From ed182d79f6c10e2def5611d6f2a007de074c890e Mon Sep 17 00:00:00 2001
From: Jesse Bannon <jesse.bannon@datarobot.com>
Date: Fri, 10 Sep 2021 22:55:15 -0700
Subject: [PATCH] main

---
 ytdl_subscribe/main.py          | 14 ++++++++------
 ytdl_subscribe/subscriptions.py | 17 ++++++++++-------
 2 files changed, 18 insertions(+), 13 deletions(-)

diff --git a/ytdl_subscribe/main.py b/ytdl_subscribe/main.py
index d84d142e..475ca515 100644
--- a/ytdl_subscribe/main.py
+++ b/ytdl_subscribe/main.py
@@ -5,17 +5,19 @@ parser = argparse.ArgumentParser(
     description="YoutubeDL-Subscribe: Download and organize your favorite media easily."
 )
 parser.add_argument(
-    "-c", "--config",
-    metavar='CONFIGPATH',
+    "-c",
+    "--config",
+    metavar="CONFIGPATH",
     type=str,
     help="path to the config yaml, uses subscriptions.yaml if not provided",
     default="subscriptions.yaml",
 )
 parser.add_argument(
-    "-s", "--subscriptions",
-    metavar='SUB',
-    nargs='+',
-    help='specific subscriptions to download, downloads all if not provided'
+    "-s",
+    "--subscriptions",
+    metavar="SUB",
+    nargs="+",
+    help="specific subscriptions to download, downloads all if not provided",
 )
 
 
diff --git a/ytdl_subscribe/subscriptions.py b/ytdl_subscribe/subscriptions.py
index 3da46004..84d778eb 100644
--- a/ytdl_subscribe/subscriptions.py
+++ b/ytdl_subscribe/subscriptions.py
@@ -199,15 +199,12 @@ class SoundcloudSubscription(Subscription):
         tracks = []
 
         if self.options.get("download_strategy") == "albums_then_tracks":
-            track_ytdl_opts = {
-                "download_archive": self.WORKING_DIRECTORY
-                + "/ytdl-download-archive.txt",
-            }
-            # Get the album tracks first, but do not download. Unfortunately we cannot use download_archive for
-            # this be
+            # Get the album info first, but do not download. This tells us which track ids belong
+            # to an album. Unfortunately we cannot use download_archive or info.json for this
             with ytdl.YoutubeDL(self.ytdl_opts) as ytd:
                 info = ytd.extract_info(base_url + "/albums", download=False)
 
+            # For each album, parse each entry in the album
             album_entries = [self.parse_album_entry(a) for a in info["entries"]]
             for album_entry in album_entries:
                 tracks += [
@@ -216,9 +213,15 @@ class SoundcloudSubscription(Subscription):
                     if not self.is_entry_skippable(e)
                 ]
 
+            # Download the tracks now, and use download_archive to cache
+            track_ytdl_opts = {
+                "download_archive": self.WORKING_DIRECTORY
+                + "/ytdl-download-archive.txt",
+            }
             with ytdl.YoutubeDL(dict(self.ytdl_opts, **track_ytdl_opts)) as ytd:
-                # Get the rest of the tracks that are part of the album tracks
                 info = ytd.extract_info(base_url + "/tracks")
+
+            # Skip parsing entries that have already been parsed when parsing albums
             album_track_ids = [t["id"] for t in tracks]
             tracks += [
                 self.parse_entry(e)