class Member
	{
		public int Id { get; set; }
		public string FirstName { get; set; }
		public string LastName { get; set; }
	}

The following class derives from the DbContext base class, and exposes our Member model:

	class EFDbContext : DbContext
	{
		public DbSet<Member> Members { get; set; }

		public EFDbContext()
		{
			Database.SetInitializer<EFDbContext>(new DbContextInitializer());
		}
	}

Note the SetInitializer call in the constructor. This allows us to seed the database with some data, using the following class:

	class DbContextInitializer : DropCreateDatabaseIfModelChanges<EFDbContext>
	{
		protected override void Seed(EFDbContext Context)
		{
			new List<Member>
			{
				new Member {
					FirstName = "Andrew",
					LastName = "Rowland"
				},
				new Member {
					FirstName = "Roland",
					LastName = "Browning"
				},
			}.ForEach(m => Context.Members.Add(m));

			base.Seed(Context);
		}
	}

Now we need a simple repository class, so we can query the database:

	class MemberRepository
	{
		private EFDbContext Context = new EFDbContext();

		public IQueryable<Member> Members
		{
			get { return Context.Members; }
		}
	}

Standard practice is to use an interface, and obtain this concrete class using an IOC container, but it will do for demonstration purposes.

Due to lazy loading, nothing would happen until we actually run a query. So finally, a console application to instantiate the repository, and display the data:

	class ConsoleApplication
	{
		static void Main(string[] args)
		{
			MemberRepository Repository = new MemberRepository();

			Repository.Members.ToList().ForEach(m => Console.WriteLine(String.Format("Member: {0} {1}", m.FirstName, m.LastName)));

			Console.ReadKey();
		}
	}

We can place all of the code into one console application. Install EF4.1, create a console application in Visual Studio 2010, paste in the class below. You may have to set the startup object in the project properties.

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Data.Entity;

namespace EFExample
{
	class Member
	{
		public int Id { get; set; }
		public string FirstName { get; set; }
		public string LastName { get; set; }
	}

	class EFDbContext : DbContext
	{
		public DbSet<Member> Members { get; set; }

		public EFDbContext()
		{
			Database.SetInitializer<EFDbContext>(new DbContextInitializer());
		}
	}

	class DbContextInitializer : DropCreateDatabaseIfModelChanges<EFDbContext>
	{
		protected override void Seed(EFDbContext Context)
		{
			new List<Member>
			{
				new Member {
					FirstName = "Andrew",
					LastName = "Rowland"
				},
				new Member {
					FirstName = "Roland",
					LastName = "Browning"
				},
			}.ForEach(m => Context.Members.Add(m));

			base.Seed(Context);
		}
	}

	class MemberRepository
	{
		private EFDbContext Context = new EFDbContext();

		public IQueryable<Member> Members
		{
			get { return Context.Members; }
		}
	}

	class ConsoleApplication
	{
		static void Main(string[] args)
		{
			MemberRepository Repository = new MemberRepository();

			Repository.Members.ToList().ForEach(m => Console.WriteLine(String.Format("Member: {0} {1}", m.FirstName, m.LastName)));

			Console.ReadKey();
		}
	}
}

Run the application, and the magic begins. If you have an instance of SQL Server Express running, a database with a single Members table will be created, populated with data, and the results displayed.

Bit of a quick post this one, so feel free to ask a question. I’m not certain I’ve added enough detail.

Recently I’ve been asked to build a site that requires a the ability to relate other pages to the one an editor is currently editing.

The plugin can be downloaded here. Note: This version is for Wolf 0.7.2 and above. The older plugin for Wolf 0.6 is also available. Once installed in the plugins directory and enabled in the administration screen, the following tab is now available when editing pages: Once installed in the plugins directory and enabled in the admininistration screen, the following tab is now available when editing pages:

See the documention screen for sample frontend code which lists the related pages.

Please be aware that I really did knock this together rather quickly, but it works for me.  If anyone finds any bugs, please let me know.

I fixed the obvious within the web.config.  For example, the following is needed to force the application to render XHTML:

<configuration>
	...
	<system.web>
		...
		<xhtmlConformance mode="Strict" />
		...
	</system.web>
	...
</configuration>

If I painfully cut and pasted the HTML into the Validator’s form field, it passed.  The problem is, is that the service is not recognised by the Web server, and so a helper file in the application is required.

In your web root, create a App_Browsers directory.  Within this directory create a file called w3c.browser.  Paste the following contents into the file:

<browsers>
	<browser id="W3C_Validator" parentID="default">
		<identification>
			<userAgent match="^W3C_Validator" />
		</identification>
		<capabilities>
			<capability name="browser" value="W3C Validator" />
			<capability name="ecmaScriptVersion" value="1.2" />
			<capability name="javascript" value="true" />
			<capability name="supportsCss" value="true" />
			<capability name="tables" value="true" />
			<capability name="tagWriter" value="System.Web.UI.HtmlTextWriter" />
			<capability name="w3cdomversion" value="1.0" />
		</capabilities>
	</browser>
</browsers>

You should now be able to pass the page by URL:

http://validator.w3.org/check?uri=http%3A%2F%2Fwww.andrewrowland.com

I generally have a number of repetitive tasks that are extremely tedious to accomplish using the browser based GUI. So I constructed a simple command line tool that allows me to script tasks using a SOAP interface known as the Tridion Business Connector.

The client takes the following arguments:

bcclient.exe /h [hostname] /u [user] /p [password] /d [domain] /x "[request XML]"

Why did I build this? Very useful when called via scripting tools. In my case, Perl.

A simple Perl script that will return the XML for a Tridion Component would look something like this:

use strict;

my $requestXML = q{
<?xml version="1.0" encoding="utf-16"?>
<tcmapi:Message version="5.0" from="AndyRowlandSoapClient" failOnError="true" xmlns:tcmapi="http://www.tridion.com/ContentManager/5.0/TCMAPI">
<tcmapi:Request ID="GetItemRequest" preserve="true">
<tcmapi:GetItem itemURI="tcm:12-345" />
</tcmapi:Request>
</tcmapi:Message>
};

$requestXML =~ s/"/\\"/g;

my $responseXML = `bcclient.exe /h MyTCMHost /u andyrowland /p mypassword /d MYDOMAIN /x "$requestXML"`;

print $responseXML;

We could add a layer of abstraction, and wrap up the SOAP call within a subroutine, something like:

sub make_request {
my $messageXML = shift;
my $requestXML = '<?xml version="1.0" encoding="utf-16"?><tcmapi:Message version="5.0" from="AndyRowlandsSoapClient" failOnError="true" xmlns:tcmapi="http://www.tridion.com/ContentManager/5.0/TCMAPI">' . $messageXML . '</tcmapi:Message>';

$requestXML =~ s/"/\\"/g;

my $responseXML = `bcclient.exe /h MyTCMHost /u andyrowland /p mypassword /d MYDOMAIN /x "$requestXML"`;

return XML::Simple::XMLin($responseXML);
}

Note the use the module XML::Simple to convert the response XML into a Perl data structure.

We can then create a succession of subroutines to perform a variery of routine tasks. To get a Tridion item:

sub get_item {
my $URI = shift;

my $messageXML = qq{
<tcmapi:Request ID="GetItemRequest" preserve="true">
<tcmapi:GetItem itemURI="$URI" />
</tcmapi:Request>
};

my $ref = make_request($messageXML);

if ($ref->{'tcmapi:Response'}{'success'} eq 'true'){
return $ref;
} else {
die "ERROR: Could not get item $URI";
}
}

Now that I’ve got the time, I’m going to start writing again. I start work again in the new year, and I’m looking forward to it. As things never stand still in development, it’ll be time to pick up a few books, and gain some new skills.

More Content Management integration coming up for me next year (Tridion, possibly Immediacy), and more ASP.Net work. Also going to brush up on my jQuery, and CodeIgniter knowledge. Really would like to do more Open Source based project work, as I’ve been mainly using proprietary applications during the pass few months. With the economy lower than a fat kids self esteem, I can’t be choosy, just glad to be working.

Have a great Christmas, and I’ll see you a few pounds heavier in the new year.

The problem I had is that both search ‘forms’ existed on one Web Form, and I was using Validation controls to check each. So when using either submit button, both search groups were validated together. Not useful when I wanted to submit each set of controls independently of each other. So there I was, coding to disable either set of validation controls when a submit button is clicked.

“Validation groups!” I hear you cry. Yes, I know that now.

Add the ValidationGroup attribute to all controls, so you can group them together, and submit them separately from other groups. Example:

<form id="form1" runat="server">

<p>
	<asp:TextBox Id="TextBox1" Runat="server" ValidationGroup="First" />
	<asp:TextBox Id="TextBox2" Runat="server" ValidationGroup="First" />
	<asp:RequiredFieldValidator Id="RequiredFieldValidator1" Runat="server" ValidationGroup="First" ErrorMessage="TextBox1 should not be blank" ControlToValidate="TextBox1" />

	<asp:Button ID="Submit1" Runat="server" ValidationGroup="First" Text="Submit 1" />
</p>

<p>
	<asp:TextBox Id="TextBox3" Runat="server" ValidationGroup="Second" />
	<asp:TextBox Id="TextBox4" Runat="server" ValidationGroup="Second" />
	<asp:RequiredFieldValidator Id="RequiredFieldValidator2" Runat="server" ErrorMessage=" TextBox3 should not be blank" ControlToValidate="TextBox3" ValidationGroup="Second" />

	<asp:Button Id="Submit2" Runat="server" ValidationGroup="Second" Text="Submit 2" />
</p>

</form>

Aye, easy when you know. Googling is a wonderful thing.

In a previous article I talked about using a library from the Zend framework to create a site search.  As you try out this functionality using the search box, above right.

I never finished the article, so I’ll explain here how I implemented this using CodeIgniter.  I’ll also detail how I built the ‘quick results’ tool (click in the search box).

Below is an example controller class that shows how I constructed the search index by pulling articles out of the database.  My actual class calls security methods that require authentication before performing any of these methods.  I left them out of the listing to keep it short and sweet.  You will have to alter the code to allow for the difference in database schema, and to suit your intended available search criteria.

<?php
class Search extends Controller {

	function Search()
	{
		parent::Controller();

		// Load Zend Lucene library.  See previous article.
		$this->load->library('zend');
		$this->zend->load( 'Zend/Search/Lucene');

		// Location of search index.  Used by all Search controller methods.
		$this->search_index = APPPATH . 'search/index';
	}

	function reindex()
	{
		// Create index.  Will delete any existing index.
		$index = Zend_Search_Lucene::create($this->search_index);

		// Obtain all articles.
		$query = $this->db->get('articles');

		// Loop through articles, adding an index for each one.
		foreach ($query->result() as $article)
		{
			// Create Lucene document for this article.
			$doc = new Zend_Search_Lucene_Document();

			// Add required fields.
			$doc->addField(Zend_Search_Lucene_Field::Text('title', $article->title));
			$doc->addField(Zend_Search_Lucene_Field::Text('subtitle', $subtitle));
			$doc->addField(Zend_Search_Lucene_Field::Text('category', $category_list));
			$doc->addField(Zend_Search_Lucene_Field::Text('path', '/article/display/' . $article->path));
			$doc->addField(Zend_Search_Lucene_Field::UnStored('content', $article->summary . $article->text));

			// Add docuument to index.
			$index->addDocument($doc);

			echo 'Added ' . $item->title . ' to index.<br />';
		}

		$index->optimize();
	}

	function optimize()
	{
		$index = Zend_Search_Lucene::open($this->search_index);
		$index->optimize();

		echo '<p>Index optimized.</p>';
	}
}

To utilise the index a result method is required:

	function result()
	{
		// Create empty array, in case there are no results.
		$data['results'] = array();

		// If a search_query parameter has been posted, search the index.
		if ($this->input->post('search_query'))
		{
			$index = Zend_Search_Lucene::open($this->search_index);

			// Get results.
			$data['results'] = $index->find($this->input->post('search_query'));
		}

		// Load view, and populate with results.
		$this->load->view('search_result_view', $data);
	}

And, of course, a view is required to display the results:

<html>
<head>
	<title>Search results</title>
</head>
<body>		

<h1>Search results</h1>

<?php if (empty($_POST['search_query'])):?>
<p>No search query submitted.</p>
<?php else:?>
	<?php if (count($results)):?>
<p><?php echo count($results) ?> result(s) returned for query: <?php echo $_POST['search_query'];?></p>
<ul>
	<?php foreach($results as $result):?>
	<li><?=anchor(site_url($result->path), $result->title);?> (<?php echo round($result->score, 2) * 100;?>%)</li>
	<?php endforeach;?>
</ul>
	<?php else:?>
<p>No results were returned for query: <?php echo $_POST['search_query'];?></p>
	<?php endif;?>
<?php endif;?>

</body>
</html>

In an attempt to challenge my OS of choice, but still keeping within my Ubuntu comfort zone, I installed Linux Mint last week.  Once slated as just a friendly, sexier version of Ubuntu, it now seems to be gaining in popularity.

I’m very impressed.  I installed the OS, development IDE (Eclipse based Aptana), Apache, PHP, MySQL, and had the development version of this site up and running within a couple of hours.  No messing around getting Flash, RealPlayer, and various audio/video codecs running.  Anything I did need to install was a simple apt-get away in the Ubuntu repositories.

I build many .net applications that require date entry, so I implement the Date Picker control often.  Unfortunately, there is a problem when it is used on a field that is validated using a .Net validation control.  Upon clicking on a date, I get the following error:

length is null or not an object

This only occurs when using Internet Explorer.  Currently, my only solution is to edit the source code.  Locate the following code in jquery-ui.js or ui.datepicker.js:

inst.input.trigger('change')

Replace it with:

if (!$.browser.msie){inst.input.trigger('change')}

This prevents the change event firing in IE.

I wouldn’t normally advocate changing source code in this manner, as it makes future upgrades tedious.  Hopefully it will be addressed in a future release.

<asp:Content ID="Content1" ContentPlaceHolderID="Content" Runat="Server">
	<asp:TextBox Id="Firstname" Runat="Server" />
</asp:Content>

The output is rendered as:

<input name="ctl00$Content$FirstName" type="text" id="ctl00_Content_FirstName" />

In the example above, the disparity in the rendered id attribute occurs because the TextBox control has been placed within a Content control, as the Page object inherits a MasterPage.

So how can you robustly reference the control using client-side JavaScript, regardless where the control resides?  Using jQuery it’s fairly trivial:

var $firstname = $("[id$=FirstName]");

This little snippet makes use of the jQuery attributeEndsWith selector, as documented here.  Bare in mind that an array is returned, as any control with an id that ends in FirstName will be selected.