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
@@ -83,6 +83,72 @@ def find_sidecar_json(image_path: str) -> Optional[str]:
    return None


+def find_archive_sidecar(archive_path: str) -> Optional[str]:
+    """
+    Find the Gallery-DL sidecar JSON file for an archive.
+
+    Gallery-DL names sidecars based on the archive's content filename,
+    which may differ from the archive filename due to prefixes added
+    during download (e.g., "02_HolyMeh July 2023.rar" -> "HolyMeh July 2023.json").
+
+    Args:
+        archive_path: Path to the archive file (e.g., "02_Filename.rar")
+
+    Returns:
+        Path to the sidecar JSON if found, None otherwise.
+    """
+    archive_path = Path(archive_path)
+    parent_dir = archive_path.parent
+    stem = archive_path.stem  # filename without extension
+
+    # Pattern 1: Exact stem match (most common case)
+    # e.g., "HolyMeh July 2023.rar" -> "HolyMeh July 2023.json"
+    stem_json = parent_dir / f"{stem}.json"
+    if stem_json.exists():
+        return str(stem_json)
+
+    # Pattern 2: Strip numeric prefix from archive name
+    # e.g., "02_HolyMeh July 2023.rar" -> "HolyMeh July 2023.json"
+    stem_no_prefix = re.sub(r'^\d+_', '', stem)
+    if stem_no_prefix != stem:
+        alt_json = parent_dir / f"{stem_no_prefix}.json"
+        if alt_json.exists():
+            return str(alt_json)
+
+    # Pattern 3: Look for any JSON in the directory that references this archive
+    # by checking the 'file_name' or 'filename' field in the JSON
+    for json_file in parent_dir.glob("*.json"):
+        try:
+            with open(json_file, "r", encoding="utf-8") as f:
+                data = json.load(f)
+
+            # Check if this JSON references our archive
+            # Gallery-DL uses 'file.file_name' or 'filename' + 'extension'
+            file_info = data.get("file", {})
+            json_filename = file_info.get("file_name", "")
+            if not json_filename:
+                # Try filename + extension pattern
+                fn = data.get("filename", "")
+                ext = data.get("extension", "")
+                if fn and ext:
+                    json_filename = f"{fn}.{ext}"
+
+            # Compare with archive filename (with or without prefix)
+            archive_filename = archive_path.name
+            archive_filename_no_prefix = re.sub(r'^\d+_', '', archive_filename)
+
+            if json_filename and (
+                json_filename == archive_filename or
+                json_filename == archive_filename_no_prefix
+            ):
+                return str(json_file)
+
+        except (json.JSONDecodeError, IOError):
+            continue
+
+    return None
+
+
 def load_sidecar_metadata(json_path: str) -> Optional[dict]:
    """Load and parse a Gallery-DL sidecar JSON file."""
    try: