Clean up and document a little

cleanmedia

@@ -3,24 +3,23 @@
 #TODO: Sanity checks: Are files on the file system that the db does not know about?
 #TODO: Sanity checks: Are there thumbnails in the db that do not have corresponding media file entries?
 
+# mediaapi_media_repository: media_id | media_origin | content_type | file_size_bytes |  creation_ts  | upload_name | base64hash | user_id
+# mediaapi_thumbnail: media_id | media_origin | content_type | file_size_bytes |  creation_ts  | width | height | resize_method
+
 from datetime import datetime, timedelta
 from pathlib import Path
-import argparse
-import logging
-import typing
+import argparse, logging, typing
 
 try:
     import psycopg2
-except ImportError as e:
-    print("Please install psycopg2")
-    exit(1)
-try:
     import yaml
 except ImportError as e:
-    print("Please install pyyaml / python3-yaml")
+    print("Please install psycopg2 and pyyaml")
     exit(1)
 
+
 def read_config(conf_file):
+    """Read in the dendrite config file and return db creds and media path"""
     try:
         with open(conf_file) as f:
             config = yaml.safe_load(f)
@@ -51,9 +50,11 @@ def read_config(conf_file):
     return (BASE_PATH, CONN_STR)
 
 
-
+#------------------------------------------------------------------------------------
 class File:
-    def __init__(self, media_repo, media_id, creation_ts, base64hash: str):
+    """Represents a file in our database together with (hopefully) a physical file and thumbnails"""
+
+    def __init__(self, media_repo: 'MediaRepository', media_id: str, creation_ts: int, base64hash: str):
         # The MediaRepository in which this file is recorded
         self.repo = media_repo
         self.media_id = media_id
@@ -73,9 +74,10 @@ class File:
         if self.fullpath() is None:
             logging.info(f"No known path for file id '{self.media_id}', cannot delete file.")
         elif not self.fullpath().is_dir():
-            logging.info(f"No known path for file id '{self.media_id}', cannot delete.")
+            logging.info(f"Path for file id '{self.media_id}' exists but is not directory, ignoring instead of deleting.")
         else:
             for file in self.fullpath().glob('*'):
+                # note: this does not handle directories in fullpath()
                 file.unlink()
             self.fullpath().rmdir()
             logging.debug(f"Deleted directory {self.fullpath()}")
@@ -96,11 +98,11 @@ class File:
 
     def has_thumbnail(self):
         with self.repo.conn.cursor() as cur:
-            # media_id | media_origin | content_type | file_size_bytes |  creation_ts  | upload_name | base64hash | user_id
             cur.execute(f"select COUNT(media_id) from mediaapi_thumbnail WHERE media_id='{self.media_id}';")
             row = cur.fetchone()
         return(row[0])
-        
+
+#----------------------------------------------------------------------
 class MediaRepository:
 
     def __init__(self, media_path: Path, connection_string: str):
@@ -117,8 +119,7 @@ class MediaRepository:
         self.connect_db();
         
     def connect_db(self):
-        #postgresql://user:pass@hostname/database?params
-        # postgres://dendrite:dendrite@localhost/dendrite?
+        #postgresql://user:pass@localhost/database?params
         if self.db_conn_string is None or not self.db_conn_string.startswith("postgres://"):
             errstr = "DB connection not a postgres one"
             logging.error(errstr)
@@ -126,20 +127,15 @@ class MediaRepository:
         self.conn = psycopg2.connect(self.db_conn_string)
 
     def get_remote_media(self):
-        cur = self.conn.cursor()
-        # media_id | media_origin | content_type | file_size_bytes |  creation_ts  | upload_name | base64hash | user_id
-        res = cur.execute("select media_id, creation_ts, base64hash from mediaapi_media_repository WHERE user_id = '';")
-        #select * from mediaapi_media_repository WHERE user_id = '';
-        files = []
-        for row in cur.fetchall():
-            # creation_ts is ms since the epoch, so we need to make sec out of it
-            f = File(self, row[0], row[1]//1000, row[2])
-            files.append(f)
-
-        cur.close()
+        with self.conn.cursor() as cur:
+            # media_id | media_origin | content_type | file_size_bytes |  creation_ts  | upload_name | base64hash | user_id
+            res = cur.execute("select media_id, creation_ts, base64hash from mediaapi_media_repository WHERE user_id = '';")
+            files = []
+            for row in cur.fetchall():
+                # creation_ts is ms since the epoch, so convert to seconds
+                f = File(self, row[0], row[1]//1000, row[2])
+                files.append(f)
         return files
-        # mediaapi_thumbnail:
-        # media_id | media_origin | content_type | file_size_bytes |  creation_ts  | width | height | resize_method
 
 def parse_options():
     loglevel=logging.INFO # default