[extractor/generic] Don't report redirect to https

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

diff --git a/yt_dlp/utils.py b/yt_dlp/utils.py
index 84a8ecd6ed28f4d3564d3e332207654586e90d66..7cba13678fc630fe566a5ca650dbc1a9626328af 100644 (file)
--- a/yt_dlp/utils.py
+++ b/yt_dlp/utils.py
@@ -418,6 +418,8 @@ def get_elements_text_and_html_by_attribute(attribute, value, html, *, tag=r'[\w
     Return the text (content) and the html (whole) of the tag with the specified
     attribute in the passed HTML document
     """
+    if not value:
+        return
 
     quote = '' if re.match(r'''[\s"'`=<>]''', value) else '?'
 
@@ -480,6 +482,7 @@ def handle_endtag(self, tag):
             raise self.HTMLBreakOnClosingTagException()
 
 
+# XXX: This should be far less strict
 def get_element_text_and_html_by_tag(tag, html):
     """
     For the first element with the specified tag in the passed HTML document
@@ -685,7 +688,8 @@ def replace_insane(char):
             return '\0_'
         return char
 
-    if restricted and is_id is NO_DEFAULT:
+    # Replace look-alike Unicode glyphs
+    if restricted and (is_id is NO_DEFAULT or not is_id):
         s = unicodedata.normalize('NFKC', s)
     s = re.sub(r'[0-9]+(?::[0-9]+)+', lambda m: m.group(0).replace(':', '_'), s)  # Handle timestamps
     result = ''.join(map(replace_insane, s))
@@ -986,16 +990,23 @@ def make_HTTPS_handler(params, **kwargs):
         context.options |= 4  # SSL_OP_LEGACY_SERVER_CONNECT
         # Allow use of weaker ciphers in Python 3.10+. See https://bugs.python.org/issue43998
         context.set_ciphers('DEFAULT')
-    elif sys.version_info < (3, 10) and ssl.OPENSSL_VERSION_INFO >= (1, 1, 1):
+    elif (
+        sys.version_info < (3, 10)
+        and ssl.OPENSSL_VERSION_INFO >= (1, 1, 1)
+        and not ssl.OPENSSL_VERSION.startswith('LibreSSL')
+    ):
         # Backport the default SSL ciphers and minimum TLS version settings from Python 3.10 [1].
         # This is to ensure consistent behavior across Python versions, and help avoid fingerprinting
         # in some situations [2][3].
         # Python 3.10 only supports OpenSSL 1.1.1+ [4]. Because this change is likely
         # untested on older versions, we only apply this to OpenSSL 1.1.1+ to be safe.
+        # LibreSSL is excluded until further investigation due to cipher support issues [5][6].
         # 1. https://github.com/python/cpython/commit/e983252b516edb15d4338b0a47631b59ef1e2536
         # 2. https://github.com/yt-dlp/yt-dlp/issues/4627
         # 3. https://github.com/yt-dlp/yt-dlp/pull/5294
         # 4. https://peps.python.org/pep-0644/
+        # 5. https://peps.python.org/pep-0644/#libressl-support
+        # 6. https://github.com/yt-dlp/yt-dlp/commit/5b9f253fa0aee996cf1ed30185d4b502e00609c4#commitcomment-89054368
         context.set_ciphers('@SECLEVEL=2:ECDH+AESGCM:ECDH+CHACHA20:ECDH+AES:DHE+AES:!aNULL:!eNULL:!aDSS:!SHA1:!AESCCM')
         context.minimum_version = ssl.TLSVersion.TLSv1_2
 
@@ -1995,9 +2006,10 @@ def system_identifier():
     with contextlib.suppress(OSError):  # We may not have access to the executable
         libc_ver = platform.libc_ver()
 
-    return 'Python %s (%s %s) - %s (%s%s)' % (
+    return 'Python %s (%s %s %s) - %s (%s%s)' % (
         platform.python_version(),
         python_implementation,
+        platform.machine(),
         platform.architecture()[0],
         platform.platform(),
         ssl.OPENSSL_VERSION,
@@ -2686,9 +2698,7 @@ def check_executable(exe, args=[]):
     return exe
 
 
-def _get_exe_version_output(exe, args, *, to_screen=None):
-    if to_screen:
-        to_screen(f'Checking exe version: {shell_quote([exe] + args)}')
+def _get_exe_version_output(exe, args):
     try:
         # STDIN should be redirected too. On UNIX-like systems, ffmpeg triggers
         # SIGTTOU if yt-dlp is run in the background.
@@ -2940,10 +2950,10 @@ def __init__(self, ydl, info_dict):
             self.is_exhausted = True
 
         requested_entries = info_dict.get('requested_entries')
-        self.is_incomplete = bool(requested_entries)
+        self.is_incomplete = requested_entries is not None
         if self.is_incomplete:
             assert self.is_exhausted
-            self._entries = [self.MissingEntry] * max(requested_entries)
+            self._entries = [self.MissingEntry] * max(requested_entries or [0])
             for i, entry in zip(requested_entries, entries):
                 self._entries[i - 1] = entry
         elif isinstance(entries, (list, PagedList, LazyList)):
@@ -3012,7 +3022,7 @@ def get_entry(i):
                     if not self.is_incomplete:
                         raise self.IndexError()
                 if entry is self.MissingEntry:
-                    raise EntryNotInPlaylist(f'Entry {i} cannot be found')
+                    raise EntryNotInPlaylist(f'Entry {i + 1} cannot be found')
                 return entry
         else:
             def get_entry(i):
@@ -5829,7 +5839,7 @@ def wrapper(self, *args, **kwargs):
         bound_args.apply_defaults()
         key = tuple(bound_args.arguments.values())[1:]
 
-        cache = vars(self).setdefault('__cached_method__cache', {}).setdefault(f.__name__, {})
+        cache = vars(self).setdefault('_cached_method__cache', {}).setdefault(f.__name__, {})
         if key not in cache:
             cache[key] = f(self, *args, **kwargs)
         return cache[key]
@@ -5837,14 +5847,23 @@ def wrapper(self, *args, **kwargs):
 
 
 class classproperty:
-    """property access for class methods"""
+    """property access for class methods with optional caching"""
+    def __new__(cls, func=None, *args, **kwargs):
+        if not func:
+            return functools.partial(cls, *args, **kwargs)
+        return super().__new__(cls)
 
-    def __init__(self, func):
+    def __init__(self, func, *, cache=False):
         functools.update_wrapper(self, func)
         self.func = func
+        self._cache = {} if cache else None
 
     def __get__(self, _, cls):
-        return self.func(cls)
+        if self._cache is None:
+            return self.func(cls)
+        elif cls not in self._cache:
+            self._cache[cls] = self.func(cls)
+        return self._cache[cls]
 
 
 class Namespace(types.SimpleNamespace):