[jsinterp] Fix nested attributes and object extraction (#12760)

Authored by: bashonly, seproDev

Co-authored-by: sepro <sepro@sepr0.com>

test/test_jsinterp.py

@@ -118,6 +118,7 @@ class TestJSInterpreter(unittest.TestCase):
         self._test('function f(){var x = 20; x = 30 + 1; return x;}', 31)
         self._test('function f(){var x = 20; x += 30 + 1; return x;}', 51)
         self._test('function f(){var x = 20; x -= 30 + 1; return x;}', -11)
+        self._test('function f(){var x = 2; var y = ["a", "b"]; y[x%y["length"]]="z"; return y}', ['z', 'b'])
 
     @unittest.skip('Not implemented')
     def test_comments(self):
@@ -403,6 +404,8 @@ class TestJSInterpreter(unittest.TestCase):
         test_result = list('test')
         tests = [
             'function f(a, b){return a.split(b)}',
+            'function f(a, b){return a["split"](b)}',
+            'function f(a, b){let x = ["split"]; return a[x[0]](b)}',
             'function f(a, b){return String.prototype.split.call(a, b)}',
             'function f(a, b){return String.prototype.split.apply(a, [b])}',
         ]
@@ -441,6 +444,9 @@ class TestJSInterpreter(unittest.TestCase):
         self._test('function f(){return "012345678".slice(-1, 1)}', '')
         self._test('function f(){return "012345678".slice(-3, -1)}', '67')
 
+    def test_splice(self):
+        self._test('function f(){var T = ["0", "1", "2"]; T["splice"](2, 1, "0")[0]; return T }', ['0', '1', '0'])
+
     def test_js_number_to_string(self):
         for test, radix, expected in [
             (0, None, '0'),

yt_dlp/jsinterp.py

@@ -188,6 +188,7 @@ _COMP_OPERATORS = {'===', '!==', '==', '!=', '<=', '>=', '<', '>'}
 _NAME_RE = r'[a-zA-Z_$][\w$]*'
 _MATCHING_PARENS = dict(zip(*zip('()', '{}', '[]')))
 _QUOTES = '\'"/'
+_NESTED_BRACKETS = r'[^[\]]+(?:\[[^[\]]+(?:\[[^\]]+\])?\])?'
 
 
 class JS_Undefined:
@@ -606,15 +607,18 @@ class JSInterpreter:
 
         m = re.match(fr'''(?x)
             (?P<assign>
-                (?P<out>{_NAME_RE})(?:\[(?P<index>[^\]]+?)\])?\s*
+                (?P<out>{_NAME_RE})(?:\[(?P<index>{_NESTED_BRACKETS})\])?\s*
                 (?P<op>{"|".join(map(re.escape, set(_OPERATORS) - _COMP_OPERATORS))})?
                 =(?!=)(?P<expr>.*)$
             )|(?P<return>
                 (?!if|return|true|false|null|undefined|NaN)(?P<name>{_NAME_RE})$
+            )|(?P<attribute>
+                (?P<var>{_NAME_RE})(?:
+                    (?P<nullish>\?)?\.(?P<member>[^(]+)|
+                    \[(?P<member2>{_NESTED_BRACKETS})\]
+                )\s*
             )|(?P<indexing>
                 (?P<in>{_NAME_RE})\[(?P<idx>.+)\]$
-            )|(?P<attribute>
-                (?P<var>{_NAME_RE})(?:(?P<nullish>\?)?\.(?P<member>[^(]+)|\[(?P<member2>[^\]]+)\])\s*
             )|(?P<function>
                 (?P<fname>{_NAME_RE})\((?P<args>.*)\)$
             )''', expr)
@@ -707,7 +711,7 @@ class JSInterpreter:
                 if obj is NO_DEFAULT:
                     if variable not in self._objects:
                         try:
-                            self._objects[variable] = self.extract_object(variable)
+                            self._objects[variable] = self.extract_object(variable, local_vars)
                         except self.Exception:
                             if not nullish:
                                 raise
@@ -847,7 +851,7 @@ class JSInterpreter:
             raise self.Exception('Cannot return from an expression', expr)
         return ret
 
-    def extract_object(self, objname):
+    def extract_object(self, objname, *global_stack):
         _FUNC_NAME_RE = r'''(?:[a-zA-Z$0-9]+|"[a-zA-Z$0-9]+"|'[a-zA-Z$0-9]+')'''
         obj = {}
         obj_m = re.search(
@@ -869,7 +873,8 @@ class JSInterpreter:
         for f in fields_m:
             argnames = f.group('args').split(',')
             name = remove_quotes(f.group('key'))
-            obj[name] = function_with_repr(self.build_function(argnames, f.group('code')), f'F<{name}>')
+            obj[name] = function_with_repr(
+                self.build_function(argnames, f.group('code'), *global_stack), f'F<{name}>')
 
         return obj