feat: trace-level matching, health watcher/worker status, timezone config

scripts/face_processor.py

@@ -13,6 +13,7 @@ Detection cost: near-zero CPU (Vision ANE)
 Embedding cost: near-zero CPU (CoreML ANE)
 """
 
+import re
 import sys
 import os
 import json
@@ -29,6 +30,7 @@ from pathlib import Path
 import coremltools as ct
 
 sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
+from redis_publisher import RedisPublisher
 
 SCRIPT_DIR = os.path.dirname(os.path.abspath(__file__))
 SWIFT_BIN = os.path.join(SCRIPT_DIR, "swift_processors", ".build", "debug", "swift_face")
@@ -49,11 +51,12 @@ def classify_pose(roll: float, yaw: float) -> str:
 
 class FaceProcessorVision:
     def __init__(self, video_path: str, output_path: str, uuid: str = "",
-                 sample_interval: int = 3):
+                 sample_interval: int = 3, publisher: RedisPublisher = None):
         self.video_path = video_path
         self.output_path = output_path
         self.uuid = uuid
         self.sample_interval = sample_interval
+        self.publisher = publisher
 
         # Load CoreML FaceNet
         self.coreml_model = None
@@ -127,7 +130,33 @@ class FaceProcessorVision:
 
         print(f"[FACE_V2] Running: {' '.join(cmd)}")
         t0 = time.time()
-        subprocess.run(cmd, check=True)
+        log_path = swift_out + ".log"
+        log_f = open(log_path, "w")
+        proc = subprocess.Popen(cmd, stdout=log_f, stderr=subprocess.STDOUT, text=True)
+        last_pct = -1
+        while proc.poll() is None:
+            time.sleep(10)
+            # Read latest log lines
+            try:
+                with open(log_path) as lf:
+                    for line in lf:
+                        line = line.strip()
+                        m = re.search(r'(\d+)% complete', line)
+                        if m:
+                            pct = int(m.group(1))
+                            if pct > last_pct:
+                                last_pct = pct
+                                if self.publisher:
+                                    self.publisher.progress("face", pct, 100, f"swift detect {pct}%")
+            except Exception:
+                pass
+        log_f.close()
+        if proc.returncode != 0:
+            stderr_out = proc.stderr.read()
+            if stderr_out:
+                print(stderr_out.strip(), file=sys.stderr)
+            raise RuntimeError(f"swift_face exited with code {proc.returncode}")
+
         elapsed = time.time() - t0
         print(f"[FACE_V2] Detection done in {elapsed:.1f}s")
 
@@ -156,6 +185,8 @@ class FaceProcessorVision:
 
         t0 = time.time()
         embed_count = 0
+        total_face_count = 0
+        last_pct = -1
 
         for frame_info in frames:
             frame_num = frame_info["frame"]
@@ -220,6 +251,12 @@ class FaceProcessorVision:
             if len(face_data["frames"]) % 100 == 0:
                 elapsed = time.time() - t0
                 print(f"[FACE_V2] {len(face_data['frames'])} frames, {embed_count} embeddings, {elapsed:.0f}s")
+                if self.publisher:
+                    pct = int(len(face_data["frames"]) * 100 / max(len(frames), 1))
+                    if pct > last_pct:
+                        last_pct = pct
+                        self.publisher.progress("face", len(face_data["frames"]), len(frames),
+                            f"{embed_count} faces", embed_count, "faces")
 
         self.video.release()
 
@@ -259,19 +296,36 @@ def main():
     parser.add_argument("--force", action="store_true")
     args = parser.parse_args()
 
+    publisher = RedisPublisher(args.uuid) if args.uuid else None
+    if publisher:
+        publisher.info("face", "FACE_START")
+
     if args.force and os.path.exists(args.output_path):
         os.remove(args.output_path)
 
     processor = FaceProcessorVision(
         args.video_path, args.output_path,
-        args.uuid, args.sample_interval
+        args.uuid, args.sample_interval, publisher
     )
 
     # Step 1: Vision detection (bbox + pose via ANE)
-    detection = processor.process_with_swift()
+    try:
+        detection = processor.process_with_swift()
+    except Exception as e:
+        if publisher:
+            publisher.error("face", f"Detection failed: {e}")
+        raise
 
     # Step 2: CoreML embedding + save
-    processor.embed_and_save(detection)
+    try:
+        processor.embed_and_save(detection)
+    except Exception as e:
+        if publisher:
+            publisher.error("face", f"Embedding failed: {e}")
+        raise
+
+    if publisher:
+        publisher.complete("face", f"{len(detection.get('frames',[]))} frames")
 
     # Clean up temp detection file
     swift_out = args.output_path.replace(".json", "_detect.json")