Merge 143f0c9f41 into 64766459e3

Authored by: Grub4K

test/test_utils.py

@@ -2059,7 +2059,22 @@ Line 1
         assert extract_basic_auth('http://user:pass@foo.bar') == ('http://foo.bar', 'Basic dXNlcjpwYXNz')
 
     @unittest.skipUnless(compat_os_name == 'nt', 'Only relevant on Windows')
-    def test_Popen_windows_escaping(self):
+    def test_windows_escaping(self):
+        tests = [
+            'test"&',
+            '%CMDCMDLINE:~-1%&',
+            'a\nb',
+            '"',
+            '\\',
+            '!',
+            '^!',
+            'a \\ b',
+            'a \\" b',
+            'a \\ b\\',
+            # We replace \r with \n
+            ('a\r\ra', 'a\n\na'),
+        ]
+
         def run_shell(args):
             stdout, stderr, error = Popen.run(
                 args, text=True, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
@@ -2067,15 +2082,18 @@ Line 1
             assert not error
             return stdout
 
-        # Test escaping
-        assert run_shell(['echo', 'test"&']) == '"test""&"\n'
-        assert run_shell(['echo', '%CMDCMDLINE:~-1%&']) == '"%CMDCMDLINE:~-1%&"\n'
-        assert run_shell(['echo', 'a\nb']) == '"a"\n"b"\n'
-        assert run_shell(['echo', '"']) == '""""\n'
-        assert run_shell(['echo', '\\']) == '\\\n'
-        # Test if delayed expansion is disabled
-        assert run_shell(['echo', '^!']) == '"^!"\n'
-        assert run_shell('echo "^!"') == '"^!"\n'
+        for argument in tests:
+            if isinstance(argument, str):
+                expected = argument
+            else:
+                argument, expected = argument
+
+            args = [sys.executable, '-c', 'import sys; print(end=sys.argv[1])', argument, 'end']
+            assert run_shell(args) == expected
+
+            escaped = shell_quote(argument, shell=True)
+            args = f'{sys.executable} -c "import sys; print(end=sys.argv[1])" {escaped} end'
+            assert run_shell(args) == expected
 
 
 if __name__ == '__main__':

yt_dlp/extractor/_extractors.py

@@ -1334,6 +1334,7 @@ from .nrk import (
     NRKTVSeriesIE,
 )
 from .nrl import NRLTVIE
+from .nts import NTSLiveIE
 from .ntvcojp import NTVCoJpCUIE
 from .ntvde import NTVDeIE
 from .ntvru import NTVRuIE

yt_dlp/extractor/nts.py

@@ -0,0 +1,60 @@
+from .common import InfoExtractor
+from ..utils import extract_attributes
+
+
+class NTSLiveIE(InfoExtractor):
+    IE_NAME = 'nts.live'
+    _VALID_URL = r'https?://(?:www\.)?nts\.live/shows/[^/]+/episodes/(?P<id>[^./?#]+)'
+    _TESTS = [
+        {
+            # embedded soundcloud
+            'url': 'https://www.nts.live/shows/yu-su/episodes/yu-su-2nd-april-2024',
+            'md5': 'b5444c04888c869d68758982de1a27d8',
+            'info_dict': {
+                'id': '1791563518',
+                'ext': 'opus',
+                'uploader_id': '995579326',
+                'title': 'Pender Street Steppers & YU SU 020424',
+                'timestamp': 1712143743,
+                'upload_date': '20240403',
+                'thumbnail': 'https://i1.sndcdn.com/artworks-qKcNO0z0AQGGbv9s-GljJCw-original.jpg',
+                'license': 'all-rights-reserved',
+                'repost_count': int,
+                'uploader_url': 'https://soundcloud.com/user-643553014',
+                'uploader': 'NTS Latest',
+                'description': 'md5:cf9d7d7997ef00a7b6046c9615b72c55',
+                'duration': 10784.157,
+            }
+        },
+        {
+            # embedded mixcloud
+            'url': 'https://www.nts.live/shows/absolute-fiction/episodes/absolute-fiction-23rd-july-2022',
+            'md5': 'TODO',
+            'info_dict': {
+                'id': 'NTSRadio_absolute-fiction-23rd-july-2022',
+                'ext': 'webm',
+                'like_count': int,
+                'title': 'Absolute Fiction - 23rd July 2022',
+                'comment_count': int,
+                'uploader_url': 'https://www.mixcloud.com/NTSRadio/',
+                'description': 'md5:a8cb9d9f040adb6525d33a6604fc759c',
+                'tags': [],
+                'duration': 3529,
+                'timestamp': 1658772398,
+                'repost_count': int,
+                'upload_date': '20220725',
+                'uploader_id': 'NTSRadio',
+                'thumbnail': 'https://thumbnailer.mixcloud.com/unsafe/1024x1024/extaudio/5/1/a/d/ae3e-1be9-4fd4-983e-9c3294226eac',
+                'uploader': 'Mixcloud NTS Radio',
+            }
+        }
+    ]
+
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
+        webpage = self._download_webpage(url, video_id)
+
+        attrs = extract_attributes(self._search_regex(
+            r'(<button[^>]+?aria-label="Play"[^>]*?>)', webpage, 'player'))
+
+        return self.url_result(attrs['data-src'])

yt_dlp/utils/_utils.py

@@ -1638,16 +1638,14 @@ def get_filesystem_encoding():
     return encoding if encoding is not None else 'utf-8'
 
 
-_WINDOWS_QUOTE_TRANS = str.maketrans({'"': '\\"', '\\': '\\\\'})
+_WINDOWS_QUOTE_TRANS = str.maketrans({'"': R'\"'})
 _CMD_QUOTE_TRANS = str.maketrans({
     # Keep quotes balanced by replacing them with `""` instead of `\\"`
     '"': '""',
-    # Requires a variable `=` containing `"^\n\n"` (set in `utils.Popen`)
+    # These require an env-variable `=` containing `"^\n\n"` (set in `utils.Popen`)
     # `=` should be unique since variables containing `=` cannot be set using cmd
     '\n': '%=%',
-    # While we are only required to escape backslashes immediately before quotes,
-    # we instead escape all of 'em anyways to be consistent
-    '\\': '\\\\',
+    '\r': '%=%',
     # Use zero length variable replacement so `%` doesn't get expanded
     # `cd` is always set as long as extensions are enabled (`/E:ON` in `utils.Popen`)
     '%': '%%cd:~,%',
@@ -1656,19 +1654,14 @@ _CMD_QUOTE_TRANS = str.maketrans({
 
 def shell_quote(args, *, shell=False):
     args = list(variadic(args))
-    if any(isinstance(item, bytes) for item in args):
-        deprecation_warning('Passing bytes to utils.shell_quote is deprecated')
-        encoding = get_filesystem_encoding()
-        for index, item in enumerate(args):
-            if isinstance(item, bytes):
-                args[index] = item.decode(encoding)
 
     if compat_os_name != 'nt':
         return shlex.join(args)
 
     trans = _CMD_QUOTE_TRANS if shell else _WINDOWS_QUOTE_TRANS
     return ' '.join(
-        s if re.fullmatch(r'[\w#$*\-+./:?@\\]+', s, re.ASCII) else s.translate(trans).join('""')
+        s if re.fullmatch(r'[\w#$*\-+./:?@\\]+', s, re.ASCII)
+        else re.sub(r'(\\+)("|$)', r'\1\1\2', s).translate(trans).join('""')
         for s in args)