In a few WordPress themes that I have developed I have modified the header.php file to include the same code. After some recent WordPress coding using filters I examined ways to see the global variables without modifying the theme files. This is exactly what WordPress hooks, filters and actions were created for.

The obvious choice was to use the add_action() function to call my function in the header area (use the wp_head action hook). The plugin can be installed in the Must Use Plugins directory (named ‘mu-plugins’) so that one doesn’t even have to log into the dashboard to enable it (though I note that this method has some caveats but they are not a problem for this).

<?php
add_action( 'wp_head', 'mu_debug_info' );

function mu_debug_info() {
  if ($_SERVER['REMOTE_ADDR'] == '192.168.1.1') {
    echo '<!-- Server Debug Info from mu-debug-info plugin.';
    echo "\n\$_SERVER: "; var_dump($_SERVER);
    echo "\n\$_GET: "; var_dump($_GET);
    echo "\n\$_POST: "; var_dump($_POST);
    echo "\n\$_SESSION: "; var_dump($_SESSION);
    echo ' -->', "\n";
  }
}
?>

The next step is to create an options panel in the dashboard to allow setting of the IP address. The code can be expanded further to include more WordPress variables.

After using a filter to tailor the post display on the 4tec home page and another filter to fix the woo_pagination error on the Irish Triathlon site I looked into a filter for the wp_nav_menu() function.

From the wp_nav_menu function page I got the source file for the function. There are 10 filters in the code. After a bit of experimenting I found just the right filter – walker_nav_menu_start_el.

So, after a little bit of regular expression coding, my menu items had lovely titles.

The next step is to expand this into a plugin where the extra title text is set in the admin area.

// Append some text to the title attribute in menu links.
add_filter( 'walker_nav_menu_start_el', 'wpb_nav_menu_item', 20, 4 );
function wpb_nav_menu_item($item_output, $item, $depth, $args) {
  $extra_title = 'EXTRA BIT';

  if (strpos($item_output, 'title=')) {
    // Append $extra_title to the existing title text.
    return preg_replace('/title=\"([^\"]*)\"/', 'title="\1 - '.$extra_title.'"', $item_output, 1);
  }
  else {
    // Create title text made up of menu text plus $extra_title.
    return str_replace('href=', 'title="'.$item->title.' - '. $extra_title.'" href=', $item_output);
  }
}

When I was developing the 4tec.ie web site recently, the client wanted the latest posts to be on the home page. The home page would be a static page so making it a blog format would require modifying the theme’s files, something I wanted to avoid.

I found a cool plugin, Display Posts Shortcode that would display posts via a shortcode. Unfortunately the html it produced made it difficult for me to style and the post thumbnails were not being displayed. Thankfully the author, Bill Erickson, provided a filter that allowed me change the html.

As I current use WordPress almost exclusively, I am always interested in contributing to the WordPress community in some way. Every now and again I look at the WordPress bug tracker to see if there are simple bugs where the patch needs testing or see if I can enhance a plugin in some way.

After I finished the 4tec.ie site I had an itch to enhance the Display Posts Shortcode plugin to allow users specify different list types (it defaults to an unordered list). Like a good citizen, instead of just suggest the change, I developed a patch that the plugin owner could review and test, and hopefully use.

--- display-posts-shortcode.orig.php	2011-10-05 01:37:32.000000000 +0100
+++ display-posts-shortcode.php	2011-10-06 10:57:08.221000000 +0100
@@ -39,6 +39,7 @@
 		'image_size' => false,
 		'taxonomy' => false,
 		'tax_term' => false,
+        'list_type' => 'ul',
 	), $atts ) );

 	$args = array(
@@ -74,7 +75,29 @@
 	$return = '';
 	$listing = new WP_Query($args);
 	if ( $listing->have_posts() ):
-		$return .= '<ul class="display-posts-listing">';
+        $container = 'ul';
+        $post_container = 'li';
+        switch ($list_type) {
+            case 'ol':
+            case 'ordered':
+            case 'ordered_list':
+              $container = 'ol';
+              $post_container = 'li';
+              break;
+            case 'none':
+              $container = 'div';
+              $post_container = 'div';
+              break;
+            case 'ul':
+            case 'unordered':
+            case 'unordered_list':
+            default:
+              $container = 'ul';
+              $post_container = 'li';
+              break;
+        }
+		$return .= '<' . $container . ' class="display-posts-listing">';
+        $post_count = 1;
 		while ( $listing->have_posts() ): $listing->the_post(); global $post;

 			if ( $image_size && has_post_thumbnail() )  $image = '<a class="image" href="'. get_permalink() .'">'. get_the_post_thumbnail($post->ID, $image_size).'</a> ';
@@ -88,13 +111,14 @@
 			if ($include_excerpt) $excerpt = ' - <span class="excerpt">' . get_the_excerpt() . '</span>';
 			else $excerpt = '';

-			$output = '<li>' . $image . $title . $date . $excerpt . '</li>';
+            $output = sprintf('<%s id="post-%d">%s%s%s%s</%s>', $post_container, $post_count, $image, $title, $date, $excerpt, $post_container );

 			$return .= apply_filters( 'display_posts_shortcode_output', $output, $atts, $image, $title, $date, $excerpt );

+            $post_count++;
 		endwhile;

-		$return .= '</ul>';
+		$return .= "</$container>";
 	endif; wp_reset_query();

 	if (!empty($return)) return $return;

Bill and I exchanged a few emails about my patch. He wanted to just provide a filter to allow a user change the tags. As filters require PHP development skills I told him this and encouraged him to provide shortcode arguments too as these could be used by someone without PHP experience.

Bill then released version 1.6 with a new shortcode argument (“wrapper”) and two new filters. While the code is very different from my patch (it’s very elegant) it essentially achieves the same result and I am happy to have nudged Bill in that direction.

I tried it out and found that you get a 404 error when you click on any of the page links. They worked okay on the test site that we used (on my web hosting account here on Blacknight). More strange was that the page links worked in the category archive pages.

I then noticed that the home page links were missing the “index.php” part of the url that is required on Register 365 hosting because they use Zeus web server (and its frustrating url rewriting scripting language).

I went trawling through the source code for the IrishTriathlon.com theme and into the WooFramework code and then deep into the WordPress source to see how they generated the page links. It got daunting quickly. I was dreading having to rewrite the code to work on Zeus web server (and have yet another reason to dislike it).

Then, right at the end of the woo_pagination function, I spotted a woo_pagination filter:

/* Allow devs to completely overwrite the output. */
$page_links = apply_filters( 'woo_pagination', $page_links );

The source code comment gave me hope!

I did a little experiment modifying $page_links and then I tried using php’s str_replace to insert the missing ‘index.php’ into the link urls. Woo hoo – it worked first time!! Phew.

// On the home page the Prev/1/2/3/Next links did not have 'index.php' in the url.
// This filter simply adds 'index.php' if necessary.
add_filter( 'woo_pagination', 'it_add_index_php_to_pagination', 20 );
function it_add_index_php_to_pagination($page_links) {
  return str_replace('com/page', 'com/index.php/page', $page_links);
}

As some point I will determine if the root cause of the issue is related to the woo_pagination function or a WordPress pagination issue. I suspect the former.

The 3 lines of code required to fix this issue (I’m not counting the comment or closing brackets lines) reminded me of a joke when I was working on the Norton Commander for Windows project back in 1996. One of the junior developers had an issue with the built-in editor. The senior developer helped him fix it – with 3 lines of code. So the junior dev always joked that “3 lines fixed my editor”. I guess you had to be there.

The script was a great help but it needed a lot of changes to work nicely and fail gracefully. I emailed the changes back to the script author. At my request he has added a credit link to the script page.

Below is a summary of the changes I made:

• Add mysql_query() query string to error messages to help track down problem query.
• Use one strtr() call instead of multiple calls to str_replace() when creating $seo urls.
• Do not add the product_id (or category_id or manufacturer_id) to the SEO url unless there is a url clash.
• Count the number of products/categories/manufacturers copied to OpenCart and display at the end.
• Don’t die() when a mysql_query() fails – store the error and display it later (for product/category/manufacturer copying).
• Put “Back” button inside ‘p’ tags.
• Added ‘manufacturer_to_store’ table to the TRUNCATE list.
• Reformat list of tables to truncate to make it easier to read.
• Added setting at top of script to disable copying of image files (my osCommerce and OpenCart databases were on different servers).

I am currently developing an online store and using OpenCart. It is quite different from osCommerce as it uses MVC architecture but it is also modular so core files (generally) don’t need to be modified.

Some OpenCart Extensions do require modifying core files and this is obviously bad practice and will make upgrades to the core code difficult.

One of the OpenCart developers, Qphoria, developed an amazing add-on that allows developers virtually modify files, including core files, with XML files. vQmod is a Virtual File Modification System. It is fantastic!

Instead of modifying core files to include an extension, you use vqmod xml file to make the changes. It’s a bit more initial work but definitely worth it.

I am using Facebook Like button extension to put a Facebook Like button on the product pages. It required modifying the product page’s template file. This would make theme upgrades painful and error prone so in steps vQmod.

I went to their Stockists Locator page to see if there were any retailers near me. The Region drop down list had a typo for Ireland (North & Rebuplic of Ireland) where “Rebuplic” should be “Republic” (swap b and p).

I used the Customer Centre section to inform them and they replied telling me that the issue would be forwarded to the web site team to fix the issue.

A few days later I checked the site and found that the list had been changed but that they had only half fixed it. Now it says “Repuplic” – the b was changed to a p but the other p wasn’t changed to a b.

I am sure that some would take this half-fix a deliberate slight against Irish people but I find that such offence takes a lot of energy.

Update: 4 November – the typos have finally been corrected.

The rest of the mile markers seemed to be placed reasonably accurately (well, mile 3 seemed a little long) but I finished the race a full minute faster than last year (29:47 vs 30:50). I was suspicious – I felt that something must be wrong.

Back home I was reading the boards.ie thread on the race and someone questioned whether the course was short. A number of others soon followed with the same conclusion, with Garmin and similar data to back it up.  My Polar HRM measured low too (it normally measures low but this time it was extra low).

One guy developed a chart for people to use their chip time to extrapolate their time for a full 5 miles.  The chart only listed whole minutes so I wrote a PHP script to extrapolate any time.

I posted this on the thread and it was well received. I then posted it to the Dublin Marathon page on Facebook and I saw a jump in visits to the page during the day.

Nice looking site.

I clicked on ‘Maps’ at the bottom of the page. Nothing happened. The url in the status bar showed a ‘#’ so I waited for a popup window. Nothing happened. I looked at the source code (no tables – good stuff) and was quite surprised to see:

<a href="#">Maps</a>

for all 15 links in the footer! (i.e. no url, no javascript for a popup, no nuthin).

I did find the directions via the Contact Us link at the top of the page. Unfortunately the directions seem to have been copied from the AA Route Planner – they are horribly verbose. The M50 and M1 toll prices are out of date too.

The page also mentions getting their by rail but omits which Leixlip station to use and directions from the station to the hotel (it’s Leixlip Confey and it’s about 1.25 miles from the station to the hotel).

It then lists three Dublin Bus routes to Leixlip but gives a link to CIE’s website instead of to the timetables of each route.

I reported the links, directions and train issues a week ago. Nothing has changed except for a typo (City Center -> City Centre) being fixed.

This omissions violates item 7 (Hidden Contact Details) in 10 reasons why your website sucks. In the list it gives Amazon.co.uk as an example. From a web development perspective, the site is horrible – look at the source code. Shudder. It appears to have been generated by Adobe GoLive 4 (AFAIK current version is GoLive 9).