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