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