feat: Add support for downloading YouTube watch history with timestamps

personal_data/fetchers/youtube.py

@@ -2,6 +2,7 @@ import csv
 import json
 import logging
 import subprocess
+import datetime
 from dataclasses import dataclass
 
 from personal_data.data import DeduplicateMode, Scraper
@@ -17,16 +18,27 @@ class YoutubeFavoritesScraper(Scraper):
     dataset_name: str = 'youtube_favorites'
     deduplicate_mode: DeduplicateMode = DeduplicateMode.BY_ALL_COLUMNS
     deduplicate_ignore_columns = []
+    watch_history: bool = False
 
     def scrape(self) -> list[dict]:
         """Use yt-dlp to fetch the list of favorited videos. This is a placeholder for invoking yt-dlp and parsing its output."""
-        result = subprocess.run(
-            [
+        if self.watch_history:
+            url = 'https://www.youtube.com/feed/history'
+            ytdlp_args = [
+                'yt-dlp',
+                '--dump-json',
+                url,
+            ]
+        else:
+            url = f'https://www.youtube.com/playlist?list={PLAYLIST_ID}'
+            ytdlp_args = [
                 'yt-dlp',
                 '--flat-playlist',
                 '--dump-json',
-                f'https://www.youtube.com/playlist?list={PLAYLIST_ID}',
-            ],
+                url,
+            ]
+        result = subprocess.run(
+            ytdlp_args,
             capture_output=True,
             text=True,
         )
@@ -34,8 +46,16 @@ class YoutubeFavoritesScraper(Scraper):
         if result.returncode != 0:
             raise RuntimeError(f'Non-zero returncode in command: {result.returncode}\n\n{result.stderr}')
     
+        output = []
         for line in result.stdout.splitlines():
             data = json.loads(line)
+            if self.watch_history:
+                if 'thumbnails' in data and data['thumbnails']:
+                    data['thumbnail'] = data['thumbnails'][-1]['url']
+                if 'timestamp' in data:
+                    data['watch_datetime'] = datetime.datetime.fromtimestamp(int(data['timestamp'])).isoformat()
+            else:
                 data['thumbnail'] = data['thumbnails'][-1]['url']
             safe_del(data, '_type', '_version', 'thumbnails')
-            yield data
+            output.append(data)
+        return output