This page looks best with JavaScript enabled

Converting site from WordPress into Hugo

 ·  ☕ 4 min read  ·  ✍️ Nikolai Antipov

What actually happened?

My wife owns a web-site which was powered by WordPress.
Some time ago we decided to switch into static solution without databases and constantly running page generating / rendering process.

Idea was great, but I came to it quite late - just a day before hosting contract should be prolonged.
Trick was in switching also into another environment where only static files can be served. Having this, the
old hosting with apache httpd, mysql database and php interpreter could be safely decommissioned saving some money.

Well, I quickly found a plugin (forgot its name; there are a few of them) for WordPress which does static pages generating effectively rendering posts and keeping the entire structure. Sounds very good.
This totally works as a temporary solution till the moment my wife adds new posts and content.

I quickly got the package of static html and css files. Then “copied” them into new environment and checked if it works.

It worked well, so I initiated DNS settings amendment and closed old hosting contract. Wow! That was a good job.

Next day, DNS records are fully propagated and old hosting was stopped. Suddenly, my wife asked me to check if her web-site is working. She received a couple of messages from friends indicating that something goes wrong :-).

I quickly checked and found out that only posts written in English are okay, others just lost (404).

It turned out, that plugin does not support encodings and slightly corrupted the content and links for post written in Russian (95%). And my quick check was done on the randomly picked English article (I am so stupid).

Yes, that was embarrassing. We placed an announcement that site is temporary under construction.

Embarrassing 'not found' announcement

Embarrassing 'not found' announcement

What I had on my table?

  • Entire backup (Yes, of course, I made a backup before ending the hosting contract);
    • All files hosted on a machine;
    • Hosting configuration files;
    • Database dumps (sql files);
  • Hardly workable web-site as number of static files;
  • Intention to “manage” all this content with one of statically-oriented solutions (e.g. Hugo, Hexo, Jekyll, etc.).

Well, files I have are mostly:

  • WordPress PHP code;
  • Huge number of images and photos;
  • Unstructured and corrupted static pages.

The only way to restore the structured content was extracting it from the database accompanying with images and photos.

Writing own converter

Here is a simple home crafted solution based on:

  • Java;
  • Spring Batch;
  • Freemarker;
  • Jsoup.

https://github.com/nantipov/wordhugopress

Maybe later I will add a dockerized version will all tools inside (e.g. java, mysql, go, hugo), but it is another story.
By the way, fill free to contribute in a project.

Configure, build and run

Well, I introduced a simple configuration within simple Spring’s settings file.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
app:
  sources:
    source-1:
      wordpress-home: /optional/path/to/public_html1
      wordpress-remote-base-url: http://optional.website1.org/
      database:
        url: jdbc:mysql://localhost:3306/database1
        username: root
        password: mysql
      tags:
        - 'regular'
      categories:
        - 'Basic posts'
    source-2:
      wordpress-home: /optional/path/to/public_html2
      wordpress-remote-base-url: http://optional.website2.org/
      target-resource-suffix: "-s"
      database:
        url: jdbc:mysql://localhost:3306/database2
        username: root
        password: mysql
      tags:
        - 'extra'
      categories:
        - 'Limited edition posts'
  target:
    hugo-site-content-items-dir: output/blog/content/posts

Sources

Here multiple sources could be declared. Local or remote or both locations could be specified, tool automatically tries to find a resource. So if you do not have a local copy, but there a workable website, just put it.

Also setting wordpress-remote-base-url helps for detecting inner-site or external links. Because inner-site links should be transformed into the new “folders” structure.

Tags, categories and other taxomonies

Special tags or categories could be assigned for specific sources. And, of course, original tokens will also be migrated.

Typical content page layout

Check file resources/templates/empty-post.ftl (Freemarker template) to get an idea about configuring the target post composition.

<#function taxname tax>
    <#if tax == "category"><#return "categories"/><#elseif tax == "post_tag"><#return "tags"/><#else><#return tax/></#if>
</#function>
---
title: '${post.title}'
author: ${post.author}
date: ${post.createdAt?iso_local}
publishdate: ${post.createdAt?iso_local}
lastmod: ${post.modifiedAt?iso_local}
draft: ${post.draft?string("true", "false")}
description: '${post.title}'
<#if post.thumbnailFilename?has_content>
cover: 'posts/${post.postDirectoryName}/${post.thumbnailFilename}'
</#if>
<#list post.taxonomy.keySet() as taxomony>
${taxname(taxomony)}: [${post.taxonomy.get(taxomony)?map(t -> "'" + t + "'")?join(", ")}]
</#list>
---

${post.processedContent}

All fields from original WordPress posts are propagated into post object. With some additions:

  • postDirectoryName - path to the target hugo directory for the specific post;
  • thumbnailFilename - path to the thumbnail file;
  • taxonomy - taxonomy data as map -> (e.g. post_tag -> ['hey', 'happy']).

Output hugo structure

Just create an empty hugo site.

1
hugo new site blog

And put its path to the tool configuration.

For example.

1
2
target:
  hugo-site-content-items-dir: output/blog/content/posts

Build the thing and run it

1
./gradlew clean build run

Working web-site

https://orangelvira.com

Working web-site

Working web-site

Share on

Nikolai Antipov
WRITTEN BY
Nikolai Antipov
Software engineer