Fixing bug with parsing video upload dates.

modules/videos/lib.py

@@ -11,7 +11,7 @@ from sqlalchemy import or_
 from sqlalchemy.orm import Session
 from yt_dlp import YoutubeDL
 
-from wrolpi import before_startup
+from wrolpi import before_startup, dates
 from wrolpi.captions import extract_captions
 from wrolpi.common import ConfigFile, get_media_directory, sanitize_link, register_refresh_cleanup, limit_concurrent
 from wrolpi.dates import Seconds
@@ -80,7 +80,7 @@ def validate_video(video: Video, channel_generate_poster: bool):
     """
     info_json_path = video.info_json_path
     json_data_missing = bool(video.file_group.title) and bool(video.duration) and bool(video.view_count) \
-                        and bool(video.url) and bool(video.upload_date)
+                        and bool(video.url)
     if info_json_path and json_data_missing is False:
         # These properties can be found in the info json.
         video_info_json = process_video_info_json(video)
@@ -101,9 +101,8 @@ def validate_video(video: Video, channel_generate_poster: bool):
         # These are the least trusted, so anything already on the video should be trusted.
         _, upload_date, source_id, title = parse_video_file_name(video_path)
         upload_date = upload_date or video.upload_date
-        if upload_date:
-            upload_date = datetime.datetime.strptime(upload_date, '%Y%M%d')
-            upload_date = upload_date.replace(tzinfo=datetime.timezone.utc)
+        if upload_date and isinstance(upload_date, str):
+            upload_date = dates.strpdate(upload_date)
         video.file_group.title = video.file_group.title or html.unescape(title)
         video.upload_date = upload_date
         video.source_id = video.source_id or source_id

modules/videos/test/test_lib.py

@@ -69,7 +69,7 @@ async def test_video_factory(test_session, video_factory, channel_factory):
 
 def test_validate_video(test_session, test_directory, video_factory, image_bytes_factory, video_file):
     """A video poster will be generated only if the channel permits."""
-    video_file = video_file.rename(test_directory / 'Channel Name_20000101_1234567890_The Title.mp4')
+    video_file = video_file.rename(test_directory / 'Channel Name_20050607_1234567890_The Title.mp4')
     vid1 = video_factory(with_video_file=video_file)
     # Clear source_id so source_id in the file will be extracted.
     vid1.source_id = None
@@ -81,7 +81,7 @@ def test_validate_video(test_session, test_directory, video_factory, image_bytes
     assert vid1.poster_path.is_file(), 'Poster path does not exist'
     assert vid1.video_path.stem == vid1.poster_path.stem
     assert vid1.poster_path.suffix == '.jpg'
-    assert vid1.upload_date == datetime.datetime(2000, 1, 1, 0, 1, tzinfo=datetime.timezone.utc)
+    assert vid1.upload_date == datetime.datetime(2005, 6, 7, 0, 0, tzinfo=datetime.timezone.utc)
     assert vid1.source_id == '1234567890'
     assert vid1.file_group.title == 'The Title'
 

wrolpi/dates.py

@@ -48,6 +48,13 @@ def strptime(dt: str) -> datetime:
     return datetime.fromisoformat(dt).astimezone(pytz.UTC)
 
 
+def strpdate(dt: str) -> datetime:
+    if '-' in dt:
+        return datetime.strptime(dt, '%Y-%m-%d').replace(tzinfo=pytz.UTC)
+
+    return datetime.strptime(dt, '%Y%m%d').replace(tzinfo=pytz.UTC)
+
+
 def strftime_ms(dt: datetime) -> str:
     return dt.strftime(DATETIME_FORMAT_MS)
 

wrolpi/test/test_dates.py

@@ -79,3 +79,13 @@ def test_timedelta_to_timestamp(td, expected):
 ])
 def test_strptime(dt, expected):
     assert dates.strptime(dt) == expected
+
+
+@pytest.mark.parametrize('dt,expected', [
+    ('20010101', datetime(2001, 1, 1, tzinfo=pytz.UTC)),
+    ('20050607', datetime(2005, 6, 7, tzinfo=pytz.UTC)),
+    ('2006-07-08', datetime(2006, 7, 8, tzinfo=pytz.UTC)),
+    ('2007-8-9', datetime(2007, 8, 9, tzinfo=pytz.UTC)),
+])
+def test_strpdate(dt, expected):
+    assert dates.strpdate(dt) == expected