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