[core] Fix support for upcoming Python 3.12 (#8130)

[yt-dlp.git] / yt_dlp / utils / _utils.py

diff --git a/yt_dlp/utils/_utils.py b/yt_dlp/utils/_utils.py
index abae0f17e4969085f50766b895d9f6269ef7c8ef..ef26de1160f802908cfb8ba4c6a5352f352c0f59 100644 (file)
--- a/yt_dlp/utils/_utils.py
+++ b/yt_dlp/utils/_utils.py
@@ -669,6 +669,7 @@ def replace_insane(char):
 
 def sanitize_path(s, force=False):
     """Sanitizes and normalizes path on Windows"""
+    # XXX: this handles drive relative paths (c:sth) incorrectly
     if sys.platform == 'win32':
         force = False
         drive_or_unc, _ = os.path.splitdrive(s)
@@ -687,7 +688,10 @@ def sanitize_path(s, force=False):
         sanitized_path.insert(0, drive_or_unc + os.path.sep)
     elif force and s and s[0] == os.path.sep:
         sanitized_path.insert(0, os.path.sep)
-    return os.path.join(*sanitized_path)
+    # TODO: Fix behavioral differences <3.12
+    # The workaround using `normpath` only superficially passes tests
+    # Ref: https://github.com/python/cpython/pull/100351
+    return os.path.normpath(os.path.join(*sanitized_path))
 
 
 def sanitize_url(url, *, scheme='http'):
@@ -1256,7 +1260,7 @@ def datetime_from_str(date_str, precision='auto', format='%Y%m%d'):
     if precision == 'auto':
         auto_precision = True
         precision = 'microsecond'
-    today = datetime_round(datetime.datetime.utcnow(), precision)
+    today = datetime_round(datetime.datetime.now(datetime.timezone.utc), precision)
     if date_str in ('now', 'today'):
         return today
     if date_str == 'yesterday':
@@ -1319,8 +1323,8 @@ def datetime_round(dt, precision='day'):
         'second': 1,
     }
     roundto = lambda x, n: ((x + n / 2) // n) * n
-    timestamp = calendar.timegm(dt.timetuple())
-    return datetime.datetime.utcfromtimestamp(roundto(timestamp, unit_seconds[precision]))
+    timestamp = roundto(calendar.timegm(dt.timetuple()), unit_seconds[precision])
+    return datetime.datetime.fromtimestamp(timestamp, datetime.timezone.utc)
 
 
 def hyphenate_date(date_str):
@@ -2464,23 +2468,6 @@ def lowercase_escape(s):
         s)
 
 
-def escape_rfc3986(s):
-    """Escape non-ASCII characters as suggested by RFC 3986"""
-    return urllib.parse.quote(s, b"%/;:@&=+$,!~*'()?#[]")
-
-
-def escape_url(url):
-    """Escape URL as suggested by RFC 3986"""
-    url_parsed = urllib.parse.urlparse(url)
-    return url_parsed._replace(
-        netloc=url_parsed.netloc.encode('idna').decode('ascii'),
-        path=escape_rfc3986(url_parsed.path),
-        params=escape_rfc3986(url_parsed.params),
-        query=escape_rfc3986(url_parsed.query),
-        fragment=escape_rfc3986(url_parsed.fragment)
-    ).geturl()
-
-
 def parse_qs(url, **kwargs):
     return urllib.parse.parse_qs(urllib.parse.urlparse(url).query, **kwargs)
 
@@ -2864,6 +2851,7 @@ def mimetype2ext(mt, default=NO_DEFAULT):
         'quicktime': 'mov',
         'webm': 'webm',
         'vp9': 'vp9',
+        'video/ogg': 'ogv',
         'x-flv': 'flv',
         'x-m4v': 'm4v',
         'x-matroska': 'mkv',