
// NOLINTNEXTLINE(bugprone-suspicious-include) -- stb_vorbis es single-file: la macro de dalt limita este TU a solo-declaracions; la implementació viu a external/stb_vorbis_impl.cpp.
#include "external/stb_vorbis.c"
// clang-format on
namespace Ja {
// --- Streaming internals (file-scope constants) ---
namespace {
// Bytes-per-sample per canal (siempre s16)
constexpr int MUSIC_BYTES_PER_SAMPLE = 2;
// Quants shorts decodifiquem per crida a get_samples_short_interleaved.
// 8192 shorts = 4096 samples/channel en estèreo ≈ 85ms de so a 48kHz.
constexpr int MUSIC_CHUNK_SHORTS = 8192;
// Umbral d'audio per davant del cursor de reproducció. Mantenim ≥ 0.5 s a
// l'SDL_AudioStream per absorbir jitter de frame i evitar underruns.
constexpr float MUSIC_LOW_WATER_SECONDS = 0.5F;
} // namespace
// --- Engine::active_ storage ---
Engine* Engine::active_ = nullptr;
auto Engine::active() noexcept -> Engine* { return active_; }
// --- Ctor/Dtor ---
Engine::Engine(const int freq, const SDL_AudioFormat format, const int num_channels) {
assert(active_ == nullptr && "Ja::Engine: més d'una instància activa no está suportat");
active_ = this;
audio_spec_ = {.format = format, .channels = num_channels, .freq = freq};
sdl_audio_device_ = SDL_OpenAudioDevice(SDL_AUDIO_DEVICE_DEFAULT_PLAYBACK, &audio_spec_);
if (sdl_audio_device_ == 0) { std::fprintf(stderr, "Ja::Engine: Failed to initialize SDL audio!\n"); }
for (auto& channel : channels_) { channel.state = ChannelState::FREE; }
}
Engine::~Engine() {
if (outgoing_music_.stream != nullptr) {
SDL_DestroyAudioStream(outgoing_music_.stream);
outgoing_music_.stream = nullptr;
}
if (sdl_audio_device_ != 0) { SDL_CloseAudioDevice(sdl_audio_device_); }
sdl_audio_device_ = 0;
if (active_ == this) { active_ = nullptr; }
}
// --- Helpers stateless (no toquen membres d'Engine) ---
namespace {
auto feedMusicChunk(Music* music) -> int {
if (music == nullptr || music->vorbis == nullptr || music->stream == nullptr) { return 0; }
short chunk[MUSIC_CHUNK_SHORTS];
const int NUM_CHANNELS = music->spec.channels;
const int SAMPLES_PER_CHANNEL = stb_vorbis_get_samples_short_interleaved(
music->vorbis,
NUM_CHANNELS,
static_cast<short*>(chunk),
MUSIC_CHUNK_SHORTS);
if (SAMPLES_PER_CHANNEL <= 0) { return 0; }
const int BYTES = SAMPLES_PER_CHANNEL * NUM_CHANNELS * MUSIC_BYTES_PER_SAMPLE;
SDL_PutAudioStreamData(music->stream, static_cast<const void*>(chunk), BYTES);
return SAMPLES_PER_CHANNEL;
}
void pumpMusic(Music* music) {
if (music == nullptr || music->vorbis == nullptr || music->stream == nullptr) { return; }
const int BYTES_PER_SECOND = music->spec.freq * music->spec.channels * MUSIC_BYTES_PER_SAMPLE;
const int LOW_WATER_BYTES = static_cast<int>(MUSIC_LOW_WATER_SECONDS * static_cast<float>(BYTES_PER_SECOND));
while (SDL_GetAudioStreamAvailable(music->stream) < LOW_WATER_BYTES) {
const int DECODED = feedMusicChunk(music);
if (DECODED > 0) { continue; }
// EOF: si queden loops, rebobinar; si no, tallar y deixar drenar.
if (music->times != 0) {
stb_vorbis_seek_start(music->vorbis);
if (music->times > 0) { music->times--; }
} else {
break;
}
}
}
void preFillOutgoing(Music* music, const int duration_ms) {
if (music == nullptr || music->vorbis == nullptr || music->stream == nullptr) { return; }
const int BYTES_PER_SECOND = music->spec.freq * music->spec.channels * MUSIC_BYTES_PER_SAMPLE;
const int NEEDED_BYTES = static_cast<int>((static_cast<std::int64_t>(duration_ms) * BYTES_PER_SECOND) / 1000);
while (SDL_GetAudioStreamAvailable(music->stream) < NEEDED_BYTES) {
const int DECODED = feedMusicChunk(music);
if (DECODED <= 0) { break; }
}
}
// Retorna el progrés lineal [0..1] d'un fade. 1.0 vol dir completat. Única
// font de la corba del fade: si es vol canviar a logarítmica/quadràtica,
// s'edita aquí i afecta fade-in i fade-out alhora.
auto fadeProgress(const FadeState& fade) -> float {
if (fade.duration_ms <= 0) { return 1.0F; }
const Uint64 ELAPSED = SDL_GetTicks() - fade.start_time;
if (ELAPSED >= static_cast<Uint64>(fade.duration_ms)) { return 1.0F; }
return static_cast<float>(ELAPSED) / static_cast<float>(fade.duration_ms);
}
} // namespace
void Engine::updateOutgoingFade() {
if (outgoing_music_.stream == nullptr || !outgoing_music_.fade.active) { return; }
// Mentre la fosa está activa, mantenim el stream con una reserva
// de samples per davant del cursor (mismo patró que pumpMusic
// para el current_music_). Así el stream no es buida ni cuando SDL
// drena més ràpid del previst en haver sounds bound a la misma
// device. Si l'OGG arriba a EOF, rebobina (la fosa pot ser més
// llarga que la pista).
if (outgoing_music_.music != nullptr && outgoing_music_.music->vorbis != nullptr) {
const Music& music = *outgoing_music_.music;
const int BYTES_PER_SECOND = music.spec.freq * music.spec.channels * MUSIC_BYTES_PER_SAMPLE;
const int LOW_WATER = static_cast<int>(MUSIC_LOW_WATER_SECONDS * static_cast<float>(BYTES_PER_SECOND));
while (SDL_GetAudioStreamAvailable(outgoing_music_.stream) < LOW_WATER) {
short chunk[MUSIC_CHUNK_SHORTS];
const int SAMPLES = stb_vorbis_get_samples_short_interleaved(
music.vorbis,
music.spec.channels,
static_cast<short*>(chunk),
MUSIC_CHUNK_SHORTS);
if (SAMPLES <= 0) {
stb_vorbis_seek_start(music.vorbis);
continue;
}
const int BYTES = SAMPLES * music.spec.channels * MUSIC_BYTES_PER_SAMPLE;
SDL_PutAudioStreamData(outgoing_music_.stream, static_cast<const void*>(chunk), BYTES);
}
}
const float PROGRESS = fadeProgress(outgoing_music_.fade);
if (PROGRESS >= 1.0F) {
SDL_DestroyAudioStream(outgoing_music_.stream);
outgoing_music_.stream = nullptr;
outgoing_music_.fade.active = false;
// Deixem el Vorbis del Music original en un estat conegut per
// a la pròxima reproducció. (playMusic también fa seek_start,
// pero fer-ho ací evita estats intermedis si algú consulta.)
if (outgoing_music_.music != nullptr && outgoing_music_.music->vorbis != nullptr) {
stb_vorbis_seek_start(outgoing_music_.music->vorbis);
}
outgoing_music_.music = nullptr;
} else {
SDL_SetAudioStreamGain(outgoing_music_.stream, outgoing_music_.fade.initial_volume * (1.0F - PROGRESS));
}
}
void Engine::updateIncomingFade() {
if (!incoming_fade_.active) { return; }
const float PROGRESS = fadeProgress(incoming_fade_);
if (PROGRESS >= 1.0F) {
incoming_fade_.active = false;
SDL_SetAudioStreamGain(current_music_->stream, music_volume_);
} else {
SDL_SetAudioStreamGain(current_music_->stream, music_volume_ * PROGRESS);
}
}
void Engine::updateCurrentMusic() {
if (current_music_ == nullptr || current_music_->state != MusicState::PLAYING) { return; }
updateIncomingFade();
pumpMusic(current_music_);
if (current_music_->times == 0 && SDL_GetAudioStreamAvailable(current_music_->stream) == 0) {
// La pista ha acabat de drenar naturalment. L'aturem primer (deixa
// l'engine en estat consistent) i entonces invoquem el callback;
// así un eventual playMusic des del callback comença net.
stopMusic();
if (on_music_ended_) { on_music_ended_(); }
}
}
void Engine::updateSoundChannels() {
for (int i = 0; i < MAX_SIMULTANEOUS_CHANNELS; ++i) {
if (channels_[i].state != ChannelState::PLAYING) { continue; }
if (channels_[i].times != 0) {
if (static_cast<Uint32>(SDL_GetAudioStreamAvailable(channels_[i].stream)) < (channels_[i].sound->length / 2)) {
SDL_PutAudioStreamData(channels_[i].stream, channels_[i].sound->buffer.get(), channels_[i].sound->length);
if (channels_[i].times > 0) { channels_[i].times--; }
}
} else if (SDL_GetAudioStreamAvailable(channels_[i].stream) == 0) {
stopChannel(i);
}
}
}
void Engine::stealCurrentIntoOutgoing(const int duration_ms) {
if (outgoing_music_.stream != nullptr) {
SDL_DestroyAudioStream(outgoing_music_.stream);
outgoing_music_.stream = nullptr;
outgoing_music_.fade.active = false;
}
if (current_music_ == nullptr || current_music_->state != MusicState::PLAYING || current_music_->stream == nullptr) {
return;
}
preFillOutgoing(current_music_, duration_ms);
outgoing_music_.stream = current_music_->stream;
// Guardem la referència al Music porque updateOutgoingFade puga
// seguir bombant Vorbis sin al stream durante tota la fosa. NO fem
// seek_start ací: la decompressió ha de continuar des d'on estava
// porque el so siga continu. El seek_start es farà cuando la fosa
// acabe (o cuando playMusic la interrompi via stopMusic).
outgoing_music_.music = current_music_;
outgoing_music_.fade = {
.active = true,
.start_time = SDL_GetTicks(),
.duration_ms = duration_ms,
.initial_volume = music_volume_,
};
current_music_->stream = nullptr;
current_music_->state = MusicState::STOPPED;
}
template <typename Fn>
void Engine::forEachTargetChannel(const int channel, Fn&& fn) {
if (channel == -1) {
for (auto& ch : channels_) { fn(ch); }
} else if (channel >= 0 && channel < MAX_SIMULTANEOUS_CHANNELS) {
fn(channels_[channel]);
}
}
// --- Engine public API ---
void Engine::update() {
updateOutgoingFade();
updateCurrentMusic();
updateSoundChannels();
}
void Engine::playMusic(Music* music, const int loop) {
if (music == nullptr || music->vorbis == nullptr) { return; }
stopMusic();
current_music_ = music;
current_music_->state = MusicState::PLAYING;
current_music_->times = loop;
stb_vorbis_seek_start(current_music_->vorbis);
current_music_->stream = SDL_CreateAudioStream(&current_music_->spec, &audio_spec_);
if (current_music_->stream == nullptr) {
std::fprintf(stderr, "Ja::Engine::playMusic: Failed to create audio stream!\n");
current_music_->state = MusicState::STOPPED;
return;
}
SDL_SetAudioStreamGain(current_music_->stream, music_volume_);
pumpMusic(current_music_);
if (!SDL_BindAudioStream(sdl_audio_device_, current_music_->stream)) {
std::fprintf(stderr, "Ja::Engine::playMusic: SDL_BindAudioStream failed!\n");
}
}
void Engine::setMusicSpeed(float ratio) {
if (current_music_ == nullptr || current_music_->stream == nullptr) { return; }
SDL_SetAudioStreamFrequencyRatio(current_music_->stream, ratio);
}
void Engine::pauseMusic() {
if (current_music_ == nullptr || current_music_->state != MusicState::PLAYING) { return; }
current_music_->state = MusicState::PAUSED;
SDL_UnbindAudioStream(current_music_->stream);
}
void Engine::resumeMusic() {
if (current_music_ == nullptr || current_music_->state != MusicState::PAUSED) { return; }
current_music_->state = MusicState::PLAYING;
SDL_BindAudioStream(sdl_audio_device_, current_music_->stream);
}
void Engine::stopMusic() {
if (outgoing_music_.stream != nullptr) {
SDL_DestroyAudioStream(outgoing_music_.stream);
outgoing_music_.stream = nullptr;
outgoing_music_.fade.active = false;
if (outgoing_music_.music != nullptr && outgoing_music_.music->vorbis != nullptr) {
stb_vorbis_seek_start(outgoing_music_.music->vorbis);
}
outgoing_music_.music = nullptr;
}
incoming_fade_.active = false;
if (current_music_ == nullptr || current_music_->state == MusicState::INVALID || current_music_->state == MusicState::STOPPED) { return; }
current_music_->state = MusicState::STOPPED;
if (current_music_->stream != nullptr) {
SDL_DestroyAudioStream(current_music_->stream);
current_music_->stream = nullptr;
}
if (current_music_->vorbis != nullptr) {
stb_vorbis_seek_start(current_music_->vorbis);
}
}
void Engine::fadeOutMusic(const int milliseconds) {
if (current_music_ == nullptr || current_music_->state != MusicState::PLAYING) { return; }
stealCurrentIntoOutgoing(milliseconds);
incoming_fade_.active = false;
}
void Engine::crossfadeMusic(Music* music, const int crossfade_ms, const int loop) {
if (music == nullptr || music->vorbis == nullptr) { return; }
stealCurrentIntoOutgoing(crossfade_ms);
current_music_ = music;
current_music_->state = MusicState::PLAYING;
current_music_->times = loop;
stb_vorbis_seek_start(current_music_->vorbis);
current_music_->stream = SDL_CreateAudioStream(&current_music_->spec, &audio_spec_);
if (current_music_->stream == nullptr) {
std::fprintf(stderr, "Ja::Engine::crossfadeMusic: Failed to create audio stream!\n");
current_music_->state = MusicState::STOPPED;
return;
}
SDL_SetAudioStreamGain(current_music_->stream, 0.0F);
pumpMusic(current_music_);
SDL_BindAudioStream(sdl_audio_device_, current_music_->stream);
incoming_fade_ = {
.active = true,
.start_time = SDL_GetTicks(),
.duration_ms = crossfade_ms,
.initial_volume = 0.0F,
};
}
auto Engine::getMusicState() const -> MusicState {
if (current_music_ == nullptr) { return MusicState::INVALID; }
return current_music_->state;
}
auto Engine::setMusicVolume(float volume) -> float {
music_volume_ = SDL_clamp(volume, 0.0F, 1.0F);
if (current_music_ != nullptr && current_music_->stream != nullptr) {
SDL_SetAudioStreamGain(current_music_->stream, music_volume_);
}
return music_volume_;
}
void Engine::setOnMusicEnded(std::function<void()> callback) {
on_music_ended_ = std::move(callback);
}
void Engine::onMusicDeleted(const Music* music) {
if (music == nullptr) { return; }
if (current_music_ == music) {
stopMusic();
current_music_ = nullptr;
}
}
// --- Sound ---
auto Engine::playSound(Sound* sound, const int loop, const int group) -> int {
if (sound == nullptr) { return -1; }
int channel = 0;
while (channel < MAX_SIMULTANEOUS_CHANNELS && channels_[channel].state != ChannelState::FREE) { channel++; }
if (channel == MAX_SIMULTANEOUS_CHANNELS) {
// No hay canal libre, reemplazamos el primero
channel = 0;
}
return playSoundOnChannel(sound, channel, loop, group);
}
auto Engine::playSoundOnChannel(Sound* sound, const int channel, const int loop, const int group) -> int {
if (sound == nullptr) { return -1; }
if (channel < 0 || channel >= MAX_SIMULTANEOUS_CHANNELS) { return -1; }
stopChannel(channel);
channels_[channel].sound = sound;
channels_[channel].times = loop;
channels_[channel].pos = 0;
channels_[channel].group = group;
channels_[channel].state = ChannelState::PLAYING;
channels_[channel].stream = SDL_CreateAudioStream(&channels_[channel].sound->spec, &audio_spec_);
if (channels_[channel].stream == nullptr) {
std::fprintf(stderr, "Ja::Engine::playSoundOnChannel: Failed to create audio stream!\n");
channels_[channel].state = ChannelState::FREE;
return -1;
}
SDL_PutAudioStreamData(channels_[channel].stream, channels_[channel].sound->buffer.get(), channels_[channel].sound->length);
SDL_SetAudioStreamGain(channels_[channel].stream, sound_volume_[group]);
SDL_BindAudioStream(sdl_audio_device_, channels_[channel].stream);
return channel;
}
void Engine::setChannelSpeed(const int channel, const float ratio) {
if (channel < 0 || channel >= MAX_SIMULTANEOUS_CHANNELS) { return; }
if (channels_[channel].stream == nullptr) { return; }
SDL_SetAudioStreamFrequencyRatio(channels_[channel].stream, ratio);
}
void Engine::pauseChannel(const int channel) {
forEachTargetChannel(channel, [](Channel& ch) {
if (ch.state == ChannelState::PLAYING) {
ch.state = ChannelState::PAUSED;
SDL_UnbindAudioStream(ch.stream);
}
});
}
void Engine::resumeChannel(const int channel) {
const SDL_AudioDeviceID DEVICE = sdl_audio_device_;
forEachTargetChannel(channel, [DEVICE](Channel& ch) {
if (ch.state == ChannelState::PAUSED) {
ch.state = ChannelState::PLAYING;
SDL_BindAudioStream(DEVICE, ch.stream);
}
});
}
void Engine::stopChannel(const int channel) {
forEachTargetChannel(channel, [this](Channel& ch) {
if (ch.state != ChannelState::FREE) {
if (ch.stream != nullptr) { SDL_DestroyAudioStream(ch.stream); }
ch.stream = nullptr;
ch.state = ChannelState::FREE;
ch.pos = 0;
ch.sound = nullptr;
if (ch.has_effect) {
ch.has_effect = false;
if (effect_channels_active_ > 0) { --effect_channels_active_; }
}
}
});
}
auto Engine::setSoundVolume(float volume, const int group) -> float {
const float V = SDL_clamp(volume, 0.0F, 1.0F);
if (group == -1) {
std::ranges::fill(sound_volume_, V);
} else if (group >= 0 && group < MAX_GROUPS) {
sound_volume_[group] = V;
} else {
return V;
}
for (auto& ch : channels_) {
if ((ch.state == ChannelState::PLAYING) || (ch.state == ChannelState::PAUSED)) {
if (group == -1 || ch.group == group) {
if (ch.stream != nullptr) {
SDL_SetAudioStreamGain(ch.stream, sound_volume_[ch.group]);
}
}
}
}
return V;
}
void Engine::onSoundDeleted(const Sound* sound) {
if (sound == nullptr) { return; }
for (int i = 0; i < MAX_SIMULTANEOUS_CHANNELS; ++i) {
if (channels_[i].sound == sound) { stopChannel(i); }
}
}
auto Engine::playProcessedOnFreeChannel(const std::vector<std::uint8_t>& bytes, const SDL_AudioSpec& spec, const int group) -> int {
// El sin de canals con efecte es valida antes de reservar slot —
// así evitem crear y destruir un stream solo per descartar el play.
if (effect_channels_active_ >= MAX_EFFECT_CHANNELS) { return -1; }
int channel = 0;
while (channel < MAX_SIMULTANEOUS_CHANNELS && channels_[channel].state != ChannelState::FREE) { ++channel; }
if (channel == MAX_SIMULTANEOUS_CHANNELS) { channel = 0; }
stopChannel(channel);
// El stream es crea contra l'spec del buffer processat (S16, ...)
// porque SDL faci el resampling sin a audio_spec_ del device.
channels_[channel].stream = SDL_CreateAudioStream(&spec, &audio_spec_);
if (channels_[channel].stream == nullptr) {
std::fprintf(stderr, "Ja::Engine::playProcessedOnFreeChannel: Failed to create audio stream!\n");
return -1;
}
channels_[channel].sound = nullptr; // El buffer no es propietat de sin Ja::Sound.
channels_[channel].times = 0;
channels_[channel].pos = 0;
const int CLAMPED_GROUP = (group >= 0 && group < MAX_GROUPS) ? group : 0;
channels_[channel].group = CLAMPED_GROUP;
channels_[channel].state = ChannelState::PLAYING;
channels_[channel].has_effect = true;
++effect_channels_active_;
SDL_PutAudioStreamData(channels_[channel].stream, bytes.data(), static_cast<int>(bytes.size()));
SDL_SetAudioStreamGain(channels_[channel].stream, sound_volume_[CLAMPED_GROUP]);
SDL_BindAudioStream(sdl_audio_device_, channels_[channel].stream);
return channel;
}
auto Engine::playSoundWithEcho(const Sound* sound, const EchoParams& params, const int group) -> int {
if (sound == nullptr) { return -1; }
auto processed = AudioEffects::applyEcho(*sound, params);
if (!processed) { return -1; }
return playProcessedOnFreeChannel(processed->bytes, processed->spec, group);
}
auto Engine::playSoundWithReverb(const Sound* sound, const ReverbParams& params, const int group) -> int {
if (sound == nullptr) { return -1; }
auto processed = AudioEffects::applyReverb(*sound, params);
if (!processed) { return -1; }
return playProcessedOnFreeChannel(processed->bytes, processed->spec, group);
}
// --- Factories y destructors (permanents) ---
auto loadMusic(const Uint8* buffer, Uint32 length) -> Music* {
if (buffer == nullptr || length == 0) { return nullptr; }
// Allocem el Music primer per aprofitar el seu `std::vector<Uint8>`
// como a propietari del OGG comprimit. stb_vorbis guarda un punter
// persistent al buffer; como que ací no el resize'jem, el .data() es
// estable durante tot el cicle de vida del music.
auto music = std::make_unique<Music>();
music->ogg_data.assign(buffer, buffer + length);
int vorbis_error = 0;
music->vorbis = stb_vorbis_open_memory(music->ogg_data.data(),
static_cast<int>(length),
&vorbis_error,
nullptr);
if (music->vorbis == nullptr) {
std::fprintf(stderr, "Ja::loadMusic: stb_vorbis_open_memory failed (error %d)\n", vorbis_error);
return nullptr;
}
const stb_vorbis_info INFO = stb_vorbis_get_info(music->vorbis);
music->spec.channels = static_cast<int>(INFO.channels);
music->spec.freq = static_cast<int>(INFO.sample_rate);
music->spec.format = SDL_AUDIO_S16;
// Pre-cálculo de la duración en ms a partir del header. stb_vorbis ya
// ha decodificat la informació necessària a `stb_vorbis_open_memory`;
// esta consulta no descodifica àudio, solo llig el comptador
// de samples. Si el sample_rate fos 0 (header malmès) deixem
// duration_ms a 0.
if (INFO.sample_rate > 0) {
const auto SAMPLES = stb_vorbis_stream_length_in_samples(music->vorbis);
music->duration_ms = static_cast<int>((static_cast<std::uint64_t>(SAMPLES) * 1000ULL) / INFO.sample_rate);
}
music->state = MusicState::STOPPED;
return music.release();
}
// Overload con filename. Resource::Cache l'usa per registrar el path dins
// del propi Ja::Music (camp `filename`); la capa Audio l'usa per recuperar
// el nom después d'un playMusic(Ja::Music*, ...) — veure PATCH-02.
auto loadMusic(const Uint8* buffer, Uint32 length, const char* filename) -> Music* {
Music* music = loadMusic(buffer, length);
if (music != nullptr && filename != nullptr) { music->filename = filename; }
return music;
}
void deleteMusic(Music* music) {
if (music == nullptr) { return; }
// Notifiquem el motor actiu porque pari la pista si es la current_music.
// Si no hay motor (shutdown-order invertit), passem: los recursos
// propis del Music es lliberen igualment a sota.
if (Engine* eng = Engine::active()) { eng->onMusicDeleted(music); }
if (music->stream != nullptr) { SDL_DestroyAudioStream(music->stream); }
if (music->vorbis != nullptr) { stb_vorbis_close(music->vorbis); }
delete music;
}
auto loadSound(std::uint8_t* buffer, std::uint32_t size) -> Sound* {
auto sound = std::make_unique<Sound>();
Uint8* raw = nullptr;
if (!SDL_LoadWAV_IO(SDL_IOFromMem(buffer, size), true, &sound->spec, &raw, &sound->length)) {
std::fprintf(stderr, "Ja::loadSound: Failed to load WAV from memory: %s\n", SDL_GetError());
return nullptr;
}
sound->buffer.reset(raw); // adopta el SDL_malloc'd buffer
return sound.release();
}
void deleteSound(Sound* sound) {
if (sound == nullptr) { return; }
if (Engine* eng = Engine::active()) { eng->onSoundDeleted(sound); }
// buffer es destrueix automàticament via RAII (SdlFreeDeleter).
delete sound;
}
} // namespace Ja
// --- stb_vorbis macro leak cleanup ---
// stb_vorbis.c filtra noms curts (L, C, R i PLAYBACK_*) al TU que el compila.
// Xocarien con parámetros de plantilla d'altres headers si estas definicions
// s'escapessin. Els netegem al final del TU per tancar la porta.
// clang-format off
#undef L
#undef C
#undef R
#undef PLAYBACK_MONO
#undef PLAYBACK_LEFT
#undef PLAYBACK_RIGHT
// clang-format on