dfdf

API/folder.py

@@ -258,6 +258,7 @@ class API_folder_store:
         cursor.execute("CREATE INDEX IF NOT EXISTS idx_files_path ON files(file_path)")
         cursor.execute("CREATE INDEX IF NOT EXISTS idx_tags_hash ON tags(hash)")
         cursor.execute("CREATE INDEX IF NOT EXISTS idx_tags_tag ON tags(tag)")
+        cursor.execute("CREATE INDEX IF NOT EXISTS idx_metadata_ext ON metadata(ext)")
         cursor.execute("CREATE INDEX IF NOT EXISTS idx_worker_id ON worker(worker_id)")
         cursor.execute("CREATE INDEX IF NOT EXISTS idx_worker_status ON worker(status)")
         cursor.execute("CREATE INDEX IF NOT EXISTS idx_worker_type ON worker(worker_type)")
@@ -1858,6 +1859,73 @@ class DatabaseAPI:
         )
         return {row[0] for row in cursor.fetchall()}
 
+    def get_file_hashes_by_ext(self, ext_value: str, limit: Optional[int] = None) -> Set[str]:
+        """Get hashes of files whose metadata ext matches the given extension.
+
+        Matches case-insensitively and ignores any leading '.' in stored ext.
+        Supports glob wildcards '*' and '?' in the query.
+        """
+        ext_clean = str(ext_value or "").strip().lower().lstrip(".")
+        ext_clean = "".join(ch for ch in ext_clean if ch.isalnum())
+        if not ext_clean:
+            return set()
+
+        cursor = self.get_cursor()
+
+        has_glob = ("*" in ext_value) or ("?" in ext_value)
+        if has_glob:
+            pattern = str(ext_value or "").strip().lower().lstrip(".")
+            pattern = pattern.replace("%", "\\%").replace("_", "\\_")
+            pattern = pattern.replace("*", "%").replace("?", "_")
+            cursor.execute(
+                """
+                SELECT DISTINCT f.hash
+                FROM files f
+                JOIN metadata m ON f.hash = m.hash
+                WHERE LOWER(LTRIM(COALESCE(m.ext, ''), '.')) LIKE ? ESCAPE '\\'
+                LIMIT ?
+                """,
+                (pattern, limit or 10000),
+            )
+        else:
+            cursor.execute(
+                """
+                SELECT DISTINCT f.hash
+                FROM files f
+                JOIN metadata m ON f.hash = m.hash
+                WHERE LOWER(LTRIM(COALESCE(m.ext, ''), '.')) = ?
+                LIMIT ?
+                """,
+                (ext_clean, limit or 10000),
+            )
+        return {row[0] for row in cursor.fetchall()}
+
+    def get_files_by_ext(self, ext_value: str, limit: Optional[int] = None) -> List[tuple]:
+        """Get files whose metadata ext matches the given extension.
+
+        Returns (hash, file_path, size, ext) tuples.
+        """
+        ext_clean = str(ext_value or "").strip().lower().lstrip(".")
+        ext_clean = "".join(ch for ch in ext_clean if ch.isalnum())
+        if not ext_clean:
+            return []
+
+        cursor = self.get_cursor()
+        cursor.execute(
+            """
+            SELECT f.hash, f.file_path,
+                   COALESCE((SELECT size FROM metadata WHERE hash = f.hash), 0) as size,
+                   COALESCE((SELECT ext FROM metadata WHERE hash = f.hash), '') as ext
+            FROM files f
+            JOIN metadata m ON f.hash = m.hash
+            WHERE LOWER(LTRIM(COALESCE(m.ext, ''), '.')) = ?
+            ORDER BY f.file_path
+            LIMIT ?
+            """,
+            (ext_clean, limit or 10000),
+        )
+        return cursor.fetchall()
+
     def get_files_with_any_url(self, limit: Optional[int] = None) -> List[tuple]:
         """Get files that have any non-empty URL metadata.