commit: 55c823634db890a328ffc23588fcd6f35d9b3ddf
parent 4050e10a4c3445c5399239567eb074acb2f65c18
Author: dirkf <fieldhouse@gmx.net>
Date: Wed, 31 Aug 2022 23:22:48 +0100
[jsinterp] Handle new YT players 113ca41c, c57c113c
* add NaN
* allow any white-space character for `after_op`
* align with yt-dlp f26af78a8ac11d9d617ed31ea5282cfaa5bcbcfa (charcodeAt and bitwise overflow)
* allow escaping in regex, fixing player c57c113c
Diffstat:
3 files changed, 65 insertions(+), 18 deletions(-)
diff --git a/test/test_jsinterp.py b/test/test_jsinterp.py
@@ -136,6 +136,11 @@ class TestJSInterpreter(unittest.TestCase):
def test_builtins(self):
jsi = JSInterpreter('''
+ function x() { return NaN }
+ ''')
+ self.assertTrue(math.isnan(jsi.call_function('x')))
+
+ jsi = JSInterpreter('''
function x() { return new Date('Wednesday 31 December 1969 18:01:26 MDT') - 0; }
''')
self.assertEqual(jsi.call_function('x'), 86000)
@@ -385,6 +390,22 @@ class TestJSInterpreter(unittest.TestCase):
''')
self.assertEqual(jsi.call_function('x').flags & ~re.U, re.I)
+ def test_char_code_at(self):
+ jsi = JSInterpreter('function x(i){return "test".charCodeAt(i)}')
+ self.assertEqual(jsi.call_function('x', 0), 116)
+ self.assertEqual(jsi.call_function('x', 1), 101)
+ self.assertEqual(jsi.call_function('x', 2), 115)
+ self.assertEqual(jsi.call_function('x', 3), 116)
+ self.assertEqual(jsi.call_function('x', 4), None)
+ self.assertEqual(jsi.call_function('x', 'not_a_number'), 116)
+
+ def test_bitwise_operators_overflow(self):
+ jsi = JSInterpreter('function x(){return -524999584 << 5}')
+ self.assertEqual(jsi.call_function('x'), 379882496)
+
+ jsi = JSInterpreter('function x(){return 1236566549 << 5}')
+ self.assertEqual(jsi.call_function('x'), 915423904)
+
if __name__ == '__main__':
unittest.main()
diff --git a/test/test_youtube_signature.py b/test/test_youtube_signature.py
@@ -112,9 +112,25 @@ _NSIG_TESTS = [
'batNX7sYqIJdkJ', 'IhOkL_zxbkOZBw',
),
(
+ 'https://www.youtube.com/s/player/009f1d77/player_ias.vflset/en_US/base.js',
+ '5dwFHw8aFWQUQtffRq', 'audescmLUzI3jw',
+ ),
+ (
'https://www.youtube.com/s/player/dc0c6770/player_ias.vflset/en_US/base.js',
'5EHDMgYLV6HPGk_Mu-kk', 'n9lUJLHbxUI0GQ',
),
+ (
+ 'https://www.youtube.com/s/player/c2199353/player_ias.vflset/en_US/base.js',
+ '5EHDMgYLV6HPGk_Mu-kk', 'AD5rgS85EkrE7',
+ ),
+ (
+ 'https://www.youtube.com/s/player/113ca41c/player_ias.vflset/en_US/base.js',
+ 'cgYl-tlYkhjT7A', 'hI7BBr2zUgcmMg',
+ ),
+ (
+ 'https://www.youtube.com/s/player/c57c113c/player_ias.vflset/en_US/base.js',
+ '-Txvy6bT5R6LqgnQNx', 'dcklJCnRUHbgSg',
+ ),
]
diff --git a/youtube_dl/jsinterp.py b/youtube_dl/jsinterp.py
@@ -23,10 +23,11 @@ from .compat import (
def _js_bit_op(op):
+ def zeroise(x):
+ return 0 if x in (None, JS_Undefined) else x
+
def wrapped(a, b):
- def zeroise(x):
- return 0 if x in (None, JS_Undefined) else x
- return op(zeroise(a), zeroise(b))
+ return op(zeroise(a), zeroise(b)) & 0xffffffff
return wrapped
@@ -44,7 +45,7 @@ def _js_arith_op(op):
def _js_div(a, b):
if JS_Undefined in (a, b) or not (a and b):
return float('nan')
- return float('inf') if not b else operator.truediv(a or 0, b)
+ return operator.truediv(a or 0, b) if b else float('inf')
def _js_mod(a, b):
@@ -260,13 +261,14 @@ class JSInterpreter(object):
counters[_MATCHING_PARENS[char]] += 1
elif char in counters:
counters[char] -= 1
- if not escaping and char in _QUOTES and in_quote in (char, None):
- if in_quote or after_op or char != '/':
- in_quote = None if in_quote and not in_regex_char_group else char
- elif in_quote == '/' and char in '[]':
- in_regex_char_group = char == '['
+ if not escaping:
+ if char in _QUOTES and in_quote in (char, None):
+ if in_quote or after_op or char != '/':
+ in_quote = None if in_quote and not in_regex_char_group else char
+ elif in_quote == '/' and char in '[]':
+ in_regex_char_group = char == '['
escaping = not escaping and in_quote and char == '\\'
- after_op = not in_quote and char in cls.OP_CHARS or (char == ' ' and after_op)
+ after_op = not in_quote and (char in cls.OP_CHARS or (char.isspace() and after_op))
if char != delim[pos] or any(counters.values()) or in_quote:
pos = skipping = 0
@@ -590,6 +592,8 @@ class JSInterpreter(object):
elif expr == 'undefined':
return JS_Undefined, should_return
+ elif expr == 'NaN':
+ return float('NaN'), should_return
elif md.get('return'):
return local_vars[m.group('name')], should_return
@@ -635,7 +639,8 @@ class JSInterpreter(object):
def assertion(cndn, msg):
""" assert, but without risk of getting optimized out """
if not cndn:
- raise ExtractorError('{member} {msg}'.format(**locals()), expr=expr)
+ memb = member
+ raise self.Exception('{member} {msg}'.format(**locals()), expr=expr)
def eval_method():
if (variable, member) == ('console', 'debug'):
@@ -737,6 +742,13 @@ class JSInterpreter(object):
return obj.index(idx, start)
except ValueError:
return -1
+ elif member == 'charCodeAt':
+ assertion(isinstance(obj, compat_str), 'must be applied on a string')
+ # assertion(len(argvals) == 1, 'takes exactly one argument') # but not enforced
+ idx = argvals[0] if isinstance(argvals[0], int) else 0
+ if idx >= len(obj):
+ return None
+ return ord(obj[idx])
idx = int(member) if isinstance(obj, list) else member
return obj[idx](argvals, allow_recursion=allow_recursion)
@@ -820,12 +832,10 @@ class JSInterpreter(object):
if mobj is None:
break
start, body_start = mobj.span()
- body, remaining = self._separate_at_paren(code[body_start - 1:])
- name = self._named_object(
- local_vars,
- self.extract_function_from_code(
- self.build_arglist(mobj.group('args')),
- body, local_vars, *global_stack))
+ body, remaining = self._separate_at_paren(code[body_start - 1:], '}')
+ name = self._named_object(local_vars, self.extract_function_from_code(
+ [x.strip() for x in mobj.group('args').split(',')],
+ body, local_vars, *global_stack))
code = code[:start] + name + remaining
return self.build_function(argnames, code, local_vars, *global_stack)
@@ -854,7 +864,7 @@ class JSInterpreter(object):
zip_longest(argnames, args, fillvalue=None))
global_stack[0].update(kwargs)
var_stack = LocalNameSpace(*global_stack)
- ret, should_abort = self.interpret_statement(code.replace('\n', ''), var_stack, allow_recursion - 1)
+ ret, should_abort = self.interpret_statement(code.replace('\n', ' '), var_stack, allow_recursion - 1)
if should_abort:
return ret
return resf
