Smokes your problems, coughs fresh air.

Tag: PHP (Page 2 of 2)

Enforcing Drupal URL aliases

I hate modules, especially core modules. I prefer code to be tightly integrated. I want it to work together. Is that too much to ask? In Drupal, most functionality has been stuffed in modules. There’s a Locale module, a Content Translation module and a Path module. What’s missing is a Working Together module.

For me, clean, meaningful URLs are a number one, two and three requirement for any website that I do. Drupal considers /node/54673 to be a cool URL. I don’t. So, as a kind of afterthought, Drupal comes with the Path module. This module allows you to set URL aliases per node.

The problem is that there’s no concept of a canonical URL. The URL alias works, but so does the node/3242 URL. Neither redirects to the other. In many cases this is not much of a problem (because regular visitors will not notice this) but for our current project it is.

We have a lot of blocks with URL dependent visibility settings. For example, for a section about investing we have a menu that is displayed on all URLs starting with /investing, such as /investing/projects and so on.

After editing a page, Drupal helpfully redirects the user to node/[nodenumber]. For us, this means that the menu is no longer displayed and even the theme will be wrong. (We use the Sections module to select a subtheme based on which section you’re in.)

Global Redirect doesn’t work

The Global Redirect module promises to solve this by allowing you to redirect node/[nodenumber] URLs to their alias if available. It kinda does, in some circumstances.

Our Drupal website sports two languages: English (EN) and Dutch (NL). English is the default language (not the fallback language; we don’t use a fallback) and doesn’t use a prefix. Dutch uses the nl prefix. Two example URLs:

URL alias Generic URL

When /node/288 is requested, the client is correctly redirected to /investing/projects, but when /node/110 is requested, no redirect takes place. It will take place when prefixing /nl, but this is completely useless since Drupal’s built-in actions such as edit don’t redirect using this prefix, and these actions were what we needed this module for in the first place.

A very simple hack that does work

We ended up tearing our hair out trying to fix Global Redirect until we decided that we could just delete the module and replace it with a RewriteRule and a simple PHP script.

Modify: .htaccess

# Put this after RewriteBase and before Drupal's default rewrite rules
RewriteRule ^(../)?node/([0-9]+)$ fixurl.php?nid=$2 [L]

Add: fixurl.php

require_once './includes/';
$result = db_query("SELECT * FROM {url_alias} WHERE src = 'node/%d' LIMIT 1", $_GET['nid']);
if ( db_error() ) die("O agony!");
$url_alias_object = db_fetch_object($result);
$destination = $url_alias_object->dst;
$result = db_query("SELECT prefix FROM {languages} WHERE language = '%s'", $url_alias_object->language);
if ( db_error() ) die("O agony!");
$prefix = db_result($result);
if ( !empty($prefix) )
  $prefix .= '/';
header("Location: /$prefix$destination",TRUE,301);

Shortcomings in our hack

The code assumes that every content page has an URL alias. For us, this is okay, because we need these pretty URLs to even have menus show up or to have the right page be displayed with the right theme.

Also, this code is specifically tailored to language code in the URL prefix. For subdomain based language selection, for example, you’d need to modify it.

WordPress feed pagination

Wiebe uses his author Atom feed to generate a list of his blog posts here on his own website. WordPress feeds only display the latest 10 entries. He has written 16 so far. What he needs is a feed with all his entries.

First, I tried if pagination works for feeds. Appending “/page/2” (as is used an non-feed lists) to a feed URL gives a 404 so I was kind of stuck there. Four days ago, after Googling for some time, unable to find a solution anywhere, I asked on the forum. Still no answer today so I tried to find out which parameters WordPress accepts in the QUERY_STRING. The WordPress Codex does explain how queries are handled but not which parameters are accepted.

Digging into wp-includes/query.php, with much trial and error, I found out that I can append ?paged=2 to the URL to get the next page. At least I got that sorted then. There are a number of much more promising parameters supported by get_posts(), but these don’t seem to be parsed by parse_query(). Next time, I’d like to find out how how to use two of these: nopaging and posts_per_page.

Wiebe could complete his list by merging together all the pages of the feed, but I’d much prefer to find a relatively painless method to produce a feed with an unlimited number of posts.



New theme

After upgrading to WordPress 2.5.x, I had to fall back on a stock theme because my old customization of the Sandbox theme no longer worked with the upgrade. But, then, it was time to redo my theme anyway. So here you’re looking at the first version of my new theme. I might have let it stabilize some more before putting it on-line, but who cares? My reader maybe? Let’s just hope he or she doesn’t use IE. 😉

Screencap of my new WP theme Screencap of my new WP theme Screencap of my new WP theme Screencap of my new WP theme Screencap of my new WP theme

Vertical navigation

Ever since the first time that I saw a blog which featured vertical navigation instead of the typical columns, I’ve wanted to implement this for myself. Well, finally…

Site-wide elements use the complete width of the page. The page content is centered in the middle at 87.5%. The identity stuff in the header and the navigation in the footer sits against a back blackground while the content area has the proven black on white for easy reading. I hope that the strong color-contrast as well as the clear difference in with between site-wide elements and page content makes it easy to keep focused on either reading or navigating without distractions.

… and a talkative footer

With this theme, I didn’t want another footer which consist of the odd logo and some loose copyright statements. I wanted a footer which you can actually read, even understand. And who cares if it takes up a little space? It’s at the bottom of the page.

Related posts

I’ve written an (unpublished, unpolished) plug-in which can generate a list of posts that are chronologically related. Traditionally, most blogs have a next/previous post link at the top and bottom of each post. This works very well if you limit your blog to one subject (which is really a very good idea anyway), but if, like mine, your blog is a little bit messy, you could say that someone who stumbled here searching for an article about Subversion is not necessarily interested in the next post if this is a photo of my baby niece.

Hence the chronologically related posts plugin. With this plugin I can say wether I want a link to the first, previous and next post in the blog, within the same category, or matching a given number of tags. (The tag matching isn’t implemented yet, though. Also, matching on meta fields would be a kick-ass ass way to support explicit sequences.)

I put the list generated by this plug-in on top of a blue background besides the various context links of the post.

Issues left

I hope to have the first major revision of my theme ready soon. Here’s a list of some issues that I might address:

  • The CSS renders a bit psychedelically in MSIE 6 (only version I tested) at the moment. Sigh… Let’s just hope that IE 7 will give better results. Then I’ll gladly drop the IE 6 support.
  • When viewing a category, the tag cloud in the navigation panel at the bottom only shows tags for that category. This has to do with the use with me calling the st_tag_cloud() from within the category template.
  • Some of the elements that I just showed to you don’t really look that good and most elements that I didn’t can be said to be … hideously ugly. 😕 Some highlights: the header (should really be a few cool images), the comment form, and the Next/Previous Page links.


I’d almost forget all about the clean, new look of the comment list. And, if you register a Gravatar, your comments will be accompanied by your avatar. Try it. Please!

Separate development/production environments for WordPress

When you’re out Googling on how to maintain a separate development environment for a WordPress installation, you will only stumble across information about how to install all kinds of WAMPP packages. Well, I don’t care about WAMP (or WAMPP). I want to be able to edit my theme, change my plugins, mess with my database locally and then deploy my changes when they’re ready and well-tested (as if I ever…)

Rails was the obvious inspiration for how to do this properly. In Rails, the whole development and deployment process is very intuitive and powerful. In WordPress documentation I never even see the awareness of the need for this separation. They usually tell you to download stuff, upload it and muck about with it on the life production server. But, I’m not the mucking-about-in-live-configurations type. I’m the I-fucked-this-up-so-often-I-want-a-staging-area type. This post is about how I managed to fulfill this wish with WordPress.

Changing the environment

The first thing I had to do was to find some way to decide which environment to go into. For some reason I decided to use Apache’s mod_rewrite to set an environment variable based on the HTTP Host header. This is in fact very illogical, but we’ll get to that later.

<IfModule mod_rewrite.c>
RewriteEngine On
RewriteBase /
RewriteCond %{HTTP_HOST} =bsblog [OR]
RewriteCond %{HTTP_HOST} =bsblog.molensteen
RewriteRule . - [env=WP_ENV:development]
RewriteCond %{HTTP_HOST}
RewriteRule . - [env=WP_ENV:production]
# Keep out of WP's own block of rewrite rules below
# BEGIN WordPress 

. - looks like a needle because it’s voodoo. The dot says I match anything and the dash says I change nothing of what I match. I do set an environment variable to whether I want to be in development or in production.

So I now have an Apache environment variable available for querying from within PHP (as if PHP doesn’t have a $_SERVER['HTTP_HOST'] variable:-? ) and I can make use of that in my wp-config.php.

Multiple configurations in one file

I love configuration files that share the program’s language; wp-config.php being simple PHP code is what make this whole thing so easy:

if ( getenv('WP_ENV') == 'production' ) {
  // ** MySQL settings ** //
  define('DB_NAME', 'blog');             // The name of the database
  define('DB_USER', 'wordpress');        // Your MySQL username
  define('DB_PASSWORD', '[my password]'); // ...and password
  define('DB_HOST', 'bigsmoke.db');        // 99% chance you won't need to change this value
  define('WP_SITEURL', '');
elseif ( getenv('WP_ENV') == 'development' ) {
  // ** MySQL settings ** //
  define('DB_NAME', 'bsblog');          // The name of the database
  define('DB_USER', 'root');            // Your MySQL username
  define('DB_PASSWORD', '[my password]'); // ...and password
  define('DB_HOST', '');       // 99% chance you won't need to change this value
  define('WP_SITEURL', 'http://bsblog');
  //define('WP_DEBUG', true);
define('WP_HOME', WP_SITEURL);
// You can have multiple installations in one database if you give each a unique prefix
$table_prefix  = 'wp_';   // Only numbers, letters, and underscores please!
// The rest of the stuff in this config file just isn't interesting

There’s a few things to note here. You have to use getenv() or $_SERVER instead of $_ENV because variables set by Apache end up in the former two. Another thing to note is that I should have just checked $_SERVER['HTTP_HOST'] instead of resorting to mod_rewrite voodoo. For the rest it’s all very straight-forward: I make some database settings depend on which environment I’m in and I set the URL accordingly.

Development URLs

I had some trouble putting the pieces back together when newer WordPress versions started doing automatic redirects for URL that didn’t match siteurl in the wp_options. This change meant that when going to http://bsblog/ (the development URL for this weblog) for example, I’d inevitably end up at

Links had always been constructed according to this setting, so I had already been planning a plug-in to transform production URLs to development URLs. But, I learned (a little late) that this is completely unnecessary since wp-config.php supports the configuration of a base URL. I had wrongly assumed that settings that weren’t in the sample config file, simply didn’t exist.

Thus, after adding WP_SITEURL and WP_SITEHOME to wp-config.php, everything was working.

Ideas to further enhance your configuration

  • Don’t limit yourself to one development environment if you have more than one development server.
  • Automate your deployment process. I use rsync for this.
  • Write a script to clone your production database to your development database. There’s no substitute for actual data.

Apache’s ForceType directive overrides AddCharset directives

Yesterday, after uploading a refreshed, some character encoding issues popped up because I had converted the website’s content from ISO-8859-1 (Latin 1) to UTF-8. (I wanted to be able to type and paste special characters from PuTTY into VIM without worrying about the particular encoding of each file.)

The Apache HTTPD at InitFour, our webhosting provider, is configured to send ISO-8859-1 by default, while the one on our test server is configured for UTF-8. This caused a little bit of a surprise when I uploaded the refreshed website and saw all characters outside the ASCII range mangled on the life website!

I quickly dug into my .htaccess file to add the AddCharset utf-8 .xhtml directive. To my surprise, this didn’t do squat. A lot of fiddling, reloading and researching later, I realized that the following section in my .htaccess file rendered the AddCharset directive irrelevant:

<Files *.xhtml>
ForceType text/html

I had to change the ForceType directive to include the charset as a MIME parameter:

<Files *.xhtml>
ForceType 'text/html; charset=UTF-8'

Now, it all seemed to work. (Except that it didn’t really because I do some ridiculously complex content negotiation stuff involving a 406 handler in PHP that virtuals the most appropriate variant when no match is found. This script didn’t send a useful Content-Type header. After first adding it to the script, I noticed that the AddDefaultCharset is actually allowed in .htaccess context—a discovery which luckily rendered the other hacks useless.)

Newer posts »

© 2024 BigSmoke

Theme by Anders NorenUp ↑