Merge branch 'master' of https://github.com/yt-dlp/yt-dlp into ytdlp

* 'master' of https://github.com/yt-dlp/yt-dlp:
  [crunchyroll:beta] Fix extractor after API change (#3801)
  Return an error code if update fails
  [utils] Fix bug in 0b9c08b
  [tiktok] Detect embeds

yt_dlp/YoutubeDL.py

@@ -33,7 +33,6 @@
     compat_str,
     compat_urllib_error,
     compat_urllib_request,
-    windows_enable_vt_mode,
 )
 from .cookies import load_cookies
 from .downloader import LDM_EXCEPTIONS, FFmpegFD, get_suitable_downloader, shorten_protocol_name
@@ -146,6 +145,7 @@
     url_basename,
     variadic,
     version_tuple,
+    windows_enable_vt_mode,
     write_json_file,
     write_string,
 )
@@ -3875,7 +3875,7 @@ def print_debug_header(self):
         def get_encoding(stream):
             ret = str(getattr(stream, 'encoding', 'missing (%s)' % type(stream).__name__))
             if not supports_terminal_sequences(stream):
-                from .compat import WINDOWS_VT_MODE  # Must be imported locally
+                from .utils import WINDOWS_VT_MODE  # Must be imported locally
                 ret += ' (No VT)' if WINDOWS_VT_MODE is False else ' (No ANSI)'
             return ret
 

yt_dlp/__init__.py

@@ -874,22 +874,16 @@ def _real_main(argv=None):
     with YoutubeDL(ydl_opts) as ydl:
         actual_use = all_urls or opts.load_info_filename
 
-        # Remove cache dir
         if opts.rm_cachedir:
             ydl.cache.remove()
 
-        # Update version
-        if opts.update_self:
+        if opts.update_self and run_update(ydl) and actual_use:
             # If updater returns True, exit. Required for windows
-            if run_update(ydl):
-                if actual_use:
-                    return 100, 'ERROR: The program must exit for the update to complete'
-                return
+            return 100, 'ERROR: The program must exit for the update to complete'
 
-        # Maybe do nothing
         if not actual_use:
             if opts.update_self or opts.rm_cachedir:
-                return
+                return ydl._download_retcode
 
             ydl.warn_if_short_id(sys.argv[1:] if argv is None else argv)
             parser.error(

yt_dlp/compat/__init__.py

@@ -1,6 +1,4 @@
-import contextlib
 import os
-import subprocess
 import sys
 import warnings
 import xml.etree.ElementTree as etree
@@ -74,17 +72,3 @@ def compat_expanduser(path):
         return userhome + path[i:]
 else:
     compat_expanduser = os.path.expanduser
-
-
-WINDOWS_VT_MODE = False if compat_os_name == 'nt' else None
-
-
-def windows_enable_vt_mode():  # TODO: Do this the proper way https://bugs.python.org/issue30075
-    if compat_os_name != 'nt':
-        return
-    global WINDOWS_VT_MODE
-    startupinfo = subprocess.STARTUPINFO()
-    startupinfo.dwFlags |= subprocess.STARTF_USESHOWWINDOW
-    with contextlib.suppress(Exception):
-        subprocess.Popen('', shell=True, startupinfo=startupinfo).wait()
-        WINDOWS_VT_MODE = True

yt_dlp/compat/_legacy.py

@@ -55,3 +55,10 @@ def compat_ctypes_WINFUNCTYPE(*args, **kwargs):
 compat_xpath = lambda xpath: xpath
 compat_zip = zip
 workaround_optparse_bug9161 = lambda: None
+
+
+def __getattr__(name):
+    if name in ('WINDOWS_VT_MODE', 'windows_enable_vt_mode'):
+        from .. import utils
+        return getattr(utils, name)
+    raise AttributeError(name)

yt_dlp/extractor/crunchyroll.py

@@ -728,11 +728,12 @@ def _get_params(self, lang):
                 headers={
                     'Authorization': auth_response['token_type'] + ' ' + auth_response['access_token']
                 })
-            bucket = policy_response['cms']['bucket']
+            cms = traverse_obj(policy_response, 'cms_beta', 'cms')
+            bucket = cms['bucket']
             params = {
-                'Policy': policy_response['cms']['policy'],
-                'Signature': policy_response['cms']['signature'],
-                'Key-Pair-Id': policy_response['cms']['key_pair_id']
+                'Policy': cms['policy'],
+                'Signature': cms['signature'],
+                'Key-Pair-Id': cms['key_pair_id']
             }
             locale = traverse_obj(initial_state, ('localization', 'locale'))
             if locale:

yt_dlp/extractor/generic.py

@@ -75,6 +75,7 @@
 from .ted import TedEmbedIE
 from .theplatform import ThePlatformIE
 from .threeqsdn import ThreeQSDNIE
+from .tiktok import TikTokIE
 from .tnaflix import TNAFlixNetworkEmbedIE
 from .tube8 import Tube8IE
 from .tunein import TuneInBaseIE
@@ -3862,6 +3863,11 @@ def _real_extract(self, url):
         if ruutu_urls:
             return self.playlist_from_matches(ruutu_urls, video_id, video_title)
 
+        # Look for Tiktok embeds
+        tiktok_urls = TikTokIE._extract_urls(webpage)
+        if tiktok_urls:
+            return self.playlist_from_matches(tiktok_urls, video_id, video_title)
+
         # Look for HTML5 media
         entries = self._parse_html5_media_entries(url, webpage, video_id, m3u8_id='hls')
         if entries:

yt_dlp/extractor/tiktok.py

@@ -1,28 +1,26 @@
 import itertools
+import json
 import random
+import re
 import string
 import time
-import json
 
 from .common import InfoExtractor
-from ..compat import (
-    compat_urllib_parse_unquote,
-    compat_urllib_parse_urlparse
-)
+from ..compat import compat_urllib_parse_unquote, compat_urllib_parse_urlparse
 from ..utils import (
     ExtractorError,
     HEADRequest,
+    LazyList,
     UnsupportedError,
     get_first,
     int_or_none,
     join_nonempty,
-    LazyList,
+    qualities,
     srt_subtitles_timecode,
     str_or_none,
     traverse_obj,
     try_get,
     url_or_none,
-    qualities,
 )
 
 
@@ -36,6 +34,10 @@ class TikTokBaseIE(InfoExtractor):
     _WEBPAGE_HOST = 'https://www.tiktok.com/'
     QUALITIES = ('360p', '540p', '720p', '1080p')
 
+    @staticmethod
+    def _create_url(user_id, video_id):
+        return f'https://www.tiktok.com/@{user_id or "_"}/video/{video_id}'
+
     def _call_api_impl(self, ep, query, manifest_app_version, video_id, fatal=True,
                        note='Downloading API JSON', errnote='Unable to download API page'):
         self._set_cookie(self._API_HOSTNAME, 'odin_tt', ''.join(random.choice('0123456789abcdef') for _ in range(160)))
@@ -361,7 +363,7 @@ def _parse_aweme_video_web(self, aweme_detail, webpage_url):
 
 
 class TikTokIE(TikTokBaseIE):
-    _VALID_URL = r'https?://www\.tiktok\.com/@[\w\.-]+/video/(?P<id>\d+)'
+    _VALID_URL = r'https?://www\.tiktok\.com/(?:embed|@(?P<user_id>[\w\.-]+)/video)/(?P<id>\d+)'
 
     _TESTS = [{
         'url': 'https://www.tiktok.com/@leenabhushan/video/6748451240264420610',
@@ -466,7 +468,7 @@ class TikTokIE(TikTokBaseIE):
         'info_dict': {
             'id': '7059698374567611694',
             'ext': 'mp4',
-            'title': 'tiktok video #7059698374567611694',
+            'title': 'TikTok video #7059698374567611694',
             'description': '',
             'uploader': 'pokemonlife22',
             'creator': 'Pokemon',
@@ -490,6 +492,11 @@ class TikTokIE(TikTokBaseIE):
         'only_matching': True
     }]
 
+    @classmethod
+    def _extract_urls(cls, webpage):
+        return [mobj.group('url') for mobj in re.finditer(
+            rf'<(?:script|iframe)[^>]+\bsrc=(["\'])(?P<url>{cls._VALID_URL})', webpage)]
+
     def _extract_aweme_app(self, aweme_id):
         try:
             aweme_detail = self._call_api('aweme/detail', {'aweme_id': aweme_id}, aweme_id,
@@ -506,7 +513,8 @@ def _extract_aweme_app(self, aweme_id):
         return self._parse_aweme_video_app(aweme_detail)
 
     def _real_extract(self, url):
-        video_id = self._match_id(url)
+        video_id, user_id = self._match_valid_url(url).group('id', 'user_id')
+        url = self._create_url(user_id, video_id)
 
         try:
             return self._extract_aweme_app(video_id)

yt_dlp/utils.py

@@ -5218,10 +5218,12 @@ def jwt_decode_hs256(jwt):
     return payload_data
 
 
+WINDOWS_VT_MODE = False if compat_os_name == 'nt' else None
+
+
 @functools.cache
 def supports_terminal_sequences(stream):
     if compat_os_name == 'nt':
-        from .compat import WINDOWS_VT_MODE  # Must be imported locally
         if not WINDOWS_VT_MODE or get_windows_version() < (10, 0, 10586):
             return False
     elif not os.getenv('TERM'):
@@ -5232,6 +5234,21 @@ def supports_terminal_sequences(stream):
         return False
 
 
+def windows_enable_vt_mode():  # TODO: Do this the proper way https://bugs.python.org/issue30075
+    if compat_os_name != 'nt':
+        return
+    global WINDOWS_VT_MODE
+    startupinfo = subprocess.STARTUPINFO()
+    startupinfo.dwFlags |= subprocess.STARTF_USESHOWWINDOW
+    try:
+        subprocess.Popen('', shell=True, startupinfo=startupinfo).wait()
+    except Exception:
+        return
+
+    WINDOWS_VT_MODE = True
+    supports_terminal_sequences.cache_clear()
+
+
 _terminal_sequences_re = re.compile('\033\\[[^m]+m')