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

   1 from __future__ import unicode_literals
   2
   3 import re
   4
   5 from .utils import (
   6     ExtractorError,
   7 )
   8
   9
  10 class JSInterpreter(object):
  11     def __init__(self, code):
  12         self.code = code
  13         self._functions = {}
  14
  15     def interpret_statement(self, stmt, local_vars, allow_recursion=20):
  16         if allow_recursion < 0:
  17             raise ExtractorError('Recursion limit reached')
  18
  19         if stmt.startswith('var '):
  20             stmt = stmt[len('var '):]
  21         ass_m = re.match(r'^(?P<out>[a-z]+)(?:\[(?P<index>[^\]]+)\])?' +
  22                          r'=(?P<expr>.*)$', stmt)
  23         if ass_m:
  24             if ass_m.groupdict().get('index'):
  25                 def assign(val):
  26                     lvar = local_vars[ass_m.group('out')]
  27                     idx = self.interpret_expression(
  28                         ass_m.group('index'), local_vars, allow_recursion)
  29                     assert isinstance(idx, int)
  30                     lvar[idx] = val
  31                     return val
  32                 expr = ass_m.group('expr')
  33             else:
  34                 def assign(val):
  35                     local_vars[ass_m.group('out')] = val
  36                     return val
  37                 expr = ass_m.group('expr')
  38         elif stmt.startswith('return '):
  39             assign = lambda v: v
  40             expr = stmt[len('return '):]
  41         else:
  42             raise ExtractorError(
  43                 'Cannot determine left side of statement in %r' % stmt)
  44
  45         v = self.interpret_expression(expr, local_vars, allow_recursion)
  46         return assign(v)
  47
  48     def interpret_expression(self, expr, local_vars, allow_recursion):
  49         if expr.isdigit():
  50             return int(expr)
  51
  52         if expr.isalpha():
  53             return local_vars[expr]
  54
  55         m = re.match(r'^(?P<in>[a-z]+)\.(?P<member>.*)$', expr)
  56         if m:
  57             member = m.group('member')
  58             val = local_vars[m.group('in')]
  59             if member == 'split("")':
  60                 return list(val)
  61             if member == 'join("")':
  62                 return u''.join(val)
  63             if member == 'length':
  64                 return len(val)
  65             if member == 'reverse()':
  66                 return val[::-1]
  67             slice_m = re.match(r'slice\((?P<idx>.*)\)', member)
  68             if slice_m:
  69                 idx = self.interpret_expression(
  70                     slice_m.group('idx'), local_vars, allow_recursion - 1)
  71                 return val[idx:]
  72
  73         m = re.match(
  74             r'^(?P<in>[a-z]+)\[(?P<idx>.+)\]$', expr)
  75         if m:
  76             val = local_vars[m.group('in')]
  77             idx = self.interpret_expression(
  78                 m.group('idx'), local_vars, allow_recursion - 1)
  79             return val[idx]
  80
  81         m = re.match(r'^(?P<a>.+?)(?P<op>[%])(?P<b>.+?)$', expr)
  82         if m:
  83             a = self.interpret_expression(
  84                 m.group('a'), local_vars, allow_recursion)
  85             b = self.interpret_expression(
  86                 m.group('b'), local_vars, allow_recursion)
  87             return a % b
  88
  89         m = re.match(
  90             r'^(?P<func>[a-zA-Z$]+)\((?P<args>[a-z0-9,]+)\)$', expr)
  91         if m:
  92             fname = m.group('func')
  93             if fname not in self._functions:
  94                 self._functions[fname] = self.extract_function(fname)
  95             argvals = [int(v) if v.isdigit() else local_vars[v]
  96                        for v in m.group('args').split(',')]
  97             return self._functions[fname](argvals)
  98         raise ExtractorError('Unsupported JS expression %r' % expr)
  99
 100     def extract_function(self, funcname):
 101         func_m = re.search(
 102             (r'(?:function %s|%s\s*=\s*function)' % (
 103                 re.escape(funcname), re.escape(funcname))) +
 104             r'\((?P<args>[a-z,]+)\){(?P<code>[^}]+)}',
 105             self.code)
 106         if func_m is None:
 107             raise ExtractorError('Could not find JS function %r' % funcname)
 108         argnames = func_m.group('args').split(',')
 109
 110         def resf(args):
 111             local_vars = dict(zip(argnames, args))
 112             for stmt in func_m.group('code').split(';'):
 113                 res = self.interpret_statement(stmt, local_vars)
 114             return res
 115         return resf
 116