Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/jsinterp.py

   1 from __future__ import unicode_literals
   2
   3 import json
   4 import operator
   5 import re
   6
   7 from .utils import (
   8     ExtractorError,
   9 )
  10
  11 _OPERATORS = [
  12     ('|', operator.or_),
  13     ('^', operator.xor),
  14     ('&', operator.and_),
  15     ('>>', operator.rshift),
  16     ('<<', operator.lshift),
  17     ('-', operator.sub),
  18     ('+', operator.add),
  19     ('%', operator.mod),
  20     ('/', operator.truediv),
  21     ('*', operator.mul),
  22 ]
  23 _ASSIGN_OPERATORS = [(op + '=', opfunc) for op, opfunc in _OPERATORS]
  24 _ASSIGN_OPERATORS.append(('=', lambda cur, right: right))
  25
  26 _NAME_RE = r'[a-zA-Z_$][a-zA-Z_$0-9]*'
  27
  28
  29 class JSInterpreter(object):
  30     def __init__(self, code, objects=None):
  31         if objects is None:
  32             objects = {}
  33         self.code = self._remove_comments(code)
  34         self._functions = {}
  35         self._objects = objects
  36
  37     def _remove_comments(self, code):
  38         return re.sub(r'(?s)/\*.*?\*/', '', code)
  39
  40     def interpret_statement(self, stmt, local_vars, allow_recursion=100):
  41         if allow_recursion < 0:
  42             raise ExtractorError('Recursion limit reached')
  43
  44         should_abort = False
  45         stmt = stmt.lstrip()
  46         stmt_m = re.match(r'var\s', stmt)
  47         if stmt_m:
  48             expr = stmt[len(stmt_m.group(0)):]
  49         else:
  50             return_m = re.match(r'return(?:\s+|$)', stmt)
  51             if return_m:
  52                 expr = stmt[len(return_m.group(0)):]
  53                 should_abort = True
  54             else:
  55                 # Try interpreting it as an expression
  56                 expr = stmt
  57
  58         v = self.interpret_expression(expr, local_vars, allow_recursion)
  59         return v, should_abort
  60
  61     def interpret_expression(self, expr, local_vars, allow_recursion):
  62         expr = expr.strip()
  63
  64         if expr == '':  # Empty expression
  65             return None
  66
  67         if expr.startswith('('):
  68             parens_count = 0
  69             for m in re.finditer(r'[()]', expr):
  70                 if m.group(0) == '(':
  71                     parens_count += 1
  72                 else:
  73                     parens_count -= 1
  74                     if parens_count == 0:
  75                         sub_expr = expr[1:m.start()]
  76                         sub_result = self.interpret_expression(
  77                             sub_expr, local_vars, allow_recursion)
  78                         remaining_expr = expr[m.end():].strip()
  79                         if not remaining_expr:
  80                             return sub_result
  81                         else:
  82                             expr = json.dumps(sub_result) + remaining_expr
  83                         break
  84             else:
  85                 raise ExtractorError('Premature end of parens in %r' % expr)
  86
  87         for op, opfunc in _ASSIGN_OPERATORS:
  88             m = re.match(r'''(?x)
  89                 (?P<out>%s)(?:\[(?P<index>[^\]]+?)\])?
  90                 \s*%s
  91                 (?P<expr>.*)$''' % (_NAME_RE, re.escape(op)), expr)
  92             if not m:
  93                 continue
  94             right_val = self.interpret_expression(
  95                 m.group('expr'), local_vars, allow_recursion - 1)
  96
  97             if m.groupdict().get('index'):
  98                 lvar = local_vars[m.group('out')]
  99                 idx = self.interpret_expression(
 100                     m.group('index'), local_vars, allow_recursion)
 101                 assert isinstance(idx, int)
 102                 cur = lvar[idx]
 103                 val = opfunc(cur, right_val)
 104                 lvar[idx] = val
 105                 return val
 106             else:
 107                 cur = local_vars.get(m.group('out'))
 108                 val = opfunc(cur, right_val)
 109                 local_vars[m.group('out')] = val
 110                 return val
 111
 112         if expr.isdigit():
 113             return int(expr)
 114
 115         var_m = re.match(
 116             r'(?!if|return|true|false)(?P<name>%s)$' % _NAME_RE,
 117             expr)
 118         if var_m:
 119             return local_vars[var_m.group('name')]
 120
 121         try:
 122             return json.loads(expr)
 123         except ValueError:
 124             pass
 125
 126         m = re.match(
 127             r'(?P<var>%s)\.(?P<member>[^(]+)(?:\(+(?P<args>[^()]*)\))?$' % _NAME_RE,
 128             expr)
 129         if m:
 130             variable = m.group('var')
 131             member = m.group('member')
 132             arg_str = m.group('args')
 133
 134             if variable in local_vars:
 135                 obj = local_vars[variable]
 136             else:
 137                 if variable not in self._objects:
 138                     self._objects[variable] = self.extract_object(variable)
 139                 obj = self._objects[variable]
 140
 141             if arg_str is None:
 142                 # Member access
 143                 if member == 'length':
 144                     return len(obj)
 145                 return obj[member]
 146
 147             assert expr.endswith(')')
 148             # Function call
 149             if arg_str == '':
 150                 argvals = tuple()
 151             else:
 152                 argvals = tuple([
 153                     self.interpret_expression(v, local_vars, allow_recursion)
 154                     for v in arg_str.split(',')])
 155
 156             if member == 'split':
 157                 assert argvals == ('',)
 158                 return list(obj)
 159             if member == 'join':
 160                 assert len(argvals) == 1
 161                 return argvals[0].join(obj)
 162             if member == 'reverse':
 163                 assert len(argvals) == 0
 164                 obj.reverse()
 165                 return obj
 166             if member == 'slice':
 167                 assert len(argvals) == 1
 168                 return obj[argvals[0]:]
 169             if member == 'splice':
 170                 assert isinstance(obj, list)
 171                 index, howMany = argvals
 172                 res = []
 173                 for i in range(index, min(index + howMany, len(obj))):
 174                     res.append(obj.pop(index))
 175                 return res
 176
 177             return obj[member](argvals)
 178
 179         m = re.match(
 180             r'(?P<in>%s)\[(?P<idx>.+)\]$' % _NAME_RE, expr)
 181         if m:
 182             val = local_vars[m.group('in')]
 183             idx = self.interpret_expression(
 184                 m.group('idx'), local_vars, allow_recursion - 1)
 185             return val[idx]
 186
 187         for op, opfunc in _OPERATORS:
 188             m = re.match(r'(?P<x>.+?)%s(?P<y>.+)' % re.escape(op), expr)
 189             if not m:
 190                 continue
 191             x, abort = self.interpret_statement(
 192                 m.group('x'), local_vars, allow_recursion - 1)
 193             if abort:
 194                 raise ExtractorError(
 195                     'Premature left-side return of %s in %r' % (op, expr))
 196             y, abort = self.interpret_statement(
 197                 m.group('y'), local_vars, allow_recursion - 1)
 198             if abort:
 199                 raise ExtractorError(
 200                     'Premature right-side return of %s in %r' % (op, expr))
 201             return opfunc(x, y)
 202
 203         m = re.match(
 204             r'^(?P<func>%s)\((?P<args>[a-zA-Z0-9_$,]+)\)$' % _NAME_RE, expr)
 205         if m:
 206             fname = m.group('func')
 207             argvals = tuple([
 208                 int(v) if v.isdigit() else local_vars[v]
 209                 for v in m.group('args').split(',')])
 210             if fname not in self._functions:
 211                 self._functions[fname] = self.extract_function(fname)
 212             return self._functions[fname](argvals)
 213
 214         raise ExtractorError('Unsupported JS expression %r' % expr)
 215
 216     def extract_object(self, objname):
 217         obj = {}
 218         obj_m = re.search(
 219             (r'(?:var\s+)?%s\s*=\s*\{' % re.escape(objname)) +
 220             r'\s*(?P<fields>([a-zA-Z$0-9]+\s*:\s*function\(.*?\)\s*\{.*?\})*)' +
 221             r'\}\s*;',
 222             self.code)
 223         fields = obj_m.group('fields')
 224         # Currently, it only supports function definitions
 225         fields_m = re.finditer(
 226             r'(?P<key>[a-zA-Z$0-9]+)\s*:\s*function'
 227             r'\((?P<args>[a-z,]+)\){(?P<code>[^}]+)}',
 228             fields)
 229         for f in fields_m:
 230             argnames = f.group('args').split(',')
 231             obj[f.group('key')] = self.build_function(argnames, f.group('code'))
 232
 233         return obj
 234
 235     def extract_function(self, funcname):
 236         func_m = re.search(
 237             r'''(?x)
 238                 (?:function\s+%s|[{;]%s\s*=\s*function)\s*
 239                 \((?P<args>[^)]*)\)\s*
 240                 \{(?P<code>[^}]+)\}''' % (
 241                 re.escape(funcname), re.escape(funcname)),
 242             self.code)
 243         if func_m is None:
 244             raise ExtractorError('Could not find JS function %r' % funcname)
 245         argnames = func_m.group('args').split(',')
 246
 247         return self.build_function(argnames, func_m.group('code'))
 248
 249     def call_function(self, funcname, *args):
 250         f = self.extract_function(funcname)
 251         return f(args)
 252
 253     def build_function(self, argnames, code):
 254         def resf(args):
 255             local_vars = dict(zip(argnames, args))
 256             for stmt in code.split(';'):
 257                 res, abort = self.interpret_statement(stmt, local_vars)
 258                 if abort:
 259                     break
 260             return res
 261         return resf