Validate and escape values in --write-link output

See https://github.com/yt-dlp/yt-dlp/security/advisories/GHSA-6v4j-43gg-vj32

Authored by: bashonly

test/test_utils.py

@@ -134,7 +134,10 @@ from yt_dlp.utils import (
     xpath_text,
     xpath_with_ns,
 )
-from yt_dlp.utils._utils import _UnsafeExtensionError
+from yt_dlp.utils._utils import (
+    _UnsafeExtensionError,
+    _desktop_entry_localestring,
+)
 from yt_dlp.utils.networking import (
     HTTPHeaderDict,
     escape_rfc3986,
@@ -1948,6 +1951,27 @@ Line 1
         self.assertEqual(
             iri_to_uri('http://导航.中国/'),
             'http://xn--fet810g.xn--fiqs8s/')
+        self.assertEqual(
+            iri_to_uri('file://example.org/run.exe', allowed_schemes=('file',)),
+            'file://example.org/run.exe')
+        self.assertRaises(ValueError, iri_to_uri, 'file://example.org/run.exe')
+
+    def test_desktop_entry_localestring(self):
+        self.assertEqual(
+            _desktop_entry_localestring('A B'),
+            'A\\sB')
+        self.assertEqual(
+            _desktop_entry_localestring('A\nB'),
+            'A\\nB')
+        self.assertEqual(
+            _desktop_entry_localestring('A\tB'),
+            'A\\tB')
+        self.assertEqual(
+            _desktop_entry_localestring('A\rB'),
+            'A\\rB')
+        self.assertEqual(
+            _desktop_entry_localestring('A\\B'),
+            'A\\\\B')
 
     def test_clean_podcast_url(self):
         self.assertEqual(clean_podcast_url('https://www.podtrac.com/pts/redirect.mp3/chtbl.com/track/5899E/traffic.megaphone.fm/HSW7835899191.mp3'), 'https://traffic.megaphone.fm/HSW7835899191.mp3')

yt_dlp/YoutubeDL.py

@@ -170,7 +170,12 @@ from .utils import (
     write_json_file,
     write_string,
 )
-from .utils._utils import _UnsafeExtensionError, _YDLLogger, _ProgressState
+from .utils._utils import (
+    _UnsafeExtensionError,
+    _YDLLogger,
+    _ProgressState,
+    _desktop_entry_localestring,
+)
 from .utils.networking import (
     HTTPHeaderDict,
     clean_headers,
@@ -3406,10 +3411,13 @@ class YoutubeDL:
 
         # Write internet shortcut files
         def _write_link_file(link_type):
+            # iri_to_uri converts to ascii, percent-escapes unsafe characters & validates scheme
+            # See https://github.com/yt-dlp/yt-dlp/security/advisories/GHSA-6v4j-43gg-vj32
             url = try_get(info_dict['webpage_url'], iri_to_uri)
             if not url:
                 self.report_warning(
-                    f'Cannot write internet shortcut file because the actual URL of "{info_dict["webpage_url"]}" is unknown')
+                    f'Cannot write internet shortcut file because the actual URL '
+                    f'of "{info_dict["webpage_url"]}" is unknown or disallowed')
                 return True
             linkfn = replace_extension(
                 self.prepare_filename(info_dict, 'link'), link_type,
@@ -3425,7 +3433,8 @@ class YoutubeDL:
                           newline='\r\n' if link_type == 'url' else '\n') as linkfile:
                     template_vars = {'url': url}
                     if link_type == 'desktop':
-                        template_vars['filename'] = linkfn[:-(len(link_type) + 1)]
+                        # See https://github.com/yt-dlp/yt-dlp/security/advisories/GHSA-6v4j-43gg-vj32
+                        template_vars['filename'] = _desktop_entry_localestring(linkfn[:-(len(link_type) + 1)])
                     linkfile.write(LINK_TEMPLATES[link_type] % template_vars)
             except OSError:
                 self.report_error(f'Cannot write internet shortcut {linkfn}')

yt_dlp/utils/_utils.py

@@ -4631,8 +4631,21 @@ LINK_TEMPLATES = {
     'webloc': DOT_WEBLOC_LINK_TEMPLATE,
 }
 
+# Ref: https://specifications.freedesktop.org/desktop-entry/latest/value-types.html
+_DESKTOP_ENTRY_TRANS = str.maketrans({
+    ' ': R'\s',
+    '\n': R'\n',
+    '\t': R'\t',
+    '\r': R'\r',
+    '\\': R'\\',
+})
 
-def iri_to_uri(iri):
+
+def _desktop_entry_localestring(s):
+    return s.translate(_DESKTOP_ENTRY_TRANS)
+
+
+def iri_to_uri(iri, *, allowed_schemes=('http', 'https')):
     """
     Converts an IRI (Internationalized Resource Identifier, allowing Unicode characters) to a URI (Uniform Resource Identifier, ASCII-only).
 
@@ -4641,6 +4654,9 @@ def iri_to_uri(iri):
 
     iri_parts = urllib.parse.urlparse(iri)
 
+    if iri_parts.scheme not in allowed_schemes:
+        raise ValueError(f'"{iri_parts.scheme}" is not in allowed_schemes: {", ".join(allowed_schemes)}')
+
     if '[' in iri_parts.netloc:
         raise ValueError('IPv6 URIs are not, yet, supported.')
         # Querying `.netloc`, when there's only one bracket, also raises a ValueError.