yt_dlp/jsinterp.py

   1 import collections
   2 import contextlib
   3 import itertools
   4 import json
   5 import math
   6 import operator
   7 import re
   8
   9 from .utils import (
  10     NO_DEFAULT,
  11     ExtractorError,
  12     function_with_repr,
  13     js_to_json,
  14     remove_quotes,
  15     truncate_string,
  16     unified_timestamp,
  17     write_string,
  18 )
  19
  20
  21 def _js_bit_op(op):
  22     def zeroise(x):
  23         return 0 if x in (None, JS_Undefined) else x
  24
  25     def wrapped(a, b):
  26         return op(zeroise(a), zeroise(b)) & 0xffffffff
  27
  28     return wrapped
  29
  30
  31 def _js_arith_op(op):
  32
  33     def wrapped(a, b):
  34         if JS_Undefined in (a, b):
  35             return float('nan')
  36         return op(a or 0, b or 0)
  37
  38     return wrapped
  39
  40
  41 def _js_div(a, b):
  42     if JS_Undefined in (a, b) or not (a and b):
  43         return float('nan')
  44     return (a or 0) / b if b else float('inf')
  45
  46
  47 def _js_mod(a, b):
  48     if JS_Undefined in (a, b) or not b:
  49         return float('nan')
  50     return (a or 0) % b
  51
  52
  53 def _js_exp(a, b):
  54     if not b:
  55         return 1  # even 0 ** 0 !!
  56     elif JS_Undefined in (a, b):
  57         return float('nan')
  58     return (a or 0) ** b
  59
  60
  61 def _js_eq_op(op):
  62
  63     def wrapped(a, b):
  64         if {a, b} <= {None, JS_Undefined}:
  65             return op(a, a)
  66         return op(a, b)
  67
  68     return wrapped
  69
  70
  71 def _js_comp_op(op):
  72
  73     def wrapped(a, b):
  74         if JS_Undefined in (a, b):
  75             return False
  76         if isinstance(a, str) or isinstance(b, str):
  77             return op(str(a or 0), str(b or 0))
  78         return op(a or 0, b or 0)
  79
  80     return wrapped
  81
  82
  83 def _js_ternary(cndn, if_true=True, if_false=False):
  84     """Simulate JS's ternary operator (cndn?if_true:if_false)"""
  85     if cndn in (False, None, 0, '', JS_Undefined):
  86         return if_false
  87     with contextlib.suppress(TypeError):
  88         if math.isnan(cndn):  # NB: NaN cannot be checked by membership
  89             return if_false
  90     return if_true
  91
  92
  93 # Ref: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Operators/Operator_Precedence
  94 _OPERATORS = {  # None => Defined in JSInterpreter._operator
  95     '?': None,
  96     '??': None,
  97     '||': None,
  98     '&&': None,
  99
 100     '|': _js_bit_op(operator.or_),
 101     '^': _js_bit_op(operator.xor),
 102     '&': _js_bit_op(operator.and_),
 103
 104     '===': operator.is_,
 105     '!==': operator.is_not,
 106     '==': _js_eq_op(operator.eq),
 107     '!=': _js_eq_op(operator.ne),
 108
 109     '<=': _js_comp_op(operator.le),
 110     '>=': _js_comp_op(operator.ge),
 111     '<': _js_comp_op(operator.lt),
 112     '>': _js_comp_op(operator.gt),
 113
 114     '>>': _js_bit_op(operator.rshift),
 115     '<<': _js_bit_op(operator.lshift),
 116
 117     '+': _js_arith_op(operator.add),
 118     '-': _js_arith_op(operator.sub),
 119
 120     '*': _js_arith_op(operator.mul),
 121     '%': _js_mod,
 122     '/': _js_div,
 123     '**': _js_exp,
 124 }
 125
 126 _COMP_OPERATORS = {'===', '!==', '==', '!=', '<=', '>=', '<', '>'}
 127
 128 _NAME_RE = r'[a-zA-Z_$][\w$]*'
 129 _MATCHING_PARENS = dict(zip(*zip('()', '{}', '[]')))
 130 _QUOTES = '\'"/'
 131
 132
 133 class JS_Undefined:
 134     pass
 135
 136
 137 class JS_Break(ExtractorError):
 138     def __init__(self):
 139         ExtractorError.__init__(self, 'Invalid break')
 140
 141
 142 class JS_Continue(ExtractorError):
 143     def __init__(self):
 144         ExtractorError.__init__(self, 'Invalid continue')
 145
 146
 147 class JS_Throw(ExtractorError):
 148     def __init__(self, e):
 149         self.error = e
 150         ExtractorError.__init__(self, f'Uncaught exception {e}')
 151
 152
 153 class LocalNameSpace(collections.ChainMap):
 154     def __setitem__(self, key, value):
 155         for scope in self.maps:
 156             if key in scope:
 157                 scope[key] = value
 158                 return
 159         self.maps[0][key] = value
 160
 161     def __delitem__(self, key):
 162         raise NotImplementedError('Deleting is not supported')
 163
 164
 165 class Debugger:
 166     import sys
 167     ENABLED = False and 'pytest' in sys.modules
 168
 169     @staticmethod
 170     def write(*args, level=100):
 171         write_string(f'[debug] JS: {"  " * (100 - level)}'
 172                      f'{" ".join(truncate_string(str(x), 50, 50) for x in args)}\n')
 173
 174     @classmethod
 175     def wrap_interpreter(cls, f):
 176         def interpret_statement(self, stmt, local_vars, allow_recursion, *args, **kwargs):
 177             if cls.ENABLED and stmt.strip():
 178                 cls.write(stmt, level=allow_recursion)
 179             try:
 180                 ret, should_ret = f(self, stmt, local_vars, allow_recursion, *args, **kwargs)
 181             except Exception as e:
 182                 if cls.ENABLED:
 183                     if isinstance(e, ExtractorError):
 184                         e = e.orig_msg
 185                     cls.write('=> Raises:', e, '<-|', stmt, level=allow_recursion)
 186                 raise
 187             if cls.ENABLED and stmt.strip():
 188                 if should_ret or not repr(ret) == stmt:
 189                     cls.write(['->', '=>'][should_ret], repr(ret), '<-|', stmt, level=allow_recursion)
 190             return ret, should_ret
 191         return interpret_statement
 192
 193
 194 class JSInterpreter:
 195     __named_object_counter = 0
 196
 197     _RE_FLAGS = {
 198         # special knowledge: Python's re flags are bitmask values, current max 128
 199         # invent new bitmask values well above that for literal parsing
 200         # TODO: new pattern class to execute matches with these flags
 201         'd': 1024,  # Generate indices for substring matches
 202         'g': 2048,  # Global search
 203         'i': re.I,  # Case-insensitive search
 204         'm': re.M,  # Multi-line search
 205         's': re.S,  # Allows . to match newline characters
 206         'u': re.U,  # Treat a pattern as a sequence of unicode code points
 207         'y': 4096,  # Perform a "sticky" search that matches starting at the current position in the target string
 208     }
 209
 210     def __init__(self, code, objects=None):
 211         self.code, self._functions = code, {}
 212         self._objects = {} if objects is None else objects
 213
 214     class Exception(ExtractorError):
 215         def __init__(self, msg, expr=None, *args, **kwargs):
 216             if expr is not None:
 217                 msg = f'{msg.rstrip()} in: {truncate_string(expr, 50, 50)}'
 218             super().__init__(msg, *args, **kwargs)
 219
 220     def _named_object(self, namespace, obj):
 221         self.__named_object_counter += 1
 222         name = f'__yt_dlp_jsinterp_obj{self.__named_object_counter}'
 223         if callable(obj) and not isinstance(obj, function_with_repr):
 224             obj = function_with_repr(obj, f'F<{self.__named_object_counter}>')
 225         namespace[name] = obj
 226         return name
 227
 228     @classmethod
 229     def _regex_flags(cls, expr):
 230         flags = 0
 231         if not expr:
 232             return flags, expr
 233         for idx, ch in enumerate(expr):
 234             if ch not in cls._RE_FLAGS:
 235                 break
 236             flags |= cls._RE_FLAGS[ch]
 237         return flags, expr[idx + 1:]
 238
 239     @staticmethod
 240     def _separate(expr, delim=',', max_split=None):
 241         OP_CHARS = '+-*/%&|^=<>!,;{}:['
 242         if not expr:
 243             return
 244         counters = {k: 0 for k in _MATCHING_PARENS.values()}
 245         start, splits, pos, delim_len = 0, 0, 0, len(delim) - 1
 246         in_quote, escaping, after_op, in_regex_char_group, in_unary_op = None, False, True, False, False
 247         for idx, char in enumerate(expr):
 248             if not in_quote and char in _MATCHING_PARENS:
 249                 counters[_MATCHING_PARENS[char]] += 1
 250             elif not in_quote and char in counters:
 251                 # Something's wrong if we get negative, but ignore it anyway
 252                 if counters[char]:
 253                     counters[char] -= 1
 254             elif not escaping:
 255                 if char in _QUOTES and in_quote in (char, None):
 256                     if in_quote or after_op or char != '/':
 257                         in_quote = None if in_quote and not in_regex_char_group else char
 258                 elif in_quote == '/' and char in '[]':
 259                     in_regex_char_group = char == '['
 260             escaping = not escaping and in_quote and char == '\\'
 261             in_unary_op = (not in_quote and not in_regex_char_group
 262                            and after_op not in (True, False) and char in '-+')
 263             after_op = char if (not in_quote and char in OP_CHARS) else (char.isspace() and after_op)
 264
 265             if char != delim[pos] or any(counters.values()) or in_quote or in_unary_op:
 266                 pos = 0
 267                 continue
 268             elif pos != delim_len:
 269                 pos += 1
 270                 continue
 271             yield expr[start: idx - delim_len]
 272             start, pos = idx + 1, 0
 273             splits += 1
 274             if max_split and splits >= max_split:
 275                 break
 276         yield expr[start:]
 277
 278     @classmethod
 279     def _separate_at_paren(cls, expr, delim=None):
 280         if delim is None:
 281             delim = expr and _MATCHING_PARENS[expr[0]]
 282         separated = list(cls._separate(expr, delim, 1))
 283         if len(separated) < 2:
 284             raise cls.Exception(f'No terminating paren {delim}', expr)
 285         return separated[0][1:].strip(), separated[1].strip()
 286
 287     def _operator(self, op, left_val, right_expr, expr, local_vars, allow_recursion):
 288         if op in ('||', '&&'):
 289             if (op == '&&') ^ _js_ternary(left_val):
 290                 return left_val  # short circuiting
 291         elif op == '??':
 292             if left_val not in (None, JS_Undefined):
 293                 return left_val
 294         elif op == '?':
 295             right_expr = _js_ternary(left_val, *self._separate(right_expr, ':', 1))
 296
 297         right_val = self.interpret_expression(right_expr, local_vars, allow_recursion)
 298         if not _OPERATORS.get(op):
 299             return right_val
 300
 301         try:
 302             return _OPERATORS[op](left_val, right_val)
 303         except Exception as e:
 304             raise self.Exception(f'Failed to evaluate {left_val!r} {op} {right_val!r}', expr, cause=e)
 305
 306     def _index(self, obj, idx, allow_undefined=False):
 307         if idx == 'length':
 308             return len(obj)
 309         try:
 310             return obj[int(idx)] if isinstance(obj, list) else obj[idx]
 311         except Exception as e:
 312             if allow_undefined:
 313                 return JS_Undefined
 314             raise self.Exception(f'Cannot get index {idx}', repr(obj), cause=e)
 315
 316     def _dump(self, obj, namespace):
 317         try:
 318             return json.dumps(obj)
 319         except TypeError:
 320             return self._named_object(namespace, obj)
 321
 322     @Debugger.wrap_interpreter
 323     def interpret_statement(self, stmt, local_vars, allow_recursion=100):
 324         if allow_recursion < 0:
 325             raise self.Exception('Recursion limit reached')
 326         allow_recursion -= 1
 327
 328         should_return = False
 329         sub_statements = list(self._separate(stmt, ';')) or ['']
 330         expr = stmt = sub_statements.pop().strip()
 331
 332         for sub_stmt in sub_statements:
 333             ret, should_return = self.interpret_statement(sub_stmt, local_vars, allow_recursion)
 334             if should_return:
 335                 return ret, should_return
 336
 337         m = re.match(r'(?P<var>(?:var|const|let)\s)|return(?:\s+|(?=["\'])|$)|(?P<throw>throw\s+)', stmt)
 338         if m:
 339             expr = stmt[len(m.group(0)):].strip()
 340             if m.group('throw'):
 341                 raise JS_Throw(self.interpret_expression(expr, local_vars, allow_recursion))
 342             should_return = not m.group('var')
 343         if not expr:
 344             return None, should_return
 345
 346         if expr[0] in _QUOTES:
 347             inner, outer = self._separate(expr, expr[0], 1)
 348             if expr[0] == '/':
 349                 flags, outer = self._regex_flags(outer)
 350                 # Avoid https://github.com/python/cpython/issues/74534
 351                 inner = re.compile(inner[1:].replace('[[', r'[\['), flags=flags)
 352             else:
 353                 inner = json.loads(js_to_json(f'{inner}{expr[0]}', strict=True))
 354             if not outer:
 355                 return inner, should_return
 356             expr = self._named_object(local_vars, inner) + outer
 357
 358         if expr.startswith('new '):
 359             obj = expr[4:]
 360             if obj.startswith('Date('):
 361                 left, right = self._separate_at_paren(obj[4:])
 362                 date = unified_timestamp(
 363                     self.interpret_expression(left, local_vars, allow_recursion), False)
 364                 if date is None:
 365                     raise self.Exception(f'Failed to parse date {left!r}', expr)
 366                 expr = self._dump(int(date * 1000), local_vars) + right
 367             else:
 368                 raise self.Exception(f'Unsupported object {obj}', expr)
 369
 370         if expr.startswith('void '):
 371             left = self.interpret_expression(expr[5:], local_vars, allow_recursion)
 372             return None, should_return
 373
 374         if expr.startswith('{'):
 375             inner, outer = self._separate_at_paren(expr)
 376             # try for object expression (Map)
 377             sub_expressions = [list(self._separate(sub_expr.strip(), ':', 1)) for sub_expr in self._separate(inner)]
 378             if all(len(sub_expr) == 2 for sub_expr in sub_expressions):
 379                 def dict_item(key, val):
 380                     val = self.interpret_expression(val, local_vars, allow_recursion)
 381                     if re.match(_NAME_RE, key):
 382                         return key, val
 383                     return self.interpret_expression(key, local_vars, allow_recursion), val
 384
 385                 return dict(dict_item(k, v) for k, v in sub_expressions), should_return
 386
 387             inner, should_abort = self.interpret_statement(inner, local_vars, allow_recursion)
 388             if not outer or should_abort:
 389                 return inner, should_abort or should_return
 390             else:
 391                 expr = self._dump(inner, local_vars) + outer
 392
 393         if expr.startswith('('):
 394             inner, outer = self._separate_at_paren(expr)
 395             inner, should_abort = self.interpret_statement(inner, local_vars, allow_recursion)
 396             if not outer or should_abort:
 397                 return inner, should_abort or should_return
 398             else:
 399                 expr = self._dump(inner, local_vars) + outer
 400
 401         if expr.startswith('['):
 402             inner, outer = self._separate_at_paren(expr)
 403             name = self._named_object(local_vars, [
 404                 self.interpret_expression(item, local_vars, allow_recursion)
 405                 for item in self._separate(inner)])
 406             expr = name + outer
 407
 408         m = re.match(r'''(?x)
 409                 (?P<try>try)\s*\{|
 410                 (?P<if>if)\s*\(|
 411                 (?P<switch>switch)\s*\(|
 412                 (?P<for>for)\s*\(
 413                 ''', expr)
 414         md = m.groupdict() if m else {}
 415         if md.get('if'):
 416             cndn, expr = self._separate_at_paren(expr[m.end() - 1:])
 417             if_expr, expr = self._separate_at_paren(expr.lstrip())
 418             # TODO: "else if" is not handled
 419             else_expr = None
 420             m = re.match(r'else\s*{', expr)
 421             if m:
 422                 else_expr, expr = self._separate_at_paren(expr[m.end() - 1:])
 423             cndn = _js_ternary(self.interpret_expression(cndn, local_vars, allow_recursion))
 424             ret, should_abort = self.interpret_statement(
 425                 if_expr if cndn else else_expr, local_vars, allow_recursion)
 426             if should_abort:
 427                 return ret, True
 428
 429         if md.get('try'):
 430             try_expr, expr = self._separate_at_paren(expr[m.end() - 1:])
 431             err = None
 432             try:
 433                 ret, should_abort = self.interpret_statement(try_expr, local_vars, allow_recursion)
 434                 if should_abort:
 435                     return ret, True
 436             except Exception as e:
 437                 # XXX: This works for now, but makes debugging future issues very hard
 438                 err = e
 439
 440             pending = (None, False)
 441             m = re.match(r'catch\s*(?P<err>\(\s*{_NAME_RE}\s*\))?\{{'.format(**globals()), expr)
 442             if m:
 443                 sub_expr, expr = self._separate_at_paren(expr[m.end() - 1:])
 444                 if err:
 445                     catch_vars = {}
 446                     if m.group('err'):
 447                         catch_vars[m.group('err')] = err.error if isinstance(err, JS_Throw) else err
 448                     catch_vars = local_vars.new_child(catch_vars)
 449                     err, pending = None, self.interpret_statement(sub_expr, catch_vars, allow_recursion)
 450
 451             m = re.match(r'finally\s*\{', expr)
 452             if m:
 453                 sub_expr, expr = self._separate_at_paren(expr[m.end() - 1:])
 454                 ret, should_abort = self.interpret_statement(sub_expr, local_vars, allow_recursion)
 455                 if should_abort:
 456                     return ret, True
 457
 458             ret, should_abort = pending
 459             if should_abort:
 460                 return ret, True
 461
 462             if err:
 463                 raise err
 464
 465         elif md.get('for'):
 466             constructor, remaining = self._separate_at_paren(expr[m.end() - 1:])
 467             if remaining.startswith('{'):
 468                 body, expr = self._separate_at_paren(remaining)
 469             else:
 470                 switch_m = re.match(r'switch\s*\(', remaining)  # FIXME
 471                 if switch_m:
 472                     switch_val, remaining = self._separate_at_paren(remaining[switch_m.end() - 1:])
 473                     body, expr = self._separate_at_paren(remaining, '}')
 474                     body = 'switch(%s){%s}' % (switch_val, body)
 475                 else:
 476                     body, expr = remaining, ''
 477             start, cndn, increment = self._separate(constructor, ';')
 478             self.interpret_expression(start, local_vars, allow_recursion)
 479             while True:
 480                 if not _js_ternary(self.interpret_expression(cndn, local_vars, allow_recursion)):
 481                     break
 482                 try:
 483                     ret, should_abort = self.interpret_statement(body, local_vars, allow_recursion)
 484                     if should_abort:
 485                         return ret, True
 486                 except JS_Break:
 487                     break
 488                 except JS_Continue:
 489                     pass
 490                 self.interpret_expression(increment, local_vars, allow_recursion)
 491
 492         elif md.get('switch'):
 493             switch_val, remaining = self._separate_at_paren(expr[m.end() - 1:])
 494             switch_val = self.interpret_expression(switch_val, local_vars, allow_recursion)
 495             body, expr = self._separate_at_paren(remaining, '}')
 496             items = body.replace('default:', 'case default:').split('case ')[1:]
 497             for default in (False, True):
 498                 matched = False
 499                 for item in items:
 500                     case, stmt = (i.strip() for i in self._separate(item, ':', 1))
 501                     if default:
 502                         matched = matched or case == 'default'
 503                     elif not matched:
 504                         matched = (case != 'default'
 505                                    and switch_val == self.interpret_expression(case, local_vars, allow_recursion))
 506                     if not matched:
 507                         continue
 508                     try:
 509                         ret, should_abort = self.interpret_statement(stmt, local_vars, allow_recursion)
 510                         if should_abort:
 511                             return ret
 512                     except JS_Break:
 513                         break
 514                 if matched:
 515                     break
 516
 517         if md:
 518             ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion)
 519             return ret, should_abort or should_return
 520
 521         # Comma separated statements
 522         sub_expressions = list(self._separate(expr))
 523         if len(sub_expressions) > 1:
 524             for sub_expr in sub_expressions:
 525                 ret, should_abort = self.interpret_statement(sub_expr, local_vars, allow_recursion)
 526                 if should_abort:
 527                     return ret, True
 528             return ret, False
 529
 530         for m in re.finditer(rf'''(?x)
 531                 (?P<pre_sign>\+\+|--)(?P<var1>{_NAME_RE})|
 532                 (?P<var2>{_NAME_RE})(?P<post_sign>\+\+|--)''', expr):
 533             var = m.group('var1') or m.group('var2')
 534             start, end = m.span()
 535             sign = m.group('pre_sign') or m.group('post_sign')
 536             ret = local_vars[var]
 537             local_vars[var] += 1 if sign[0] == '+' else -1
 538             if m.group('pre_sign'):
 539                 ret = local_vars[var]
 540             expr = expr[:start] + self._dump(ret, local_vars) + expr[end:]
 541
 542         if not expr:
 543             return None, should_return
 544
 545         m = re.match(fr'''(?x)
 546             (?P<assign>
 547                 (?P<out>{_NAME_RE})(?:\[(?P<index>[^\]]+?)\])?\s*
 548                 (?P<op>{"|".join(map(re.escape, set(_OPERATORS) - _COMP_OPERATORS))})?
 549                 =(?!=)(?P<expr>.*)$
 550             )|(?P<return>
 551                 (?!if|return|true|false|null|undefined|NaN)(?P<name>{_NAME_RE})$
 552             )|(?P<indexing>
 553                 (?P<in>{_NAME_RE})\[(?P<idx>.+)\]$
 554             )|(?P<attribute>
 555                 (?P<var>{_NAME_RE})(?:(?P<nullish>\?)?\.(?P<member>[^(]+)|\[(?P<member2>[^\]]+)\])\s*
 556             )|(?P<function>
 557                 (?P<fname>{_NAME_RE})\((?P<args>.*)\)$
 558             )''', expr)
 559         if m and m.group('assign'):
 560             left_val = local_vars.get(m.group('out'))
 561
 562             if not m.group('index'):
 563                 local_vars[m.group('out')] = self._operator(
 564                     m.group('op'), left_val, m.group('expr'), expr, local_vars, allow_recursion)
 565                 return local_vars[m.group('out')], should_return
 566             elif left_val in (None, JS_Undefined):
 567                 raise self.Exception(f'Cannot index undefined variable {m.group("out")}', expr)
 568
 569             idx = self.interpret_expression(m.group('index'), local_vars, allow_recursion)
 570             if not isinstance(idx, (int, float)):
 571                 raise self.Exception(f'List index {idx} must be integer', expr)
 572             idx = int(idx)
 573             left_val[idx] = self._operator(
 574                 m.group('op'), self._index(left_val, idx), m.group('expr'), expr, local_vars, allow_recursion)
 575             return left_val[idx], should_return
 576
 577         elif expr.isdigit():
 578             return int(expr), should_return
 579
 580         elif expr == 'break':
 581             raise JS_Break()
 582         elif expr == 'continue':
 583             raise JS_Continue()
 584         elif expr == 'undefined':
 585             return JS_Undefined, should_return
 586         elif expr == 'NaN':
 587             return float('NaN'), should_return
 588
 589         elif m and m.group('return'):
 590             return local_vars.get(m.group('name'), JS_Undefined), should_return
 591
 592         with contextlib.suppress(ValueError):
 593             return json.loads(js_to_json(expr, strict=True)), should_return
 594
 595         if m and m.group('indexing'):
 596             val = local_vars[m.group('in')]
 597             idx = self.interpret_expression(m.group('idx'), local_vars, allow_recursion)
 598             return self._index(val, idx), should_return
 599
 600         for op in _OPERATORS:
 601             separated = list(self._separate(expr, op))
 602             right_expr = separated.pop()
 603             while True:
 604                 if op in '?<>*-' and len(separated) > 1 and not separated[-1].strip():
 605                     separated.pop()
 606                 elif not (separated and op == '?' and right_expr.startswith('.')):
 607                     break
 608                 right_expr = f'{op}{right_expr}'
 609                 if op != '-':
 610                     right_expr = f'{separated.pop()}{op}{right_expr}'
 611             if not separated:
 612                 continue
 613             left_val = self.interpret_expression(op.join(separated), local_vars, allow_recursion)
 614             return self._operator(op, left_val, right_expr, expr, local_vars, allow_recursion), should_return
 615
 616         if m and m.group('attribute'):
 617             variable, member, nullish = m.group('var', 'member', 'nullish')
 618             if not member:
 619                 member = self.interpret_expression(m.group('member2'), local_vars, allow_recursion)
 620             arg_str = expr[m.end():]
 621             if arg_str.startswith('('):
 622                 arg_str, remaining = self._separate_at_paren(arg_str)
 623             else:
 624                 arg_str, remaining = None, arg_str
 625
 626             def assertion(cndn, msg):
 627                 """ assert, but without risk of getting optimized out """
 628                 if not cndn:
 629                     raise self.Exception(f'{member} {msg}', expr)
 630
 631             def eval_method():
 632                 if (variable, member) == ('console', 'debug'):
 633                     if Debugger.ENABLED:
 634                         Debugger.write(self.interpret_expression(f'[{arg_str}]', local_vars, allow_recursion))
 635                     return
 636
 637                 types = {
 638                     'String': str,
 639                     'Math': float,
 640                 }
 641                 obj = local_vars.get(variable, types.get(variable, NO_DEFAULT))
 642                 if obj is NO_DEFAULT:
 643                     if variable not in self._objects:
 644                         try:
 645                             self._objects[variable] = self.extract_object(variable)
 646                         except self.Exception:
 647                             if not nullish:
 648                                 raise
 649                     obj = self._objects.get(variable, JS_Undefined)
 650
 651                 if nullish and obj is JS_Undefined:
 652                     return JS_Undefined
 653
 654                 # Member access
 655                 if arg_str is None:
 656                     return self._index(obj, member, nullish)
 657
 658                 # Function call
 659                 argvals = [
 660                     self.interpret_expression(v, local_vars, allow_recursion)
 661                     for v in self._separate(arg_str)]
 662
 663                 if obj == str:
 664                     if member == 'fromCharCode':
 665                         assertion(argvals, 'takes one or more arguments')
 666                         return ''.join(map(chr, argvals))
 667                     raise self.Exception(f'Unsupported String method {member}', expr)
 668                 elif obj == float:
 669                     if member == 'pow':
 670                         assertion(len(argvals) == 2, 'takes two arguments')
 671                         return argvals[0] ** argvals[1]
 672                     raise self.Exception(f'Unsupported Math method {member}', expr)
 673
 674                 if member == 'split':
 675                     assertion(argvals, 'takes one or more arguments')
 676                     assertion(len(argvals) == 1, 'with limit argument is not implemented')
 677                     return obj.split(argvals[0]) if argvals[0] else list(obj)
 678                 elif member == 'join':
 679                     assertion(isinstance(obj, list), 'must be applied on a list')
 680                     assertion(len(argvals) == 1, 'takes exactly one argument')
 681                     return argvals[0].join(obj)
 682                 elif member == 'reverse':
 683                     assertion(not argvals, 'does not take any arguments')
 684                     obj.reverse()
 685                     return obj
 686                 elif member == 'slice':
 687                     assertion(isinstance(obj, list), 'must be applied on a list')
 688                     assertion(len(argvals) == 1, 'takes exactly one argument')
 689                     return obj[argvals[0]:]
 690                 elif member == 'splice':
 691                     assertion(isinstance(obj, list), 'must be applied on a list')
 692                     assertion(argvals, 'takes one or more arguments')
 693                     index, howMany = map(int, (argvals + [len(obj)])[:2])
 694                     if index < 0:
 695                         index += len(obj)
 696                     add_items = argvals[2:]
 697                     res = []
 698                     for i in range(index, min(index + howMany, len(obj))):
 699                         res.append(obj.pop(index))
 700                     for i, item in enumerate(add_items):
 701                         obj.insert(index + i, item)
 702                     return res
 703                 elif member == 'unshift':
 704                     assertion(isinstance(obj, list), 'must be applied on a list')
 705                     assertion(argvals, 'takes one or more arguments')
 706                     for item in reversed(argvals):
 707                         obj.insert(0, item)
 708                     return obj
 709                 elif member == 'pop':
 710                     assertion(isinstance(obj, list), 'must be applied on a list')
 711                     assertion(not argvals, 'does not take any arguments')
 712                     if not obj:
 713                         return
 714                     return obj.pop()
 715                 elif member == 'push':
 716                     assertion(argvals, 'takes one or more arguments')
 717                     obj.extend(argvals)
 718                     return obj
 719                 elif member == 'forEach':
 720                     assertion(argvals, 'takes one or more arguments')
 721                     assertion(len(argvals) <= 2, 'takes at-most 2 arguments')
 722                     f, this = (argvals + [''])[:2]
 723                     return [f((item, idx, obj), {'this': this}, allow_recursion) for idx, item in enumerate(obj)]
 724                 elif member == 'indexOf':
 725                     assertion(argvals, 'takes one or more arguments')
 726                     assertion(len(argvals) <= 2, 'takes at-most 2 arguments')
 727                     idx, start = (argvals + [0])[:2]
 728                     try:
 729                         return obj.index(idx, start)
 730                     except ValueError:
 731                         return -1
 732                 elif member == 'charCodeAt':
 733                     assertion(isinstance(obj, str), 'must be applied on a string')
 734                     assertion(len(argvals) == 1, 'takes exactly one argument')
 735                     idx = argvals[0] if isinstance(argvals[0], int) else 0
 736                     if idx >= len(obj):
 737                         return None
 738                     return ord(obj[idx])
 739
 740                 idx = int(member) if isinstance(obj, list) else member
 741                 return obj[idx](argvals, allow_recursion=allow_recursion)
 742
 743             if remaining:
 744                 ret, should_abort = self.interpret_statement(
 745                     self._named_object(local_vars, eval_method()) + remaining,
 746                     local_vars, allow_recursion)
 747                 return ret, should_return or should_abort
 748             else:
 749                 return eval_method(), should_return
 750
 751         elif m and m.group('function'):
 752             fname = m.group('fname')
 753             argvals = [self.interpret_expression(v, local_vars, allow_recursion)
 754                        for v in self._separate(m.group('args'))]
 755             if fname in local_vars:
 756                 return local_vars[fname](argvals, allow_recursion=allow_recursion), should_return
 757             elif fname not in self._functions:
 758                 self._functions[fname] = self.extract_function(fname)
 759             return self._functions[fname](argvals, allow_recursion=allow_recursion), should_return
 760
 761         raise self.Exception(
 762             f'Unsupported JS expression {truncate_string(expr, 20, 20) if expr != stmt else ""}', stmt)
 763
 764     def interpret_expression(self, expr, local_vars, allow_recursion):
 765         ret, should_return = self.interpret_statement(expr, local_vars, allow_recursion)
 766         if should_return:
 767             raise self.Exception('Cannot return from an expression', expr)
 768         return ret
 769
 770     def extract_object(self, objname):
 771         _FUNC_NAME_RE = r'''(?:[a-zA-Z$0-9]+|"[a-zA-Z$0-9]+"|'[a-zA-Z$0-9]+')'''
 772         obj = {}
 773         obj_m = re.search(
 774             r'''(?x)
 775                 (?<!this\.)%s\s*=\s*{\s*
 776                     (?P<fields>(%s\s*:\s*function\s*\(.*?\)\s*{.*?}(?:,\s*)?)*)
 777                 }\s*;
 778             ''' % (re.escape(objname), _FUNC_NAME_RE),
 779             self.code)
 780         if not obj_m:
 781             raise self.Exception(f'Could not find object {objname}')
 782         fields = obj_m.group('fields')
 783         # Currently, it only supports function definitions
 784         fields_m = re.finditer(
 785             r'''(?x)
 786                 (?P<key>%s)\s*:\s*function\s*\((?P<args>(?:%s|,)*)\){(?P<code>[^}]+)}
 787             ''' % (_FUNC_NAME_RE, _NAME_RE),
 788             fields)
 789         for f in fields_m:
 790             argnames = f.group('args').split(',')
 791             name = remove_quotes(f.group('key'))
 792             obj[name] = function_with_repr(self.build_function(argnames, f.group('code')), f'F<{name}>')
 793
 794         return obj
 795
 796     def extract_function_code(self, funcname):
 797         """ @returns argnames, code """
 798         func_m = re.search(
 799             r'''(?xs)
 800                 (?:
 801                     function\s+%(name)s|
 802                     [{;,]\s*%(name)s\s*=\s*function|
 803                     (?:var|const|let)\s+%(name)s\s*=\s*function
 804                 )\s*
 805                 \((?P<args>[^)]*)\)\s*
 806                 (?P<code>{.+})''' % {'name': re.escape(funcname)},
 807             self.code)
 808         code, _ = self._separate_at_paren(func_m.group('code'))
 809         if func_m is None:
 810             raise self.Exception(f'Could not find JS function "{funcname}"')
 811         return [x.strip() for x in func_m.group('args').split(',')], code
 812
 813     def extract_function(self, funcname):
 814         return function_with_repr(
 815             self.extract_function_from_code(*self.extract_function_code(funcname)),
 816             f'F<{funcname}>')
 817
 818     def extract_function_from_code(self, argnames, code, *global_stack):
 819         local_vars = {}
 820         while True:
 821             mobj = re.search(r'function\((?P<args>[^)]*)\)\s*{', code)
 822             if mobj is None:
 823                 break
 824             start, body_start = mobj.span()
 825             body, remaining = self._separate_at_paren(code[body_start - 1:])
 826             name = self._named_object(local_vars, self.extract_function_from_code(
 827                 [x.strip() for x in mobj.group('args').split(',')],
 828                 body, local_vars, *global_stack))
 829             code = code[:start] + name + remaining
 830         return self.build_function(argnames, code, local_vars, *global_stack)
 831
 832     def call_function(self, funcname, *args):
 833         return self.extract_function(funcname)(args)
 834
 835     def build_function(self, argnames, code, *global_stack):
 836         global_stack = list(global_stack) or [{}]
 837         argnames = tuple(argnames)
 838
 839         def resf(args, kwargs={}, allow_recursion=100):
 840             global_stack[0].update(itertools.zip_longest(argnames, args, fillvalue=None))
 841             global_stack[0].update(kwargs)
 842             var_stack = LocalNameSpace(*global_stack)
 843             ret, should_abort = self.interpret_statement(code.replace('\n', ' '), var_stack, allow_recursion - 1)
 844             if should_abort:
 845                 return ret
 846         return resf