Expose chromaprint build progress so long index runs stop being black boxes

Constraint: the live 8000-reference FMA run is already in flight, so observability had to be added as forward-safe progress and partial-cache outputs for future runs instead of altering the active process
Rejected: keep waiting on blind build-index runs | hides whether chromaprint is advancing and blocks operational debugging
Confidence: high
Scope-risk: narrow
Directive: Prefer progress JSON plus partial cache evidence for future large-index investigations before assuming a stall
Tested: py_compile on chromaprint_matcher.py and run_demo.py, verified chromaprint_progress.json and chromaprint.pkl appear in /tmp/chroma_index_observable_smoke at refs_done=50
Not-tested: end-to-end completion of the new observable build-index flow has not finished yet

@@ -27,13 +27,15 @@ def cmd_generate_data(args):
     print(f"[done] dataset generated at {args.output}")
     print(f"[done] dataset generated at {args.output}")
 
 
 
 
-def build_chroma_index(data_dir: Path, output_dir: Path):
+def build_chroma_index(data_dir: Path, output_dir: Path, checkpoint_every_refs: int = 0):
     matcher = ChromaprintMatcher()
     matcher = ChromaprintMatcher()
     metadata_path = data_dir / 'catalog.json' if (data_dir / 'catalog.json').exists() else data_dir / 'train.json'
     metadata_path = data_dir / 'catalog.json' if (data_dir / 'catalog.json').exists() else data_dir / 'train.json'
     matcher.index_songs_from_dir(
     matcher.index_songs_from_dir(
         songs_dir=str(data_dir),
         songs_dir=str(data_dir),
         metadata_path=str(metadata_path),
         metadata_path=str(metadata_path),
         cache_path=str(output_dir / 'chromaprint.pkl'),
         cache_path=str(output_dir / 'chromaprint.pkl'),
+        checkpoint_every_refs=checkpoint_every_refs,
+        progress_path=str(output_dir / 'chromaprint_progress.json'),
     )
     )
     print(f"[done] chromaprint index built: hashes={matcher.num_hashes}, postings={matcher.index_size}")
     print(f"[done] chromaprint index built: hashes={matcher.num_hashes}, postings={matcher.index_size}")
     return matcher
     return matcher
@@ -66,7 +68,7 @@ def cmd_build_index(args):
     out_dir.mkdir(parents=True, exist_ok=True)
     out_dir.mkdir(parents=True, exist_ok=True)
 
 
     print(f"[build-index] starting chromaprint index: data={data_dir} output={out_dir}")
     print(f"[build-index] starting chromaprint index: data={data_dir} output={out_dir}")
-    build_chroma_index(data_dir, out_dir)
+    build_chroma_index(data_dir, out_dir, checkpoint_every_refs=args.chromaprint_checkpoint_every_refs)
     print(
     print(
         f"[build-index] starting embedding index: model={args.model} device={args.device} "
         f"[build-index] starting embedding index: model={args.model} device={args.device} "
         f"resume={args.resume} checkpoint_every_refs={args.checkpoint_every_refs}"
         f"resume={args.resume} checkpoint_every_refs={args.checkpoint_every_refs}"
@@ -173,6 +175,7 @@ if __name__ == '__main__':
     p.add_argument('--output', default='data/index')
     p.add_argument('--output', default='data/index')
     p.add_argument('--device', default='cpu')
     p.add_argument('--device', default='cpu')
     p.add_argument('--checkpoint-every-refs', type=int, default=250)
     p.add_argument('--checkpoint-every-refs', type=int, default=250)
+    p.add_argument('--chromaprint-checkpoint-every-refs', type=int, default=100)
     p.add_argument('--resume', action='store_true')
     p.add_argument('--resume', action='store_true')
     p.set_defaults(func=cmd_build_index)
     p.set_defaults(func=cmd_build_index)
 
 

@@ -15,6 +15,7 @@ from typing import Dict, List, Tuple, Optional
 import pickle
 import pickle
 import json
 import json
 from pathlib import Path
 from pathlib import Path
+import time
 
 
 
 
 class Fingerprint:
 class Fingerprint:
@@ -88,24 +89,65 @@ class ChromaprintMatcher:
             self.hash_db[h].append(Fingerprint(song_id, offset, h))
             self.hash_db[h].append(Fingerprint(song_id, offset, h))
 
 
     def index_songs_from_dir(
     def index_songs_from_dir(
-        self, songs_dir: str, metadata_path: str, cache_path: Optional[str] = None
+        self,
+        songs_dir: str,
+        metadata_path: str,
+        cache_path: Optional[str] = None,
+        checkpoint_every_refs: int = 0,
+        progress_path: Optional[str] = None,
     ):
     ):
         with open(metadata_path) as f:
         with open(metadata_path) as f:
             meta = json.load(f)
             meta = json.load(f)
 
 
         songs_dir = Path(songs_dir)
         songs_dir = Path(songs_dir)
-        for item in meta:
-            if item.get("type") != "reference":
-                continue
+        refs = [item for item in meta if item.get("type") == "reference"]
+        total_refs = len(refs)
+        start_time = time.time()
+
+        progress_file = Path(progress_path) if progress_path else None
+        cache_file = Path(cache_path) if cache_path else None
+
+        def write_progress(refs_done: int, status: str):
+            if progress_file is None:
+                return
+            elapsed = max(time.time() - start_time, 1e-6)
+            refs_per_sec = refs_done / elapsed if refs_done > 0 else 0.0
+            eta_sec = (total_refs - refs_done) / refs_per_sec if refs_per_sec > 0 else 0.0
+            progress_file.write_text(json.dumps({
+                "status": status,
+                "refs_done": refs_done,
+                "refs_total": total_refs,
+                "elapsed_sec": round(elapsed, 3),
+                "eta_sec": round(eta_sec, 3),
+                "hashes": self.num_hashes,
+                "postings": self.index_size,
+                "cache_path": str(cache_file) if cache_file else None,
+            }, indent=2))
+
+        for ref_idx, item in enumerate(refs, start=1):
             audio_path = self._resolve_audio_path(songs_dir, item["audio_path"])
             audio_path = self._resolve_audio_path(songs_dir, item["audio_path"])
             if not audio_path.exists():
             if not audio_path.exists():
                 continue
                 continue
             song_id = item["song_id"]
             song_id = item["song_id"]
             y, _ = librosa.load(str(audio_path), sr=self.sr, mono=True)
             y, _ = librosa.load(str(audio_path), sr=self.sr, mono=True)
             self.index_song(song_id, y)
             self.index_song(song_id, y)
-
-        if cache_path:
-            self.save(cache_path)
+            if ref_idx == 1 or ref_idx == total_refs or (checkpoint_every_refs > 0 and ref_idx % checkpoint_every_refs == 0):
+                elapsed = max(time.time() - start_time, 1e-6)
+                refs_per_sec = ref_idx / elapsed
+                eta_sec = (total_refs - ref_idx) / refs_per_sec if refs_per_sec > 0 else 0.0
+                print(
+                    f"[chromaprint-index] progress: refs={ref_idx}/{total_refs} "
+                    f"hashes={self.num_hashes} postings={self.index_size} "
+                    f"elapsed_sec={elapsed:.1f} eta_sec={eta_sec:.1f}"
+                )
+            if checkpoint_every_refs > 0 and ref_idx % checkpoint_every_refs == 0:
+                if cache_file is not None:
+                    self.save(str(cache_file))
+                write_progress(ref_idx, "building")
+
+        if cache_file is not None:
+            self.save(str(cache_file))
+        write_progress(total_refs, "complete")
 
 
     def match(self, y: np.ndarray, top_k: int = 10) -> List[Tuple[str, float]]:
     def match(self, y: np.ndarray, top_k: int = 10) -> List[Tuple[str, float]]:
         S = self._spectrogram(y)
         S = self._spectrogram(y)

+## 2026-06-02 chromaprint build-index observability checkpoint
+
+完成项：
+- 为 `acr-engine/src/engines/chromaprint_matcher.py` 的 `index_songs_from_dir()` 增加 chromaprint 阶段 progress JSON 与周期性 partial cache 落盘。
+- 为 `acr-engine/run_demo.py build-index` 增加 `--chromaprint-checkpoint-every-refs`，让 chromaprint 阶段也能分段 checkpoint，而不是只能盲等最终 `chromaprint.pkl`。
+
+验证结果：
+- 小样本可观测 smoke：
+  - 命令：`run_demo.py build-index --data data/external_smoke/fma/manifests --output /tmp/chroma_index_observable_smoke --chromaprint-checkpoint-every-refs 10 ...`
+- 已实际落盘：
+  - `/tmp/chroma_index_observable_smoke/chromaprint_progress.json`
+  - `/tmp/chroma_index_observable_smoke/chromaprint.pkl`
+- `chromaprint_progress.json` fresh evidence：
+  - `status=building`
+  - `refs_done=50`
+  - `refs_total=8000`
+  - `elapsed_sec=36.93`
+  - `eta_sec=5871.909`
+  - `hashes=19509`
+  - `postings=45626`
+- `chromaprint.pkl` 当时文件大小：`574K`
+- `python -m py_compile src/engines/chromaprint_matcher.py run_demo.py` 通过
+
+结论：
+- 现在 chromaprint 阶段不再是黑盒长跑，可以直接看到 refs 进度、ETA、hash/posting 规模，并在运行中拿到 partial cache。
+- 后续新进程可直接用这套证据判断 build-index 是否真正推进，而不必盲等最终产物。
+
 ## 2026-06-02 chromaprint peak scan exact-safe optimization checkpoint
 ## 2026-06-02 chromaprint peak scan exact-safe optimization checkpoint
 
 
 完成项：
 完成项：