Chip Cullen

Migrating From Wordpress to Hugo

I wanted to share some of the experiences that I had migrating this blog from a Wordpress site to a Hugo site. This is more of a technical "how I did it" as opposed to why I did it.

Background

After deciding to move to Hugo (after many years on Wordpress) it took some trial and error. I had about 120ish posts over 8 years of blogging. I had used a variety of text formats, code examples and imagery along the way.

Getting content out of Wordpress

Per the Hugo documentation site, there were (at the time) two options for migrating content out of Wordpress:

(There apparently is a 3rd python-based export option for Wordpress, but I have no experience with it.)

Deciding what content to transfer

Now, I say my data export was mostly successful, but there was one large hurdle:

My original Wordpress posts had largely already been authored in markdown. I think that the xml export, and consequential blog2md parsing, didn't know what to do with that. So, my posts came over as large blobs of text that ran together.

I knew that I was going to have to manually clean up every post that I wanted to move over. It was at this point that I decided to do a content inventory to remove low-value, or no-longer-relevant blog posts.

At the end of that, I knew that I had just over 50 blog posts that I was going to have to clean up, which was much more approachable.

Cleaning posts up

So, the blog text was already a big blob, but there were aspects of the data migration that had worked well. Specifically, the meta data block for each post was accurate.

Most critically, the URL's that I had ended up with for each post were correctly transferred as the filename for my posts.

After much trial and error, though, I found that the easiest solution for cleaning up a blog post file was probably the dumbest one: I would keep my Wordpress site open in my browser, and just copy & paste that markdown into my new Hugo markdown files. It was the fastest way, and the least error prone. This only worked because I had already been authoring in Markdown for years.

Moving images

Probably the biggest pain point in migration was dealing with images. Wordpress had inserted lots of absolute URLs to images that I had to manually copy over. Again, I did kind of a dumb obvious thing:

![alt text for the image](../images/filename.png)

This relative URL is based on the assumption that my blog posts were not buried in a 'subdirectory' permalink structure.

Which leads me to …

Dealing with URLs

One of my objectives was to make sure my existing blog posts would still be reachable at their existing URL's. That took some tweaking of the permalink configuration for Hugo.

It's pretty easy — this is what I ended up with in my config.toml file:

  [permalinks]
    post = "/:slug/"

Now, my files already had the correct names in terms of corresponding to urls (i.e., /blog-post-url and been recreated as blog-post-url.md), so I could have easily used:

  post = "/:filename/"

But I wanted to have an easy time adjusting URL's in future, and I figured manipulating metadata in the post front matter was an easier overall experience. So I had to add slug to all of my blog post files. That wasn't a big deal.

Moving comments

As I had already been using Disqus for my comments on my Wordpress site, it was pretty easy to migrate comments, as long as the URL's matched.

Note: When I migrated to 11ty in late 2022, I removed comments altogether from this blog.

Monitoring 404's

Because I was on a serveless hosting infrastructure, I wasn't going to have access to logs that showed when users experienced 404 pages. I found a great blog post on how you can set up custom reports in Google Analytics so that you can monitor how many times users either:

A Word on Hosting

I also moved web hosts - not that I was unhappy with my last host, but I had given Netlify a go and was super excited. I really like how easy it was to push to a Github repo and have Netlify just ... work.

The Hugo documentation has a very good guide on getting set up with Netlify, which was what I followed.

One gotcha that I ran into - because I was migrating an existing site, which I wanted to keep up while I was working on this, I initially setup Netlify with out a custom domain. So, I could reach my in-progress site at https://chipcullen.netlify.com.

In order to get Hugo to work with that, though, I had to set my config.toml like this:

  baseURL = "https://chipcullen.netlify.com/"

However, when I finally switched my domain over to point at Netlify, my home page would load, but all of my posts would resolve under the Netlify development URL. A coworker of mine who is much smarter than me spotted the fact that I had left my config.toml file like I had above, and that I would need to update it to my actual domain:

  baseURL = "https://chipcullen.com/"

And all was right with the world.