Starting to grasp closures in general and some groovy features.
Given the following code:
class Mailer {
void to(final String to) { println "to $to" }
void from(final String from) { println "from $from" }
static void send(Closure configuration) {
Mailer mailer = new Mailer()
mailer.with configuration
}
}
class MailSender {
static void sendMessage() {
Mailer.send {
to 'them'
from 'me'
}
}
}
MailSender.sendMessage()
What happens under the hood when you pass a closure to Mailer.send method?
Does to and from are passed as arguments from the Closure point of view? Which types the Closure maps them?
And then inside the Mailer.send method at the moment the Mailer object calls mailer.with receiving the configuration object, the object maps them into method calls. Groovy does this by reflection?
Groovy can dynamically define the delegate of a closure and even the this object.
with is setting the delegate and executing the closure. This is a verbose way to achieve the same:
def math = {
given 4
sum 5
print
}
class PrintMath {
def initial
def given(val) {
initial = val
}
def sum(val) {
initial += val
}
def getPrint() {
println initial
return initial
}
}
math.delegate = new PrintMath()
math.resolveStrategy = Closure.DELEGATE_ONLY
assert math() == 9
What happens under the hood when you pass a closure to Mailer.send method?
It receives a not-yet-executed block of code.
Does to and from are passed as arguments from the Closure point of view?
No, it is better thinking of them as an anonymous class/lambda in java, or a function(){} in javascript.
Which types the Closure maps them?
None, they are method calls waiting to be executed. They can be delegated to different objects, though.
And then inside the Mailer.send method at the moment the Mailer object calls mailer.with receiving the configuration object, the object maps them into method calls. Groovy does this by reflection?
You can decompile a Groovy class file to see what is going on. IIRC, Groovy currently uses a "reflector" strategy (with an arrayOfCallSite caching) to make calls faster OR it can use invokedynamic.
The closure math in the code above will result in this class:
// .. a lot of techno-babble
public Object doCall(Object it) {
CallSite[] arrayOfCallSite = $getCallSiteArray();
arrayOfCallSite[0].callCurrent(this, Integer.valueOf(4));
arrayOfCallSite[1].callCurrent(this, Integer.valueOf(5));
return arrayOfCallSite[2].callGroovyObjectGetProperty(this);
return null;
}
Related
I've got a class that is wrapping objects with a proxy using Groovy's metaClass "getProperty" and "invokeMethod" methods. The proxy/wrapper lets me do lazy loading on the objects, only grab the data once the getter for a property has been called.
It works well when I actually use the getters, but I ran into a problem when I started trying to use Jackson to serialize the object (a secondary use case). I thought that Jackson would spin through all of the getter methods, which would invoke each of the getters and load all the data. But unlike when I call getters myself, when Jackson uses reflection to call the methods, my interceptor methods are not invoked, so all my data is still null.
I trace the call to my object in the Jackson library to this bit of code in the BeanPropertyWriter. The _accessorMethod.invoke(bean) call invokes the 'getter' method, but not the code in my proxy. I was able to recreate it in a unit test too outside of Jackson too.
Is there something I should implement in my Groovy proxy class that would capture those requests via the Method objects in reflection?
This is what my proxying code looks like:
static <T> T lazyProxyFor(T object) {
Class clazz = object.class
object.metaClass.getProperty = { String name ->
println "getting property $name"
def currentVal = clazz.metaClass.getMetaProperty(name).getProperty(delegate)
if (!currentVal) {
setProperty('testing') //normally this goes and gets data from my service
}
return currentVal
}
object.metaClass.invokeMethod = { String name, Object args ->
println "calling $name"
if (name ==~ /get[A-Z][a-zA-Z0-9_]*/ && args?.size() == 0) {
object[name.drop(3).uncapitalize()] // call the getProperty method
} else {
clazz.metaClass.getMetaMethod(name, args)?.invoke(delegate, args) //like the underlying object process the
}
}
return object
}
and this is the code I was testing with:
Address address = lazyProxyFor(Address)
address.class.declaredMethods.each{ method ->
if (method.name.startsWith('get')) {
method.invoke(address) //<-- if I run with this version I get no output
//address."$method.name"() //<-- if I run with this version I get two messages, one for each
//getter method and then the underlying property, presummably because
//groovy is more involved here, instead of reflection
}
}
The code snippet is from the book < Groovy in action 2nd >, with minor modifications.
1 this code works as expected
package test
class InspectMe {
int outer(){
return inner()
}
int inner(){
return 1
}
}
def tracer = new TracingInterceptor(writer: new StringWriter())
def proxyMetaClass = ProxyMetaClass.getInstance(InspectMe)
proxyMetaClass.interceptor = tracer
InspectMe inspectMe = new InspectMe()
inspectMe.metaClass = proxyMetaClass
inspectMe.outer()
println(tracer.writer.toString())
output:
before test.InspectMe.outer()
before test.InspectMe.inner()
after test.InspectMe.inner()
after test.InspectMe.outer()
2 but this code's output is different
package test
class InspectMe {
int outer(){
return inner()
}
int inner(){
return 1
}
}
def tracer = new TracingInterceptor(writer: new StringWriter())
def proxyMetaClass = ProxyMetaClass.getInstance(InspectMe)
proxyMetaClass.interceptor = tracer
InspectMe inspectMe = new InspectMe()
proxyMetaClass.use(inspectMe){
inspectMe.outer()
}
println(tracer.writer.toString())
output:
before test.InspectMe.outer()
after test.InspectMe.outer()
It seems TracingInterceptor dosen't intercept inner methods in the second code.
Maybe it's normal behavior, But it seems to me like a bug.
Can somebody please explain this?
I don't know if this is a bug or not, but I can explain why this different behavior happens. Let's start with analyzing what InspectMe.outer() method implementation looks like at the bytecode level (we decompile .class file):
//
// Source code recreated from a .class file by IntelliJ IDEA
// (powered by Fernflower decompiler)
//
import groovy.lang.GroovyObject;
import groovy.lang.MetaClass;
import org.codehaus.groovy.runtime.BytecodeInterface8;
import org.codehaus.groovy.runtime.callsite.CallSite;
import org.codehaus.groovy.runtime.typehandling.DefaultTypeTransformation;
public class InspectMe implements GroovyObject {
public InspectMe() {
CallSite[] var1 = $getCallSiteArray();
MetaClass var2 = this.$getStaticMetaClass();
this.metaClass = var2;
}
public int outer() {
CallSite[] var1 = $getCallSiteArray();
return !__$stMC && !BytecodeInterface8.disabledStandardMetaClass() ? this.inner() : DefaultTypeTransformation.intUnbox(var1[0].callCurrent(this));
}
public int inner() {
CallSite[] var1 = $getCallSiteArray();
return 1;
}
}
As you can see, the outer() method tests the following predicate
!__$stMC && !BytecodeInterface8.disabledStandardMetaClass()
and if it evaluates to true, it invokes directly this.inner() method avoiding Groovy's MOP (meta-object protocol) layer (no metaclass involved in this case). Otherwise, it invokes var1[0].callCurrent(this) which means that inner() method gets invoked through Groovy's MOP with metaclass and interceptor involved in its execution.
The two examples you have shown in the question present a different way of setting metaclass field. In the first case:
def tracer = new TracingInterceptor(writer: new StringWriter())
def proxyMetaClass = ProxyMetaClass.getInstance(InspectMe)
proxyMetaClass.interceptor = tracer
InspectMe inspectMe = new InspectMe()
inspectMe.metaClass = proxyMetaClass // <-- setting metaClass with DefaultGroovyMethods
inspectMe.outer()
println(tracer.writer.toString())
we are invoking inspectMe.setMetaClass(proxyMetaClass) method using Groovy's MOP layer. This method gets added to InspectMe class by DefaultGroovyMethods.setMetaClass(GroovyObject self, MetaClass metaClass).
Now, if we take a quick look at how this setMetaClass method is implemented we will find something interesting:
/**
* Set the metaclass for a GroovyObject.
* #param self the object whose metaclass we want to set
* #param metaClass the new metaclass value
* #since 2.0.0
*/
public static void setMetaClass(GroovyObject self, MetaClass metaClass) {
// this method was introduced as to prevent from a stack overflow, described in GROOVY-5285
if (metaClass instanceof HandleMetaClass)
metaClass = ((HandleMetaClass)metaClass).getAdaptee();
self.setMetaClass(metaClass);
disablePrimitiveOptimization(self);
}
private static void disablePrimitiveOptimization(Object self) {
Field sdyn;
Class c = self.getClass();
try {
sdyn = c.getDeclaredField(Verifier.STATIC_METACLASS_BOOL);
sdyn.setBoolean(null, true);
} catch (Throwable e) {
//DO NOTHING
}
}
It invokes at the end private method disablePrimitiveOptimization(self). This method is responsible for assigning true to __$stMC class field (the constant Verifier.STATIC_METACLASS_BOOL stores __$stMC value). What does it mean in our case? It means that the predicate in outer() method:
return !__$stMC && !BytecodeInterface8.disabledStandardMetaClass() ? this.inner() : DefaultTypeTransformation.intUnbox(var1[0].callCurrent(this));
evaluates to false, because __$stMC is set to true. And in this case inner() method gets executed via MOP with metaClass and interceptor.
OK, but it explains the first case that works as expected. What happens in the second case?
def tracer = new TracingInterceptor(writer: new StringWriter())
def proxyMetaClass = ProxyMetaClass.getInstance(InspectMe)
proxyMetaClass.interceptor = tracer
InspectMe inspectMe = new InspectMe()
proxyMetaClass.use(inspectMe){
inspectMe.outer()
}
println(tracer.writer.toString())
Firstly, we need to check what does proxyMetaClass.use() look like:
/**
* Use the ProxyMetaClass for the given Closure.
* Cares for balanced setting/unsetting ProxyMetaClass.
*
* #param closure piece of code to be executed with ProxyMetaClass
*/
public Object use(GroovyObject object, Closure closure) {
// grab existing meta (usually adaptee but we may have nested use calls)
MetaClass origMetaClass = object.getMetaClass();
object.setMetaClass(this);
try {
return closure.call();
} finally {
object.setMetaClass(origMetaClass);
}
}
It's pretty simple - it replaces metaClass for the time of closure execution and it sets the old metaClass back when closure's execution completes. Sounds like something similar to the first case, right? Not necessarily. This is Java code and it invokes object.setMetaClass(this) method directly (the object variable is of type GroovyObject which contains setMetaClass method). It means that the field __$stMC is not set to true (the default value is false), so the predicate in outer() method has to evaluate:
BytecodeInterface8.disabledStandardMetaClass()
If we run the second example we will see that this method call returns false:
And that is why the whole expression
!__$stMC && !BytecodeInterface8.disabledStandardMetaClass()
evaluates to true and the branch that invokes this.inner() directly gets executed.
Conclusion
I don't know if it was intended or not, but as you can see dynamic setMetaClass method disables primitive optimizations and continues using MOP, while ProxyMetaClass.use() sets the metaClass keeping primitive optimizations enabled and caused a direct method call. I guess this example shows a corner case no one thought about when implementing ProxyMetaClass class.
UPDATE
It seems like the difference between these two methods exists because ProxyMetaClass.use() was implemented in 2005 for Groovy 1.x and it got updated for the last time in 2009. This __$stMC field was added in 2011 and the DefaultGroovyMethods.setMetaClass(GroovyObject object, Closure cl) was introduced in 2012 according to its javadoc that says this method is available since Groovy 2.0.
In Kotlin, there is the apply method:
inline fun <T> T.apply(block: T.() -> Unit): T (source)
Calls the specified function block with this value as its receiver and returns this value.
This allows you to configure an object like the following:
val myObject = MyObject().apply {
someProperty = "this value"
myMethod()
}
myObject would be the MyObject after the apply {} call.
Groovy has the with method, which is similar:
public static <T,U> T with(U self,
#DelegatesTo(value=DelegatesTo.Target.class,target="self",strategy=1)
Closure<T> closure
)
Allows the closure to be called for the object reference self.
...
And an example from the doc:
def b = new StringBuilder().with {
append('foo')
append('bar')
return it
}
assert b.toString() == 'foobar'
The part with the Groovy method is always having to use return it to return the delegate of the with call, which makes the code considerably more verbose.
Is there an equivalent to the Kotlin apply in Groovy?
The function is called tap and is part of Groovy 2.5. See discussions about the naming in merge request.
Other than that, only foo.with{ bar=baz; it } can be used. You can retrofit your own doto, tap, apply, ... via metaprogramming.
The following C# program does what I expect, which is to output "First," "Second", "Third." However, when I change the type of foo in Main to dynamic, it raises an exception that says:
"Cannot implicitly convert type 'MyProgram.Program' to 'System.Collections.IEnumerable'. An explicit conversion exists (are you missing a cast?)"
Why does changing the type to dynamic break the code in this way?
Thanks!
using System;
namespace TestForEach
{
class Program
{
private int idx = -1;
public Program GetEnumerator() {
return this;
}
public string Current
{
get {
string[] arr = { "First", "Second", "Third" };
return arr[idx];
}
}
public Boolean MoveNext()
{
return ++idx < 3;
}
static void Main(string[] args)
{
Program foo = new Program();
foreach (var i in foo)
{
System.Console.WriteLine(i);
}
System.Console.ReadKey();
}
}
}
My guess is that it is because your Program implements the appropriate methods in order to be considered an iterator by the compiler.
An iterator is invoked from client code by using a foreach statement. For example, you can create an iterator for a class that returns the elements in reverse order, or that performs an operation on each element before the iterator returns it. When you create an iterator for your class or struct, you do not have to implement the whole IEnumerator interface. When the compiler detects your iterator, it will automatically generate the Current, MoveNext and Dispose methods of the IEnumerator or IEnumerator(Of T) interface.
When using the dynamic keyword, the compiler is unable to detect that Program is being used in in iterable context. Because of this, it does not generate the appropriate code in order for it to be used in a foreach loop.
I'm surprised it even compiles. The Program class needs to implement IEnumerable.
I have a closure within an object Foo and inside the closure i define a method called 'myStaticMethod' that I want to resolve once the closure is called outside the object Foo. I also happen to have 'on purpose' a static method within my object Foo with the same name. When I call the closure i set the 'resolve strategy' to DELEGATE_ONLY to intercept the call to myStaticMethod that is defined within the closure.
I tried to achieve that through missingMethod but the method is never intercepted. When i make the Foo.myStaticMethod non static, the method is intercepted. I don't quite understand why this is happening though my resolve strategy is set to DELEGATE_ONLY. having the Foo.myStaticMethod static or not shouldn't matter or I am missing something
class Foo {
static myclosure = {
myStaticMethod()
}
static def myStaticMethod() {}
}
class FooTest {
def c = Foo.myclosure
c.resolveStrategy = Closure.DELEGATE_ONLY
c.call()
def missingMethod(String name, def args) {
println $name
}
}
To solve the problem, I ended up overriding the invokeMethod right before calling the closure in FooTests
Foo.metaClass.'static'.invokeMethod = { String name, args ->
println "Static Builder processing $name "
}
While trying to solve this problem, i discovered a very weird way to intercept missing static methods. Might be useful to some of you in the future.
static $static_methodMissing(String name, args) {
println "Missing static $name"
}
-Ken
Static methods unfortunately aren't intercepted by the closure property resolution. The only way that I know to intercept those is to override the static metaClass invokeMethod on the class that owns the closure, ex:
class Foo {
static myclosure = {
myStaticMethod()
}
static myStaticMethod() {
return false
}
}
Foo.metaClass.'static'.invokeMethod = { String name, args ->
println "in static invokeMethod for $name"
return true
}
def closure = Foo.myclosure
assert true == closure()