tuning job log views and deep scan archive processing.

2026-02-02 18:48:01 -05:00
parent 042a69f9c3
commit 09883960d4
9 changed files with 1078 additions and 103 deletions
@@ -846,8 +846,69 @@ def is_first_volume(path: str) -> bool:

    return True

+def extract_archive_name(filename: str) -> str:
+    """
+    Extract a clean archive name from a filename by removing common prefixes.
+
+    Handles patterns like:
+    - "43387617_attachment_83109081_October 2020 Rewards.rar" -> "October 2020 Rewards"
+    - "01_Archive Name.zip" -> "Archive Name"
+    - "Archive Name.rar" -> "Archive Name"
+
+    Returns the cleaned name without extension.
+    """
+    import re
+    from pathlib import Path
+
+    # Remove extension
+    stem = Path(filename).stem
+
+    # Pattern 1: Gallery-DL style "{id}_attachment_{id}_{name}" or "{id}_{name}"
+    # Match one or more numeric_prefix patterns at the start
+    cleaned = re.sub(r'^(\d+_)+(attachment_)?(\d+_)?', '', stem)
+
+    # Pattern 2: Simple numeric prefix "01_name" or "001_name"
+    if cleaned == stem:  # No match from pattern 1
+        cleaned = re.sub(r'^\d+_', '', stem)
+
+    # If we stripped everything or got empty, fall back to original stem
+    if not cleaned.strip():
+        cleaned = stem
+
+    return cleaned.strip()
+
+
+def compute_archive_tag_name(artist: str, archive_filename: str) -> str:
+    """
+    Generate an archive tag name in the format: '{artist}:{archive_name}'
+
+    Examples:
+    - artist="InCaseArt", filename="43387617_attachment_83109081_October 2020 Rewards.rar"
+      -> "InCaseArt:October 2020 Rewards"
+
+    If a tag with the same name already exists, appends a numeric suffix.
+    """
+    archive_name = extract_archive_name(archive_filename)
+    base_tag = f"{artist}:{archive_name}"
+
+    # Check if tag already exists
+    existing = Tag.query.filter_by(name=base_tag, kind='archive').first()
+    if not existing:
+        return base_tag
+
+    # Tag exists, add numeric suffix
+    n = 2
+    while True:
+        candidate = f"{base_tag} ({n})"
+        if not Tag.query.filter_by(name=candidate, kind='archive').first():
+            return candidate
+        n += 1
+
+
 def compute_next_archive_tag_name(artist: str, width: int = ARCHIVE_NUM_WIDTH) -> str:
    """
+    DEPRECATED: Use compute_archive_tag_name() instead.
+
    Return a unique incremental tag name like: 'archive:{artist}/0001', '0002', ...
    - Looks only at Tag(kind='archive') with names starting 'archive:{artist}/'
    - Finds max numeric suffix and returns next number (zero-padded).