yt_dlp/jsinterp.py

   1 import collections
   2 import contextlib
   3 import json
   4 import operator
   5 import re
   6
   7 from .utils import ExtractorError, remove_quotes
   8
   9 _OPERATORS = [
  10     ('|', operator.or_),
  11     ('^', operator.xor),
  12     ('&', operator.and_),
  13     ('>>', operator.rshift),
  14     ('<<', operator.lshift),
  15     ('-', operator.sub),
  16     ('+', operator.add),
  17     ('%', operator.mod),
  18     ('/', operator.truediv),
  19     ('*', operator.mul),
  20 ]
  21 _ASSIGN_OPERATORS = [(op + '=', opfunc) for op, opfunc in _OPERATORS]
  22 _ASSIGN_OPERATORS.append(('=', (lambda cur, right: right)))
  23
  24 _NAME_RE = r'[a-zA-Z_$][a-zA-Z_$0-9]*'
  25
  26 _MATCHING_PARENS = dict(zip('({[', ')}]'))
  27
  28
  29 class JS_Break(ExtractorError):
  30     def __init__(self):
  31         ExtractorError.__init__(self, 'Invalid break')
  32
  33
  34 class JS_Continue(ExtractorError):
  35     def __init__(self):
  36         ExtractorError.__init__(self, 'Invalid continue')
  37
  38
  39 class LocalNameSpace(collections.ChainMap):
  40     def __setitem__(self, key, value):
  41         for scope in self.maps:
  42             if key in scope:
  43                 scope[key] = value
  44                 return
  45         self.maps[0][key] = value
  46
  47     def __delitem__(self, key):
  48         raise NotImplementedError('Deleting is not supported')
  49
  50
  51 class JSInterpreter:
  52     def __init__(self, code, objects=None):
  53         if objects is None:
  54             objects = {}
  55         self.code = code
  56         self._functions = {}
  57         self._objects = objects
  58         self.__named_object_counter = 0
  59
  60     def _named_object(self, namespace, obj):
  61         self.__named_object_counter += 1
  62         name = f'__yt_dlp_jsinterp_obj{self.__named_object_counter}'
  63         namespace[name] = obj
  64         return name
  65
  66     @staticmethod
  67     def _separate(expr, delim=',', max_split=None):
  68         if not expr:
  69             return
  70         counters = {k: 0 for k in _MATCHING_PARENS.values()}
  71         start, splits, pos, delim_len = 0, 0, 0, len(delim) - 1
  72         for idx, char in enumerate(expr):
  73             if char in _MATCHING_PARENS:
  74                 counters[_MATCHING_PARENS[char]] += 1
  75             elif char in counters:
  76                 counters[char] -= 1
  77             if char != delim[pos] or any(counters.values()):
  78                 pos = 0
  79                 continue
  80             elif pos != delim_len:
  81                 pos += 1
  82                 continue
  83             yield expr[start: idx - delim_len]
  84             start, pos = idx + 1, 0
  85             splits += 1
  86             if max_split and splits >= max_split:
  87                 break
  88         yield expr[start:]
  89
  90     @staticmethod
  91     def _separate_at_paren(expr, delim):
  92         separated = list(JSInterpreter._separate(expr, delim, 1))
  93         if len(separated) < 2:
  94             raise ExtractorError(f'No terminating paren {delim} in {expr}')
  95         return separated[0][1:].strip(), separated[1].strip()
  96
  97     def interpret_statement(self, stmt, local_vars, allow_recursion=100):
  98         if allow_recursion < 0:
  99             raise ExtractorError('Recursion limit reached')
 100
 101         sub_statements = list(self._separate(stmt, ';'))
 102         stmt = (sub_statements or ['']).pop()
 103         for sub_stmt in sub_statements:
 104             ret, should_abort = self.interpret_statement(sub_stmt, local_vars, allow_recursion - 1)
 105             if should_abort:
 106                 return ret
 107
 108         should_abort = False
 109         stmt = stmt.lstrip()
 110         stmt_m = re.match(r'var\s', stmt)
 111         if stmt_m:
 112             expr = stmt[len(stmt_m.group(0)):]
 113         else:
 114             return_m = re.match(r'return(?:\s+|$)', stmt)
 115             if return_m:
 116                 expr = stmt[len(return_m.group(0)):]
 117                 should_abort = True
 118             else:
 119                 # Try interpreting it as an expression
 120                 expr = stmt
 121
 122         v = self.interpret_expression(expr, local_vars, allow_recursion)
 123         return v, should_abort
 124
 125     def interpret_expression(self, expr, local_vars, allow_recursion):
 126         expr = expr.strip()
 127         if expr == '':  # Empty expression
 128             return None
 129
 130         if expr.startswith('{'):
 131             inner, outer = self._separate_at_paren(expr, '}')
 132             inner, should_abort = self.interpret_statement(inner, local_vars, allow_recursion - 1)
 133             if not outer or should_abort:
 134                 return inner
 135             else:
 136                 expr = json.dumps(inner) + outer
 137
 138         if expr.startswith('('):
 139             inner, outer = self._separate_at_paren(expr, ')')
 140             inner = self.interpret_expression(inner, local_vars, allow_recursion)
 141             if not outer:
 142                 return inner
 143             else:
 144                 expr = json.dumps(inner) + outer
 145
 146         if expr.startswith('['):
 147             inner, outer = self._separate_at_paren(expr, ']')
 148             name = self._named_object(local_vars, [
 149                 self.interpret_expression(item, local_vars, allow_recursion)
 150                 for item in self._separate(inner)])
 151             expr = name + outer
 152
 153         m = re.match(r'try\s*', expr)
 154         if m:
 155             if expr[m.end()] == '{':
 156                 try_expr, expr = self._separate_at_paren(expr[m.end():], '}')
 157             else:
 158                 try_expr, expr = expr[m.end() - 1:], ''
 159             ret, should_abort = self.interpret_statement(try_expr, local_vars, allow_recursion - 1)
 160             if should_abort:
 161                 return ret
 162             return self.interpret_statement(expr, local_vars, allow_recursion - 1)[0]
 163
 164         m = re.match(r'catch\s*\(', expr)
 165         if m:
 166             # We ignore the catch block
 167             _, expr = self._separate_at_paren(expr, '}')
 168             return self.interpret_statement(expr, local_vars, allow_recursion - 1)[0]
 169
 170         m = re.match(r'for\s*\(', expr)
 171         if m:
 172             constructor, remaining = self._separate_at_paren(expr[m.end() - 1:], ')')
 173             if remaining.startswith('{'):
 174                 body, expr = self._separate_at_paren(remaining, '}')
 175             else:
 176                 m = re.match(r'switch\s*\(', remaining)  # FIXME
 177                 if m:
 178                     switch_val, remaining = self._separate_at_paren(remaining[m.end() - 1:], ')')
 179                     body, expr = self._separate_at_paren(remaining, '}')
 180                     body = 'switch(%s){%s}' % (switch_val, body)
 181                 else:
 182                     body, expr = remaining, ''
 183             start, cndn, increment = self._separate(constructor, ';')
 184             if self.interpret_statement(start, local_vars, allow_recursion - 1)[1]:
 185                 raise ExtractorError(
 186                     f'Premature return in the initialization of a for loop in {constructor!r}')
 187             while True:
 188                 if not self.interpret_expression(cndn, local_vars, allow_recursion):
 189                     break
 190                 try:
 191                     ret, should_abort = self.interpret_statement(body, local_vars, allow_recursion - 1)
 192                     if should_abort:
 193                         return ret
 194                 except JS_Break:
 195                     break
 196                 except JS_Continue:
 197                     pass
 198                 if self.interpret_statement(increment, local_vars, allow_recursion - 1)[1]:
 199                     raise ExtractorError(
 200                         f'Premature return in the initialization of a for loop in {constructor!r}')
 201             return self.interpret_statement(expr, local_vars, allow_recursion - 1)[0]
 202
 203         m = re.match(r'switch\s*\(', expr)
 204         if m:
 205             switch_val, remaining = self._separate_at_paren(expr[m.end() - 1:], ')')
 206             switch_val = self.interpret_expression(switch_val, local_vars, allow_recursion)
 207             body, expr = self._separate_at_paren(remaining, '}')
 208             items = body.replace('default:', 'case default:').split('case ')[1:]
 209             for default in (False, True):
 210                 matched = False
 211                 for item in items:
 212                     case, stmt = (i.strip() for i in self._separate(item, ':', 1))
 213                     if default:
 214                         matched = matched or case == 'default'
 215                     elif not matched:
 216                         matched = case != 'default' and switch_val == self.interpret_expression(case, local_vars, allow_recursion)
 217                     if not matched:
 218                         continue
 219                     try:
 220                         ret, should_abort = self.interpret_statement(stmt, local_vars, allow_recursion - 1)
 221                         if should_abort:
 222                             return ret
 223                     except JS_Break:
 224                         break
 225                 if matched:
 226                     break
 227             return self.interpret_statement(expr, local_vars, allow_recursion - 1)[0]
 228
 229         # Comma separated statements
 230         sub_expressions = list(self._separate(expr))
 231         expr = sub_expressions.pop().strip() if sub_expressions else ''
 232         for sub_expr in sub_expressions:
 233             self.interpret_expression(sub_expr, local_vars, allow_recursion)
 234
 235         for m in re.finditer(rf'''(?x)
 236                 (?P<pre_sign>\+\+|--)(?P<var1>{_NAME_RE})|
 237                 (?P<var2>{_NAME_RE})(?P<post_sign>\+\+|--)''', expr):
 238             var = m.group('var1') or m.group('var2')
 239             start, end = m.span()
 240             sign = m.group('pre_sign') or m.group('post_sign')
 241             ret = local_vars[var]
 242             local_vars[var] += 1 if sign[0] == '+' else -1
 243             if m.group('pre_sign'):
 244                 ret = local_vars[var]
 245             expr = expr[:start] + json.dumps(ret) + expr[end:]
 246
 247         for op, opfunc in _ASSIGN_OPERATORS:
 248             m = re.match(rf'''(?x)
 249                 (?P<out>{_NAME_RE})(?:\[(?P<index>[^\]]+?)\])?
 250                 \s*{re.escape(op)}
 251                 (?P<expr>.*)$''', expr)
 252             if not m:
 253                 continue
 254             right_val = self.interpret_expression(m.group('expr'), local_vars, allow_recursion)
 255
 256             if m.groupdict().get('index'):
 257                 lvar = local_vars[m.group('out')]
 258                 idx = self.interpret_expression(m.group('index'), local_vars, allow_recursion)
 259                 if not isinstance(idx, int):
 260                     raise ExtractorError(f'List indices must be integers: {idx}')
 261                 cur = lvar[idx]
 262                 val = opfunc(cur, right_val)
 263                 lvar[idx] = val
 264                 return val
 265             else:
 266                 cur = local_vars.get(m.group('out'))
 267                 val = opfunc(cur, right_val)
 268                 local_vars[m.group('out')] = val
 269                 return val
 270
 271         if expr.isdigit():
 272             return int(expr)
 273
 274         if expr == 'break':
 275             raise JS_Break()
 276         elif expr == 'continue':
 277             raise JS_Continue()
 278
 279         var_m = re.match(
 280             r'(?!if|return|true|false|null)(?P<name>%s)$' % _NAME_RE,
 281             expr)
 282         if var_m:
 283             return local_vars[var_m.group('name')]
 284
 285         with contextlib.suppress(ValueError):
 286             return json.loads(expr)
 287
 288         m = re.match(
 289             r'(?P<in>%s)\[(?P<idx>.+)\]$' % _NAME_RE, expr)
 290         if m:
 291             val = local_vars[m.group('in')]
 292             idx = self.interpret_expression(m.group('idx'), local_vars, allow_recursion)
 293             return val[idx]
 294
 295         for op, opfunc in _OPERATORS:
 296             separated = list(self._separate(expr, op))
 297             if len(separated) < 2:
 298                 continue
 299             right_val = separated.pop()
 300             left_val = op.join(separated)
 301             left_val, should_abort = self.interpret_statement(
 302                 left_val, local_vars, allow_recursion - 1)
 303             if should_abort:
 304                 raise ExtractorError(f'Premature left-side return of {op} in {expr!r}')
 305             right_val, should_abort = self.interpret_statement(
 306                 right_val, local_vars, allow_recursion - 1)
 307             if should_abort:
 308                 raise ExtractorError(f'Premature right-side return of {op} in {expr!r}')
 309             return opfunc(left_val or 0, right_val)
 310
 311         m = re.match(
 312             r'(?P<var>%s)(?:\.(?P<member>[^(]+)|\[(?P<member2>[^]]+)\])\s*' % _NAME_RE,
 313             expr)
 314         if m:
 315             variable = m.group('var')
 316             member = remove_quotes(m.group('member') or m.group('member2'))
 317             arg_str = expr[m.end():]
 318             if arg_str.startswith('('):
 319                 arg_str, remaining = self._separate_at_paren(arg_str, ')')
 320             else:
 321                 arg_str, remaining = None, arg_str
 322
 323             def assertion(cndn, msg):
 324                 """ assert, but without risk of getting optimized out """
 325                 if not cndn:
 326                     raise ExtractorError(f'{member} {msg}: {expr}')
 327
 328             def eval_method():
 329                 nonlocal member
 330                 if variable == 'String':
 331                     obj = str
 332                 elif variable in local_vars:
 333                     obj = local_vars[variable]
 334                 else:
 335                     if variable not in self._objects:
 336                         self._objects[variable] = self.extract_object(variable)
 337                     obj = self._objects[variable]
 338
 339                 if arg_str is None:
 340                     # Member access
 341                     if member == 'length':
 342                         return len(obj)
 343                     return obj[member]
 344
 345                 # Function call
 346                 argvals = [
 347                     self.interpret_expression(v, local_vars, allow_recursion)
 348                     for v in self._separate(arg_str)]
 349
 350                 if obj == str:
 351                     if member == 'fromCharCode':
 352                         assertion(argvals, 'takes one or more arguments')
 353                         return ''.join(map(chr, argvals))
 354                     raise ExtractorError(f'Unsupported string method {member}')
 355
 356                 if member == 'split':
 357                     assertion(argvals, 'takes one or more arguments')
 358                     assertion(argvals == [''], 'with arguments is not implemented')
 359                     return list(obj)
 360                 elif member == 'join':
 361                     assertion(isinstance(obj, list), 'must be applied on a list')
 362                     assertion(len(argvals) == 1, 'takes exactly one argument')
 363                     return argvals[0].join(obj)
 364                 elif member == 'reverse':
 365                     assertion(not argvals, 'does not take any arguments')
 366                     obj.reverse()
 367                     return obj
 368                 elif member == 'slice':
 369                     assertion(isinstance(obj, list), 'must be applied on a list')
 370                     assertion(len(argvals) == 1, 'takes exactly one argument')
 371                     return obj[argvals[0]:]
 372                 elif member == 'splice':
 373                     assertion(isinstance(obj, list), 'must be applied on a list')
 374                     assertion(argvals, 'takes one or more arguments')
 375                     index, howMany = map(int, (argvals + [len(obj)])[:2])
 376                     if index < 0:
 377                         index += len(obj)
 378                     add_items = argvals[2:]
 379                     res = []
 380                     for i in range(index, min(index + howMany, len(obj))):
 381                         res.append(obj.pop(index))
 382                     for i, item in enumerate(add_items):
 383                         obj.insert(index + i, item)
 384                     return res
 385                 elif member == 'unshift':
 386                     assertion(isinstance(obj, list), 'must be applied on a list')
 387                     assertion(argvals, 'takes one or more arguments')
 388                     for item in reversed(argvals):
 389                         obj.insert(0, item)
 390                     return obj
 391                 elif member == 'pop':
 392                     assertion(isinstance(obj, list), 'must be applied on a list')
 393                     assertion(not argvals, 'does not take any arguments')
 394                     if not obj:
 395                         return
 396                     return obj.pop()
 397                 elif member == 'push':
 398                     assertion(argvals, 'takes one or more arguments')
 399                     obj.extend(argvals)
 400                     return obj
 401                 elif member == 'forEach':
 402                     assertion(argvals, 'takes one or more arguments')
 403                     assertion(len(argvals) <= 2, 'takes at-most 2 arguments')
 404                     f, this = (argvals + [''])[:2]
 405                     return [f((item, idx, obj), this=this) for idx, item in enumerate(obj)]
 406                 elif member == 'indexOf':
 407                     assertion(argvals, 'takes one or more arguments')
 408                     assertion(len(argvals) <= 2, 'takes at-most 2 arguments')
 409                     idx, start = (argvals + [0])[:2]
 410                     try:
 411                         return obj.index(idx, start)
 412                     except ValueError:
 413                         return -1
 414
 415                 if isinstance(obj, list):
 416                     member = int(member)
 417                 return obj[member](argvals)
 418
 419             if remaining:
 420                 return self.interpret_expression(
 421                     self._named_object(local_vars, eval_method()) + remaining,
 422                     local_vars, allow_recursion)
 423             else:
 424                 return eval_method()
 425
 426         m = re.match(r'^(?P<func>%s)\((?P<args>[a-zA-Z0-9_$,]*)\)$' % _NAME_RE, expr)
 427         if m:
 428             fname = m.group('func')
 429             argvals = tuple(
 430                 int(v) if v.isdigit() else local_vars[v]
 431                 for v in self._separate(m.group('args')))
 432             if fname in local_vars:
 433                 return local_vars[fname](argvals)
 434             elif fname not in self._functions:
 435                 self._functions[fname] = self.extract_function(fname)
 436             return self._functions[fname](argvals)
 437
 438         if expr:
 439             raise ExtractorError('Unsupported JS expression %r' % expr)
 440
 441     def extract_object(self, objname):
 442         _FUNC_NAME_RE = r'''(?:[a-zA-Z$0-9]+|"[a-zA-Z$0-9]+"|'[a-zA-Z$0-9]+')'''
 443         obj = {}
 444         obj_m = re.search(
 445             r'''(?x)
 446                 (?<!this\.)%s\s*=\s*{\s*
 447                     (?P<fields>(%s\s*:\s*function\s*\(.*?\)\s*{.*?}(?:,\s*)?)*)
 448                 }\s*;
 449             ''' % (re.escape(objname), _FUNC_NAME_RE),
 450             self.code)
 451         fields = obj_m.group('fields')
 452         # Currently, it only supports function definitions
 453         fields_m = re.finditer(
 454             r'''(?x)
 455                 (?P<key>%s)\s*:\s*function\s*\((?P<args>[a-z,]+)\){(?P<code>[^}]+)}
 456             ''' % _FUNC_NAME_RE,
 457             fields)
 458         for f in fields_m:
 459             argnames = f.group('args').split(',')
 460             obj[remove_quotes(f.group('key'))] = self.build_function(argnames, f.group('code'))
 461
 462         return obj
 463
 464     def extract_function_code(self, funcname):
 465         """ @returns argnames, code """
 466         func_m = re.search(
 467             r'''(?x)
 468                 (?:function\s+%s|[{;,]\s*%s\s*=\s*function|var\s+%s\s*=\s*function)\s*
 469                 \((?P<args>[^)]*)\)\s*
 470                 (?P<code>\{(?:(?!};)[^"]|"([^"]|\\")*")+\})''' % (
 471                 re.escape(funcname), re.escape(funcname), re.escape(funcname)),
 472             self.code)
 473         code, _ = self._separate_at_paren(func_m.group('code'), '}')  # refine the match
 474         if func_m is None:
 475             raise ExtractorError('Could not find JS function %r' % funcname)
 476         return func_m.group('args').split(','), code
 477
 478     def extract_function(self, funcname):
 479         return self.extract_function_from_code(*self.extract_function_code(funcname))
 480
 481     def extract_function_from_code(self, argnames, code, *global_stack):
 482         local_vars = {}
 483         while True:
 484             mobj = re.search(r'function\((?P<args>[^)]*)\)\s*{', code)
 485             if mobj is None:
 486                 break
 487             start, body_start = mobj.span()
 488             body, remaining = self._separate_at_paren(code[body_start - 1:], '}')
 489             name = self._named_object(
 490                 local_vars,
 491                 self.extract_function_from_code(
 492                     [str.strip(x) for x in mobj.group('args').split(',')],
 493                     body, local_vars, *global_stack))
 494             code = code[:start] + name + remaining
 495         return self.build_function(argnames, code, local_vars, *global_stack)
 496
 497     def call_function(self, funcname, *args):
 498         return self.extract_function(funcname)(args)
 499
 500     def build_function(self, argnames, code, *global_stack):
 501         global_stack = list(global_stack) or [{}]
 502
 503         def resf(args, **kwargs):
 504             global_stack[0].update({
 505                 **dict(zip(argnames, args)),
 506                 **kwargs
 507             })
 508             var_stack = LocalNameSpace(*global_stack)
 509             for stmt in self._separate(code.replace('\n', ''), ';'):
 510                 ret, should_abort = self.interpret_statement(stmt, var_stack)
 511                 if should_abort:
 512                     break
 513             return ret
 514         return resf