Search and replace in MySQL

Currently, I want to wrap all acronyms – »PHP«, for example – in a span element to display them as small capitals. So I searched for a way to search and replace in MySQL, and there it is: REGEXP_REPLACE.

For example, the following command wraps WOFF in the post titles in the wanted markup (it is a good idea to create a backup before):

UPDATE `wp_posts` SET `post_title` = REGEXP_REPLACE(post_title COLLATE utf8mb4_bin, 'WOFF', '<span class="smcp">WOFF</span>' ) WHERE `ID` = 4691

The first param of REGEXP_REPLACE is the string to search in – we set the column’s name we want to search in. The second parameter is the searched string and the third the string to replace the searched string with. Setting COLLATE utf8mb4_bin (you may need to adjust the collation if you use another one) is important because otherwise, the search would be case insensitive.

Like written in the MariaDB documentation (and as the name suggests) you can also use regular expressions with REGEXP_REPLACE. For my German version of the weekly recap, I used the following:

UPDATE `wp_posts` SET `post_title` = REGEXP_REPLACE(post_title COLLATE utf8mb4_bin, 'KW([0-9]+)', '<span class="smcp">KW</span>\\1' ) WHERE `ID` = 4691

\\1 inserts the part inside the brackets from the searched string into the replace string: the week number.

Better WordPress performance through the use of Elasticsearch

Recently I needed to improve the performance of a WooCommerce installation with multiple 10,000 products. On regular WordPress pages, performance was already good – the performance issues occurred on product archive pages. The post »What does it take to scale WooCommerce?« by Chris Lema contained the tip to use Elasticsearch, that can be integrated via ElasticPress that also supports WooCommerce. After that, WordPress uses Elasticsearch instead of the MySQL database for querying content.

Read and tested, and the result: a significant improvement of load time, now below one second ⚡

So if you are having issues with load speed on larger installations, where the database seems to be the bottleneck, try using Elasticsearch 🙂