зеркало из https://github.com/mozilla/gecko-dev.git
123 строки
6.7 KiB
HTML
123 строки
6.7 KiB
HTML
<!DOCTYPE html PUBLIC "-//w3c//dtd html 4.0 transitional//en">
|
|
<html>
|
|
<head>
|
|
|
|
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
|
|
|
|
<meta name="Author" content="Norris Boyd">
|
|
|
|
<meta name="GENERATOR" content="Mozilla/4.72 [en]C-NSCP (WinNT; U) [Netscape]">
|
|
|
|
<meta name="KeyWords" content="Rhino, JavaScript, Java">
|
|
<title>Serialization</title>
|
|
</head>
|
|
<body bgcolor="#ffffff">
|
|
|
|
<script src="owner.js"></script>
|
|
<center>
|
|
<h1>Serialization</h1>
|
|
</center>
|
|
|
|
<script>document.write(owner());</script> <br>
|
|
<script>
|
|
var d = new Date(document.lastModified);
|
|
document.write((d.getMonth()+1)+"/"+d.getDate()+"/"+d.getFullYear());
|
|
document.write('<br>');
|
|
</script>
|
|
<center>
|
|
<hr width="100%"></center>
|
|
|
|
<p>Beginning with Rhino 1.5 Release 3 it is possible to serialize JavaScript
|
|
objects, including functions and scripts. However, serialization of
|
|
code in compilation mode has some significant limitations.. Serialization
|
|
provides a way to save the state of an object and write it out to a file
|
|
or send it across a network connection. <br>
|
|
</p>
|
|
<h2>Simple serialization example</h2>
|
|
The Rhino shell has two new top-level functions, serialize and deserialize.
|
|
They're intended mainly as examples of the use of serialization:<br>
|
|
<pre>$ java org.mozilla.javascript.tools.shell.Main<br>js> function f() { return 3; }<br>js> serialize(f, "f.ser")<br>js> quit()<br><br>$ java org.mozilla.javascript.tools.shell.Main<br>js> f = deserialize("f.ser")<br><br>function f() {<br> return 3;<br>}<br><br>js> f()<br>3<br>js></pre>
|
|
<pre></pre>
|
|
<pre></pre>
|
|
<pre></pre>
|
|
<pre></pre>
|
|
<pre></pre>
|
|
<pre></pre>
|
|
<pre></pre>
|
|
<pre></pre>
|
|
<pre></pre>
|
|
Here we see a simple case of a function being serialized to a file and then
|
|
read into a new instance of Rhino and called. <br>
|
|
<br>
|
|
<h2>Rhino serialization APIs</h2>
|
|
Two new classes, ScriptableOutputStream and ScriptableInputStream, were introduced
|
|
to handle serialization of Rhino classes. These classes extend ObjectOutputStream
|
|
and ObjectInputStream respectively. Writing an object to a file can be done
|
|
in a few lines of Java code:<br>
|
|
<pre>FileOutputStream fos = new FileOutputStream(filename);<br>ScriptableOutputStream out = new ScriptableOutputStream(fos, scope);<br>out.writeObject(obj);<br>out.close();</pre>
|
|
<p>Here filename is the file to write to, obj is the object or function to
|
|
write, and scope is the top-level scope containing obj. </p>
|
|
<p>Reading the serialized object back into memory is similarly simple:</p>
|
|
<pre>FileInputStream fis = new FileInputStream(filename);<br>ObjectInputStream in = new ScriptableInputStream(fis, scope);<br>Object deserialized = in.readObject();<br>in.close();<br></pre>
|
|
<p>Again, we need the scope to create our serialization stream class. </p>
|
|
<p>So why do we need these specialized stream classes instead of simply using
|
|
ObjectOutputStream and ObjectInputStream? To understand the answer we must
|
|
know what goes on behind the scenes when Rhino serializes objects. </p>
|
|
<h2>How Rhino serialization works</h2>
|
|
By default, Java serialization of an object also serializes objects that
|
|
are referred to by that object. Upon deserialization the initial object and
|
|
the objects it refers to are all created and the references between the objects
|
|
are resolved. <br>
|
|
<br>
|
|
However, for JavaScript this creates a problem. JavaScript objects contain
|
|
references to prototypes and to parent scopes. Default serialization would
|
|
serialize the object or function we desired but would also serialize Object.prototype
|
|
or even possibly the entire top-level scope and everything it refers to!
|
|
We want to be able to serialize a JavaScript object and then deserialize
|
|
it into a new scope and have all of the references from the deserialized
|
|
object to prototypes and parent scopes resolved correctly to refer to objects
|
|
in the new scope. <br>
|
|
<br>
|
|
ScriptableOutputStream takes a scope as a parameter to its constructor. If
|
|
in the process of serialization it encounters a reference to the scope it
|
|
will serialize a marker that will be resolved to the new scope upon deserialization.
|
|
It is also possible to add names of objects to a list in the ScriptableOutputStream
|
|
object. These objects will also be saved as markers upon serialization and
|
|
resolved in the new scope upon deserialization. Use the addExcludedName method
|
|
of ScriptableOutputStream to add new names. By default, ScriptableOutputStream
|
|
excludes all the names defined using Context.initStandardObjects.<br>
|
|
<br>
|
|
If you are using Rhino serialization in an environment where you always define,
|
|
say, a constructor "Foo", you should add the following code before calling
|
|
writeObject:<br>
|
|
<pre>out.addExcludedName("Foo");<br>out.addExcludedName("Foo.prototype");<br></pre>
|
|
This code will prevent Foo and Foo.prototype from being serialized and will
|
|
cause references to Foo or Foo.prototype to be resolved to the objects in
|
|
the new scope upon deserialization. Exceptions will be thrown if Foo or Foo.prototype
|
|
cannot be found the scopes used in either ScriptableOutputStream or ScriptableInputStream.<br>
|
|
<br>
|
|
<h2>Rhino serialization in compilation mode</h2>
|
|
Serialization works well with objects and with functions and scripts in
|
|
interpretive mode. However, you can run into problems with serialization
|
|
of compiled functions and scripts:<br>
|
|
<pre>$ cat test.js<br>function f() { return 3; }<br>serialize(f, "f.ser");<br>g = deserialize("f.ser");<br>print(g());<br>$ java org.mozilla.javascript.tools.shell.Main -opt -1 test.js<br>3<br>$ java org.mozilla.javascript.tools.shell.Main test.js<br>js: uncaught JavaScript exception: java.lang.ClassNotFoundException: c1<br></pre>
|
|
<p>The problem is that Java serialization has no built-in way to serialize
|
|
Java classes themselves. (It might be possible to save the Java bytecodes
|
|
in an array and then load the class upon deserialization, but at best that
|
|
would eat up a lot of memory for just this feature.) One way around this
|
|
is to compile the functions using the jsc tool: </p>
|
|
<pre>$ cat f.js<br>function f() { return 3; }<br>$ java -classpath js.jar org.mozilla.javascript.tools.jsc.Main f.js<br>$ cat test2.js<br>loadClass("f");<br>serialize(f, "f.ser");<br>g = deserialize("f.ser");<br>print(g());<br>$ java -classpath 'js.jar;.' org.mozilla.javascript.tools.shell.Main test2.js<br>3<br></pre>
|
|
<p> Now the function f is compiled to a Java class, but that class is
|
|
then made available in the classpath so serialization works. This isn't that
|
|
interesting an example since compiling a function to a class and then loading
|
|
it accomplishes the same as serializing an interpreted function, but it becomes
|
|
more relevant if you wish to serialize JavaScript objects that have references
|
|
to compiled functions. </p>
|
|
<h3>
|
|
<hr width="100%"><br>
|
|
<a href="index.html">back to top</a>
|
|
</h3>
|
|
|
|
</body>
|
|
</html>
|