Merge 24954d0542 into 74e90dd9b8

test/test_utils.py

@@ -99,11 +99,13 @@ from yt_dlp.utils import (
     remove_start,
     render_table,
     replace_extension,
+    datetime_round,
     rot47,
     sanitize_filename,
     sanitize_path,
     sanitize_url,
     shell_quote,
+    strftime_or_none,
     smuggle_url,
     str_to_int,
     strip_jsonp,
@@ -407,6 +409,23 @@ class TestUtil(unittest.TestCase):
         self.assertEqual(datetime_from_str('now+1day', precision='hour'), datetime_from_str('now+24hours', precision='auto'))
         self.assertEqual(datetime_from_str('now+23hours', precision='hour'), datetime_from_str('now+23hours', precision='auto'))
 
+    def test_datetime_round(self):
+        self.assertEqual(datetime_round(dt.datetime.strptime('1820-05-12T01:23:45Z', '%Y-%m-%dT%H:%M:%SZ')),
+                         dt.datetime(1820, 5, 12, tzinfo=dt.timezone.utc))
+        self.assertEqual(datetime_round(dt.datetime.strptime('1969-12-31T23:34:45Z', '%Y-%m-%dT%H:%M:%SZ'), 'hour'),
+                         dt.datetime(1970, 1, 1, 0, tzinfo=dt.timezone.utc))
+        self.assertEqual(datetime_round(dt.datetime.strptime('2024-12-25T01:23:45Z', '%Y-%m-%dT%H:%M:%SZ'), 'minute'),
+                         dt.datetime(2024, 12, 25, 1, 24, tzinfo=dt.timezone.utc))
+        self.assertEqual(datetime_round(dt.datetime.strptime('2024-12-25T01:23:45.123Z', '%Y-%m-%dT%H:%M:%S.%fZ'), 'second'),
+                         dt.datetime(2024, 12, 25, 1, 23, 45, tzinfo=dt.timezone.utc))
+        self.assertEqual(datetime_round(dt.datetime.strptime('2024-12-25T01:23:45.678Z', '%Y-%m-%dT%H:%M:%S.%fZ'), 'second'),
+                         dt.datetime(2024, 12, 25, 1, 23, 46, tzinfo=dt.timezone.utc))
+
+    def test_strftime_or_none(self):
+        self.assertEqual(strftime_or_none(-4722192000), '18200512')
+        self.assertEqual(strftime_or_none(0), '19700101')
+        self.assertEqual(strftime_or_none(1735084800), '20241225')
+
     def test_daterange(self):
         _20century = DateRange('19000101', '20000101')
         self.assertFalse('17890714' in _20century)

yt_dlp/YoutubeDL.py

@@ -2692,11 +2692,7 @@ class YoutubeDL:
                 ('modified_timestamp', 'modified_date'),
         ):
             if info_dict.get(date_key) is None and info_dict.get(ts_key) is not None:
-                # Working around out-of-range timestamp values (e.g. negative ones on Windows,
-                # see http://bugs.python.org/issue1646728)
-                with contextlib.suppress(ValueError, OverflowError, OSError):
-                    upload_date = dt.datetime.fromtimestamp(info_dict[ts_key], dt.timezone.utc)
-                    info_dict[date_key] = upload_date.strftime('%Y%m%d')
+                info_dict[date_key] = strftime_or_none(info_dict[ts_key])
 
         if not info_dict.get('release_year'):
             info_dict['release_year'] = traverse_obj(info_dict, ('release_date', {lambda x: int(x[:4])}))

yt_dlp/utils/_utils.py

@@ -1369,6 +1369,13 @@ def datetime_add_months(dt_, months):
     return dt_.replace(year, month, day)
 
 
+def datetime_from_timestamp(timestamp):
+    # Calling dt.datetime.fromtimestamp with negative timestamps throws error in Windows
+    # Ref: https://github.com/yt-dlp/yt-dlp/issues/5185, https://github.com/python/cpython/issues/94414,
+    # https://github.com/yt-dlp/yt-dlp/issues/6706#issuecomment-1496842642
+    return (dt.datetime.fromtimestamp(0, dt.timezone.utc) + dt.timedelta(seconds=timestamp))
+
+
 def datetime_round(dt_, precision='day'):
     """
     Round a datetime object's time to a specific precision
@@ -1376,6 +1383,7 @@ def datetime_round(dt_, precision='day'):
     if precision == 'microsecond':
         return dt_
 
+    time_scale = 1_000_000
     unit_seconds = {
         'day': 86400,
         'hour': 3600,
@@ -1383,8 +1391,8 @@ def datetime_round(dt_, precision='day'):
         'second': 1,
     }
     roundto = lambda x, n: ((x + n / 2) // n) * n
-    timestamp = roundto(calendar.timegm(dt_.timetuple()), unit_seconds[precision])
-    return dt.datetime.fromtimestamp(timestamp, dt.timezone.utc)
+    timestamp = roundto(calendar.timegm(dt_.timetuple()) + dt_.microsecond / time_scale, unit_seconds[precision])
+    return datetime_from_timestamp(timestamp)
 
 
 def hyphenate_date(date_str):
@@ -2051,12 +2059,7 @@ def strftime_or_none(timestamp, date_format='%Y%m%d', default=None):
     datetime_object = None
     try:
         if isinstance(timestamp, (int, float)):  # unix timestamp
-            # Using naive datetime here can break timestamp() in Windows
-            # Ref: https://github.com/yt-dlp/yt-dlp/issues/5185, https://github.com/python/cpython/issues/94414
-            # Also, dt.datetime.fromtimestamp breaks for negative timestamps
-            # Ref: https://github.com/yt-dlp/yt-dlp/issues/6706#issuecomment-1496842642
-            datetime_object = (dt.datetime.fromtimestamp(0, dt.timezone.utc)
-                               + dt.timedelta(seconds=timestamp))
+            datetime_object = datetime_from_timestamp(timestamp)
         elif isinstance(timestamp, str):  # assume YYYYMMDD
             datetime_object = dt.datetime.strptime(timestamp, '%Y%m%d')
         date_format = re.sub(  # Support %s on windows