Merge commit 'ed5c44e7b74ac77f87ca5ed6cb5e964a0c6a0678'

* commit 'ed5c44e7b74ac77f87ca5ed6cb5e964a0c6a0678':
  [compat] Replace deficient ChainMap class in Py3.3 and earlier * fix version check
  [compat] Replace deficient ChainMap class in Py3.3 and earlier
  [jsinterp] Improve try/catch/finally support
  [jsinterp] Fix bug in operator precedence * from yt-dlp/yt-dlp@164b03c * added tests
  [YouTube] Improve error check for n-sig processing
  [core] Avoid processing empty format list after removing bad formats * also ensure compat encoding of error strings
  [utils] Ensure RFC3986 encoding result is unicode
  [infoq] Avoid crash if the page has no `mp3Form`
  [uktvplay] Support domain without .uktv
  [jsinterp] Clean up and pull yt-dlp style * add compat_re_Pattern * improve compat_collections_chain_map * use class JS_Undefined * remove unused code
  [jsinterp] Handle regexp literals and throw/catch execution (ytdl-org#31182)
  [jsinterp] Improve JS language support (ytdl-org#31175)

test/test_jsinterp.py

@@ -8,7 +8,12 @@
 import unittest
 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
 
-from youtube_dl.jsinterp import JSInterpreter
+import math
+import re
+
+from youtube_dl.compat import compat_re_Pattern
+
+from youtube_dl.jsinterp import JS_Undefined, JSInterpreter
 
 
 class TestJSInterpreter(unittest.TestCase):
@@ -48,6 +53,9 @@ def test_operators(self):
         jsi = JSInterpreter('function f(){return 1 << 5;}')
         self.assertEqual(jsi.call_function('f'), 32)
 
+        jsi = JSInterpreter('function f(){return 2 ** 5}')
+        self.assertEqual(jsi.call_function('f'), 32)
+
         jsi = JSInterpreter('function f(){return 19 & 21;}')
         self.assertEqual(jsi.call_function('f'), 17)
 
@@ -57,6 +65,18 @@ def test_operators(self):
         jsi = JSInterpreter('function f(){return []? 2+3: 4;}')
         self.assertEqual(jsi.call_function('f'), 5)
 
+        jsi = JSInterpreter('function f(){return 1 == 2}')
+        self.assertEqual(jsi.call_function('f'), False)
+
+        jsi = JSInterpreter('function f(){return 0 && 1 || 2;}')
+        self.assertEqual(jsi.call_function('f'), 2)
+
+        jsi = JSInterpreter('function f(){return 0 ?? 42;}')
+        self.assertEqual(jsi.call_function('f'), 0)
+
+        jsi = JSInterpreter('function f(){return "life, the universe and everything" < 42;}')
+        self.assertFalse(jsi.call_function('f'))
+
     def test_array_access(self):
         jsi = JSInterpreter('function f(){var x = [1,2,3]; x[0] = 4; x[0] = 5; x[2.0] = 7; return x;}')
         self.assertEqual(jsi.call_function('f'), [5, 2, 7])
@@ -175,6 +195,30 @@ def test_try(self):
         ''')
         self.assertEqual(jsi.call_function('x'), 10)
 
+    def test_catch(self):
+        jsi = JSInterpreter('''
+        function x() { try{throw 10} catch(e){return 5} }
+        ''')
+        self.assertEqual(jsi.call_function('x'), 5)
+
+    def test_finally(self):
+        jsi = JSInterpreter('''
+        function x() { try{throw 10} finally {return 42} }
+        ''')
+        self.assertEqual(jsi.call_function('x'), 42)
+        jsi = JSInterpreter('''
+        function x() { try{throw 10} catch(e){return 5} finally {return 42} }
+        ''')
+        self.assertEqual(jsi.call_function('x'), 42)
+
+    def test_nested_try(self):
+        jsi = JSInterpreter('''
+        function x() {try {
+            try{throw 10} finally {throw 42}
+            } catch(e){return 5} }
+        ''')
+        self.assertEqual(jsi.call_function('x'), 5)
+
     def test_for_loop_continue(self):
         jsi = JSInterpreter('''
         function x() { a=0; for (i=0; i-10; i++) { continue; a++ } return a }
@@ -187,6 +231,14 @@ def test_for_loop_break(self):
         ''')
         self.assertEqual(jsi.call_function('x'), 0)
 
+    def test_for_loop_try(self):
+        jsi = JSInterpreter('''
+        function x() {
+            for (i=0; i-10; i++) { try { if (i == 5) throw i} catch {return 10} finally {break} };
+            return 42 }
+        ''')
+        self.assertEqual(jsi.call_function('x'), 42)
+
     def test_literal_list(self):
         jsi = JSInterpreter('''
         function x() { return [1, 2, "asdf", [5, 6, 7]][3] }
@@ -203,6 +255,11 @@ def test_comma(self):
         ''')
         self.assertEqual(jsi.call_function('x'), 7)
 
+        jsi = JSInterpreter('''
+        function x() { return (l=[0,1,2,3], function(a, b){return a+b})((l[1], l[2]), l[3]) }
+        ''')
+        self.assertEqual(jsi.call_function('x'), 5)
+
     def test_void(self):
         jsi = JSInterpreter('''
         function x() { return void 42; }
@@ -215,6 +272,119 @@ def test_return_function(self):
         ''')
         self.assertEqual(jsi.call_function('x')([]), 1)
 
+    def test_null(self):
+        jsi = JSInterpreter('''
+        function x() { return null; }
+        ''')
+        self.assertIs(jsi.call_function('x'), None)
+
+        jsi = JSInterpreter('''
+        function x() { return [null > 0, null < 0, null == 0, null === 0]; }
+        ''')
+        self.assertEqual(jsi.call_function('x'), [False, False, False, False])
+
+        jsi = JSInterpreter('''
+        function x() { return [null >= 0, null <= 0]; }
+        ''')
+        self.assertEqual(jsi.call_function('x'), [True, True])
+
+    def test_undefined(self):
+        jsi = JSInterpreter('''
+        function x() { return undefined === undefined; }
+        ''')
+        self.assertTrue(jsi.call_function('x'))
+
+        jsi = JSInterpreter('''
+        function x() { return undefined; }
+        ''')
+        self.assertIs(jsi.call_function('x'), JS_Undefined)
+
+        jsi = JSInterpreter('''
+        function x() { let v; return v; }
+        ''')
+        self.assertIs(jsi.call_function('x'), JS_Undefined)
+
+        jsi = JSInterpreter('''
+        function x() { return [undefined === undefined, undefined == undefined, undefined < undefined, undefined > undefined]; }
+        ''')
+        self.assertEqual(jsi.call_function('x'), [True, True, False, False])
+
+        jsi = JSInterpreter('''
+        function x() { return [undefined === 0, undefined == 0, undefined < 0, undefined > 0]; }
+        ''')
+        self.assertEqual(jsi.call_function('x'), [False, False, False, False])
+
+        jsi = JSInterpreter('''
+        function x() { return [undefined >= 0, undefined <= 0]; }
+        ''')
+        self.assertEqual(jsi.call_function('x'), [False, False])
+
+        jsi = JSInterpreter('''
+        function x() { return [undefined > null, undefined < null, undefined == null, undefined === null]; }
+        ''')
+        self.assertEqual(jsi.call_function('x'), [False, False, True, False])
+
+        jsi = JSInterpreter('''
+        function x() { return [undefined === null, undefined == null, undefined < null, undefined > null]; }
+        ''')
+        self.assertEqual(jsi.call_function('x'), [False, True, False, False])
+
+        jsi = JSInterpreter('''
+        function x() { let v; return [42+v, v+42, v**42, 42**v, 0**v]; }
+        ''')
+        for y in jsi.call_function('x'):
+            self.assertTrue(math.isnan(y))
+
+        jsi = JSInterpreter('''
+        function x() { let v; return v**0; }
+        ''')
+        self.assertEqual(jsi.call_function('x'), 1)
+
+        jsi = JSInterpreter('''
+        function x() { let v; return [v>42, v<=42, v&&42, 42&&v]; }
+        ''')
+        self.assertEqual(jsi.call_function('x'), [False, False, JS_Undefined, JS_Undefined])
+
+        jsi = JSInterpreter('function x(){return undefined ?? 42; }')
+        self.assertEqual(jsi.call_function('x'), 42)
+
+    def test_object(self):
+        jsi = JSInterpreter('''
+        function x() { return {}; }
+        ''')
+        self.assertEqual(jsi.call_function('x'), {})
+
+        jsi = JSInterpreter('''
+        function x() { let a = {m1: 42, m2: 0 }; return [a["m1"], a.m2]; }
+        ''')
+        self.assertEqual(jsi.call_function('x'), [42, 0])
+
+        jsi = JSInterpreter('''
+        function x() { let a; return a?.qq; }
+        ''')
+        self.assertIs(jsi.call_function('x'), JS_Undefined)
+
+        jsi = JSInterpreter('''
+        function x() { let a = {m1: 42, m2: 0 }; return a?.qq; }
+        ''')
+        self.assertIs(jsi.call_function('x'), JS_Undefined)
+
+    def test_regex(self):
+        jsi = JSInterpreter('''
+        function x() { let a=/,,[/,913,/](,)}/; }
+        ''')
+        self.assertIs(jsi.call_function('x'), None)
+
+        jsi = JSInterpreter('''
+        function x() { let a=/,,[/,913,/](,)}/; return a; }
+        ''')
+        self.assertIsInstance(jsi.call_function('x'), compat_re_Pattern)
+
+        jsi = JSInterpreter('''
+        function x() { let a=/,,[/,913,/](,)}/i; return a; }
+        ''')
+        self.assertEqual(jsi.call_function('x').flags & ~re.U, re.I)
+
 
 if __name__ == '__main__':
     unittest.main()

test/test_youtube_signature.py

@@ -12,10 +12,11 @@
 import re
 import string
 
+from youtube_dl.compat import compat_str, compat_urlretrieve
+
 from test.helper import FakeYDL
 from youtube_dl.extractor import YoutubeIE
 from youtube_dl.jsinterp import JSInterpreter
-from youtube_dl.compat import compat_str, compat_urlretrieve
 
 _SIG_TESTS = [
     (
@@ -102,6 +103,18 @@
         'https://www.youtube.com/s/player/4c3f79c5/player_ias.vflset/en_US/base.js',
         'TDCstCG66tEAO5pR9o', 'dbxNtZ14c-yWyw',
     ),
+    (
+        'https://www.youtube.com/s/player/c81bbb4a/player_ias.vflset/en_US/base.js',
+        'gre3EcLurNY2vqp94', 'Z9DfGxWP115WTg',
+    ),
+    (
+        'https://www.youtube.com/s/player/1f7d5369/player_ias.vflset/en_US/base.js',
+        'batNX7sYqIJdkJ', 'IhOkL_zxbkOZBw',
+    ),
+    (
+        'https://www.youtube.com/s/player/dc0c6770/player_ias.vflset/en_US/base.js',
+        '5EHDMgYLV6HPGk_Mu-kk', 'n9lUJLHbxUI0GQ',
+    ),
 ]
 
 

youtube_dl/YoutubeDL.py

@@ -721,7 +721,7 @@ def prepare_filename(self, info_dict):
                 filename = encodeFilename(filename, True).decode(preferredencoding())
             return sanitize_path(filename)
         except ValueError as err:
-            self.report_error('Error in output template: ' + str(err) + ' (encoding: ' + repr(preferredencoding()) + ')')
+            self.report_error('Error in output template: ' + error_to_compat_str(err) + ' (encoding: ' + repr(preferredencoding()) + ')')
             return None
 
     def _match_entry(self, info_dict, incomplete):
@@ -1570,9 +1570,6 @@ def sanitize_numeric_fields(info):
         else:
             formats = info_dict['formats']
 
-        if not formats:
-            raise ExtractorError('No video formats found!')
-
         def is_wellformed(f):
             url = f.get('url')
             if not url:
@@ -1585,7 +1582,10 @@ def is_wellformed(f):
             return True
 
         # Filter out malformed formats for better extraction robustness
-        formats = list(filter(is_wellformed, formats))
+        formats = list(filter(is_wellformed, formats or []))
+
+        if not formats:
+            raise ExtractorError('No video formats found!')
 
         formats_dict = {}
 
@@ -2085,7 +2085,7 @@ def compatible_formats(formats):
                 try:
                     self.post_process(filename, info_dict)
                 except (PostProcessingError) as err:
-                    self.report_error('postprocessing: %s' % str(err))
+                    self.report_error('postprocessing: %s' % error_to_compat_str(err))
                     return
                 self.record_download_archive(info_dict)
                 # avoid possible nugatory search for further items (PR #26638)

youtube_dl/compat.py

@@ -3007,8 +3007,11 @@ def unpack(self, string):
 # new class in collections
 try:
     from collections import ChainMap as compat_collections_chain_map
+    # Py3.3's ChainMap is deficient
+    if sys.version_info < (3, 4):
+        raise ImportError
 except ImportError:
-    # Py < 3.3
+    # Py <= 3.3
     class compat_collections_chain_map(compat_collections_abc.MutableMapping):
 
         maps = [{}]
@@ -3026,18 +3029,34 @@ def __setitem__(self, k, v):
             self.maps[0].__setitem__(k, v)
             return
 
-        def __delitem__(self, k):
+        def __contains__(self, k):
+            return any((k in m) for m in self.maps)
+
+        def __delitem(self, k):
             if k in self.maps[0]:
                 del self.maps[0][k]
                 return
             raise KeyError(k)
 
+        def __delitem__(self, k):
+            self.__delitem(k)
+
         def __iter__(self):
             return itertools.chain(*reversed(self.maps))
 
         def __len__(self):
             return len(iter(self))
 
+        # to match Py3, don't del directly
+        def pop(self, k, *args):
+            if self.__contains__(k):
+                off = self.__getitem__(k)
+                self.__delitem(k)
+                return off
+            elif len(args) > 0:
+                return args[0]
+            raise KeyError(k)
+
         def new_child(self, m=None, **kwargs):
             m = m or {}
             m.update(kwargs)
@@ -3048,6 +3067,9 @@ def parents(self):
             return compat_collections_chain_map(*(self.maps[1:]))
 
 
+# Pythons disagree on the type of a pattern (RegexObject, _sre.SRE_Pattern, Pattern, ...?)
+compat_re_Pattern = type(re.compile(''))
+
 if sys.version_info < (3, 3):
     def compat_b64decode(s, *args, **kwargs):
         if isinstance(s, compat_str):
@@ -3113,6 +3135,7 @@ def compat_ctypes_WINFUNCTYPE(*args, **kwargs):
     'compat_os_name',
     'compat_parse_qs',
     'compat_print',
+    'compat_re_Pattern',
     'compat_realpath',
     'compat_setenv',
     'compat_shlex_quote',

youtube_dl/extractor/infoq.py

@@ -1,6 +1,9 @@
 # coding: utf-8
 
 from __future__ import unicode_literals
+from ..utils import (
+    ExtractorError,
+)
 
 from ..compat import (
     compat_b64decode,
@@ -90,7 +93,11 @@ def _extract_http_video(self, webpage):
         }]
 
     def _extract_http_audio(self, webpage, video_id):
-        fields = self._form_hidden_inputs('mp3Form', webpage)
+        try:
+            fields = self._form_hidden_inputs('mp3Form', webpage)
+        except ExtractorError:
+            fields = {}
+
         http_audio_url = fields.get('filename')
         if not http_audio_url:
             return []

youtube_dl/extractor/uktvplay.py

@@ -5,7 +5,7 @@
 
 
 class UKTVPlayIE(InfoExtractor):
-    _VALID_URL = r'https?://uktvplay\.uktv\.co\.uk/(?:.+?\?.*?\bvideo=|([^/]+/)*watch-online/)(?P<id>\d+)'
+    _VALID_URL = r'https?://uktvplay\.(?:uktv\.)?co\.uk/(?:.+?\?.*?\bvideo=|([^/]+/)*watch-online/)(?P<id>\d+)'
     _TESTS = [{
         'url': 'https://uktvplay.uktv.co.uk/shows/world-at-war/c/200/watch-online/?video=2117008346001',
         'info_dict': {