From 9751bf28b981d6c437f44517147ef2dcdeebac78 Mon Sep 17 00:00:00 2001
From: Christian Groschupp <christian@groschupp.org>
Date: Thu, 27 Mar 2025 10:18:52 +0100
Subject: [PATCH] feat: calculate saved space

---
 cleanmedia.py | 46 ++++++++++++++++++++++++++++++++++++----------
 1 file changed, 36 insertions(+), 10 deletions(-)

diff --git a/cleanmedia.py b/cleanmedia.py
index 2a00720..58825d6 100755
--- a/cleanmedia.py
+++ b/cleanmedia.py
@@ -52,12 +52,14 @@ class File:
         media_id: MediaID,
         creation_ts: Timestamp,
         base64hash: Base64Hash,
+        file_size: int,
     ) -> None:
         """Initialize a File object."""
         self.repo = media_repo
         self.media_id = media_id
         self.create_date = datetime.fromtimestamp(creation_ts)
         self.base64hash = base64hash
+        self.file_size = file_size
 
     @cached_property
     def fullpath(self) -> Path | None:
@@ -163,7 +165,7 @@ class MediaRepository:
         for db_type, conn_string in connection_strings.items():
             self.conn[db_type] = self.connect_db(conn_string)
 
-    def execute(self, query: str, params=(), db_type: str = "media_api"):
+    def _execute(self, query: str, params=(), db_type: str = "media_api"):
         paramstyle = getattr(self.conn, "paramstyle", "format")
         query = self._adjust_paramstyle(query, paramstyle)
         cur = self.conn[db_type].cursor()
@@ -203,7 +205,7 @@ class MediaRepository:
 
     def get_single_media(self, mxid: MediaID) -> File | None:
         """Retrieve a single media file by ID."""
-        cur = self.execute(
+        cur = self._execute(
                 "SELECT media_id, creation_ts, base64hash from mediaapi_media_repository WHERE media_id = %s;",
                 (mxid,),
             )
@@ -219,7 +221,7 @@ class MediaRepository:
         Returns:
             List of File objects
         """
-        cur = self.execute(
+        cur = self._execute(
             "SELECT media_id, creation_ts, base64hash FROM mediaapi_media_repository WHERE user_id = %s;",
             (user_id,),
         )
@@ -234,12 +236,12 @@ class MediaRepository:
         Returns:
             List of File objects
         """
-        query = """SELECT media_id, creation_ts, base64hash
+        query = """SELECT media_id, creation_ts, base64hash, file_size_bytes
                     FROM mediaapi_media_repository"""
         if not local:
             query += " WHERE user_id = ''"
-        cur = self.execute(query)
-        return [File(self, row[0], row[1] // 1000, row[2]) for row in cur.fetchall()]
+        cur = self._execute(query)
+        return [File(self, row[0], row[1] // 1000, row[2], row[3]) for row in cur.fetchall()]
 
     def get_avatar_images(self) -> List[MediaID]:
         """Get media IDs of current avatar images.
@@ -247,7 +249,7 @@ class MediaRepository:
         Returns:
             List of media IDs
         """
-        cur = self.execute("SELECT avatar_url FROM userapi_profiles WHERE avatar_url > '';", db_type="user_api")
+        cur = self._execute("SELECT avatar_url FROM userapi_profiles WHERE avatar_url > '';", db_type="user_api")
         media_ids = []
         for (url,) in cur.fetchall():
             try:
@@ -259,7 +261,7 @@ class MediaRepository:
 
     def sanity_check_thumbnails(self) -> None:
         """Check for orphaned thumbnail entries in database."""
-        cur = self.execute(
+        cur = self._execute(
             """SELECT COUNT(media_id) FROM mediaapi_thumbnail
                 WHERE NOT EXISTS (SELECT media_id FROM mediaapi_media_repository);""",
         )
@@ -292,8 +294,9 @@ class MediaRepository:
             for f in self.get_all_media(local)
             if f.media_id not in self._avatar_media_ids and f.create_date < cutoff_date
         ]
-
+        file_size_counter = 0
         for file in files_to_delete:
+            file_size_counter += file.file_size
             if dryrun:
                 logging.info(f"Would delete file {file.media_id} at {file.fullpath}")
                 if not file.exists():
@@ -302,7 +305,7 @@ class MediaRepository:
                 file.delete()
 
         action = "Would have deleted" if dryrun else "Deleted"
-        logging.info("%s %d files", action, len(files_to_delete))
+        logging.info("%s %d files, in total %s", action, len(files_to_delete), sizeof_fmt(file_size_counter))
         return len(files_to_delete)
 
 
@@ -348,6 +351,29 @@ def read_config(conf_file: Union[str, Path]) -> Tuple[Path, str, dict[str, str]]
 
     return Path(base_path), conns
 
+def sizeof_fmt(num: Union[int, float], suffix: str = "B") -> str:
+    """
+    Convert a number of bytes (or other units) into a human-readable format using binary prefixes.
+
+    Args:
+        num (int | float): The number to format, typically a size in bytes.
+        suffix (str): The suffix to append (default is 'B' for bytes).
+
+    Returns:
+        str: A string representing the human-readable size (e.g. '1.5MiB', '42.0KiB').
+
+    Notes:
+        Uses binary (base-1024) units: Ki, Mi, Gi, Ti, etc.
+        Automatically chooses the appropriate unit based on the input value.
+    """
+    BASE = 1024.0
+
+    for unit in ("", "Ki", "Mi", "Gi", "Ti", "Pi", "Ei", "Zi"):
+        if abs(num) < BASE:
+            return f"{num:3.1f}{unit}{suffix}"
+        num /= 1024.0
+    return f"{num:.1f}Yi{suffix}"
+
 
 def parse_options() -> argparse.Namespace:
     """Parse command line arguments.