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 and char in _QUOTES and in_quote in (char, None):
 249                 if in_quote or after_op or char != '/':
 250                     in_quote = None if in_quote and not in_regex_char_group else char
 251             elif in_quote == '/' and char in '[]':
 252                 in_regex_char_group = char == '['
 253             escaping = not escaping and in_quote and char == '\\'
 254             after_op = not in_quote and char in OP_CHARS or (char.isspace() and after_op)
 255
 256             if char != delim[pos] or any(counters.values()) or in_quote:
 257                 pos = 0
 258                 continue
 259             elif pos != delim_len:
 260                 pos += 1
 261                 continue
 262             yield expr[start: idx - delim_len]
 263             start, pos = idx + 1, 0
 264             splits += 1
 265             if max_split and splits >= max_split:
 266                 break
 267         yield expr[start:]
 268
 269     @classmethod
 270     def _separate_at_paren(cls, expr, delim):
 271         separated = list(cls._separate(expr, delim, 1))
 272         if len(separated) < 2:
 273             raise cls.Exception(f'No terminating paren {delim}', expr)
 274         return separated[0][1:].strip(), separated[1].strip()
 275
 276     def _operator(self, op, left_val, right_expr, expr, local_vars, allow_recursion):
 277         if op in ('||', '&&'):
 278             if (op == '&&') ^ _js_ternary(left_val):
 279                 return left_val  # short circuiting
 280         elif op == '??':
 281             if left_val not in (None, JS_Undefined):
 282                 return left_val
 283         elif op == '?':
 284             right_expr = _js_ternary(left_val, *self._separate(right_expr, ':', 1))
 285
 286         right_val = self.interpret_expression(right_expr, local_vars, allow_recursion)
 287         if not _OPERATORS.get(op):
 288             return right_val
 289
 290         try:
 291             return _OPERATORS[op](left_val, right_val)
 292         except Exception as e:
 293             raise self.Exception(f'Failed to evaluate {left_val!r} {op} {right_val!r}', expr, cause=e)
 294
 295     def _index(self, obj, idx, allow_undefined=False):
 296         if idx == 'length':
 297             return len(obj)
 298         try:
 299             return obj[int(idx)] if isinstance(obj, list) else obj[idx]
 300         except Exception as e:
 301             if allow_undefined:
 302                 return JS_Undefined
 303             raise self.Exception(f'Cannot get index {idx}', repr(obj), cause=e)
 304
 305     def _dump(self, obj, namespace):
 306         try:
 307             return json.dumps(obj)
 308         except TypeError:
 309             return self._named_object(namespace, obj)
 310
 311     @Debugger.wrap_interpreter
 312     def interpret_statement(self, stmt, local_vars, allow_recursion=100):
 313         if allow_recursion < 0:
 314             raise self.Exception('Recursion limit reached')
 315         allow_recursion -= 1
 316
 317         should_return = False
 318         sub_statements = list(self._separate(stmt, ';')) or ['']
 319         expr = stmt = sub_statements.pop().strip()
 320
 321         for sub_stmt in sub_statements:
 322             ret, should_return = self.interpret_statement(sub_stmt, local_vars, allow_recursion)
 323             if should_return:
 324                 return ret, should_return
 325
 326         m = re.match(r'(?P<var>(?:var|const|let)\s)|return(?:\s+|(?=["\'])|$)|(?P<throw>throw\s+)', stmt)
 327         if m:
 328             expr = stmt[len(m.group(0)):].strip()
 329             if m.group('throw'):
 330                 raise JS_Throw(self.interpret_expression(expr, local_vars, allow_recursion))
 331             should_return = not m.group('var')
 332         if not expr:
 333             return None, should_return
 334
 335         if expr[0] in _QUOTES:
 336             inner, outer = self._separate(expr, expr[0], 1)
 337             if expr[0] == '/':
 338                 flags, outer = self._regex_flags(outer)
 339                 inner = re.compile(inner[1:], flags=flags)
 340             else:
 341                 inner = json.loads(js_to_json(f'{inner}{expr[0]}', strict=True))
 342             if not outer:
 343                 return inner, should_return
 344             expr = self._named_object(local_vars, inner) + outer
 345
 346         if expr.startswith('new '):
 347             obj = expr[4:]
 348             if obj.startswith('Date('):
 349                 left, right = self._separate_at_paren(obj[4:], ')')
 350                 expr = unified_timestamp(
 351                     self.interpret_expression(left, local_vars, allow_recursion), False)
 352                 if not expr:
 353                     raise self.Exception(f'Failed to parse date {left!r}', expr)
 354                 expr = self._dump(int(expr * 1000), local_vars) + right
 355             else:
 356                 raise self.Exception(f'Unsupported object {obj}', expr)
 357
 358         if expr.startswith('void '):
 359             left = self.interpret_expression(expr[5:], local_vars, allow_recursion)
 360             return None, should_return
 361
 362         if expr.startswith('{'):
 363             inner, outer = self._separate_at_paren(expr, '}')
 364             # Look for Map first
 365             sub_expressions = [list(self._separate(sub_expr.strip(), ':', 1)) for sub_expr in self._separate(inner)]
 366             if all(len(sub_expr) == 2 for sub_expr in sub_expressions):
 367                 def dict_item(key, val):
 368                     val = self.interpret_expression(val, local_vars, allow_recursion)
 369                     if re.match(_NAME_RE, key):
 370                         return key, val
 371                     return self.interpret_expression(key, local_vars, allow_recursion), val
 372
 373                 return dict(dict_item(k, v) for k, v in sub_expressions), should_return
 374
 375             inner, should_abort = self.interpret_statement(inner, local_vars, allow_recursion)
 376             if not outer or should_abort:
 377                 return inner, should_abort or should_return
 378             else:
 379                 expr = self._dump(inner, local_vars) + outer
 380
 381         if expr.startswith('('):
 382             inner, outer = self._separate_at_paren(expr, ')')
 383             inner, should_abort = self.interpret_statement(inner, local_vars, allow_recursion)
 384             if not outer or should_abort:
 385                 return inner, should_abort or should_return
 386             else:
 387                 expr = self._dump(inner, local_vars) + outer
 388
 389         if expr.startswith('['):
 390             inner, outer = self._separate_at_paren(expr, ']')
 391             name = self._named_object(local_vars, [
 392                 self.interpret_expression(item, local_vars, allow_recursion)
 393                 for item in self._separate(inner)])
 394             expr = name + outer
 395
 396         m = re.match(rf'''(?x)
 397             (?P<try>try|finally)\s*|
 398             (?P<catch>catch\s*(?P<err>\(\s*{_NAME_RE}\s*\)))|
 399             (?P<switch>switch)\s*\(|
 400             (?P<for>for)\s*\(|''', expr)
 401         if m and m.group('try'):
 402             if expr[m.end()] == '{':
 403                 try_expr, expr = self._separate_at_paren(expr[m.end():], '}')
 404             else:
 405                 try_expr, expr = expr[m.end() - 1:], ''
 406             try:
 407                 ret, should_abort = self.interpret_statement(try_expr, local_vars, allow_recursion)
 408                 if should_abort:
 409                     return ret, True
 410             except JS_Throw as e:
 411                 local_vars[self._EXC_NAME] = e.error
 412             except Exception as e:
 413                 # XXX: This works for now, but makes debugging future issues very hard
 414                 local_vars[self._EXC_NAME] = e
 415             ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion)
 416             return ret, should_abort or should_return
 417
 418         elif m and m.group('catch'):
 419             catch_expr, expr = self._separate_at_paren(expr[m.end():], '}')
 420             if self._EXC_NAME in local_vars:
 421                 catch_vars = local_vars.new_child({m.group('err'): local_vars.pop(self._EXC_NAME)})
 422                 ret, should_abort = self.interpret_statement(catch_expr, catch_vars, allow_recursion)
 423                 if should_abort:
 424                     return ret, True
 425
 426             ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion)
 427             return ret, should_abort or should_return
 428
 429         elif m and m.group('for'):
 430             constructor, remaining = self._separate_at_paren(expr[m.end() - 1:], ')')
 431             if remaining.startswith('{'):
 432                 body, expr = self._separate_at_paren(remaining, '}')
 433             else:
 434                 switch_m = re.match(r'switch\s*\(', remaining)  # FIXME
 435                 if switch_m:
 436                     switch_val, remaining = self._separate_at_paren(remaining[switch_m.end() - 1:], ')')
 437                     body, expr = self._separate_at_paren(remaining, '}')
 438                     body = 'switch(%s){%s}' % (switch_val, body)
 439                 else:
 440                     body, expr = remaining, ''
 441             start, cndn, increment = self._separate(constructor, ';')
 442             self.interpret_expression(start, local_vars, allow_recursion)
 443             while True:
 444                 if not _js_ternary(self.interpret_expression(cndn, local_vars, allow_recursion)):
 445                     break
 446                 try:
 447                     ret, should_abort = self.interpret_statement(body, local_vars, allow_recursion)
 448                     if should_abort:
 449                         return ret, True
 450                 except JS_Break:
 451                     break
 452                 except JS_Continue:
 453                     pass
 454                 self.interpret_expression(increment, local_vars, allow_recursion)
 455             ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion)
 456             return ret, should_abort or should_return
 457
 458         elif m and m.group('switch'):
 459             switch_val, remaining = self._separate_at_paren(expr[m.end() - 1:], ')')
 460             switch_val = self.interpret_expression(switch_val, local_vars, allow_recursion)
 461             body, expr = self._separate_at_paren(remaining, '}')
 462             items = body.replace('default:', 'case default:').split('case ')[1:]
 463             for default in (False, True):
 464                 matched = False
 465                 for item in items:
 466                     case, stmt = (i.strip() for i in self._separate(item, ':', 1))
 467                     if default:
 468                         matched = matched or case == 'default'
 469                     elif not matched:
 470                         matched = (case != 'default'
 471                                    and switch_val == self.interpret_expression(case, local_vars, allow_recursion))
 472                     if not matched:
 473                         continue
 474                     try:
 475                         ret, should_abort = self.interpret_statement(stmt, local_vars, allow_recursion)
 476                         if should_abort:
 477                             return ret
 478                     except JS_Break:
 479                         break
 480                 if matched:
 481                     break
 482             ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion)
 483             return ret, should_abort or should_return
 484
 485         # Comma separated statements
 486         sub_expressions = list(self._separate(expr))
 487         if len(sub_expressions) > 1:
 488             for sub_expr in sub_expressions:
 489                 ret, should_abort = self.interpret_statement(sub_expr, local_vars, allow_recursion)
 490                 if should_abort:
 491                     return ret, True
 492             return ret, False
 493
 494         for m in re.finditer(rf'''(?x)
 495                 (?P<pre_sign>\+\+|--)(?P<var1>{_NAME_RE})|
 496                 (?P<var2>{_NAME_RE})(?P<post_sign>\+\+|--)''', expr):
 497             var = m.group('var1') or m.group('var2')
 498             start, end = m.span()
 499             sign = m.group('pre_sign') or m.group('post_sign')
 500             ret = local_vars[var]
 501             local_vars[var] += 1 if sign[0] == '+' else -1
 502             if m.group('pre_sign'):
 503                 ret = local_vars[var]
 504             expr = expr[:start] + self._dump(ret, local_vars) + expr[end:]
 505
 506         if not expr:
 507             return None, should_return
 508
 509         m = re.match(fr'''(?x)
 510             (?P<assign>
 511                 (?P<out>{_NAME_RE})(?:\[(?P<index>[^\]]+?)\])?\s*
 512                 (?P<op>{"|".join(map(re.escape, set(_OPERATORS) - _COMP_OPERATORS))})?
 513                 =(?!=)(?P<expr>.*)$
 514             )|(?P<return>
 515                 (?!if|return|true|false|null|undefined|NaN)(?P<name>{_NAME_RE})$
 516             )|(?P<indexing>
 517                 (?P<in>{_NAME_RE})\[(?P<idx>.+)\]$
 518             )|(?P<attribute>
 519                 (?P<var>{_NAME_RE})(?:(?P<nullish>\?)?\.(?P<member>[^(]+)|\[(?P<member2>[^\]]+)\])\s*
 520             )|(?P<function>
 521                 (?P<fname>{_NAME_RE})\((?P<args>.*)\)$
 522             )''', expr)
 523         if m and m.group('assign'):
 524             left_val = local_vars.get(m.group('out'))
 525
 526             if not m.group('index'):
 527                 local_vars[m.group('out')] = self._operator(
 528                     m.group('op'), left_val, m.group('expr'), expr, local_vars, allow_recursion)
 529                 return local_vars[m.group('out')], should_return
 530             elif left_val in (None, JS_Undefined):
 531                 raise self.Exception(f'Cannot index undefined variable {m.group("out")}', expr)
 532
 533             idx = self.interpret_expression(m.group('index'), local_vars, allow_recursion)
 534             if not isinstance(idx, (int, float)):
 535                 raise self.Exception(f'List index {idx} must be integer', expr)
 536             idx = int(idx)
 537             left_val[idx] = self._operator(
 538                 m.group('op'), self._index(left_val, idx), m.group('expr'), expr, local_vars, allow_recursion)
 539             return left_val[idx], should_return
 540
 541         elif expr.isdigit():
 542             return int(expr), should_return
 543
 544         elif expr == 'break':
 545             raise JS_Break()
 546         elif expr == 'continue':
 547             raise JS_Continue()
 548         elif expr == 'undefined':
 549             return JS_Undefined, should_return
 550         elif expr == 'NaN':
 551             return float('NaN'), should_return
 552
 553         elif m and m.group('return'):
 554             return local_vars.get(m.group('name'), JS_Undefined), should_return
 555
 556         with contextlib.suppress(ValueError):
 557             return json.loads(js_to_json(expr, strict=True)), should_return
 558
 559         if m and m.group('indexing'):
 560             val = local_vars[m.group('in')]
 561             idx = self.interpret_expression(m.group('idx'), local_vars, allow_recursion)
 562             return self._index(val, idx), should_return
 563
 564         for op in _OPERATORS:
 565             separated = list(self._separate(expr, op))
 566             right_expr = separated.pop()
 567             while True:
 568                 if op in '?<>*-' and len(separated) > 1 and not separated[-1].strip():
 569                     separated.pop()
 570                 elif not (separated and op == '?' and right_expr.startswith('.')):
 571                     break
 572                 right_expr = f'{op}{right_expr}'
 573                 if op != '-':
 574                     right_expr = f'{separated.pop()}{op}{right_expr}'
 575             if not separated:
 576                 continue
 577             left_val = self.interpret_expression(op.join(separated), local_vars, allow_recursion)
 578             return self._operator(op, left_val, right_expr, expr, local_vars, allow_recursion), should_return
 579
 580         if m and m.group('attribute'):
 581             variable, member, nullish = m.group('var', 'member', 'nullish')
 582             if not member:
 583                 member = self.interpret_expression(m.group('member2'), local_vars, allow_recursion)
 584             arg_str = expr[m.end():]
 585             if arg_str.startswith('('):
 586                 arg_str, remaining = self._separate_at_paren(arg_str, ')')
 587             else:
 588                 arg_str, remaining = None, arg_str
 589
 590             def assertion(cndn, msg):
 591                 """ assert, but without risk of getting optimized out """
 592                 if not cndn:
 593                     raise self.Exception(f'{member} {msg}', expr)
 594
 595             def eval_method():
 596                 if (variable, member) == ('console', 'debug'):
 597                     if Debugger.ENABLED:
 598                         Debugger.write(self.interpret_expression(f'[{arg_str}]', local_vars, allow_recursion))
 599                     return
 600
 601                 types = {
 602                     'String': str,
 603                     'Math': float,
 604                 }
 605                 obj = local_vars.get(variable, types.get(variable, NO_DEFAULT))
 606                 if obj is NO_DEFAULT:
 607                     if variable not in self._objects:
 608                         try:
 609                             self._objects[variable] = self.extract_object(variable)
 610                         except self.Exception:
 611                             if not nullish:
 612                                 raise
 613                     obj = self._objects.get(variable, JS_Undefined)
 614
 615                 if nullish and obj is JS_Undefined:
 616                     return JS_Undefined
 617
 618                 # Member access
 619                 if arg_str is None:
 620                     return self._index(obj, member, nullish)
 621
 622                 # Function call
 623                 argvals = [
 624                     self.interpret_expression(v, local_vars, allow_recursion)
 625                     for v in self._separate(arg_str)]
 626
 627                 if obj == str:
 628                     if member == 'fromCharCode':
 629                         assertion(argvals, 'takes one or more arguments')
 630                         return ''.join(map(chr, argvals))
 631                     raise self.Exception(f'Unsupported String method {member}', expr)
 632                 elif obj == float:
 633                     if member == 'pow':
 634                         assertion(len(argvals) == 2, 'takes two arguments')
 635                         return argvals[0] ** argvals[1]
 636                     raise self.Exception(f'Unsupported Math method {member}', expr)
 637
 638                 if member == 'split':
 639                     assertion(argvals, 'takes one or more arguments')
 640                     assertion(len(argvals) == 1, 'with limit argument is not implemented')
 641                     return obj.split(argvals[0]) if argvals[0] else list(obj)
 642                 elif member == 'join':
 643                     assertion(isinstance(obj, list), 'must be applied on a list')
 644                     assertion(len(argvals) == 1, 'takes exactly one argument')
 645                     return argvals[0].join(obj)
 646                 elif member == 'reverse':
 647                     assertion(not argvals, 'does not take any arguments')
 648                     obj.reverse()
 649                     return obj
 650                 elif member == 'slice':
 651                     assertion(isinstance(obj, list), 'must be applied on a list')
 652                     assertion(len(argvals) == 1, 'takes exactly one argument')
 653                     return obj[argvals[0]:]
 654                 elif member == 'splice':
 655                     assertion(isinstance(obj, list), 'must be applied on a list')
 656                     assertion(argvals, 'takes one or more arguments')
 657                     index, howMany = map(int, (argvals + [len(obj)])[:2])
 658                     if index < 0:
 659                         index += len(obj)
 660                     add_items = argvals[2:]
 661                     res = []
 662                     for i in range(index, min(index + howMany, len(obj))):
 663                         res.append(obj.pop(index))
 664                     for i, item in enumerate(add_items):
 665                         obj.insert(index + i, item)
 666                     return res
 667                 elif member == 'unshift':
 668                     assertion(isinstance(obj, list), 'must be applied on a list')
 669                     assertion(argvals, 'takes one or more arguments')
 670                     for item in reversed(argvals):
 671                         obj.insert(0, item)
 672                     return obj
 673                 elif member == 'pop':
 674                     assertion(isinstance(obj, list), 'must be applied on a list')
 675                     assertion(not argvals, 'does not take any arguments')
 676                     if not obj:
 677                         return
 678                     return obj.pop()
 679                 elif member == 'push':
 680                     assertion(argvals, 'takes one or more arguments')
 681                     obj.extend(argvals)
 682                     return obj
 683                 elif member == 'forEach':
 684                     assertion(argvals, 'takes one or more arguments')
 685                     assertion(len(argvals) <= 2, 'takes at-most 2 arguments')
 686                     f, this = (argvals + [''])[:2]
 687                     return [f((item, idx, obj), {'this': this}, allow_recursion) for idx, item in enumerate(obj)]
 688                 elif member == 'indexOf':
 689                     assertion(argvals, 'takes one or more arguments')
 690                     assertion(len(argvals) <= 2, 'takes at-most 2 arguments')
 691                     idx, start = (argvals + [0])[:2]
 692                     try:
 693                         return obj.index(idx, start)
 694                     except ValueError:
 695                         return -1
 696                 elif member == 'charCodeAt':
 697                     assertion(isinstance(obj, str), 'must be applied on a string')
 698                     assertion(len(argvals) == 1, 'takes exactly one argument')
 699                     idx = argvals[0] if isinstance(argvals[0], int) else 0
 700                     if idx >= len(obj):
 701                         return None
 702                     return ord(obj[idx])
 703
 704                 idx = int(member) if isinstance(obj, list) else member
 705                 return obj[idx](argvals, allow_recursion=allow_recursion)
 706
 707             if remaining:
 708                 ret, should_abort = self.interpret_statement(
 709                     self._named_object(local_vars, eval_method()) + remaining,
 710                     local_vars, allow_recursion)
 711                 return ret, should_return or should_abort
 712             else:
 713                 return eval_method(), should_return
 714
 715         elif m and m.group('function'):
 716             fname = m.group('fname')
 717             argvals = [self.interpret_expression(v, local_vars, allow_recursion)
 718                        for v in self._separate(m.group('args'))]
 719             if fname in local_vars:
 720                 return local_vars[fname](argvals, allow_recursion=allow_recursion), should_return
 721             elif fname not in self._functions:
 722                 self._functions[fname] = self.extract_function(fname)
 723             return self._functions[fname](argvals, allow_recursion=allow_recursion), should_return
 724
 725         raise self.Exception(
 726             f'Unsupported JS expression {truncate_string(expr, 20, 20) if expr != stmt else ""}', stmt)
 727
 728     def interpret_expression(self, expr, local_vars, allow_recursion):
 729         ret, should_return = self.interpret_statement(expr, local_vars, allow_recursion)
 730         if should_return:
 731             raise self.Exception('Cannot return from an expression', expr)
 732         return ret
 733
 734     def extract_object(self, objname):
 735         _FUNC_NAME_RE = r'''(?:[a-zA-Z$0-9]+|"[a-zA-Z$0-9]+"|'[a-zA-Z$0-9]+')'''
 736         obj = {}
 737         obj_m = re.search(
 738             r'''(?x)
 739                 (?<!this\.)%s\s*=\s*{\s*
 740                     (?P<fields>(%s\s*:\s*function\s*\(.*?\)\s*{.*?}(?:,\s*)?)*)
 741                 }\s*;
 742             ''' % (re.escape(objname), _FUNC_NAME_RE),
 743             self.code)
 744         if not obj_m:
 745             raise self.Exception(f'Could not find object {objname}')
 746         fields = obj_m.group('fields')
 747         # Currently, it only supports function definitions
 748         fields_m = re.finditer(
 749             r'''(?x)
 750                 (?P<key>%s)\s*:\s*function\s*\((?P<args>(?:%s|,)*)\){(?P<code>[^}]+)}
 751             ''' % (_FUNC_NAME_RE, _NAME_RE),
 752             fields)
 753         for f in fields_m:
 754             argnames = f.group('args').split(',')
 755             obj[remove_quotes(f.group('key'))] = self.build_function(argnames, f.group('code'))
 756
 757         return obj
 758
 759     def extract_function_code(self, funcname):
 760         """ @returns argnames, code """
 761         func_m = re.search(
 762             r'''(?xs)
 763                 (?:
 764                     function\s+%(name)s|
 765                     [{;,]\s*%(name)s\s*=\s*function|
 766                     (?:var|const|let)\s+%(name)s\s*=\s*function
 767                 )\s*
 768                 \((?P<args>[^)]*)\)\s*
 769                 (?P<code>{.+})''' % {'name': re.escape(funcname)},
 770             self.code)
 771         code, _ = self._separate_at_paren(func_m.group('code'), '}')
 772         if func_m is None:
 773             raise self.Exception(f'Could not find JS function "{funcname}"')
 774         return [x.strip() for x in func_m.group('args').split(',')], code
 775
 776     def extract_function(self, funcname):
 777         return self.extract_function_from_code(*self.extract_function_code(funcname))
 778
 779     def extract_function_from_code(self, argnames, code, *global_stack):
 780         local_vars = {}
 781         while True:
 782             mobj = re.search(r'function\((?P<args>[^)]*)\)\s*{', code)
 783             if mobj is None:
 784                 break
 785             start, body_start = mobj.span()
 786             body, remaining = self._separate_at_paren(code[body_start - 1:], '}')
 787             name = self._named_object(local_vars, self.extract_function_from_code(
 788                 [x.strip() for x in mobj.group('args').split(',')],
 789                 body, local_vars, *global_stack))
 790             code = code[:start] + name + remaining
 791         return self.build_function(argnames, code, local_vars, *global_stack)
 792
 793     def call_function(self, funcname, *args):
 794         return self.extract_function(funcname)(args)
 795
 796     def build_function(self, argnames, code, *global_stack):
 797         global_stack = list(global_stack) or [{}]
 798         argnames = tuple(argnames)
 799
 800         def resf(args, kwargs={}, allow_recursion=100):
 801             global_stack[0].update(itertools.zip_longest(argnames, args, fillvalue=None))
 802             global_stack[0].update(kwargs)
 803             var_stack = LocalNameSpace(*global_stack)
 804             ret, should_abort = self.interpret_statement(code.replace('\n', ' '), var_stack, allow_recursion - 1)
 805             if should_abort:
 806                 return ret
 807         return resf