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