Well, yes and no. The solution is very simple, and hardly worth blogging. But google turned up NOTHING for this. I spent a good 20 minutes googling before giving up and diving into Zend Frameworks API Reference. The code to get the module, controller and action from a route name is;

Source code

1. $objRoute = Zend_Controller_Front::getInstance()->getRouter();
2.  
3. $objRoute =  $objRoute->getRoute('exampleRoute');
4.  
5. var_dump( $objRoute->getDefaults() );

This would produce the following;

Source code

1. array
2.   'module' => string 'default' (length=7)
3.   'controller' => string 'index' (length=5)
4.   'action' => string 'index' (length=5)

Seems a little too simple, but I couldn’t find any reference to it.

The one I came accross recently was this error;

Source code

1. Fatal error: spl_autoload() [function.spl-autoload]: Class Doctrine_Event could not be loaded in /library/Doctrine/Record.php on line 803

Bit of a bizare one, and had me scratching my head for a good hour or so. It turns out the problem was the way Zend and APC both use sessions, and they overwrite each others shutdown mechanisms. This then cascades down and causes the autoloading problem.

The solution to this is PHP‘s register_showdown_function() executes a function on shutdown. With this, you can manually shut down the Zend Session session handler, resolving this conflict.

To accomplish this, simply add this method into your application Bootstrap.php

Source code

1. protected function _initShutdownSession() {
2.     register_shutdown_function( array( 'Zend_Session', 'writeClose' ), true );
3.   }

With this, you should be able to get APC up and caching effectively with Zend Framework and Doctrine 1.2.

After a little of searching and investigating, I found out what was causing this. The front-end development had been done by our Front End Developer, and handed over to myself to work on. Within his generic CSS file was the property;

Source code

1. overflow-y:scroll;

He, quite rightly, added this in so on IE browsers, if the content length changed dynamically, the page didn’t jump when the scrollbar appeared. Removing this properly from the pages stylesheet fixed the offset bug with jQuery UI Sortable.

Not quite sure why, I intend to investigate why this is happening within the coming weeks, but for now I hope posting this will make somebody else’s life a little easier!

Zend Translate is a great piece of kit, allowing you to switch between translation adapters with the flick of a few keys. GetText is my sin of choice, and that is what I’ll be talking about here. GetText itself can scan over projects and pick up all the translations, and drop them into your PO files. This is fantastic, as long as you stick to the extremely strict methods that GetText defines. But if I’m using Smarty translation blocks, how do I get GetText to pick them up? Because am I hell adding them all in by hand!

Well, the nasty answer is I couldn’t find a way to actually do it, without actually forking GetText. If anyone does know of a way, please feel free to post a comment with a better solution. First off lets check out the Smartyblock code that I used throughout my View to pass the translation strings back.

Source code

1. <?PHP
2.  
3.   function smarty_block_t($params, $content, $smarty, &$r, $t) {
4.  
5.     if( ! is_null($content) ) {
6.  
7.       $translate = Zend_Registry::get( 'Zend_Translate' );
8.  
9.       return stripslashes($translate->_($content));
10.  
11.     }
12.  
13.   }

This basically calls in Zend Registry (which was set up within the application bootstrap) and uses it to translate the passed string. Tthe code within my View would then look like the following;

Source code

1. {t}Lorem ipsum dolor sit amet, consectetur adipiscing elit.{/t}

Its not a new concept, its been used by Smarty enthusiasts for many a year. In my opinion it is a hell of a lot easier to read than the default Zend/GetText templating/translating mechanisms supplied by default.

So far this will translate any text given, as long as it is within the existing translation file. Great, but if the site has 50,000 {t} blocks, entering them in by hand is very impracticable.

My solution to this was to write my own adapter for Zend Translate, which extended GetText.

Please consider the code on the following .PHPS file.

Source code

http://dev.ricky-stevens.com/getText/getText.phps

As you can see, all this is really doing is high-jacking the translate method, and if the application is within the ‘development’ it checks if the translation exists. If it doesn’t exist, it attempts to add it. The parent translate method is then called to carry out the translation with GetText.

For my example, your folder structure should resemble the following;

Source code

1. /application
2. /bin
3. /languages
4.   /files
5.     /lang_de.mo
6.     /lang_de.po
7.     /lang_en.mo
8.     /lang_en.po
9.     /lang_fr.mo
10.     /lang_fr.po
11.   /translations
12. /library
13. /public

The ‘translations’ file is a text file which needs to be writeable by the application. The /files directory will also need to be writeable.

The last step is to load in the translation adapter inside the application Bootstrap. Please see the following method;

Source code

1. protected function _initLocale(){
2.  
3.   date_default_timezone_set( 'Europe/London' );
4.     Zend_Locale::setDefault( 'en_GB' );
5.  
6.     $objLocale = new Zend_Locale('en_GB');
7.     Zend_Registry::set( 'Zend_Locale' , $objLocale );
8.  
9.     $translate = new Zend_Translate(
10.       'My_Translate_Adapter_Gettext',
11.       APPLICATION_PATH . '/../../languages/files/',
12.       'auto',
13.       array(
14.           'disableNotices'    => true,
15.           'logUntranslated'   => false,
16.           'scan'              => Zend_Translate::LOCALE_FILENAME
17.       )
18.     );
19.  
20.   Zend_Form::setDefaultTranslator($translate);
21.  
22.   Zend_Validate_Abstract::setDefaultTranslator($translate);
23.  
24.     Zend_Registry::set( 'Zend_Translate' , $translate );
25.  
26.     return $objLocale;
27.  
28. }

This should now add in all translations from the View {t} blocks, Zend Form and Zend Validate. They will be added to all the GetText files within the /files directory. If anyone has any improvements or modifications on this, please feel free to leave a comment.

Hopefully this made somebody else’s life easier!

Recently I had a situation where I had a referer URL, and if the refer existed I wanted to be able to forward them to it. If it didn’t exist I needed to be able to forward them to a default route. This meant using the Zend Redirector Helpers method gotoRoute(). Problem was one of these was a Route, and one of these was just a raw URL. So I needed to convert this raw URL to its corresponding route name. Took a bit of time in the Zend Framework API Docs, but eventually I found the method I needed.

As I needed the instance of the router inside a Controller Plugin, rather than a Controller itself, I had to create an instance of the Front Controller.

Source code

1. Zend_Controller_Front::getInstance()
2.       ->getRouter()
3.       ->getCurrentRouteName();

This will return the Route Name of the page the script is called on.

For this I created an example project using Zend Framework 1.10. I then removed the Zend library, as I saw no point in making you download it just to see the example code. The database connectivity for the project I took this from was using Doctrine 1.2. The connectivity inside the library files still refers the Doctrine models, but I saw no need to include these as the arrays and datasets they pass back should be easy to re-create with your own models.

The files added to implement and activate the ACL are:

application/Bootstrap.php – Added in an initiate method to load the ACL controller plugin.

library/My/Controller/Plugin/ACL.php – The ACL controller plugin, checks and manages access levels

library/My/ACL/Exception.php – Common ACL exception extending Zend_ACL_Exception

library/My/ACL/Factory.php – Factory pattern for loading ACL properties

library/My/ACL/Resources.php – Resource creation script

This is all untested, as I just grabbed it from a working project and dropped it into a fresh Zend Framework install to provide an example of an ACL setup. If there any problems or bugs with this implementation, just let me know and I’ll sort them out.

The archive file with the install can be downloaded here.

If you have any questions or queries, feel free to leave a comment.

The copy(); method of the Doctrine Record only cloned itself, not any of its active relationships. See the following copy code:

Source code

1. $objRecord = Doctrine_Core::getTable('Table')->find(1);
2.  
3. $objRecord->RelationshipOne;
4.  
5. $objRecord->RelationshipTwo;
6.  
7. $objRecord2 = $objRecord->copy();
8.  
9. $objRecord2->save();

This copy will only clone the initial Record, but ignores its relationships. According to the documentation, it will clone all records unless a ‘false‘ flag is passed. The documentation does not mention that by default, the flag is ‘false’, and will need to specifrically set to ‘true’ for the relationships to be copied. See the following code:

Source code

1. $objRecord = Doctrine_Core::getTable('Table')->find(1);
2.  
3. $objRecord->RelationshipOne;
4.  
5. $objRecord->RelationshipTwo;
6.  
7. $objRecord2 = $objRecord->copy( true );
8.  
9. $objRecord2->save();

This will now copy/clone the active relationships to the new record too.

The basic situation was I had a large array of Flash animations, which the client wanted to have animating from the page load. The user could then click on one of these flash movies to open up a JavaScript box with the appropriate content. In this project, I had no FLA’s. The client had outsourced the flash work to another company who would not release their source code, just the compiled SWF’s. My suspicion is they either ripped them off, or were embarrassed enough by the quality of the files they wouldn’t release them. But thats another story! The result was it left me with a large headache. With no way of using or hooking into the flash onClick event, my solution had to be purely HTML/CSS/JavaScript.

For my flash insertion I used SWFObject which would replace a “no-flash” image with the appropriate markup to display the flash movie.

Firstly I followed the suggested methods and added {wmode : “transparent” } to my flash params. This caused a bit of a performance hit as it stopped the use of hardware acceleration, but a working prototype was better than no prototype at all! See the following code:

Attempt 1

Source code

1. <script type="text/javascript" src="js/jquery.min.js"></script>
2. <script type="text/javascript" src="js/swfobject.js"></script>
3.  
4. <script type="text/javascript">
5.  
6.   $(document).ready(function() {
7.  
8.     swfobject.embedSWF("flash/movie.swf", "no-flash", "195", "120",
9.     "9.0.0", "flash/expressInstall.swf", {}, {wmode:"transparent"}, {});
10.  
11.     $('#div').live('click',function() {
12.       alert('click event');
13.     });
14.  
15.   });
16.  
17. </script>
18.  
19. <style type="text/css">
20.   #div {
21.     position:absolute;
22.     width: 195px;
23.     height: 120px;
24.     z-index: 5;
25.   }
26.   #div object, #div embed {
27.     position:absolute;
28.     width: 195px;
29.     height: 120px;
30.     z-index: 1;
31.   }
32. </style>
33.  
34. <div id="div">
35.   <img id="no-flash" src="images/no-flash.gif" alt="" />
36. </div>

This had the desired effect in Firefox and Opera, with the alert being triggered, but absolutely no luck with Safari/Chrome and Internet Explorer. The div with the event trigger on it was behind the flash movie. No good! The next approach was to create another div, and position it over the top of the first one. Logical, but I wasn’t really too hopeful. Consider the following code:

Attempt 2

Source code

1. <script type="text/javascript" src="js/jquery.min.js"></script>
2. <script type="text/javascript" src="js/swfobject.js"></script>
3.  
4. <script type="text/javascript">
5.  
6.   $(document).ready(function() {
7.  
8.     swfobject.embedSWF("flash/movie.swf", "no-flash", "195", "120",
9.     "9.0.0", "flash/expressInstall.swf", {}, {wmode:"transparent"}, {});
10.  
11.     $('#div').live('click',function() {
12.       alert('click event');
13.     });
14.  
15.     $('#div2').live('click',function() {
16.       alert('click event on Div2');
17.     });
18.  
19.   });
20.  
21. </script>
22.  
23. <style type="text/css">
24.   #div {
25.     position:absolute;
26.     width: 195px;
27.     height: 120px;
28.     z-index: 5;
29.   }
30.   #div object, #div embed {
31.     position:absolute;
32.     width: 195px;
33.     height: 120px;
34.     z-index: 1;
35.   }
36.   #div2 {
37.                 position:absolute;
38.                 width: 195px;
39.                 height: 120px;
40.                 z-index: 10;
41.                 top: 0;
42.                 left: 0;
43.   }
44. </style>
45.  
46. <div id="div">
47.   <img id="no-flash" src="images/no-flash.gif" alt="" />
48. </div>
49.  
50. <div id="div2"></div>

Oddly, this was slightly more successful. It worked perfectly in the Webkit browsers, but not so well in Internet Explorer. On Internet Explorer 8, the click event would be initiated if you clicked on a transparent part of the movie, but clicking on an opaque part of the movie didn’t fire the trigger. Predictably, IE7 and IE6 just ignored it.

This was about as far as Googling could take me as it was just the same useless answers over and over again. At this stage, I gave up on that and started looking for my own solution to the problem.

Logically thinking, Flash has its own loading point on the page which is after the rest of the DOM elements, so therefore should be above any other element. But after a bit of reading and  practical research, I discovered it wasn’t flash itself that had its own loading point, but the HTML <object> tag itself. This seemed like a step in the right direction, so I stuck with it. They key came when I discovered the various properties of the ‘Type’ attribute of the tag. This allowed me to set the MIME type of the object, and use it for for other purposes than just Flash, allowing me to position another Object tag ontop of the Object tag housing the flash movie.

Consider the following code, which should solve this godforsaken headache forever.

Attempt 3 – Solution

Source code

1. <script type="text/javascript" src="js/jquery.min.js"></script>
2. <script type="text/javascript" src="js/swfobject.js"></script>
3.  
4. <script type="text/javascript">
5.  
6.   $(document).ready(function() {
7.  
8.     swfobject.embedSWF("flash/movie.swf", "no-flash", "195", "120",
9.     "9.0.0", "flash/expressInstall.swf", {}, {wmode:"transparent"}, {});
10.  
11.     $('#div2').live('click',function() {
12.       alert('triggered a click');
13.     });
14.   });
15. </script>
16.  
17. <style type="text/css">
18.   #div {
19.     position:relative;
20.   }
21.   #div2 {
22.     position:absolute;
23.     z-index: 1;
24.   }
25. </style>  
26.  
27. <div id="div">
28.   <div id="div2" style="width:195px;height:120px;">
29.     <object type="img/gif">
30.       <img src="images/null.gif" width="100%" height="100%" />
31.     </object>
32.   </div>   
33.  
34.   <img id="no-flash" src="images/no-flash.gif" alt="" />  
35.  
36.  </div>

With this, you can see I’ve put the #div2 div within the first div this time, but above the image that is being replaced by SWFObject. Inside this div is an Object tag with the MIME type set to “img/gif”, and then obviously just an image within the object tag. This image is just a 1×1 transparent gif. Therefore the Object with the image will be above the other due to DOM order and z-indexes.

This will now successfully trigger the alert on all the browsers I require.. FireFox, Safari, Chrome, Opera, Internet Explorer 8, Internet Explorer 7 and Internet Explorer 6.

Hopefully this will make some peoples lives that little bit easier, and avoid the headache I went through with searching for answers.

If you have any questions or observations, please feel free to leave a comment.

Consider the following code:

Source code

1. <?PHP
2.  
3.   $arrPossible = array('a','b','c','d','e','f');
4.  
5.   $arrDisable = array('a','c','d');
6.  
7.   foreach( $arrPossible as $key => $val ) {
8.     if( in_array( $val, $arrDisable ) ) {
9.       unset( $arrPossible[ $key ] );
10.     }
11.   }
12.  
13.   var_export($arrPossible);
14.  
15.   /*
16.    *  array(
17.    *    1 => 'b'
18.    *    4 => 'e'
19.    *    5 => 'f'
20.    *  )
21.    */
22.  
23. ?>

As you can see, this technique leaves the original array keys in place, meaning they no longer associate the the offset of the array item. The way to rebuild this array and assign new offset keys is to use PHP‘s array_values() function. See the following code:

Source code

1. <?PHP
2.  
3.   $arrPossible = array('a','b','c','d','e','f');
4.  
5.   $arrDisable = array('a','c','d');
6.  
7.   foreach( $arrPossible as $key => $val ) {
8.     if( in_array( $val, $arrDisable ) ) {
9.       unset( $arrPossible[ $key ] );
10.     }
11.   }
12.  
13.   var_export( array_values( $arrPossible ) );
14.  
15.   /*
16.    *  array(
17.    *    0 => 'b'
18.    *    1 => 'e'
19.    *    2 => 'f'
20.    *  )
21.    */
22.  
23. ?>

Obviously this technique won’t apply to everyone, as many may wish to assign the unmatched array items to a new array. This works if there is only one routine to disable the array items, but when you’re working with a large number of routines to manipulate an array, I find it more efficient to modify the existing array and then rebuild its indexes, than keep creating new arrays.

I’ll disregard the parts of his article regarding scanning and readability as that is his opinion and not one for me to rebuke. I will also disregard the point of tabbing along a hyphenated string, as while this is a valid point, it all comes down to your development technique and what is easiest for you. Personally, its not easier for me, but that’s another debate entirely!

My problem comes, when developers such as myself, develop across multiple languages on a day to day basis. In a typical day I might be doing HTML, CSS, JavaScript, PHP, SQL and Java. Of all of these languages, the ONLY language which supports hyphenated delimiters is CSS. This would be fine if these languages were stand-alone entities which didn’t cross over, but that isn’t the case.

For example, if I’m creating a HTML form, that will post to a PHP script, I have to take into account naming conventions. If I were to use HTML‘s naming convention of single words with no delimiters, it would be fine in the markup itself, but all of my PHP superglobal variables would be single words. If I had a complex form where an element required more than one word for its name, I would need to disregard the HTML convention, and use a proper naming convention. If I was using CamelCase in my PHP code, it would therefore make sense to use CamelCase for my HTML form element names, as even though it isn’t quite accurate in the markup, it would be 100% consistent in the PHP code.

Another example, which directly relates to CSS and hyphen delimited naming conventions is the use of JavaScript within a web page. When using JavaScript, I’m constantly using CSS selectors from the DOM to attach events. JavaScript does not support hyphens in its variable names. This means that as I’m constantly refering to CSS selectors, creating new JavaScript variables, assigning new CSS classes and removing existing CSS classes, I would then have 2 different naming conventions embedded throughout the entire JavaScript file. I might then have ANOTHER set of naming conventions going on for my PHP code, which would be be used for AJAX calls. While none of this is going to end the world, if we’re talking about consistencies this is definitely worth being thrown into the debate.

Camel Case, as well as underscore delimiters, are consistently usable through all of these languages, while hyphenated delimiters are not. When possible, I like to keep my conventions and techniques consistent through the entire project, as I believe this makes it easier for me to keep to grip with my code, and when someone takes a look at the project for the first time they will know that the convention they see on the first page will be the same on any other page within the system.

So while Harry Roberts has made a good point about the naming convention of CSS, I honestly believe it is an impractical naming convention for a system that consists of anything other than static CSS pages. If you want any kind of consistency through a big project, you need to take into account the abilities and conventions of all the languages in the system, not just jump at the first language you see and say “Yes, I’ll do it that way!”