Again but better

* fix signature code extraction
* raise if n function returns input value
* add new tests from yt-dlp

Co-authored-by: bashonly

test/test_jsinterp.py

@@ -516,11 +516,10 @@ class TestJSInterpreter(unittest.TestCase):
                    ['t', 'e', 's', 't'])
         self._test('function f(){return "t-e-s-t".split(/[es-]+/)}',
                    ['t', 't'])
-        # from MDN: surrogate pairs aren't handled: case 1 fails
+        # from MDN: surrogate pairs aren't handled: case 1 fails, and case 2 beats Py3.2
-        # self._test('function f(){return "😄😄".split(/(?:)/)}',
-        #            ['\ud83d', '\ude04', '\ud83d', '\ude04'])
-        # case 2 beats Py3.2: it gets the case 1 result
         if sys.version_info >= (2, 6) and not ((3, 0) <= sys.version_info < (3, 3)):
+            # self._test('function f(){return "😄😄".split(/(?:)/)}',
+            #            ['\ud83d', '\ude04', '\ud83d', '\ude04'])
             self._test('function f(){return "😄😄".split(/(?:)/u)}',
                        ['😄', '😄'])
 

test/test_youtube_signature.py

@@ -206,10 +206,6 @@ _NSIG_TESTS = [
         'https://www.youtube.com/s/player/3bb1f723/player_ias.vflset/en_US/base.js',
         'gK15nzVyaXE9RsMP3z', 'ZFFWFLPWx9DEgQ',
     ),
-    (
-        'https://www.youtube.com/s/player/f8f53e1a/player_ias.vflset/en_US/base.js',
-        'VTQOUOv0mCIeJ7i8kZB', 'kcfD8wy0sNLyNQ',
-    ),
 ]
 
 

youtube_dl/extractor/youtube.py

@@ -1665,29 +1665,37 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
 
     def _extract_n_function_name(self, jscode):
         func_name, idx = self._search_regex(
-            # (y=NuD(),Mw(k),q=k.Z[y]||null)&&(q=narray[idx](q),k.set(y,q),k.V||NuD(''))}};
             # (R="nn"[+J.Z],mW(J),N=J.K[R]||null)&&(N=narray[idx](N),J.set(R,N))}};
-            # or:  (b=String.fromCharCode(110),c=a.get(b))&&c=narray[idx](c)
+            # new: (b=String.fromCharCode(110),c=a.get(b))&&c=nfunc[idx](c)
-            # or:  (b="nn"[+a.D],c=a.get(b))&&(c=narray[idx](c)
+            # or:  (b="nn"[+a.D],c=a.get(b))&&(c=nfunc[idx](c)
-            # or:  (PL(a),b=a.j.n||null)&&(b=narray[idx](b)
+            # or:  (PL(a),b=a.j.n||null)&&(b=nfunc[idx](b)
             # or:  (b="nn"[+a.D],vL(a),c=a.j[b]||null)&&(c=narray[idx](c),a.set(b,c),narray.length||nfunc("")
-            # old: (b=a.get("n"))&&(b=narray[idx](b)(?P<c>[a-z])\s*=\s*[a-z]\s*
+            # old: (b=a.get("n"))&&(b=nfunc[idx](b)(?P<c>[a-z])\s*=\s*[a-z]\s*
             # older: (b=a.get("n"))&&(b=nfunc(b)
             r'''(?x)
-                # (expr, ...,
+                \((?:[\w$()\s]+,)*?\s*      # (
-                \((?:(?:\s*[\w$]+\s*=)?(?:[\w$"+\.\s(\[]+(?:[)\]]\s*)?),)*
+                (?P<b>[a-zA-Z])\s*=\s*      # b=, R=
-                  # b=...
+                (?:
-                  (?P<b>[\w$]+)\s*=\s*(?!(?P=b)[^\w$])[\w$]+\s*(?:(?:
+                    (?:                     # expect ,c=a.get(b) (etc)
-                    \.\s*[\w$]+ |
+                        String\s*\.\s*fromCharCode\s*\(\s*110\s*\)|
-                    \[\s*[\w$]+\s*\] |
+                        "n+"\[\s*\+?s*[\w$.]+\s*]
-                    \.\s*get\s*\(\s*[\w$"]+\s*\)
+                    )\s*(?:,[\w$()\s]+(?=,))*|
-                  )\s*){,2}(?:\s*\|\|\s*null(?=\s*\)))?\s*
+                       (?P<old>[\w$]+)      # a (old[er])
-                \)\s*&&\s*\(        # ...)&&(
+                   )\s*
-                # b = nfunc, b = narray[idx]
+                   (?(old)
-                (?P=b)\s*=\s*(?P<nfunc>[\w$]+)\s*
+                                            # b.get("n")
-                    (?:\[\s*(?P<idx>[\w$]+)\s*\]\s*)?
+                       (?:\.\s*[\w$]+\s*|\[\s*[\w$]+\s*]\s*)*?
-                    # (...)
+                       (?:\.\s*n|\[\s*"n"\s*]|\.\s*get\s*\(\s*"n"\s*\))
-                    \(\s*[\w$]+\s*\)
+                       |                    # ,c=a.get(b)
+                       ,\s*(?P<c>[a-zA-Z])\s*=\s*[a-zA-Z]\s*
+                       (?:\.\s*[\w$]+\s*|\[\s*[\w$]+\s*]\s*)*?
+                       (?:\[\s*(?P=b)\s*]|\.\s*get\s*\(\s*(?P=b)\s*\))
+                   )
+                                            # interstitial junk
+                   \s*(?:\|\|\s*null\s*)?(?:\)\s*)?&&\s*(?:\(\s*)?
+               (?(c)(?P=c)|(?P=b))\s*=\s*   # [c|b]=
+                                            # nfunc|nfunc[idx]
+                   (?P<nfunc>[a-zA-Z_$][\w$]*)(?:\s*\[(?P<idx>\d+)\])?\s*\(\s*[\w$]+\s*\)
             ''', jscode, 'Initial JS player n function name', group=('nfunc', 'idx'),
             default=(None, None))
         # thx bashonly: yt-dlp/yt-dlp/pull/10611