yt_dlp/jsinterp.py

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