youtube_dl/jsinterp.py

   1 from __future__ import unicode_literals
   2
   3 import json
   4 import re
   5
   6 from .utils import (
   7     ExtractorError,
   8 )
   9
  10
  11 class JSInterpreter(object):
  12     def __init__(self, code):
  13         self.code = code
  14         self._functions = {}
  15         self._objects = {}
  16
  17     def interpret_statement(self, stmt, local_vars, allow_recursion=20):
  18         if allow_recursion < 0:
  19             raise ExtractorError('Recursion limit reached')
  20
  21         if stmt.startswith('var '):
  22             stmt = stmt[len('var '):]
  23         ass_m = re.match(r'^(?P<out>[a-z]+)(?:\[(?P<index>[^\]]+)\])?' +
  24                          r'=(?P<expr>.*)$', stmt)
  25         if ass_m:
  26             if ass_m.groupdict().get('index'):
  27                 def assign(val):
  28                     lvar = local_vars[ass_m.group('out')]
  29                     idx = self.interpret_expression(
  30                         ass_m.group('index'), local_vars, allow_recursion)
  31                     assert isinstance(idx, int)
  32                     lvar[idx] = val
  33                     return val
  34                 expr = ass_m.group('expr')
  35             else:
  36                 def assign(val):
  37                     local_vars[ass_m.group('out')] = val
  38                     return val
  39                 expr = ass_m.group('expr')
  40         elif stmt.startswith('return '):
  41             assign = lambda v: v
  42             expr = stmt[len('return '):]
  43         else:
  44             # Try interpreting it as an expression
  45             expr = stmt
  46             assign = lambda v: v
  47
  48         v = self.interpret_expression(expr, local_vars, allow_recursion)
  49         return assign(v)
  50
  51     def interpret_expression(self, expr, local_vars, allow_recursion):
  52         if expr.isdigit():
  53             return int(expr)
  54
  55         if expr.isalpha():
  56             return local_vars[expr]
  57
  58         try:
  59             return json.loads(expr)
  60         except ValueError:
  61             pass
  62
  63         m = re.match(
  64             r'^(?P<var>[$a-zA-Z0-9_]+)\.(?P<member>[^(]+)(?:\(+(?P<args>[^()]*)\))?$',
  65             expr)
  66         if m:
  67             variable = m.group('var')
  68             member = m.group('member')
  69             arg_str = m.group('args')
  70
  71             if variable in local_vars:
  72                 obj = local_vars[variable]
  73             else:
  74                 if variable not in self._objects:
  75                     self._objects[variable] = self.extract_object(variable)
  76                 obj = self._objects[variable]
  77
  78             if arg_str is None:
  79                 # Member access
  80                 if member == 'length':
  81                     return len(obj)
  82                 return obj[member]
  83
  84             assert expr.endswith(')')
  85             # Function call
  86             if arg_str == '':
  87                 argvals = tuple()
  88             else:
  89                 argvals = tuple([
  90                     self.interpret_expression(v, local_vars, allow_recursion)
  91                     for v in arg_str.split(',')])
  92
  93             if member == 'split':
  94                 assert argvals == ('',)
  95                 return list(obj)
  96             if member == 'join':
  97                 assert len(argvals) == 1
  98                 return argvals[0].join(obj)
  99             if member == 'reverse':
 100                 assert len(argvals) == 0
 101                 obj.reverse()
 102                 return obj
 103             if member == 'slice':
 104                 assert len(argvals) == 1
 105                 return obj[argvals[0]:]
 106             if member == 'splice':
 107                 assert isinstance(obj, list)
 108                 index, howMany = argvals
 109                 res = []
 110                 for i in range(index, min(index + howMany, len(obj))):
 111                     res.append(obj.pop(index))
 112                 return res
 113
 114             return obj[member](argvals)
 115
 116         m = re.match(
 117             r'^(?P<in>[a-z]+)\[(?P<idx>.+)\]$', expr)
 118         if m:
 119             val = local_vars[m.group('in')]
 120             idx = self.interpret_expression(
 121                 m.group('idx'), local_vars, allow_recursion - 1)
 122             return val[idx]
 123
 124         m = re.match(r'^(?P<a>.+?)(?P<op>[%])(?P<b>.+?)$', expr)
 125         if m:
 126             a = self.interpret_expression(
 127                 m.group('a'), local_vars, allow_recursion)
 128             b = self.interpret_expression(
 129                 m.group('b'), local_vars, allow_recursion)
 130             return a % b
 131
 132         m = re.match(
 133             r'^(?P<func>[a-zA-Z$]+)\((?P<args>[a-z0-9,]+)\)$', expr)
 134         if m:
 135             fname = m.group('func')
 136             argvals = tuple([
 137                 int(v) if v.isdigit() else local_vars[v]
 138                 for v in m.group('args').split(',')])
 139             if fname not in self._functions:
 140                 self._functions[fname] = self.extract_function(fname)
 141             return self._functions[fname](argvals)
 142         raise ExtractorError('Unsupported JS expression %r' % expr)
 143
 144     def extract_object(self, objname):
 145         obj = {}
 146         obj_m = re.search(
 147             (r'(?:var\s+)?%s\s*=\s*\{' % re.escape(objname)) +
 148             r'\s*(?P<fields>([a-zA-Z$0-9]+\s*:\s*function\(.*?\)\s*\{.*?\})*)' +
 149             r'\}\s*;',
 150             self.code)
 151         fields = obj_m.group('fields')
 152         # Currently, it only supports function definitions
 153         fields_m = re.finditer(
 154             r'(?P<key>[a-zA-Z$0-9]+)\s*:\s*function'
 155             r'\((?P<args>[a-z,]+)\){(?P<code>[^}]+)}',
 156             fields)
 157         for f in fields_m:
 158             argnames = f.group('args').split(',')
 159             obj[f.group('key')] = self.build_function(argnames, f.group('code'))
 160
 161         return obj
 162
 163     def extract_function(self, funcname):
 164         func_m = re.search(
 165             (r'(?:function %s|[{;]%s\s*=\s*function)' % (
 166                 re.escape(funcname), re.escape(funcname))) +
 167             r'\((?P<args>[a-z,]+)\){(?P<code>[^}]+)}',
 168             self.code)
 169         if func_m is None:
 170             raise ExtractorError('Could not find JS function %r' % funcname)
 171         argnames = func_m.group('args').split(',')
 172
 173         return self.build_function(argnames, func_m.group('code'))
 174
 175     def build_function(self, argnames, code):
 176         def resf(args):
 177             local_vars = dict(zip(argnames, args))
 178             for stmt in code.split(';'):
 179                 res = self.interpret_statement(stmt, local_vars)
 180             return res
 181         return resf