Gap 12: soundfont render qua native bridge — render_soundfont_offline (FluidSynth C API) thay pyfluidsynth khi thiếu lib; render_engine soundfont branch fallback bridge (test_64, 176 passed)

2026-08-23 08:52:57 +07:00
parent 6b2c9a039b
commit 2b5f396133
3 changed files with 186 additions and 58 deletions
@@ -126,6 +126,65 @@ def _preset_b64(preset_id, preset_path, preset_data_b64) -> str:
    return ""


+def _run_notes_job(job: dict, label: str) -> tuple:
+    """Chạy job MIDI render qua native_bridge --render → (out_path, duration_sec).
+
+    Dùng chung cho VSTi (render_offline) + SoundFont (render_soundfont_offline).
+    Lỗi → HTTPException (501 thiếu bridge, 500 lỗi render — file rác bị xóa).
+    """
+    exe = find_bridge_exe()
+    if not exe:
+        raise HTTPException(
+            status_code=501,
+            detail="native_bridge không khả dụng — đặt SF_BRIDGE_PATH hoặc cài daw_vst_bridge",
+        )
+    os.makedirs(settings.PROCESSED_DIR, exist_ok=True)
+    out_path = os.path.join(settings.PROCESSED_DIR, f"preview_{uuid.uuid4().hex[:10]}.wav")
+    job_path = out_path + ".job.json"
+    try:
+        with open(job_path, "w", encoding="utf-8") as fh:
+            json.dump(job, fh)
+        r = subprocess.run([exe, "--render", job_path, "--out", out_path],
+                           capture_output=True, text=True, timeout=120,
+                           creationflags=subprocess.CREATE_NO_WINDOW if os.name == "nt" else 0)
+        log = ((r.stdout or "") + (r.stderr or "")).strip()
+        if r.returncode != 0 or not os.path.isfile(out_path):
+            if os.path.exists(out_path):
+                try:
+                    os.remove(out_path)
+                except Exception:
+                    pass
+            if r.returncode == 2:
+                raise HTTPException(status_code=500,
+                                    detail=f"Bridge không load được plugin {label}. {log[-500:]}")
+            raise HTTPException(status_code=500,
+                                detail=f"Render thất bại (rc={r.returncode}). {log[-500:]}")
+        normalize_peak_file(out_path)
+        return out_path, sf.info(out_path).frames / float(job.get("sample_rate", 44100))
+    except HTTPException:
+        raise
+    except subprocess.TimeoutExpired:
+        if os.path.exists(out_path):
+            try:
+                os.remove(out_path)
+            except Exception:
+                pass
+        raise HTTPException(status_code=500, detail="Render quá lâu (timeout 120s)")
+    except Exception as e:
+        if os.path.exists(out_path):
+            try:
+                os.remove(out_path)
+            except Exception:
+                pass
+        raise HTTPException(status_code=500, detail=f"Render thất bại: {e}")
+    finally:
+        if os.path.exists(job_path):
+            try:
+                os.remove(job_path)
+            except Exception:
+                pass
+
+
 def render_offline(instrument_id, notes, bpm, sample_rate, preset_id=None,
                   preset_path=None, preset_data_b64=None,
                   soundfont_bank=None, soundfont_program=None) -> tuple:
@@ -141,12 +200,6 @@ def render_offline(instrument_id, notes, bpm, sample_rate, preset_id=None,
            detail=f"Không tìm thấy VSTi: {instrument_id} — chưa scan thấy plugin này. "
                   "Kiểm tra Plugins Manager → Scan.",
        )
-    exe = find_bridge_exe()
-    if not exe:
-        raise HTTPException(
-            status_code=501,
-            detail="native_bridge không khả dụng — đặt SF_BRIDGE_PATH hoặc cài daw_vst_bridge",
-        )
    job = {
        "instrument_type": itype,
        "plugin_path": path,
@@ -169,51 +222,31 @@ def render_offline(instrument_id, notes, bpm, sample_rate, preset_id=None,
        job["soundfont_bank"] = int(soundfont_bank) & 0x7F
    if soundfont_program is not None:
        job["soundfont_program"] = int(soundfont_program) & 0x7F
-    os.makedirs(settings.PROCESSED_DIR, exist_ok=True)
-    out_path = os.path.join(settings.PROCESSED_DIR, f"preview_{uuid.uuid4().hex[:10]}.wav")
-    job_path = out_path + ".job.json"
-    try:
-        with open(job_path, "w", encoding="utf-8") as fh:
-            json.dump(job, fh)
-        r = subprocess.run([exe, "--render", job_path, "--out", out_path],
-                           capture_output=True, text=True, timeout=120,
-                           creationflags=subprocess.CREATE_NO_WINDOW if os.name == "nt" else 0)
-        log = ((r.stdout or "") + (r.stderr or "")).strip()
-        if r.returncode != 0 or not os.path.isfile(out_path):
-            if os.path.exists(out_path):
-                try:
-                    os.remove(out_path)
-                except Exception:
-                    pass
-            if r.returncode == 2:
-                raise HTTPException(status_code=500,
-                                    detail=f"Bridge không load được plugin {instrument_id}. {log[-500:]}")
-            raise HTTPException(status_code=500,
-                                detail=f"Render thất bại (rc={r.returncode}). {log[-500:]}")
-        normalize_peak_file(out_path)
-        return out_path, sf.info(out_path).frames / float(sample_rate)
-    except HTTPException:
-        raise
-    except subprocess.TimeoutExpired:
-        if os.path.exists(out_path):
-            try:
-                os.remove(out_path)
-            except Exception:
-                pass
-        raise HTTPException(status_code=500, detail="Render quá lâu (timeout 120s)")
-    except Exception as e:
-        if os.path.exists(out_path):
-            try:
-                os.remove(out_path)
-            except Exception:
-                pass
-        raise HTTPException(status_code=500, detail=f"Render thất bại: {e}")
-    finally:
-        if os.path.exists(job_path):
-            try:
-                os.remove(job_path)
-            except Exception:
-                pass
+    return _run_notes_job(job, instrument_id)
+
+
+def render_soundfont_offline(sf_path, notes, bpm, sample_rate, bank=0, program=0) -> tuple:
+    """Render MIDI notes bằng SoundFont (.sf2/.sf3) qua native_bridge --render.
+
+    Thay pyfluidsynth khi thiếu FluidSynth lib: bridge nhúng FluidSynth C API
+    (NativeInstrumentEngine.cpp, InstrumentType::SOUNDFONT_SF2_SF3) — cùng job
+    shape với render_offline nhưng instrument_type=2 + plugin_path là file SF2.
+    Lỗi → HTTPException (501 thiếu bridge, 500 lỗi render)."""
+    job = {
+        "instrument_type": 2,
+        "plugin_path": os.fspath(sf_path),
+        "sample_rate": int(sample_rate),
+        "bpm": float(bpm),
+        "soundfont_bank": int(bank) & 0x7F,
+        "soundfont_program": int(program) & 0x7F,
+        "notes": [{
+            "pitch": int(n.get("pitch", 60)),
+            "velocity": float(n.get("velocity", 0.8)),  # bridge job: float 0..1 (RenderJob.cpp *127)
+            "start_beat": float(n.get("start_beat", 0)),
+            "duration_beats": float(n.get("duration_beats", 1)),
+        } for n in notes],
+    }
+    return _run_notes_job(job, os.fspath(sf_path))


 def render_fx_chain(input_wav, fx_chain, sample_rate, out_path=None):
@@ -603,11 +603,11 @@ class PythonRenderEngine:
                                    )
                            elif soundfont_id or (instrument_id and instrument_id.startswith("sf_")):
                                sf_path = _find_sf2_path(soundfont_id or instrument_id)
-                                # 3-level fallback: selected SF → default SF → oscillator synth
-                                if not sf_path or not os.path.exists(sf_path) or not HAS_PYFLUIDSYNTH:
-                                    if not sf_path or not os.path.exists(sf_path):
-                                        logger.warning(f"[RenderEngine] SoundFont not found for {soundfont_id or instrument_id}, trying default")
-                                    sf_path = _find_default_sf2() if HAS_PYFLUIDSYNTH else ""
+                                # 4-level fallback: selected SF → default SF → native bridge → oscillator synth.
+                                # Gap 12: pyfluidsynth thiếu FluidSynth lib → bridge render SF2 thật thay vì oscillator.
+                                if not sf_path or not os.path.exists(sf_path):
+                                    logger.warning(f"[RenderEngine] SoundFont not found for {soundfont_id or instrument_id}, trying default")
+                                    sf_path = _find_default_sf2()

                                if sf_path and os.path.exists(sf_path) and HAS_PYFLUIDSYNTH:
                                    import fluidsynth as _fs
@@ -651,9 +651,39 @@ class PythonRenderEngine:
                                        _cursor = end_s
                                    synth_buffer = midi_data
                                    _fs.delete_fluid_synth(_fl)
+                                elif sf_path and os.path.exists(sf_path) and settings.RENDER_ENGINE == "bridge":
+                                    # Gap 12: native bridge nhúng FluidSynth C API — render SF2/SF3 thật
+                                    # khi pyfluidsynth thiếu (RENDER_ENGINE=bridge mặc định).
+                                    try:
+                                        from app.core import native_render
+                                        out_path, _dur = native_render.render_soundfont_offline(
+                                            sf_path=sf_path,
+                                            notes=[{
+                                                "pitch": ev["note"],
+                                                "velocity": float(ev["velocity"]) / 127.0,
+                                                "start_beat": ev["start_beat"],
+                                                "duration_beats": ev["duration_beats"],
+                                            } for ev in midi_events],
+                                            bpm=bpm,
+                                            sample_rate=self.sample_rate,
+                                            bank=soundfont_bank,
+                                            program=soundfont_program,
+                                        )
+                                        synth_buffer, _ = sf.read(out_path, dtype="float32")
+                                        if synth_buffer.ndim == 1:
+                                            synth_buffer = np.vstack([synth_buffer, synth_buffer])
+                                        else:
+                                            synth_buffer = synth_buffer.T
+                                        try:
+                                            os.remove(out_path)
+                                        except Exception:
+                                            pass
+                                    except Exception as e:
+                                        logger.warning("[RenderEngine] native soundfont render failed: %s", e)
+                                        synth_buffer = render_midi_events_to_audio(
+                                            midi_events=midi_events, sr=self.sample_rate, bpm=bpm, instrument='synth'
+                                        )
                                else:
-                                    if not HAS_PYFLUIDSYNTH:
-                                        logger.warning("[RenderEngine] pyfluidsynth not available, falling back to oscillator synth")
                                    synth_buffer = render_midi_events_to_audio(
                                        midi_events=midi_events, sr=self.sample_rate, bpm=bpm, instrument='synth'
                                    )
@@ -32,6 +32,8 @@ FX_BRIDGE = os.path.join(REPO, "native_bridge", "build", "Release", "fx_vst_brid
 DAW_BRIDGE = os.path.join(REPO, "native_bridge", "build", "Release", "daw_vst_bridge.exe")
 ADELAY = os.path.join(REPO, "native_bridge", "build", "VST3", "Release", "adelay.vst3")
 SR = 44100
+REAL_SF2 = os.path.join(REPO, "app", "storage", "soundfonts", "518e850f-a5d3-4790-b1f9-0c90c203c524.sf2")
+


@pytest.fixture(scope="module")
@@ -741,3 +743,66 @@ def test_63_fxrt_ws_latency_push(monkeypatch):
        assert d.get("cmd") == "latency", d
        assert d.get("total") == 1024, d
        assert d.get("latencies", {}).get("0") == 1024, d
+
+# ── 5.10 Gap 12: SoundFont thật qua native bridge (thay pyfluidsynth) ─────
+def test_64_soundfont_render_via_bridge(monkeypatch, tmp_path):
+    """Gap 12: pyfluidsynth thiếu FluidSynth lib → native bridge render SF2 thật
+    (bridge nhúng FluidSynth C API, NativeInstrumentEngine.cpp).
+
+    (1) render_soundfont_offline trực tiếp: non-silent, đúng duration, program
+    khác → audio khác (program_select áp).
+    (2) render_project với HAS_PYFLUIDSYNTH=False: track MIDI soundfont ra âm
+    thật — không fallback oscillator synth."""
+    if not os.path.isfile(DAW_BRIDGE):
+        pytest.skip("daw_vst_bridge.exe chưa build (native_bridge/build/Release)")
+    if not os.path.isfile(REAL_SF2):
+        pytest.skip("không có SF2 mẫu trong app/storage/soundfonts")
+    monkeypatch.setenv("SF_BRIDGE_PATH", DAW_BRIDGE)
+
+    from app.core import native_render
+    notes = [{"pitch": 60, "velocity": 0.8, "start_beat": 0.0, "duration_beats": 2.0}]
+    out0, dur0 = native_render.render_soundfont_offline(REAL_SF2, notes, 120.0, SR, bank=0, program=0)
+    try:
+        d0, sr0 = _read_wav(out0)
+        assert sr0 == SR, f"sr {sr0}"
+        assert d0.shape[1] >= 44000, f"duration ngắn: {d0.shape[1]} (2 beats @120 = 44100)"
+        assert float(np.max(np.abs(d0))) > 0.01, "SF2 render im lặng"
+    finally:
+        os.remove(out0)
+    out40, _ = native_render.render_soundfont_offline(REAL_SF2, notes, 120.0, SR, bank=0, program=40)
+    try:
+        d40, _ = _read_wav(out40)
+        n = min(d0.shape[1], d40.shape[1])
+        assert float(np.max(np.abs(d40[:, :n] - d0[:, :n]))) > 1e-3, "program_select không áp — program 0 == 40"
+    finally:
+        os.remove(out40)
+
+    # (2) engine-level: bắt buộc branch bridge (pyfluidsynth vắng mặt)
+    import app.core.vst_engine as ve
+    import app.core.render_engine as re_mod
+    monkeypatch.setattr(ve, "HAS_PYFLUIDSYNTH", False)
+    monkeypatch.setattr(re_mod, "HAS_PYFLUIDSYNTH", False)
+    monkeypatch.setattr(re_mod, "_find_sf2_path", lambda sf_id: REAL_SF2)
+    from app.api.v1 import plugins as pl
+    monkeypatch.setattr(pl.settings, "RENDER_ENGINE", "bridge")
+    from app.core.render_engine import PythonRenderEngine
+    engine = PythonRenderEngine()
+
+    track = _midi_track()
+    track["instrument_id"] = ""
+    track["instrument_source"] = "soundfont"
+    track["soundfont_id"] = "518e850f-a5d3-4790-b1f9-0c90c203c524"
+    track["volume_db"] = -12.0
+    project = {
+        "metadata": {"bpm": 120.0, "time_signature_numerator": 4},
+        "main_session": {
+            "length_bars": 1.0,
+            "tracks": [track],
+            "master": {"fx_chain": [], "bypass": False, "volume_db": 0.0},
+        },
+        "section_store": {},
+    }
+    out = str(tmp_path / "sf2_engine.wav")
+    engine.render_project(project, out)
+    data, _ = _read_wav(out)
+    assert float(np.max(np.abs(data))) > 0.01, "engine render soundfont im lặng — fallback oscillator?"