more

batch_generate.py

@@ -62,6 +62,44 @@ def sanitize_for_filename(name: str) -> str:
     return name
 
 
+def strip_date_prefix(filename: str) -> str:
+    """
+    Strip the date prefix (YYYY_MM_DD_) from a filename.
+    If the filename doesn't have a date prefix, return it unchanged.
+    
+    Example: '2026_02_19_verbs_en_de_A1.json' -> 'verbs_en_de_A1.json'
+    """
+    # Match date pattern at the start: YYYY_MM_DD_
+    date_pattern = r'^\d{4}_\d{2}_\d{2}_'
+    return re.sub(date_pattern, '', filename)
+
+
+def find_existing_file_ignoring_date(output_dir: str, target_filename: str) -> str | None:
+    """
+    Check if a file with the same name (ignoring date prefix) already exists.
+    
+    Args:
+        output_dir: The directory to search in
+        target_filename: The target filename (may include date prefix)
+    
+    Returns:
+        The path to the existing file if found, None otherwise.
+    """
+    # Strip the date prefix from our target filename
+    target_without_date = strip_date_prefix(target_filename)
+    
+    if not os.path.isdir(output_dir):
+        return None
+    
+    # Search for files matching the pattern
+    for existing_file in os.listdir(output_dir):
+        existing_without_date = strip_date_prefix(existing_file)
+        if existing_without_date == target_without_date:
+            return os.path.join(output_dir, existing_file)
+    
+    return None
+
+
 def generate_output_filename(
     entry: Dict[str, Any],
     code_map: Dict[int, str],
@@ -271,7 +309,7 @@ def main() -> None:
     else:
         already = sum(
             1 for b in active
-            if os.path.isfile(os.path.join(output_dir, b["output_filename"]))
+            if find_existing_file_ignoring_date(output_dir, b["output_filename"]) is not None
         )
         if already:
             print(f"  Resuming   : {already} existing file(s) will be skipped  (use --force to override)")
@@ -331,8 +369,10 @@ def main() -> None:
             print(f"  ETA        : {eta_str}  ({int(avg_time_per_item)}s/item)")
 
         # Skip if already generated (unless --force)
-        if not args.force and os.path.isfile(vocab_file_path):
-            print(f"  ✔ Already exists — skipping  (use --force to regenerate)")
+        existing_file = find_existing_file_ignoring_date(output_dir, output_filename)
+        if not args.force and existing_file is not None:
+            existing_filename = os.path.basename(existing_file)
+            print(f"  ✔ Already exists ({existing_filename}) — skipping  (use --force to regenerate)")
             print("-" * 60)
             skipped_existing += 1
             continue