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