Merge pull request #10 from MattBlackOnly/testing

Add an option to trim metadata length via environmental variable

tubetube/helpers.py

@@ -1,4 +1,5 @@
 import re
+import yt_dlp
 
 
 def parse_video_id(url):
@@ -13,3 +14,19 @@ def parse_video_id(url):
         if match:
             return match.group(1)
     return None
+
+
+class TrimDescriptionPP(yt_dlp.postprocessor.PostProcessor):
+    def run(self, info):
+        description = info.get("description", "")
+
+        if description:
+            trimmed_description = description[:250]
+            self.to_screen(f"Original description length: {len(description)}")
+            self.to_screen(f"Trimmed description to 250 chars: {trimmed_description}")
+
+            info["description"] = trimmed_description
+        else:
+            self.to_screen("No description found to trim.")
+
+        return [], info

tubetube/yt_downloader.py

@@ -7,7 +7,7 @@
 import random
 import yt_dlp
 from settings import DownloadCancelledException
-from helpers import parse_video_id
+import helpers
 
 
 class DownloadManager:
@@ -26,6 +26,9 @@ def __init__(self):
         self.verbose_ytdlp = bool(os.getenv("VERBOSE_LOGS", "False").lower() == "true")
         logging.info(f"Verbose logging for yt-dlp set to: {self.verbose_ytdlp}")
 
+        self.trim_metadata = bool(os.getenv("TRIM_METADATA", "False").lower() == "true")
+        logging.info(f"Trim Metadata set to: {self.trim_metadata}")
+
         self.thread_count = int(os.getenv("THREAD_COUNT", "4"))
         logging.info(f"Thread Count: {self.thread_count}")
 
@@ -69,7 +72,7 @@ def add_to_queue(self, item_info):
             url = re.sub(r"&list=.*", "", url)
 
         with self.lock:
-            parsed_identifier = parse_video_id(url)
+            parsed_identifier = helpers.parse_video_id(url)
             if any(item["url"] == url or item["video_identifier"] == parsed_identifier for item in self.all_items.values()):
                 logging.info(f"URL {url} is already in the queue or being downloaded.")
                 self.socketio.emit("toast", {"title": "Duplicate URL", "body": f"The video '{url}' is already in the queue or being processed."})
@@ -208,6 +211,8 @@ def _download_item(self, download_id):
         try:
             logging.info(f'Starting {threading.current_thread().name} Download: {item.get("title")}')
             ydl = yt_dlp.YoutubeDL(ydl_opts)
+            if self.trim_metadata:
+                ydl.add_post_processor(helpers.TrimDescriptionPP(), when="before_dl")
             result = ydl.download([item["url"]])
             item["progress"] = "Done" if result == 0 else "Incomplete"
             item["status"] = "Complete"