From ea993bcf96a0414e7d24ff6c0448fea8ddca395d Mon Sep 17 00:00:00 2001
From: Honney OC <honney@gmx.net>
Date: Sun, 13 Sep 2026 23:58:04 +0000
Subject: [PATCH] fix: waveform cache

---
 Makefile          |  6 +++---
 waveform_cache.py | 43 +++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 46 insertions(+), 3 deletions(-)
 create mode 100644 waveform_cache.py

diff --git a/Makefile b/Makefile
index 12c142e..0cb33a9 100644
--- a/Makefile
+++ b/Makefile
@@ -81,10 +81,10 @@ extract:
 	@if [ -x $(PY) ]; then $(PY) extract.py; else python3 extract.py; fi
 
 sync:
-	@if [ -x $(PY) ]; then $(PY) apply_sync.py --config $(CONFIG) --lag $(LAG); else python3 apply_sync.py --config $(CONFIG) --lag $(LAG); fi
+	@if [ -x $(PY) ]; then PYTHONPATH=. $(PY) apply_sync.py --config $(CONFIG) --lag $(LAG); else PYTHONPATH=. python3 apply_sync.py --config $(CONFIG) --lag $(LAG); fi
 
 verify:
-	@if [ -x $(PY) ]; then LC_NUMERIC=C QT_STYLE_OVERRIDE=Fusion $(PY) verify_sync.py --config $(CONFIG) --lag $(LAG); else LC_NUMERIC=C QT_STYLE_OVERRIDE=Fusion python3 verify_sync.py --config $(CONFIG) --lag $(LAG); fi
+	@if [ -x $(PY) ]; then LC_NUMERIC=C PYTHONPATH=. QT_STYLE_OVERRIDE=Fusion $(PY) verify_sync.py --config $(CONFIG) --lag $(LAG); else LC_NUMERIC=C PYTHONPATH=. QT_STYLE_OVERRIDE=Fusion python3 verify_sync.py --config $(CONFIG) --lag $(LAG); fi
 
 ast:
 	@AUDIO_A=$$(python3 -c "import tomllib; c=tomllib.load(open('config.toml','rb')); print(c['extract'].get('file_a',''))" 2>/dev/null); \
@@ -100,7 +100,7 @@ ast:
 	else \
 		echo "=> extracted wavs exist ($$OUT/$${STEM_A}_audio.wav, $$OUT/$${STEM_B}_audio.wav), skipping extract"; \
 	fi
-	@if [ -x $(PY) ]; then LC_NUMERIC=C QT_STYLE_OVERRIDE=Fusion $(PY) sync_audio_test.py --config $(CONFIG); else LC_NUMERIC=C QT_STYLE_OVERRIDE=Fusion python3 sync_audio_test.py --config $(CONFIG); fi
+	@if [ -x $(PY) ]; then LC_NUMERIC=C PYTHONPATH=. QT_STYLE_OVERRIDE=Fusion $(PY) sync_audio_test.py --config $(CONFIG); else LC_NUMERIC=C PYTHONPATH=. QT_STYLE_OVERRIDE=Fusion python3 sync_audio_test.py --config $(CONFIG); fi
 
 clean:
 	rm -rf $(VENV) __pycache__ .pytest_cache
diff --git a/waveform_cache.py b/waveform_cache.py
new file mode 100644
index 0000000..bcd56c7
--- /dev/null
+++ b/waveform_cache.py
@@ -0,0 +1,43 @@
+"""File-based cache for loaded/resampled WAV data.
+
+Stores .npy files in `cache/waveforms/` keyed by source path + target_sr + mono.
+Invalidated when source file mtime changes.
+"""
+from __future__ import annotations
+import hashlib
+import os
+import numpy as np
+
+CACHE_DIR = os.path.join(os.path.dirname(os.path.abspath(__file__)), "cache", "waveforms")
+
+
+def _cache_key(path: str, target_sr: int, mono: bool) -> str:
+    real = os.path.realpath(path)
+    mtime = os.path.getmtime(real)
+    raw = f"{real}|{target_sr}|{mono}|{mtime:.6f}"
+    return hashlib.sha256(raw.encode()).hexdigest()[:16]
+
+
+def cached_load_wav(path: str, target_sr: int = 48000, mono: bool = True,
+                    _loader=None) -> tuple[np.ndarray, int]:
+    """Load WAV with disk cache. Falls back to _loader(data, sr) if cache miss."""
+    key = _cache_key(path, target_sr, mono)
+    npy_path = os.path.join(CACHE_DIR, f"{key}.npy")
+    sr_path = os.path.join(CACHE_DIR, f"{key}.sr")
+
+    if os.path.exists(npy_path) and os.path.exists(sr_path):
+        try:
+            data = np.load(npy_path, allow_pickle=False)
+            sr = int(np.load(sr_path, allow_pickle=False).item())
+            return data, sr
+        except Exception:
+            pass  # corrupted cache, rebuild
+
+    if _loader is None:
+        from sync_audio import load_wav as _loader
+
+    data, sr = _loader(path, target_sr=target_sr, mono=mono)
+    os.makedirs(CACHE_DIR, exist_ok=True)
+    np.save(npy_path, data, allow_pickle=False)
+    np.save(sr_path, np.array(sr), allow_pickle=False)
+    return data, sr