notebookVb

30 SběrDat/verify_photos.py

@@ -214,6 +214,19 @@ def verify_file(local_path: Path, row) -> list[tuple[str, str, str]]:
     return errors
 
 
+def _progress(msg: str) -> None:
+    """In-place jednořádkový progress na stdout (přepisuje stejný řádek)."""
+    # \r = návrat na začátek řádku, \033[K = smaž zbytek řádku
+    sys.stdout.write(f"\r{msg}\033[K")
+    sys.stdout.flush()
+
+
+def _progress_clear() -> None:
+    """Ukončí progress řádek nulou + newline (aby další log.info šel na čistý řádek)."""
+    sys.stdout.write("\n")
+    sys.stdout.flush()
+
+
 # ── Zpracování ───────────────────────────────────────────────────────────────
 
 
@@ -231,7 +244,8 @@ def reset_state(conn, dry_run: bool) -> None:
     log.info("Reset done. Cleared verified_at on %d rows.", affected)
 
 
-def process_batch(conn, batch_size: int, dry_run: bool) -> int:
+def process_batch(conn, batch_size: int, dry_run: bool,
+                  batch_num: int = 0, total_so_far: int = 0) -> int:
     where = "" if REVERIFY_ALL else "WHERE verified_at IS NULL"
     with conn.cursor(cursor_factory=psycopg2.extras.RealDictCursor) as cur:
         cur.execute(
@@ -249,23 +263,29 @@ def process_batch(conn, batch_size: int, dry_run: bool) -> int:
     if not rows:
         return 0
 
+    batch_total = len(rows)
     processed = 0
-    for row in rows:
+    crit_n = warn_n = info_n = 0
+    t_start = time.time()
+
+    for idx, row in enumerate(rows, 1):
         photo_id = row["id"]
         local_path = db_path_to_local(row["file_path"])
 
         try:
             errors = verify_file(local_path, row)
         except Exception:
+            # Vyčistit progress a logovat exception celý
+            _progress_clear()
             log.exception("Unexpected error verifying id=%d (%s)", photo_id, local_path)
             errors = [("critical", "read_error", "Unexpected exception during verification")]
 
         if dry_run:
             for sev, code, msg in errors:
+                _progress_clear()
                 log.info("[DRY RUN] id=%d %s/%s: %s", photo_id, sev, code, msg)
         else:
             with conn.cursor() as cur:
-                # Smažeme staré záznamy pro tuto fotku — vždy chceme čerstvý snapshot.
                 cur.execute("DELETE FROM photo_errors WHERE photo_id = %s", (photo_id,))
                 if errors:
                     psycopg2.extras.execute_values(
@@ -277,14 +297,24 @@ def process_batch(conn, batch_size: int, dry_run: bool) -> int:
                 conn.commit()
 
         processed += 1
+        for sev, _, _ in errors:
+            if sev == "critical": crit_n += 1
+            elif sev == "warning": warn_n += 1
+            elif sev == "info":    info_n += 1
 
-        if errors:
+        # Single-line progress
-            crit = sum(1 for s, _, _ in errors if s == "critical")
+        elapsed = time.time() - t_start
-            warn = sum(1 for s, _, _ in errors if s == "warning")
+        rate = processed / elapsed if elapsed > 0 else 0
-            info = sum(1 for s, _, _ in errors if s == "info")
+        name = local_path.name[:40]
-            log.info("id=%d: %d errors (crit=%d, warn=%d, info=%d) — %s",
+        _progress(
-                     photo_id, len(errors), crit, warn, info, local_path.name)
+            f"[batch {batch_num}] {idx}/{batch_total}  "
+            f"total={total_so_far + processed}  "
+            f"id={photo_id}  "
+            f"crit={crit_n} warn={warn_n} info={info_n}  "
+            f"{rate:.1f}/s  {name}"
+        )
 
+    _progress_clear()
     return processed
 
 
@@ -322,7 +352,8 @@ def main():
 
             batch_num += 1
             t0 = time.time()
-            count = process_batch(conn, remaining, args.dry_run)
+            count = process_batch(conn, remaining, args.dry_run,
+                                  batch_num=batch_num, total_so_far=total_processed)
             elapsed = time.time() - t0
 
             if count == 0: