Urubatan’s Weblog
msgbartop
Just a little about development! Think before you write!
msgbarbottom

24 Jan 08 Writing rails plugins – tips and tricks – plugins with generators

Following the sequence I started here, but changing the proposed order, let’s talk a little about Rails Generators.

Generators are one of the coolest things in Rails!
Of course you have already seem them, remember the first screencast you’ve seen about rails? Yes, that one where the guy wrote an entire CRUD application[bb] in two minutes …

Generators are a way to, generate code :D

But the generated code can be what you want, and rails already has a great support for writing generators, because it is already used in the rails core.

OK, but why would you want to write a generator?
Think about the project that is starting today, remember all that CRUD forms you’ll need to create, or that almost equal start point for every page of the application …
Now think about all the applications for your company …
Got it?

Ok, I agree that “scaffolding” a generic CRUD is not really useful, but if that scaffold is done following your company standards, it can improve a lot the development performance …

Now, I guess you are thinking: Good, generators[bb] seem cool, but stop the easy talk e show me the code!

So, let’s play a little …
Wi’ll start creating a new rails project: rails plugins102
Now from the project directory run: script\generate plugin my_generator
(if you think this is kind of a flash back, maybe you read my previous post about creating rails plugins :D )

Now that you have already created your brand new plugin, let’s write some code!
To create a generator you need to create a generators directory inside your plugin’s directory, a directory with the name you want for your generator (for example test_gen), and a templates directory inside the previous one, for example, it will look like this:

  • my_generator
    • lib
    • tasks
    • test
    • generators
      • test_gen
        • templates

Ok, you do not really need a plugin to create a generator, you can of course, place your brand new generator inside the following directories:

  • RAILS_ROOT/lib/generators
  • RAILS_ROOT/vendor/generators
  • RAILS_ROOT/vendor/plugins/plugin_name/generators
  • USER_HOME/.rails/generators
  • gems ending in _generator

But I think that a plugin is the easiest way to do it, and it will be easier to deploy to your applications too.
Of course that a GEM would be a better way, because you can install only once per machine, and does not need it installed on the server, but I have never created a GEM before, so let’s stay with the plugin for a wile …

to start coding the generator, we’ll need to create a file named [generator_name]_generator.rb inside the generator’s directory, in my case, I’ve created the file: my_generator/generators/test_gen/test_gen_generator.rb

Now inside that file, create the class to define your generator, it must be named according to the file name, in my case TestGenGenerator, and this class must extend one of the base classes for Rails generators:
Rails::Generator::Base or Rails::Generator::NamedBase, I’ll use NamedBase (The base for the controller generator), and I’ll create a generator for a model and a migration, just to show how to do it …

The NamedBase is the perfect base class for generators that expect parameters in the form: Name [param1] [param2] …
For all the others, Base is a best start point …

The starting point for our code is:

1
2
3
4
5
6
7

class TestGenGenerator < Rails::Generator::NamedBase
  def manifest
    record do |m|
 
    end
  end
end

In this class, all we need to do is to set up the generator’s manifest (more on this later) and to setup any local variables to be used by our templates (more on this later too).

with only this, we can already run: ruby script\generate test_gen asdas_dasda asd:ash (randon letters as parameters for now).
The NamedBase will automatically set up some variables for us, and the values for the given parameters will be:

  • class_name -> AsdasDasda
  • class_nesting ->
  • class_nesting_depth -> 0
  • class_path ->
  • file_path -> asdas_dasda
  • name -> asdas_dasda
  • plural_name -> asdas_dasdas
  • singular_name -> asdas_dasda
  • table_name -> asdas_dasdas
  • attributes -> #<Rails::Generator::GeneratedAttribute:0×3716418>
  • args -> asd:ash

of course this generator now will generate nothing, because our manifest is empty, so let’s build a simple example to see how it really works …
so, lets create inside the template directory, a directory called “dummy”, and a blank file called “log.log” inside of it and let’s change the manifest method to some thing like this:

1
2
3
4
5

  def manifest
    record do |m|
      m.file 'dummy/log.log', "log/#{file_path}.log"
    end
  end

This code will tell the generator, to copy our newly created, blank file, to $APP_ROOT/log/asdas_dasda.log if we run the generator with the same parameters as before …
but just copying files from one place to another is not a very cool thing to be done, so let’s play a little with ERB, and let’s create a migration for our plugin, so we need to change again the manifest method as follow:

1
2
3
4
5
6
7
8

  def manifest
    @migration_name = "Create#{class_name}"
    @migration_action = "add"
    record do |m|
      m.file 'dummy/log.log', "log/#{file_path}.log"
      m.migration_template 'lib/mymigration.rb',"db/migrate", :migration_file_name => "create_#{file_path}"
    end
  end

As you can see in the code, I’m telling the manifest that I have a template named mymigration.rb inside the directory lib in my templates directory, it will be processed and the result will be placed inside the directory “db/migrate” and will be called “000_create_asdas_dasda.rb” (the 000 will be replaced with the latest migration number plus one as the normal “generate migration” command do.

to be able to access the attributes migration_name and migration_action in the template, we need create the accessor methods for them, it is as easy as adding the following line to the TestGenGenerator class (outside the manifest method):

1

attr_accessor :migration_name, :migration_action

My migration template is the following:

1
2
3
4
5
6
7
8
9
10
11

class <%= migration_name.underscore.camelize %> < ActiveRecord::Migration
  def self.up<% attributes.each do |attribute| %>
    <%= migration_action %>_column :<%= table_name %>, :<%= attribute.name %><% if migration_action == 'add' %>, :<%= attribute.type %><% end -%>
  <%- end %>
  end
 
  def self.down<% attributes.reverse.each do |attribute| %>
    <%= migration_action == 'add' ? 'remove' : 'add' %>_column :<%= table_name %>, :<%= attribute.name %><% if migration_action == 'remove' %>, :<%= attribute.type %><% end -%>
  <%- end %>
  end
end

It is an ERB template that will generate a ruby file.

Now we have already a very cool working plugin, but what else can we do?
to answer this question, I’ll quote a little the documentation of the Rails::Generator::Commands::Create class, that means, what can you do within the manifest method:

  • class_collisions – Check whether the given class names are already taken by Ruby or Rails. In the future, expand to check other namespaces such as the rest of the user‘s app.
  • directory – Create a directory including any missing parent directories. Always directories which exist.
  • file – Copy a file from source to destination with collision checking.
  • identical? – Checks if the source and the destination file are identical. If passed a block then the source file is a template that needs to first be evaluated before being compared to the destination.
  • migration_template – When creating a migration, it knows to find the first available file in db/migrate and use the migration.rb template.
  • readme – Display a README.
  • route_resources – add a route to the routes.rb
  • template – Generate a file for a Rails application using an ERuby template. Looks up and evaluates a template by name and writes the result.

I think that is it, you can ask more questions if you want, I’ll try to answer all, if any :D

for more resources on Rails generators you can follow this links:
http://wiki.rubyonrails.org/rails/pages/UnderstandingGenerators
http://www.aidanf.net/node/33
http://api.rubyonrails.org/classes/Rails/Generator/Base.html
http://api.rubyonrails.org/classes/Rails/Generator/NamedBase.html
http://api.rubyonrails.org/classes/Rails/Generator/Commands/Create.html

I hope this little step by step help some one!
The next one will be about writing tests for your plugins! and the fourth I have not started to think about yet :D

If you enjoyed this post, make sure you subscribe to my RSS feed!

Tags: , , , , ,

Filed in productivity, rails with 11 Comments

15 Jan 08 Writing rails plugins – tips and tricks – plugins with view helpers

One of the beauty of Ruby on Rails are the Plugins …
Ruby On Rails itself is already a great framework, but the combination of Rails plugins with the Ruby Open Classes is a killer feature!
This little combination enables you to create you own “tags” to be used in your views and layouts.
And it is really easy to create this kind of helper in Rails.

The non plugin way is to create a method in one of the Helper classes (the ones in app/helpers directory), for example, if all the forms in your application are inside a table, with one column for the label and one column for the real field, you can create a helper to simplify the code in your views like this:

  1. rails plugins101
  2. cd plugins101
  3. ./script/generate scaffold example name:string url:string

The standard very simple form for the “Example” model we created for this example, in the application standards would be:

1
2
3
4
5
6
7

<% form_for(@example) do |f| %>
  <table>
    <tr><td><label for="example_name">Name</label></td><td><%= f.text_field :name %></td></tr>
    <tr><td><label for="example_name">Url</label></td><td><%= f.text_field :url %></td></tr>
    <tr><td colspan="2"><%= f.submit "Update" %></td></tr>
  </table>
<% end %>

but if we edit the file app/application_helper.rb and add the following method there:

1
2
3

   def textfield label, object, property, options = {}
    %Q{<tr><td><label for="#{object.to_s}_#{property.to_s}">#{label}</label></td><td>#{text_field object, property, options}</td></tr>}
  end

we could simplify a lot the code only with the help of this “helper”, the new form code would be like this:

1
2
3
4
5
6
7

<% form_for(@example) do |f| %>
  <table>
    <%= textfield "Name", :example, :name %>
    <%= textfield "Url", :example, :url %>
    <tr><td colspan="2"><%= f.submit "Update" %></td></tr>
  </table>
<% end %>

for this little form, this is too much work for some less characters in the view, but if you think about your entire application it would help you a lot!

Now think bigger, this ugly table/form standard I’ve created is the standard for the entire company, all applications in the company follow this standard.

So all developers, always write the forms using the first option (the ugly one, with lots of HTML).
and one day, a new designer tells that you need to add one new class in the “TR” tag in all forms for all applications, and you start crying :D
Or you choose to use the second option (with the helper method) and just need to change one line of code for each application in the company!
Now you are a hero, right?
But if you are smarter than this? what about creating a plugin, that all applications can use, and define this tag inside the plugin?
That way you change only one line of code, test only once, and all applications will be fixed at the same time!

And as always with Rails, this is a lot easy to do!
Just follow this steps:

  1. ./script/generate plugin life_saver
  2. Edit the file vendor/plugins/life_saver/lib/life_saver.rb and put the following content in it: 
    1
2
3
4
5
6

    # LifeSaver
module LifeSaver
  def textfield label, object, property, options = {}
    %Q{<tr><td><label for=#{object.to_s}_#{property.to_s}>#{label}</label></td><td>#{text_field object, property, options}</td></tr>}
  end
end
  3. Edit the file vendor/plugins/life_saver/init.rb 
    1
2

    # Include hook code here
ActionView::Base.send :include, LifeSaver
  4. Remove the method from the application_helper.rb

Now you have just created your first Ruby On Rails plugin!
And yes, the code is the same as the one used in the application_helper.rb, the only trick here is in the init.rb file of the plugin, that line of code is including all methods of the module “LifeSaver” in the base class for all Rails views, the ActionView::Base class …

Now, if you did this, instead of just adding a helper method to each application, tell your boss how much time you have saved for the company with this little story here, and ask for a rise :D

I hope this little step by step help some one!
The next one will be about writing tests for your plugins, and the third one will talk about plugins with generators, the forth you’ll have to come back here to discover :D

If you enjoyed this post, make sure you subscribe to my RSS feed!

Tags: , , , , ,

Filed in rails with 2 Comments

04 Dec 07 Flex/ActionScript3 – Object to XML

Working with Flex and Java as backend, I decided that the best choice for the project I’m working now is to simply post XML from Flex and get XML back from Java (Flex works really well with XML).

But there is a problem with that, creating a XML with one or two values is easy, but if you need to serialize a very large object tree you just got yourself a lot of work.

So I’ve wrote this very simple Flex XML Serializer, and decided to post it here, because some one might need some thing similar :D 

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23

package teste
{
	import mx.utils.ObjectUtil;
 
	public class Utils
	{
		public static function objectToXml(obj : Object, name : String) : XML{
			var result : XML;
			var info:Object = ObjectUtil.getClassInfo(obj);
			if(name==null)
				name = info.name;
			result = new XML("<" + name + "></"+ name + ">");
			for each (var qn : QName in info.properties){
				var val : Object = obj[qn.toString()];
				if(ObjectUtil.isSimple(val))
					result[qn.toString()] = val;
				else
					result.appendChild(objectToXml(val,qn.toString()));
			}
			return result;			
		}
	}
}

To use is is really simple:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15

<?xml version="1.0" encoding="utf-8"?>
<mx:Application xmlns:mx="http://www.adobe.com/2006/mxml" layout="absolute" initialize="appInit()">
<mx:Script>
	<![CDATA[
		import teste.Utils;
		import mx.utils.ObjectUtil;
		import mx.controls.Alert;
		public function appInit() : void {
			var obj : Object = {name:'teste',address:{street:'rua',number:20}};
			Alert.show(Utils.objectToXml(obj,'teste').toXMLString());
		}
	]]>
</mx:Script>
 
</mx:Application>

The generated XML will look like this:

1
2
3
4
5
6
7

<teste>
  <address>
    <number>20</number>
    <street>rua</street>
  </address>
  <name>teste</name>
</teste>

This might not be the best way to work, but I liked this solution for the problem I had :D

One thing I changed from this code to the code that is in production today is that I’m using XML attributes now, and as you can see below the change was big from the first example:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23

package teste
{
	import mx.utils.ObjectUtil;
 
	public class Utils
	{
		public static function objectToXml(obj : Object, name : String) : XML{
			var result : XML;
			var info:Object = ObjectUtil.getClassInfo(obj);
			if(name==null)
				name = info.name;
			result = new XML("<" + name + "></"+ name + ">");
			for each (var qn : QName in info.properties){
				var val : Object = obj[qn.toString()];
				if(ObjectUtil.isSimple(val))
					result['@' + qn.toString()] = val;
				else
					result.appendChild(objectToXml(val,qn.toString()));
			}
			return result;			
		}
	}
}

Could not find the difference? the only change was the “‘@’ +” at line 16, and I was using attributes for the values :D
So, what do you think about this solution?
The next step is to write a deserializer and we’ll make XML rock the Flex world (just kidding :D )

If you enjoyed this post, make sure you subscribe to my RSS feed!

Tags: , , ,

Filed in java with 2 Comments

31 Oct 07 Java + Flex = Cool application and reliable back end!

Probably you have already heard about Adobe Flex.
Flex is a SDK to develop Rich Internet Applications that will run within the Flash Player, and Flash Player is present in almost all browser and all desktop platforms today.
Flex applications generate an SWF file and this file will communicate with a back end server, this backend server can be a Flash Lifecycle Server, but you can use java and open source for it too.

This two/tree examples work without any paid software, you will need only:
Flex SDK – that is free, and the version 3 will be open source.
A servlet container – I’m using Tomcat
– For the first example that is all
And for the other two examples you’ll need to download
OpenAMF – A flex remoting implementation in Java
RemoteObjectAMF0 – an implementation of the RemoteObject tag that supports the version 0 of AMF protocol (OpenAMF does not supports the version 3 yet)

So, lets start, create a java web project with your favorite IDE, and create a Servlet named TestServlet, I’ll use jaxb to render XML from the servlet, but you can use anything else, if you want to use jaxb too, the schema I created for this example is the following:

1
2
3
4
5
6
7
8
9
10
11
12
13
14

<?xml version="1.0" encoding="UTF-8"?>
<schema xmlns="http://www.w3.org/2001/XMLSchema" targetNamespace="http://www.example.com/example" xmlns:tns="http://www.example.com/example" elementFormDefault="qualified">
    <element name="TestSVo" type="tns:TestList"></element>
        <complexType name="TestSVoType">
    	<attribute name="id" type="int"></attribute>
    	<attribute name="name" type="string"></attribute>
    	<attribute name="other" type="string"></attribute>
    </complexType>
    <complexType name="TestList">
    	<sequence minOccurs="1" maxOccurs="unbounded">
    		<element name="all" type="tns:TestSVoType"></element>
    	</sequence>
    </complexType>
</schema>

This schema is for a XML like the following:

1
2
3
4
5

<TestSVo>
<all id="0" name="foo" other="bar"/>
<all id="1" name="foo1" other="bar1"/>
...
</TestSVo>

Generate the stubs for the XML generation using JaxB or use any other tool to generate a XML with this schema.

Like any other Java application, I’ll start with my web.xml

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35

<?xml version="1.0" encoding="UTF-8"?>
<web-app version="2.5" xmlns="http://java.sun.com/xml/ns/javaee"
	xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
	xsi:schemaLocation="http://java.sun.com/xml/ns/javaee http://java.sun.com/xml/ns/javaee/web-app_2_5.xsd ">
	<servlet>
		<servlet-name>testServlet</servlet-name>
		<servlet-class>....servlet.TestServlet</servlet-class>
	</servlet>
	<servlet>
		<servlet-name>AdvancedGateway</servlet-name>
		<servlet-class>org.openamf.AdvancedGateway</servlet-class>
		<init-param>
			<description>
				Location of the OpenAMF config file.
			</description>
			<param-name>OPENAMF_CONFIG</param-name>
			<param-value>/WEB-INF/openamf-config.xml</param-value>
		</init-param>
		<load-on-startup>1</load-on-startup>
	</servlet>
	<servlet-mapping>
		<servlet-name>AdvancedGateway</servlet-name>
		<url-pattern>/gateway2</url-pattern>
	</servlet-mapping>
	<servlet-mapping>
		<servlet-name>testServlet</servlet-name>
		<url-pattern>/TestServlet</url-pattern>
	</servlet-mapping>
	<session-config>
		<session-timeout>30</session-timeout>
	</session-config>
	<welcome-file-list>
		<welcome-file>teste.html</welcome-file>
	</welcome-file-list>
</web-app>

for the first example, the only needed servlet is the test servlet, for the other examples the AdvancedGateway servlet is used …

Now write the following code in the servlet:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62

import java.io.IOException;
 
import javax.servlet.ServletException;
import javax.servlet.http.HttpServletRequest;
import javax.servlet.http.HttpServletResponse;
import javax.xml.bind.JAXBContext;
import javax.xml.bind.JAXBElement;
import javax.xml.bind.JAXBException;
import javax.xml.bind.Marshaller;
import javax.xml.bind.Unmarshaller;
 
/**
 * Servlet implementation class for Servlet: TestServlet
 * 
 */
public class TestServlet extends javax.servlet.http.HttpServlet implements javax.servlet.Servlet
{
	static final long serialVersionUID = 1L;
	private JAXBContext jc;
	private Marshaller marc;
	private Unmarshaller unmarc;
	private ObjectFactory factory;
 
	public TestServlet()
	{
		super();
		factory = new ObjectFactory();
		try {
			jc = JAXBContext.newInstance("....servlet");
			marc = jc.createMarshaller();
			unmarc = jc.createUnmarshaller();
		} catch (JAXBException e) {
			e.printStackTrace();
		}
 
	}
 
	protected void doGet(HttpServletRequest request, HttpServletResponse response) throws ServletException, IOException
	{
		float v = Float.parseFloat(request.getParameter("v"));
		float v1 = Float.parseFloat(request.getParameter("v1"));
		response.getWriter().print(v * v1);
	}
 
	protected void doPost(HttpServletRequest request, HttpServletResponse response) throws ServletException,
		IOException
	{
		System.out.println("List called");
		TestList res = new TestList();
		for(int i=0;i<40;i++){
			res.getAll().add(new TestSVoType(i,"Téste ã " + i));
		}
		JAXBElement<TestList> elem = factory.createTestSVo(res);
		System.out.println(elem.toString());
		try {
			marc.marshal(elem, response.getWriter());
		} catch (JAXBException e) {
			// TODO Auto-generated catch block
			e.printStackTrace();
		}
	}
}

the doGet method of this servlet is a calculator, it just multiply the two parameters and prints out the result.
The doPost prints a XML like the one above, automatically generated using the JAXB API.

Now the flex part:
Create a file named xmltest.mxml in the home folder of your webapp with the following content.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29

<?xml version="1.0" encoding="utf-8"?>
<mx:Application xmlns:mx="http://www.adobe.com/2006/mxml"
	 layout="vertical"	backgroundColor="#F4F4F4">
	 <mx:HTTPService id="calc" method="GET" url="http://localhost:8080/teste/TestServlet" >
	 </mx:HTTPService>
	 <mx:HTTPService id="list" method="POST" url="http://localhost:8080/teste/TestServlet" >
	 </mx:HTTPService>
	<mx:Script>
		<![CDATA[
			import mx.controls.DataGrid;
            import mx.controls.Alert;
        ]]>
	</mx:Script>
	<mx:Panel width="80%" height="531" title="Teste">
		<mx:Label width="100%" color="blue"
			text="Type two numbers and press calculate." />
		<mx:TextInput id="v1" text="3" />
		<mx:TextInput id="v2" text="5" />
		<mx:Label id="lbl"  text="{calc.lastResult}"/>
		<mx:Button label="Calculate" click="calc.send({v:v1.text,v1:v2.text});" />
		<mx:DataGrid dataProvider="{list.lastResult.TestSVo.all}" width="100%" height="195" change="Alert.show(DataGrid(event.currentTarget).selectedItem.other)">
			<mx:columns>
				<mx:DataGridColumn headerText="Id" dataField="id" />
				<mx:DataGridColumn headerText="Nome" dataField="name" />
			</mx:columns>
		</mx:DataGrid>
		<mx:Button label="List" click="list.send({v:v1.text,v1:v2.text})" />
	</mx:Panel>
</mx:Application>

The interface with the java code is in the two mx:HTTPService lines, the first one invokes the URL with a GET request, calling the calculator method
Then the first button (Calculate) is activated, the fist service is called [calc.send(parameters)] and the text on the first label will be updated because of the value binding on the last result …
When the seccond button (List) fires a post [list.send(params)] and the data grid will process the returned XML and display the lines …
That is all!
Gotcha: if you call list.send without parameters the HTTPService tag fires a GET method, this is the reason for the unused parameters there, at least with Flex 2 this happened every time

This aproach is very simple and does not need much code from any side, but it does not looks like OOP very much …

And flex has a great remoting support, so what do you think agout creating a POJO in Java and use it as a “remote object” from your flex code?
Like the idea?
So let’s create the TestService bellow with the same two operations (calc and list)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19

import java.util.ArrayList;
import java.util.List;
 
public class TestService
{
	public float calc(float v, float v1){
		System.out.format("%f X %f\n", v, v1);
		return v * v1;
	}
 
	public List<TestVo> list(){
		System.out.println("List called");
		ArrayList<TestVo> res = new ArrayList<TestVo>();
		for(int i=0;i<40;i++){
			res.add(new TestVo(i,"Testé  ã " + i));
		}
		return res;
	}
}

And now, we do not need all that XML stuf, a plain old VO will do the JOB …

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

public class TestVo
{
	private String name;
	private int id;
	public TestVo(int id, String name)
	{
		this.id = id;
		this.name = name;
	}
	public String getName()
	{
		return name;
	}
	public int getId()
	{
		return id;
	}
	public void setName(String name)
	{
		this.name = name;
	}
	public void setId(int id)
	{
		this.id = id;
	}
}

To use this service from Flash, we need to configure the OpenAMF framework using the file WEB-INF/openamf-config.xml, the name and location of this file was configured as a parameter to the AdvancedGateway in web.xml file …

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31

<?xml version="1.0" encoding="UTF-8"?>
<config>
	<amf-serializer>
			<force-lower-case-keys>false</force-lower-case-keys>
	</amf-serializer>
	<invoker>
		<name>Java</name>
		<class>org.openamf.invoker.JavaServiceInvoker</class>
	</invoker>
	<custom-class-mapping>
		<java-class>....TestVo</java-class>
		<custom-class>TestVo</custom-class>
	</custom-class-mapping>
	<service>
		<name>TestService</name>
		<service-location>....TestService</service-location>
		<invoker-ref>Java</invoker-ref>
		<method>
			<name>calc</name>
			<parameter>
				<type>*</type>
			</parameter>
		</method>
		<method>
			<name>list</name>
			<parameter>
				<type>*</type>
			</parameter>
		</method>
	</service>
</config>

I have removed almost all the code from this file
There is the Serializer/Deserializer configured, the only registered invoker is the Java invoker, there is one custom-class-mapping for the VO and a Service definition for the service.
If you do not want to declare all methods for each service, you can use * as the name, and it will match any method …
The OpenAMF frameworks has many more configuration options, you can call EJBs, WebServices, any java class, JMX beans and Spring Beans from your flex code without any problem …

We can not use the standard mx:RemoteObject to call this service, because OpenAMF does not support AMF3, only AMF0 and I do not know how to configure the AMF protocol version for the mx:RemoteObject tag …
So the first example, uses only a NetConnection instance:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57

<?xml version="1.0" encoding="utf-8"?>
<mx:Application xmlns:mx="http://www.adobe.com/2006/mxml"
	layout="vertical"
	backgroundColor="#F4F4F4"
	initialize="init()">
	<mx:Script>
		<![CDATA[
            import mx.controls.Alert;
            import mx.rpc.events.ResultEvent;
            import mx.collections.ArrayCollection;   
            [Bindable]
            public var lista:ArrayCollection; 
            public var nc : NetConnection;
 
            function init() : void{
            	nc = new NetConnection();
            	nc.objectEncoding = ObjectEncoding.AMF3;
            	nc.addEventListener(NetStatusEvent.NET_STATUS,netStatus);
            	nc.connect("http://localhost:8080/teste/gateway2");
            }
            function netStatus(event : NetStatusEvent) : void {
            	Alert.show(event.info.code);
            }
            function onRetornaLista(event):void{
            	var e : Array = event;
            	lista = new ArrayCollection(e);
            }       
            function onRetornaCalc(event):void{
            	lbl.text = String(event);
            }
            function calc(v1 : Number, v2 : Number) : void {
            	var r : Responder = new Responder(onRetornaCalc);
            	nc.call("TestService.calc",r,v1,v2);
            }
            function list() : void{
            	var r : Responder = new Responder(onRetornaLista);
            	nc.call("TestService.list",r);
            }
 
        ]]>
	</mx:Script>
	<mx:Panel width="80%" height="531" title="Teste">
		<mx:Label width="100%" color="blue"
			text="Type two numbers and press calculate." />
		<mx:TextInput id="v1" text="3" />
		<mx:TextInput id="v2" text="5" />
		<mx:Label id="lbl"  />
		<mx:Button label="Calculate" click="calc(Number(v1.text),Number(v2.text))" />
		<mx:DataGrid dataProvider="{lista}" width="100%" height="195">
			<mx:columns>
				<mx:DataGridColumn headerText="Id" dataField="id" />
				<mx:DataGridColumn headerText="Nome" dataField="name" />
			</mx:columns>
		</mx:DataGrid>
		<mx:Button label="List" click="list()" />
	</mx:Panel>
</mx:Application>

the init method just initialized the NetConnection object (nc).
When the first button is called (Calculate) the calc method is called, in this method we call the TestService.calc method in the server, using the NetConnection object.

The new thing here is that we need the Responders to process the result from the server call …
I think this approach is better than the last one, but still not looking very good for me.

So I looked a little around and found the RemoteObjectAMF0 library …
It is a little ActionScript library, that enables you to use tags to call your service from flex using the AMF0 protocol like this:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48

<?xml version="1.0" encoding="utf-8"?>
<mx:Application xmlns:mx="http://www.adobe.com/2006/mxml"
	xmlns:renaun="com.renaun.rpc.*" layout="vertical"
	backgroundColor="#F4F4F4">
	<mx:Script>
		<![CDATA[
            import mx.controls.Alert;
            import mx.rpc.events.ResultEvent;
            import mx.collections.ArrayCollection;   
            [Bindable]
            public var lista:ArrayCollection; 
 
            NetConnection.defaultObjectEncoding = ObjectEncoding.AMF0;
 
            function onRetornaLista(event:ResultEvent):void{
            	lista = ArrayCollection(event.result);
            }       
            function onRetornaCalc(event:ResultEvent):void{
            	lbl.text = String(event.result);
            }
        ]]>
	</mx:Script>
	<renaun:RemoteObjectAMF0
		endpoint="http://localhost:8080/teste/gateway2" id="TST"
		source="....TestService" showBusyCursor="true"
		makeObjectsBindable="true"
		fault="Alert.show(String(event.fault)), 'Error'">
		<renaun:methods>
			<renaun:method name="list" result="onRetornaLista(event)"></renaun:method>
			<renaun:method name="calc" result="onRetornaCalc(event)"></renaun:method>
		</renaun:methods>
	</renaun:RemoteObjectAMF0>
	<mx:Panel width="80%" height="531" title="Teste">
		<mx:Label width="100%" color="blue"
			text="Type two numbers and press calculate." />
		<mx:TextInput id="v1" text="3" />
		<mx:TextInput id="v2" text="5" />
		<mx:Label id="lbl"  />
		<mx:Button label="Calculate" click="TST.calc(Number(v1.text),Number(v2.text))" />
		<mx:DataGrid dataProvider="{lista}" width="100%" height="195">
			<mx:columns>
				<mx:DataGridColumn headerText="Id" dataField="id" />
				<mx:DataGridColumn headerText="Nome" dataField="name" />
			</mx:columns>
		</mx:DataGrid>
		<mx:Button label="List" click="TST.list()" />
	</mx:Panel>
</mx:Application>

Still not perfect, since I did not liked the idea of using the full file name to call a service, I think my final solution will be to create an dynamic proxy for the services, but this last one looks good enought for me :D

I hope this post helps some one, I spend two days looking for solutions for this problem, because the current project does not have enought cash for a Life Cycle Server license …

PS.: to compile the mxml files you can use the mxmlc command from the free flex SDK
PS2.: you need to fix the URLs to the full path for your context, and your flex app must be downloaded from the same domain
PS3.: the é and ã letters are not encoding problems, I put it there to test if there would be any encoding problems and there were none :D

If you enjoyed this post, make sure you subscribe to my RSS feed!

Tags: , , , , , ,

Filed in java with 21 Comments

24 Oct 07 A cool chat example created with Ruby On Rails and Juggernaut

Versão em portugues aqui
Before we start with the chat, you need to have the requisites installed, juggernaut need the gems json and eventmachine installed, so, run the following command before reading the rest of this example …

  • gem install -y json eventmachine

Ok, now we are ready to go!
First of all, create a rails application, and install the Juggernaut plugin with this commands:

  • rails -d sqlite3 chattest
  • cd chattest
  • script/plugin install svn://rubyforge.org//var/svn/juggernaut/trunk/juggernaut

Juggernaut uses an external Flash XML Push Server to do the reverse ajax magic, and we’ll need to configure that server, so, let’s edit the file: config/juggernaut.yml
I’ve put in the snipet bellow just the lines I changed in the file

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15

PUSH_PORT: 8080
...
DEFAULT_CHANNELS: 
- "chat"
...
PUSH_HELPER_HOST: "localhost"
...
SECRET: "481516232342edededededed"
...
LOGIN_GET_URL: "http://localhost:3000/session/login"
LOGOUT_GET_URL: "http://localhost:3000/session/logout"
...
SESSION_ID: "_chattest_session_id"
...
BASE64: true

I had to change the PUSH_PORT because I’m using a linux box and the application does not run as root, so I was not able to use the default 443 port, and I do not think that 443 is a good port choice because it is the default HTTPS port.
Make sure you change the PUSH_HELPER_HOST to the same host name as the one you are using to access the application, localhost will do the job in the development environment, but remember to change it when you publish your site in a production environment.
the LOGIN_GET_URL and LOGOUT_GET_URL are used to notify the application about clients arriving and leaving, we will really use only the leaving notification.
the SESSION_ID must be the same as the defined cookie name for your application, it is defined in the application controller for rails 1.2.x and will be moved to environment.rb for rails 2.0
and BASE64 must be set to true if we want to use the rails helpers to generate the javascript for us.

Now let’s start the layout for the application. Create a file named pubic/stylesheets/public.css with the following content:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30

body {
  background-color: white;
}
#users {
  float: left;
  width: 200px;
  height: 400px;
  border-style: inset;
  overflow: auto;
  color: white;
  background-color: gray;
}
#dasd {
  height: 400px;
  margin-left: 5px;
  border-style: inset;
  overflow: auto;
  color: white;
  background-color: gray;
}
#controls {
  clear: both;
  padding: 0 0 0 0;
  height: 55px;
  vertical-align: top;
  border-style: inset;
  overflow: auto;
  color: white;
  background-color: gray;
}

and a file named: app/views/layouts/application.rhtml with the following content.

1
2
3
4
5
6
7
8
9
10

<html>
  <head>
    <title>Chat Test</title>
    <%= stylesheet_link_tag 'public' %>
    <%= javascript_include_tag :defaults %>
  </head>
  <body>
    <%= yield %>
  </body>
</html>

in this file it is important to add the stylesheet and the default javascript includes.

With the layout ready (ok, I know it is pretty ugly, but I’m a developer not a webdesigner so, for production, ask a designer in your team for a new layout :D ) let’s generate the needed files and database tables with the following four commands.

  • script/generate model online_user username:string session_id:string last_seen:date online:boolean
  • script/generate controller session
  • script/generate controller chat index
  • rake db:migrate

Every thing ready, we just need to edit some files …
Open the OnlineUser model (app/model/online_user.rb) and change the content to something like the following

1
2
3
4

class OnlineUser < ActiveRecord::Base
	validates_presence_of :username, :session_id, :last_seen
	validates_uniqueness_of :username, :if => Proc.new {|user| user.online }
end

It is just a few validations, not really needed, this was my first idea for the chat, I’ve changed it a little but still works.

Now let’s code the main view of the application in the file:app/views/chat/index.rhtml

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

<!-- Register with Juggernault -->
<%= listen_to_juggernaut_channels [:generic],session.session_id %>
<!-- The Users List -->
<div id="users">
  <ul id="users_list"></ul>
</div>
<!-- The messages pane -->
<div id="dasd"></div>
<!-- The controls pane (login and send messages) -->
<div id="controls"><%= render :partial => 'login' %></div>
<!-- An util javascript to scroll the messages window -->
<script type="text/javascript">
  function scrollMessages(){
    $('dasd').scrollTop = $('dasd').scrollHeight;
  }
</script>

That is all, just tree DIVs, the tag to initialize juggernaut and a simple javascript to scroll the messages DIV to the last sent message.
the messages DIV is named dasd because I was testing some conflicts and forgot to change it back :D

As seen in the page above, we need a login partial, and we’ll need a controls partial too, so let’s code the controls partial (app/views/chat/_controls.rhtml) with the following code:

1
2
3
4
5
6

<% form_remote_tag(
      :url => { :action => :say },
      :complete => "$('message').value = '';$('message').focus();" ) do %>
      <%= text_field_tag( 'message', '', { :size => 90, :id => 'message'} ) %>
      <%= submit_tag "Send" %>
  <% end %>

It has only a remote form tag and two fields, after the form is submited the message field is cleared and the focus os placed back in that field so the user can type another message.

And the login partial (app/views/chat/_login.rhtml):

1
2
3
4
5
6
7

<%= "#{@message}<br/>" if @message %><% form_remote_tag(
      :url => { :action => :login },
      :complete => "$('username').value = ''",
      :after => "$('login').disabled = true" ) do %>
      <%= text_field_tag( 'username', '', { :size => 90, :id => 'username'} ) %>
      <%= submit_tag "Join", :id => 'login' %>
  <% end %>

Very similar to the controls partial, but it shows a message to the user if the chosen nick name is already taken.

the views are all set, and now we need the application logic, as seen in the views, we need a chat controller with two methods: login and say
Let’s take a look at the chat controller (app/controllers/chat_controller.rb):

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59

class ChatController < ApplicationController
  #this method does not need to exist, but I like to see it here, it only needs to render the index.rhtml view
  def index
  end
  def login
    #creates a new OnlineUser record, this is used to store who are the users that are online now
	@user = OnlineUser.new
	@user.username = Juggernaut.html_and_string_escape params[:username]
	@user.session_id = session.session_id
	@user.online = true
	@user.last_seen = Time.now
        #if we can save, it means that there is no other user with the same nick online, so this user can join the chat
	if @user.save
          #let's save the username in the session for future reference
	  session[:username] = @user.username
          #if there are online users, fill the users box for the new user know who is online
	  @users = OnlineUser.find(:all, :conditions => ["online = true and id != ?", @user.id]) 
          if @users.size >0 
            data = render_to_string(:update) do |page|
              @users.each {|u|
                page.insert_html :bottom, :users_list, %Q{<li id="user_#{u.username}">#{u.username}</li>}
              }
	    end
            #send the javascript only to the new user
            Juggernaut.send_to(@user.session_id, data)
          end
          #create a javascript call to add the new user to the end of the online users list
          data = render_to_string(:update) do |page|
            page.insert_html :bottom, :users_list, %Q{<li id="user_#{@user.username}">#{@user.username}</li>}
            page.insert_html :bottom, :dasd, "<b>user #{@user.username} just joined the chat</b><br/>"
          end
          #add the new user to the chat channel
          Juggernaut.add_channel(@user.session_id, 'chat')
          #send the javascript to all users in the chat channel
          Juggernaut.send_data(data, 'chat')
	  render(:update) do |page|
	    page.replace_html 'controls', :partial => "controls"
          end
	else
          @message = 'This nick name is already in use, please choose another'
	  render(:update) do |page|
	    page.replace_html 'controls', :partial => "login"
	  end
	end
  end
 
  def say
    #escape the message, that way the user can not harm others sending HTML ot JavaScript commands
    message = "#{session[:username]}: #{Juggernaut.html_and_string_escape(params[:message])}"
    #create a javascript to add the new message to the end of the messages screen and scroll the div to the bottom
    data = render_to_string(:update) do |page|
      page.insert_html :bottom, :dasd, "#{message}<br/>"
      page.call "scrollMessages"
    end
    #send the message to all users
    Juggernaut.send_data(data, 'chat')
    render :nothing => true
  end
end

The method say is really simple, so we’ll start explaining it:
It first build a new message appending the escaped original message to the user name, then it creates the javascript to add the message to the bottom of the messages div using the Rails JavaScriptBuilder and the render_to_string method, that returns a string instead of rendering the code directly to the client.
then it sends the message to all users subscribed to the “chat” channel, and all the users will execute that javascript.

Now a little about the login method:

  • from the line 7 to 11 we just configure the new OnlineUser model.
  • if the user is saved to the database it means that there is no other user with the same nick name, so we can proceed with the logic
  • from line 17 to 23 we just create a javascript to populate the users DIV for the new user with the users already online
  • and in the line 25 we send this javascript just for the new user.
  • from line 28 to 31 we create a javascript adding the new user to the users DIV and add a message to the messages window telling all users that a new user joined the chat.
  • in the line 33 we add the new user to the “chat” channel
  • in the line 35 the generated javascript is sent to all users (now including the new arriver)
  • and in from the line 36 to 38 the traditional rails ajax helper is used to change the content of the controls DIV to the “controls” partial (the one that enables the user to send messages to other users.

This is almost all logic needed for this chat application, the only missing thing is removing the nick of users that are no more online from the users DIV from the other users, and we’ll do that in the session controller (app/controllers/session_controller.rb)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25

class SessionController < ApplicationController
  #Called when a user disconnect (a refresh in the browser causes this to be called too)
  def logout
    #search for the user record using the session_id
    @u = OnlineUser.find_by_session_id(session.session_id)
    reset_session
    #if a user was found
    if @u
      username = @u.username
      #remove it from the database
      @u.destroy
      #remove from the online users list from all users, and tell others this user left the chat
      data = render_to_string(:update) do |page|
        page.remove "user_#{username}"
        page.insert_html :bottom, :dasd, "<b>User #{username} left the chat</b><br/>"
      end
      Juggernaut.send_data(data,'chat')
    end                
    render :nothing => true
  end
 
  def login
    render :nothing => true
  end
end

The login method does nothing, but we have some code in the logout method …

  • in the line 5 we lookup the user record from the database
  • if a user is found
  • we destroy it in the line 8
  • from line 13 to 17 we create a javascript to remove the user from the users DIV and send a message to all users telling that the user has left the chat

That is all folks, we just need to run the application
to run this application we need to start the rails application server as usual, and then start the push server.
to do this, just run the following two commands:

  • script/server
  • script/push_server

and access your newly build chat with the URL: http://localhost:3000/chat and play a little around.
I’ve built this example while studding the juggernaut lib, so it is possible that there is a easy way to do this, but I think this is a good start point :D
The start idea was not to use a database, but I could not find anything like the ServletContext in java for rails (an application context), I’ll try to use ENV to store the online user names, but I could not make it work until now.
Any tips for improving this example will be very welcome.

If you enjoyed this post, make sure you subscribe to my RSS feed!

Tags: , , , , , , , , , ,

Filed in rails with 28 Comments

16 Oct 07 Ruby out of the Rails – Nitro and Og

The Ruby language, started to grow inside the enterprises and to be “the topic” in many blogs after the Rails framework showed up, but Rails is not the only option for developing web applications with Ruby, there are other frameworks, and one of “the others” is called “Nitro Framework“, thsi one has almost the same age as Rails but it has a lot less documentation and AFAIK a lot less users too.

This post is about my first 30 minutes with Nitro.

One of the better things about rails, and one of the things that you can not do with Nitro is just to change a class while the server is running, and see the change after a refresh in the browser window, but I’ll talk about these problems later …

The Nitro framework, goes through a path different from Rails, while Rails tell you where to put your models, where to put your controllers, where to put your views, …

What I think is a great feature, since when you start working in a new project you already know where every thing is, Nitro let you put your stuff just where you want it to be.

What is a nice feature too, but if I want to put things where I want them to be, I’ll use Java as I’m doing for the last 5 years :D

Nitro is almost a two framework in one, with Nitro you can develop MVC style applications like Rails, but you can write page oriented applications like in PHP or ASP too, and the best thing is that you can mix the two styles too, of course it will increase the complexity of your application, but it can make it a lot easier to work too, you can write what you need using the easier way to accomplish that task.
To start playing with Nitro just run the following commands:

  • gem install -y nitro (this one will install Nitro and all its dependencies)
  • gen app rocketpower (now let’s create a Nitro application called “rocketpower”m I know they tell you in the home page that it is not rocket science but the name is fun :D )
  • cd rocketpower
  • ruby run.rb (I needed to edit the file for it to work, just open it in any editor and add “require ‘rubygems’” in the second line)
  • point your browser to http://locahost:9000

Every thing ready, you are running your first Nitro application!

Until now, I think that nitro is more complicated than Rails, probably the lack of documentation contributed a lot to this, the terrible examples have their share in this responsibility too, but Nitro has some cool things too.

If you want to create a page based application, you just need to create some “.xhtml” files and place your Ruby code between <?r and ?>, or if you just want to display some string, do it just like in any Ruby String, placing your code between #{ … }.

Since nitro does not have a fixed directory structure, you need to “require” some of your Ruby files from inside the “run.rb” code, I think that this is the cause of the lack of automatic reloading of your code.

Nitro uses Og for persistence, and Og does not need your persistent objects to extend any class, it will persist any object that uses any of it’s “property definition helpers”, itlooks for an automatic created method called “serializable_attributes”, I think it is more the “ruby way”, since it uses “Duck Typing” instead of hierarchy.

Nitro does not have any thing like “migrations”, at least I did not find it yet, and I really like Rails Migrations.

I found the source code for Nitro and Og easier to read and understand than the Rails code, but the code for the Rails applications is a lot cleaner and easier to read.

One thing that I found cool about nitro is that they use the explicit parameters for almost all methods, when rails use hashes for almost every thing, both approaches have advantages …

The example I wanted to write for this post will be published another time, because I need to study a little more about Nitro and Og to develop any thing that I’m not ashamed of publishing.

They do not follow a standard directory structure even in the sample applications, in some of them the code is in the source directory, in others it is in the app directory, the only standard is that the public things (style sheets, images, …) are in a directory called public.

For now, they (Nitro and Og) looks like they need more work, they stopped in time for a while, and the development restarted now with the “Rails Boom“.

Nitro has some cool things, but at least for now, I’ll use Rails for my projects (at least the ones that aren’t in Java) :D

Portuguese version Here.

If you enjoyed this post, make sure you subscribe to my RSS feed!

Tags: , , , , , ,

Filed in rails with 7 Comments
sidebartop

Recommend this blog

Recommend Me
Add to Technorati Favorites
sidebarbottom
sidebartop

Sponsored links

sidebarbottom
sidebartop

Recent Posts

sidebarbottom
sidebartop

Tags

sidebarbottom
sidebartop

Recent Comments

sidebarbottom
sidebartop

Cool Software

Basecamp Highrise Backpack
sidebarbottom
sidebartop

Time machine

sidebarbottom
sidebartop

License

Creative Commons License
Blog do Urubatan by Rodrigo Urubatan Ferreira Jardim is licensed under a Creative Commons Attribution-Share Alike 2.5 Brazil License.
Permissions beyond the scope of this license may be available at http://www.urubatan.info/contact/.
sidebarbottom
© Urubatan’s Weblog / Design: Smashing Wordpress Themes