Imported Upstream version 2014.11.21

[youtubedl] / youtube_dl / jsinterp.py

diff --git a/youtube_dl/jsinterp.py b/youtube_dl/jsinterp.py
index 3bbb07704128cf1ab27197ca1abbeeaec086b36f..b4617fbad0fc40323a129ce1218f9f97590c89bb 100644 (file)
--- a/youtube_dl/jsinterp.py
+++ b/youtube_dl/jsinterp.py
@@ -1,5 +1,6 @@
 from __future__ import unicode_literals
 
 from __future__ import unicode_literals
 

+import json

 import re
 
 from .utils import (
 import re
 
 from .utils import (


@@ -11,6 +12,7 @@ class JSInterpreter(object):
     def __init__(self, code):
         self.code = code
         self._functions = {}
     def __init__(self, code):
         self.code = code
         self._functions = {}

+        self._objects = {}

 
     def interpret_statement(self, stmt, local_vars, allow_recursion=20):
         if allow_recursion < 0:
 
     def interpret_statement(self, stmt, local_vars, allow_recursion=20):
         if allow_recursion < 0:


@@ -39,8 +41,9 @@ class JSInterpreter(object):
             assign = lambda v: v
             expr = stmt[len('return '):]
         else:
             assign = lambda v: v
             expr = stmt[len('return '):]
         else:

-            raise ExtractorError(
-                'Cannot determine left side of statement in %r' % stmt)
+            # Try interpreting it as an expression
+            expr = stmt
+            assign = lambda v: v

 
         v = self.interpret_expression(expr, local_vars, allow_recursion)
         return assign(v)
 
         v = self.interpret_expression(expr, local_vars, allow_recursion)
         return assign(v)


@@ -52,23 +55,63 @@ class JSInterpreter(object):
         if expr.isalpha():
             return local_vars[expr]
 
         if expr.isalpha():
             return local_vars[expr]
 

-        m = re.match(r'^(?P<in>[a-z]+)\.(?P<member>.*)$', expr)
+        try:
+            return json.loads(expr)
+        except ValueError:
+            pass
+
+        m = re.match(
+            r'^(?P<var>[$a-zA-Z0-9_]+)\.(?P<member>[^(]+)(?:\(+(?P<args>[^()]*)\))?$',
+            expr)

         if m:
         if m:

+            variable = m.group('var')

             member = m.group('member')
             member = m.group('member')

-            val = local_vars[m.group('in')]
-            if member == 'split("")':
-                return list(val)
-            if member == 'join("")':
-                return ''.join(val)
-            if member == 'length':
-                return len(val)
-            if member == 'reverse()':
-                return val[::-1]
-            slice_m = re.match(r'slice\((?P<idx>.*)\)', member)
-            if slice_m:
-                idx = self.interpret_expression(
-                    slice_m.group('idx'), local_vars, allow_recursion - 1)
-                return val[idx:]
+            arg_str = m.group('args')
+
+            if variable in local_vars:
+                obj = local_vars[variable]
+            else:
+                if variable not in self._objects:
+                    self._objects[variable] = self.extract_object(variable)
+                obj = self._objects[variable]
+
+            if arg_str is None:
+                # Member access
+                if member == 'length':
+                    return len(obj)
+                return obj[member]
+
+            assert expr.endswith(')')
+            # Function call
+            if arg_str == '':
+                argvals = tuple()
+            else:
+                argvals = tuple([
+                    self.interpret_expression(v, local_vars, allow_recursion)
+                    for v in arg_str.split(',')])
+
+            if member == 'split':
+                assert argvals == ('',)
+                return list(obj)
+            if member == 'join':
+                assert len(argvals) == 1
+                return argvals[0].join(obj)
+            if member == 'reverse':
+                assert len(argvals) == 0
+                obj.reverse()
+                return obj
+            if member == 'slice':
+                assert len(argvals) == 1
+                return obj[argvals[0]:]
+            if member == 'splice':
+                assert isinstance(obj, list)
+                index, howMany = argvals
+                res = []
+                for i in range(index, min(index + howMany, len(obj))):
+                    res.append(obj.pop(index))
+                return res
+
+            return obj[member](argvals)

 
         m = re.match(
             r'^(?P<in>[a-z]+)\[(?P<idx>.+)\]$', expr)
 
         m = re.match(
             r'^(?P<in>[a-z]+)\[(?P<idx>.+)\]$', expr)


@@ -90,13 +133,33 @@ class JSInterpreter(object):
             r'^(?P<func>[a-zA-Z$]+)\((?P<args>[a-z0-9,]+)\)$', expr)
         if m:
             fname = m.group('func')
             r'^(?P<func>[a-zA-Z$]+)\((?P<args>[a-z0-9,]+)\)$', expr)
         if m:
             fname = m.group('func')

+            argvals = tuple([
+                int(v) if v.isdigit() else local_vars[v]
+                for v in m.group('args').split(',')])

             if fname not in self._functions:
                 self._functions[fname] = self.extract_function(fname)
             if fname not in self._functions:
                 self._functions[fname] = self.extract_function(fname)

-            argvals = [int(v) if v.isdigit() else local_vars[v]
-                       for v in m.group('args').split(',')]

             return self._functions[fname](argvals)
         raise ExtractorError('Unsupported JS expression %r' % expr)
 
             return self._functions[fname](argvals)
         raise ExtractorError('Unsupported JS expression %r' % expr)
 

+    def extract_object(self, objname):
+        obj = {}
+        obj_m = re.search(
+            (r'(?:var\s+)?%s\s*=\s*\{' % re.escape(objname)) +
+            r'\s*(?P<fields>([a-zA-Z$0-9]+\s*:\s*function\(.*?\)\s*\{.*?\})*)' +
+            r'\}\s*;',
+            self.code)
+        fields = obj_m.group('fields')
+        # Currently, it only supports function definitions
+        fields_m = re.finditer(
+            r'(?P<key>[a-zA-Z$0-9]+)\s*:\s*function'
+            r'\((?P<args>[a-z,]+)\){(?P<code>[^}]+)}',
+            fields)
+        for f in fields_m:
+            argnames = f.group('args').split(',')
+            obj[f.group('key')] = self.build_function(argnames, f.group('code'))
+
+        return obj
+

     def extract_function(self, funcname):
         func_m = re.search(
             (r'(?:function %s|[{;]%s\s*=\s*function)' % (
     def extract_function(self, funcname):
         func_m = re.search(
             (r'(?:function %s|[{;]%s\s*=\s*function)' % (


@@ -107,10 +170,12 @@ class JSInterpreter(object):
             raise ExtractorError('Could not find JS function %r' % funcname)
         argnames = func_m.group('args').split(',')
 
             raise ExtractorError('Could not find JS function %r' % funcname)
         argnames = func_m.group('args').split(',')
 

+        return self.build_function(argnames, func_m.group('code'))
+
+    def build_function(self, argnames, code):

         def resf(args):
             local_vars = dict(zip(argnames, args))
         def resf(args):
             local_vars = dict(zip(argnames, args))

-            for stmt in func_m.group('code').split(';'):
+            for stmt in code.split(';'):

                 res = self.interpret_statement(stmt, local_vars)
             return res
         return resf
                 res = self.interpret_statement(stmt, local_vars)
             return res
         return resf

-