From 171163b3d917951d8366be1fb7707c6f67b3710e Mon Sep 17 00:00:00 2001
From: MasterofJOKers <joker@someserver.de>
Date: Fri, 8 Mar 2024 16:21:32 +0100
Subject: [PATCH] Add mirror-yt.py - a wrapper around yt-dlp

This script is supposed to run regularly to mirror videos of certain
YouTube channels locally for archiving purposes.
---
 mirror-yt.py | 66 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 66 insertions(+)
 create mode 100644 mirror-yt.py

diff --git a/mirror-yt.py b/mirror-yt.py
new file mode 100644
index 0000000..6601edb
--- /dev/null
+++ b/mirror-yt.py
@@ -0,0 +1,66 @@
+#!/usr/bin/env python3
+from pathlib import Path
+import subprocess
+
+import click
+
+
+@click.command()
+@click.option('--allow-unknown', is_flag=True,
+              help='Error out if one of the directories from the given list does not exist')
+@click.option('--base-dir', type=Path, default='.', show_default=True,
+              help='Base directory where the channel directories get created')
+@click.argument('CHANNEL_FILE', required=True)
+def download(allow_unknown, base_dir, channel_file):
+    """Mirror channels from YouTube
+
+    The list of channels needs to be provided via CHANNEL_FILE. The format of the file should be one channel per line.
+    each line consisting of the directory name and the URL to download the videos joined by a semicolon.
+    """
+    base_dir = base_dir.resolve()
+
+    channels = {}
+    with open(channel_file) as f:
+        for line in f:
+            line = line.strip()
+            if line.startswith('#'):
+                continue
+            
+            dir_name, url = line.split(';')
+            channels[dir_name] = url
+
+    if not allow_unknown:
+        error = False
+        for dir_name in channels:
+            if (base_dir / dir_name).is_dir():
+                continue
+
+            print(f"Error: '{dir_name}' is an unknown channel")
+            error = True
+
+        if error:
+            raise click.ClickException("Unknown channels and '--allow-unknown' not given")
+
+    for dir_name, url in channels.items():
+        target_dir = base_dir / dir_name
+
+        print(f"Starting download for {dir_name}")
+
+        if not target_dir.exists():
+            target_dir.mkdir()
+
+        cmd = [
+            'yt-dlp',
+            '-o', '%(upload_date)s - %(title)s - %(id)s.%(ext)s',
+            '--sleep-interval', '30',
+            '--max-sleep-interval', '300',
+            '--download-archive', target_dir / '.archive',
+            '--break-on-existing',
+            '--write-thumbnail',
+            url
+        ]
+        subprocess.run(cmd, check=True, cwd=target_dir) 
+
+
+if __name__ == '__main__':
+    download()