add deletion of single media

Put the main workhorse (cleaning out old media) in a function of the
MediaRepository, so that we can introduce alternative functions, e.g.
cleaning specific single media.

cleanmedia

@@ -112,7 +112,17 @@ class MediaRepository:
             raise ValueError(errstr)
         return psycopg2.connect(self.db_conn_string)
 
-    def get_media(self, local: bool = False) -> List[File]:
+    def get_single_media(self, mxid: str) -> Optional[File]:
+        """Return `File` or `None`"""
+        with self.conn.cursor() as cur:
+            sql_str = "SELECT media_id, creation_ts, base64hash from mediaapi_media_repository WHERE media_id = %s;"
+            cur.execute(sql_str,(mxid,))
+            row = cur.fetchone()
+            if row is None: return None
+            # creation_ts is ms since the epoch, so convert to seconds
+            return File(self, row[0], row[1] // 1000, row[2])
+
+    def get_all_media(self, local: bool = False) -> List[File]:
         """Return List[File] of remote media or ALL media if local==True"""
         with self.conn.cursor() as cur:
             # media_id | media_origin | content_type | file_size_bytes |  creation_ts  | upload_name | base64hash | user_id
@@ -154,6 +164,41 @@ class MediaRepository:
             if row is not None and row[0]:
                 logging.error("You have {} thumbnails in your db that do not refer to media. This needs fixing (we don't do that)!".format(row[0]))
 
+    def clean_media_files(self, days: int, local: bool = False, dryrun: bool = False) -> int:
+        """Clean out old media files from this repository
+
+        :params:
+           :days: (int) delete media files older than N days.
+           :local: (bool) Also delete media originating from local users
+           :dryrun: (bool) Do not actually delete any files (just count)
+        :returns: (int) The number of files that were/would be deleted
+        """
+        # Preps
+        if local:
+            # populate the cache of current avt img. so we don't delete them
+            logging.warning("AVATAR")
+            mr.get_avatar_images()
+
+        cleantime = datetime.today() - timedelta(days=days)
+        logging.info("Deleting remote media older than %s", cleantime)
+        num_deleted = 0
+        files = mr.get_all_media(local)
+        for file in [f for f in files if f.media_id not in mr._avatar_media_ids]:
+            if file.create_date < cleantime:
+                num_deleted += 1
+                if dryrun:  # the great pretender
+                    logging.info(f"Pretending to delete file id {file.media_id} on path {file.fullpath}.")
+                    if not file.exists():
+                        logging.info(f"File id {file.media_id} does not physically exist (path {file.fullpath}).")
+                else:
+                    file.delete()
+        info_str = "Deleted %d files during the run."
+        if dryrun:
+            info_str = "%d files would have been deleted during the run."
+        logging.info(info_str, num_deleted)
+
+        return num_deleted
+
 
 # --------------------------------------------------------------
 def read_config(conf_file: Union[str, Path]) -> Tuple[Path, str]:
@@ -195,6 +240,8 @@ def parse_options() -> argparse.Namespace:
         prog='cleanmedia',
         description='Deletes 30 day old remote media files from dendrite servers')
     parser.add_argument('-c', '--config', default="config.yaml", help="location of the dendrite.yaml config file.")
+    parser.add_argument('-m', '--mxid', dest="mxid",
+                        help="Just delete media <MXID>. (no cleanup otherwise)")
     parser.add_argument('-t', '--days', dest="days",
                         default="30", type=int,
                         help="Keep remote media for <DAYS> days.")
@@ -217,28 +264,16 @@ if __name__ == '__main__':
     args = parse_options()
     (MEDIA_PATH, CONN_STR) = read_config(args.config)
     mr = MediaRepository(MEDIA_PATH, CONN_STR)
-    # Sanity checks
+
-    mr.sanity_check_thumbnails()  # warn in case of superfluous thumbnails
+
-    # Preps
+    if args.mxid:
-    if args.local:
+        # Just clean a single media
-        # populate the cache of current avt img. so we don't delete them
+        file = mr.get_single_media(args.mxid)
-        mr.get_avatar_images()
+        logging.info("Found media with id", args.mxid)
-    # ------real main part------------
+        if file and not args.dryrun:
-    cleantime = datetime.today() - timedelta(days=args.days)
+            file.delete()
-    logging.info("Deleting remote media older than %s", cleantime)
-    num_deleted = 0
-    files = mr.get_media(args.local)
-    for file in [f for f in files if f.media_id not in mr._avatar_media_ids]:
-        if file.create_date < cleantime:
-            num_deleted += 1
-            if args.dryrun:  # the great pretender
-                logging.info(f"Pretending to delete file id {file.media_id} on path {file.fullpath}.")
-                if not file.exists():
-                    logging.info(f"File id {file.media_id} does not physically exist (path {file.fullpath}).")
-            else:
-                file.delete()
-    if args.dryrun:
-        logging.info("%d files would have been deleted during the run.",
-                     num_deleted)
     else:
-        logging.info("Deleted %d files during the run.", num_deleted)
+        # Sanity checks
+        mr.sanity_check_thumbnails()  # warn in case of superfluous thumbnails
+        # Clean out of files
+        mr.clean_media_files(args.days, args.local, args.dryrun)