From 1edafbfac89f76b50b6b6d1dee2129553be7c5ff Mon Sep 17 00:00:00 2001
From: Sam Rolfe <samuelrolfe@gmail.com>
Date: Mon, 10 Aug 2026 06:01:51 +1000
Subject: [PATCH] Add 10s per-file timeout to scan_once (prevents hung file
 blocking large scans)

---
 photo_ingest.py | 19 ++++++++++++++++++-
 1 file changed, 18 insertions(+), 1 deletion(-)

diff --git a/photo_ingest.py b/photo_ingest.py
index 3ba8e3c..2dd8708 100644
--- a/photo_ingest.py
+++ b/photo_ingest.py
@@ -12,6 +12,8 @@ Performance: dhash-only rows (~6ms/file); phash only when check_near_dups.
 """
 
 from pathlib import Path
+import signal
+import time as _time
 
 from prefect import flow, task
 from PIL import Image
@@ -49,15 +51,23 @@ def scan_once(base_dir: str, source: str, batch_size: int, check_near_dups: bool
     pending = []
     t0 = time.time()
 
+    def _timeout_handler(signum, frame):
+        raise TimeoutError(f"file took >{FILE_TIMEOUT}s")
+
+    FILE_TIMEOUT = 10  # seconds per file — guards against hung/corrupt files
+    signal.signal(signal.SIGALRM, _timeout_handler)
+
     for p_str in walk_files(base_dir):
         p = Path(p_str)
         try:
+            signal.alarm(FILE_TIMEOUT)
             known = conn.execute(
                 "SELECT 1 FROM image_hashes WHERE path=? UNION SELECT 1 FROM known_paths WHERE path=?",
                 (p_str, p_str),
             ).fetchone()
             if known:
                 totals["known"] += 1
+                signal.alarm(0)
                 continue
 
             sha = db.sha256_file(p_str)
@@ -68,6 +78,7 @@ def scan_once(base_dir: str, source: str, batch_size: int, check_near_dups: bool
                 conn.execute(
                     "INSERT OR IGNORE INTO known_paths (path, sha256) VALUES (?,?)", (p_str, sha))
                 totals["exact_dups"] += 1
+                signal.alarm(0)
                 continue
 
             hx = db.hash_image(p)
@@ -76,18 +87,24 @@ def scan_once(base_dir: str, source: str, batch_size: int, check_near_dups: bool
             pending.append((sha, hx["phash"], hx["dhash"], p.stat().st_size,
                             w, h, str(p), source))
             totals["new"] += 1
+            signal.alarm(0)
 
             if len(pending) >= batch_size:
                 conn.executemany(
                     "INSERT INTO image_hashes (sha256, phash, dhash, file_size, width, height, path, source) "
                     "VALUES (?,?,?,?,?,?,?,?)", pending)
                 conn.commit()
-                print(f"  checkpoint: {totals['new']} new ({time.time()-t0:.0f}s)", flush=True)
+                print(f"  checkpoint: {totals['new']} new ({_time.time()-t0:.0f}s)", flush=True)
                 pending = []
 
+        except TimeoutError as e:
+            totals["skipped"] += 1
+            print(f"  TIMEOUT {p.name}: {e}", flush=True)
         except Exception as e:
             totals["skipped"] += 1
             print(f"  SKIP {p.name}: {type(e).__name__}: {e}", flush=True)
+        finally:
+            signal.alarm(0)
 
     if pending:
         conn.executemany(