[jsinterp] Fix bitwise operations (#15985)

Authored by: bashonly

test/test_jsinterp.py

@@ -9,7 +9,12 @@ sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
 
 import math
 
-from yt_dlp.jsinterp import JS_Undefined, JSInterpreter, js_number_to_string
+from yt_dlp.jsinterp import (
+    JS_Undefined,
+    JSInterpreter,
+    int_to_int32,
+    js_number_to_string,
+)
 
 
 class NaN:
@@ -101,8 +106,8 @@ class TestJSInterpreter(unittest.TestCase):
         self._test('function f(){return 5 ^ 9;}', 12)
         self._test('function f(){return 0.0 << NaN}', 0)
         self._test('function f(){return null << undefined}', 0)
-        # TODO: Does not work due to number too large
-        # self._test('function f(){return 21 << 4294967297}', 42)
+        self._test('function f(){return -12616 ^ 5041}', -8951)
+        self._test('function f(){return 21 << 4294967297}', 42)
 
     def test_array_access(self):
         self._test('function f(){var x = [1,2,3]; x[0] = 4; x[0] = 5; x[2.0] = 7; return x;}', [5, 2, 7])
@@ -447,6 +452,22 @@ class TestJSInterpreter(unittest.TestCase):
     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_int_to_int32(self):
+        for inp, exp in [
+            (0, 0),
+            (1, 1),
+            (-1, -1),
+            (-8951, -8951),
+            (2147483647, 2147483647),
+            (2147483648, -2147483648),
+            (2147483649, -2147483647),
+            (-2147483649, 2147483647),
+            (-2147483648, -2147483648),
+            (-16799986688, 379882496),
+            (39570129568, 915423904),
+        ]:
+            assert int_to_int32(inp) == exp
+
     def test_js_number_to_string(self):
         for test, radix, expected in [
             (0, None, '0'),

yt_dlp/extractor/xhamster.py

@@ -3,6 +3,7 @@ import re
 import urllib.parse
 
 from .common import InfoExtractor
+from ..jsinterp import int_to_int32
 from ..utils import (
     ExtractorError,
     clean_html,
@@ -20,73 +21,69 @@ from ..utils import (
 )
 
 
-def to_signed_32(n):
-    return n % ((-1 if n < 0 else 1) * 2**32)
-
-
 class _ByteGenerator:
     def __init__(self, algo_id, seed):
         try:
             self._algorithm = getattr(self, f'_algo{algo_id}')
         except AttributeError:
             raise ExtractorError(f'Unknown algorithm ID "{algo_id}"')
-        self._s = to_signed_32(seed)
+        self._s = int_to_int32(seed)
 
     def _algo1(self, s):
         # LCG (a=1664525, c=1013904223, m=2^32)
         # Ref: https://en.wikipedia.org/wiki/Linear_congruential_generator
-        s = self._s = to_signed_32(s * 1664525 + 1013904223)
+        s = self._s = int_to_int32(s * 1664525 + 1013904223)
         return s
 
     def _algo2(self, s):
         # xorshift32
         # Ref: https://en.wikipedia.org/wiki/Xorshift
-        s = to_signed_32(s ^ (s << 13))
-        s = to_signed_32(s ^ ((s & 0xFFFFFFFF) >> 17))
-        s = self._s = to_signed_32(s ^ (s << 5))
+        s = int_to_int32(s ^ (s << 13))
+        s = int_to_int32(s ^ ((s & 0xFFFFFFFF) >> 17))
+        s = self._s = int_to_int32(s ^ (s << 5))
         return s
 
     def _algo3(self, s):
         # Weyl Sequence (k≈2^32*φ, m=2^32) + MurmurHash3 (fmix32)
         # Ref: https://en.wikipedia.org/wiki/Weyl_sequence
         # https://commons.apache.org/proper/commons-codec/jacoco/org.apache.commons.codec.digest/MurmurHash3.java.html
-        s = self._s = to_signed_32(s + 0x9e3779b9)
-        s = to_signed_32(s ^ ((s & 0xFFFFFFFF) >> 16))
-        s = to_signed_32(s * to_signed_32(0x85ebca77))
-        s = to_signed_32(s ^ ((s & 0xFFFFFFFF) >> 13))
-        s = to_signed_32(s * to_signed_32(0xc2b2ae3d))
-        return to_signed_32(s ^ ((s & 0xFFFFFFFF) >> 16))
+        s = self._s = int_to_int32(s + 0x9e3779b9)
+        s = int_to_int32(s ^ ((s & 0xFFFFFFFF) >> 16))
+        s = int_to_int32(s * int_to_int32(0x85ebca77))
+        s = int_to_int32(s ^ ((s & 0xFFFFFFFF) >> 13))
+        s = int_to_int32(s * int_to_int32(0xc2b2ae3d))
+        return int_to_int32(s ^ ((s & 0xFFFFFFFF) >> 16))
 
     def _algo4(self, s):
         # Custom scrambling function involving a left rotation (ROL)
-        s = self._s = to_signed_32(s + 0x6d2b79f5)
-        s = to_signed_32((s << 7) | ((s & 0xFFFFFFFF) >> 25))  # ROL 7
-        s = to_signed_32(s + 0x9e3779b9)
-        s = to_signed_32(s ^ ((s & 0xFFFFFFFF) >> 11))
-        return to_signed_32(s * 0x27d4eb2d)
+        s = self._s = int_to_int32(s + 0x6d2b79f5)
+        s = int_to_int32((s << 7) | ((s & 0xFFFFFFFF) >> 25))  # ROL 7
+        s = int_to_int32(s + 0x9e3779b9)
+        s = int_to_int32(s ^ ((s & 0xFFFFFFFF) >> 11))
+        return int_to_int32(s * 0x27d4eb2d)
 
     def _algo5(self, s):
         # xorshift variant with a final addition
-        s = to_signed_32(s ^ (s << 7))
-        s = to_signed_32(s ^ ((s & 0xFFFFFFFF) >> 9))
-        s = to_signed_32(s ^ (s << 8))
-        s = self._s = to_signed_32(s + 0xa5a5a5a5)
+        s = int_to_int32(s ^ (s << 7))
+        s = int_to_int32(s ^ ((s & 0xFFFFFFFF) >> 9))
+        s = int_to_int32(s ^ (s << 8))
+        s = self._s = int_to_int32(s + 0xa5a5a5a5)
         return s
 
     def _algo6(self, s):
         # LCG (a=0x2c9277b5, c=0xac564b05) with a variable right shift scrambler
-        s = self._s = to_signed_32(s * to_signed_32(0x2c9277b5) + to_signed_32(0xac564b05))
-        s2 = to_signed_32(s ^ ((s & 0xFFFFFFFF) >> 18))
+        s = self._s = int_to_int32(s * int_to_int32(0x2c9277b5) + int_to_int32(0xac564b05))
+        s2 = int_to_int32(s ^ ((s & 0xFFFFFFFF) >> 18))
         shift = (s & 0xFFFFFFFF) >> 27 & 31
-        return to_signed_32((s2 & 0xFFFFFFFF) >> shift)
+        return int_to_int32((s2 & 0xFFFFFFFF) >> shift)
 
     def _algo7(self, s):
         # Weyl Sequence (k=0x9e3779b9) + custom multiply-xor-shift mixing function
-        s = self._s = to_signed_32(s + to_signed_32(0x9e3779b9))
-        e = to_signed_32(s ^ (s << 5))
-        e = to_signed_32(e * to_signed_32(0x7feb352d))
-        e = to_signed_32(e ^ ((e & 0xFFFFFFFF) >> 15))
-        return to_signed_32(e * to_signed_32(0x846ca68b))
+        s = self._s = int_to_int32(s + int_to_int32(0x9e3779b9))
+        e = int_to_int32(s ^ (s << 5))
+        e = int_to_int32(e * int_to_int32(0x7feb352d))
+        e = int_to_int32(e ^ ((e & 0xFFFFFFFF) >> 15))
+        return int_to_int32(e * int_to_int32(0x846ca68b))
 
     def __next__(self):
         return self._algorithm(self._s) & 0xFF

yt_dlp/jsinterp.py

@@ -18,6 +18,14 @@ from .utils import (
 )
 
 
+def int_to_int32(n):
+    """Converts an integer to a signed 32-bit integer"""
+    n &= 0xFFFFFFFF
+    if n & 0x80000000:
+        return n - 0x100000000
+    return n
+
+
 def _js_bit_op(op):
     def zeroise(x):
         if x in (None, JS_Undefined):
@@ -28,7 +36,7 @@ def _js_bit_op(op):
         return int(float(x))
 
     def wrapped(a, b):
-        return op(zeroise(a), zeroise(b)) & 0xffffffff
+        return int_to_int32(op(int_to_int32(zeroise(a)), int_to_int32(zeroise(b))))
 
     return wrapped