YouTube fixed

personal_data/fetchers/gitea.py

@@ -4,18 +4,13 @@ from collections.abc import Iterator, Mapping
 from typing import Any
 
 from personal_data.data import DeduplicateMode, Scraper
+from ..util import safe_del
 
 from .. import secrets
 
 logger = logging.getLogger(__name__)
 
 
-def safe_del(d: dict, *keys: str):
-    for key in keys:
-        if key in d:
-            del d[key]
-
-
 def to_data_point(p: dict[str, Any]) -> Mapping[str, Any]:
     p['owner'] = p['owner']['login']
     safe_del(p, 'permissions', 'internal_tracker')

personal_data/fetchers/youtube.py

@@ -5,57 +5,37 @@ import subprocess
 from dataclasses import dataclass
 
 from personal_data.data import DeduplicateMode, Scraper
+from ..util import safe_del
 
 logger = logging.getLogger(__name__)
 
+PLAYLIST_ID='PLAfDVJvDKCvOMvfoTL7eW8GkWNJwd90eV'
+#PLAYLIST_ID='LL'
 
 @dataclass(frozen=True)
 class YoutubeFavoritesScraper(Scraper):
     dataset_name: str = 'youtube_favorites'
     deduplicate_mode: DeduplicateMode = DeduplicateMode.BY_ALL_COLUMNS
+    deduplicate_ignore_columns = []
 
-    def fetch_data(self) -> list[dict]:
+    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."""
-        try:
-            # Replace 'YOUR_FAVORITES_ID' with your actual favorites playlist ID.
         result = subprocess.run(
             [
                 'yt-dlp',
                 '--flat-playlist',
                 '--dump-json',
-                    'https://www.youtube.com/playlist?list=YOUR_FAVORITES_ID',
+                f'https://www.youtube.com/playlist?list={PLAYLIST_ID}',
             ],
             capture_output=True,
-                check=True,
             text=True,
         )
-            return [json.loads(line) for line in result.stdout.splitlines()]
-        except Exception:
-            logger.exception('Failed to fetch YouTube favorites')
-            raise
 
-    def to_csv(self, videos: list[dict]) -> str:
-        """Convert the list of videos to CSV format."""
-        headers = ['id', 'title', 'url', 'upload_date']
-        rows = [headers] + [
-            [
-                video.get('id'),
-                video.get('title'),
-                video.get('url'),
-                video.get('upload_date'),
-            ]
-            for video in videos
-        ]
-        from io import StringIO
+        if result.returncode != 0:
+            raise RuntimeError(f'Non-zero returncode in command: {result.returncode}\n\n{result.stderr}')
 
-        sio = StringIO()
-        csv.writer(sio).writerows(rows)
-        return sio.getvalue()
-
-    def run(self) -> None:
-        videos = self.fetch_data()
-        csv_data = self.to_csv(videos)
-        logger.info('Fetched and converted %d videos to CSV', len(videos))
-        with open('youtube_favorites.csv', 'w', encoding='utf-8') as f:
-            f.write(csv_data)
-            logger.info('CSV file written to youtube_favorites.csv')
+        for line in result.stdout.splitlines():
+            data = json.loads(line)
+            data['thumbnail'] = data['thumbnails'][-1]['url']
+            safe_del(data, '_type', '_version', 'thumbnails')
+            yield data

personal_data/util.py

@@ -14,6 +14,13 @@ from . import csv_import, data
 logger = logging.getLogger(__name__)
 
 
+def safe_del(d: dict, *keys: str):
+    for key in keys:
+        if key in d:
+            del d[key]
+
+
+
 def equals_without_fields(
     a: Mapping[str, Any],
     b: Mapping[str, Any],