Project

wslave

0.0
Low commit activity in last 3 years
There's a lot of open issues
No release in over a year
Word Slave includes the wslave command and a control library to generate a "best practice" WordPress installation and includes a pre-rolled Docker setup for running a development server and a Capistrano setup for deployment.
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
 Dependencies

Runtime

= 3.17.0
~> 0.7, = 0.7.0
~> 1.11, = 1.11.0
~> 1.1, = 1.1.1
~> 1.2, = 1.2.1
 Project Readme

WSlave

WSlave, short for "word slave", is a WordPress control tool.

"Word Slave" is an archaic term that refers to someone being bound to [a slave to] the words of a contract. Since most WordPress work is usually done under contract the naming seemed doubly-appropriate. The name make no reference to actual slavery, and we have no particular intention of changing the name; but if, even after reading this explanation, you feel the name is inappropriate, then please suggest an alternative and we may consider changing the name.

Details

Word Slave sets up a WordPress installation in a fairly generic layout. Basically the layout a lot of people, including "officials" in the Word Press developer community seem to be using combined with a sort of Rails like layout below 'public' that houses the control tools and configuration files. In somewhat-generally-accepted "best practice" form, the WordPress folder is self contained as a Git sub-module in the 'public' folder. Contents and themes are stored in a separate wp-content folder also in 'public'. Overrides are pre-set in the wpconfig.php file and a vanilla .htaccess with general settings that match the layout are automatically placed during installation. On deployment, wpconfig.php is dynamically generated. Salts are also generated during setup, and stored locally. Basically unless you have a very specific reason you'll never have to touch any WordPress control file; everything is done for you with the installed tool chain.

Development server functionality is provided by Docker, so you don't need to install a ○AMP stack on your machine. Rake tasks are provided for if you want to pull the database and files inserted into the installation [uploaded, attached, etc.] from the control panel within the development VM. Files pulled out of the VM can be set as seeds, and will be seeded to the live or staging site upon deployment.

Deployment is provided by Capistrano because it's basically the lowest overhead deployment tool and doesn't try to do things like provisioning - which is overkill as most WordPress sites are put up on shared hosts etc.

Requirements

Short list: ruby, git, docker with docker-compose, *nix/POSIX compliant terminal/shell.

Detailed list

  1. A user who belongs to the 'www-data' group (only required to set write permissions for Docker).
  2. A newer version of ruby and usable ruby gems setup. Ruby packages on some systems may not allow gem installation. We generally recommend RVM.
  3. A properly installed Git.
  4. A full Docker installation including docker-compose.
  5. A working and accessable installation of PHP composer.
  6. Preferably a POSIX compliant shell with standard tools such as BASH/DASH/ZSH in Linux, OS-X, or MSYS2 or WSL2 in Windows.

Re: Windows

wslave is not actively maintained on Windows, but as of this writing it has been tested and runs in WSL2 on both Pro and non-Pro versions, or for Pro versions in MSYS2, CMD with POSIX tools enabled (ls, cd, mkdir, chown, chmod...), or PowerShell.

Running under WSL2 is generally easier regardless of if you have a Pro version of Windows: simply install Ruby, Docker, etc., and add your user to the www-data group (as per the standard Linux setup).

Running under MSYS2 requires you install Ruby, PHP, etc. within MSYS2 and set your environment variables properly to allow docker and docker-compose to be run from within MSYS2.

Running from CMD or PowerShell requires a regular Windows Ruby installations and probably some tweaking of your environment. This is the most complex setup as it requires a lot of system specific environment settings and there's a variety of issues that can arrise due to CMD and PowerShell not really being properly POSIX compliant / completely compatible with *nix style tools.

All things considered running under WSL2 is recommended as it maintains the higest compatibility with the lowest ammount of setting and tweaking environment variables - and it doesn't require a Pro version of Windows to run the container VMs.

Installation

In a terminal:

gem install wslave

Depending on how you installed Ruby/Ruby Gems you may need to do this with root privileges or sudo.

Usage

Create a project in an empty folder:

wslave new

or create a new project with a name:

wslave new myblog

If you are going to be developing wslave, you can specify the path of your local wslave and this will be automatically set in the Gemfile of the generated project (NOTE: this path is relative to the project, not where you are running the command):

wslave new myblog --wspath ../wslave

If you are generating multiple projects, you can save on time and bandwidth by specifying the location of an already cloned WordPress distribution, which will copy from that location instead of doing a full clone from GitHub:

wslave new myblog --wppath /shared/wordpress

If you want to specify a specific verison of WordPress to use, you can do so like this:

wslave new myblog --version 5.2

You now have a variety of rake tasks and a pre-configured Capistrano configuration.
First, try starting up a local development server:

wslave server start

This will start up 3 docker containers: one running a MariaDB instance, one running an Apache + PHP server instance mapped to localhost:8000, and one running an nginx + PHP-FPM instance mapped to localhost:8001.

If you have stale containers running it's possible you'll have issues, so we made a flag to kill orphaned containers. Just run the server command with -f:

wslave server start -f

You can edit themes and files in public/wp-content. This folder is mounted within the container, so changes should be immediate as if you were running the server on your host OS. Edit your themes and plugins as you like, and be sure to use git to manage your sources. When you're done with the dev server, type this to shut it down:

wslave server stop

If you want to clear out all the data saved in the database/cache of the dev container type:

wslave server stop -v

This will reinitialize everything, and you will loose your work if you haven't backed up the container information locally.

To back up the dev database locally, we use a rake task:

rake db:dev:backup

And to set this as the "active" database, which will be seeded during deployment and to new dev containers, type:

rake db:dev:activate

You can check this into your repository with git, and when another dev grabs your repository and starts a dev container they'll have a copy of your dev database.

Now let's deploy. First off edit the following files:

  • config/database.yml
  • config/definitions.yml

Put in the databse, host, user, etc. information for your deploy target(s). If you don't have a staging server simply ignore those fields. Now, type:

cap production deploy:initial

The "deploy:initial" sub task needs to be run the first time, after that simply use "deploy" to update. Initial sets up some remote variables. We'll try to make this automatic in the future but for now this is just how it is.

If you encountered any problems, try adjusting your configuration files. If you're being constantly asked for your SSH key then you need to read up on adding your public key to the authorized hosts on your server. Protip: This is really easy with the "ssh-copy-id" command on GNU/Linux.

Now, if you edit your theme etc locally, simply commit your changes to git and push them to master of your repository host. Then type:

cap production deploy

and your changes will be pushed to your production server.

To back up your production server, type:

cap production backup

This will copy all the uploaded and changed files in everything but the themes directory of wp-content, and will create a backup of the databse in db/production. To make that production databse the active database, type:

rake db:production:activate

Now, when you start a fresh dev container, all the files and database from the production server will show up within the dev container.

This concludes the short guide. If there's enough interest we'll expand it or perhaps make a detailed video tutorial.

Cloning an Existing wslave Project

Since Wordpress has specific versioning and permissions that need to be set, wslave comes with the wslave sync command to sync project files and set permissions for you. Simply run this command after cloning the repository and running bundle install. Keep in mind you will need a user account that is a member of the 'www-data' account on your local system if you are using a *nix OS.

!WARNING!

Currently there are multiple issues with depenedencies for what is currently the stable version (9) of Sage. Because of this, wslave defaults to using the in-development version. If you need to use Sage 9 we recommend you don't use the wslave shortcuts and helpers for Sage.
※This warning will be removed when Sage 10 is officially released.

Updating

wslave will update to the newest version when you run bundle update. Before you update, you should probably make a commit to your git repository. After updating, you can update your files to the latest wslave version by running wslave update

Porting an Existing WordPress Installation to wslave

As long as you have some basic tools you can quickly port most WordPress installations into a wslave managed project fairly easily. Specifically, you'll need the wp-content folder and a dump file of the database (of a specific format, NOT the "Database Backup" you get from some web-based management consoles).

NOTE: You will need SSH shell access to the server currently hosting the site. You will need to have your SSH public key registered on the server to smoothly perform any automatic operations/perform Capistrano tasks.

The general steps to migrate an existing WordPress installation are:

  1. Create a new wslave project with wslave new my-project-name.
  2. Obtain the wp-content folder, either with a tool like rsync, sftp, scp, or even just ftp. If you are only obtaining a theme from the wp-content folder you only need to copy that theme into the wslave generated public/wp-content/themes folder, but if you have uploads / specific plugins / customized language files / etc. you'll need to either merge the wp-content folder from your existing project with the public/wp-content folder generated by wslave, or remove the public/wp-content folder generated by wslave and copy the wp-content folder into public/ to replace the wslave generated version.
  3. Obtain a databse dump from the running site. You can do this manually, or attempt to let wslave do it for you.
    • Automatic Method:
      1. Edit the config/definitions.yml and config/database.yml files and fill in all details for the "production" profile to reflect the details of the currently running site. Specifically the user, production information under host in definitions.yml, and the username, password, host, and databse name for production in database.yml. HINT: You can find database login information in the wp-config.php file in the public root of the WordPress installation on the server.
      2. Run cap production db:backup. This create a snapshot dump of the database and copy it to db/production/wordpress.sql.
      3. Make the dump you just obtained the active database by running rake db:production:activate.
      4. Be sure to alter any settings you've made to config/definitions.yml and config/database.yml which will not be used for your future production deployment. If your web server or database details will be different when you re-deploy the site in the future you should change these now as to not accidentally deploy over the running site.
    • Manual Method: You will need to obtain a snapshot dump of your running database either with the mysqldump tool or in a way that dumps the database into a compatible .sql format. Usually this would be done with a command like: mysqldump --opt --user=USERNAME --password=PASSWORD --host=HOST DATABASENAME > wordpress.sql replacing the all-caps values with the appropriate values for your database. Then, you can copy this file to db/active/wordpress.sql in the wslave managed project you've created.

If you do not have shell access this whole process will be quite different. You'll probably need to obtain the wordpress files either through plain FTP or through a proprietary tool, and you may need to create a backup of the WordPress database using an internal tool or plugin and then re-import it into your new wslave managed project by starting a development server and using the same internal tool or plugin to import the database information.

As each server and setup are different, and there are many "managed" services for WordPress which have specific restirctions, this process could be very different for you and the above guide should only be considered a generic example.

Notes on Apache

The provided configuration aims to support as many configurations as possible. At the moment .htaccess is provided by the wslave base files and wp-config.php is automatically generated both during the configuration and deployment processes. Currently there is no way to override this, and running wslave update or deploying will end up overwriting these files. Up until now we haven't found any instance of having to override this, but we expect there is need to do so, so please submit an issue with what and how you're trying to modify so we can plan and implement such functionality.

.htaccess

The .htaccess file provided has directives which do the following:

  • Silently redirects the site apex to the wordpress index.php file.
  • [※New] Redirects the wp-admin pages to wordpress/wp-admin. (Newer versions of WP do -not- route through the main index, so this redirect was required.)
  • Disallowes direct access to files. ※This is a generally recommended security measure.
  • Forwards URL paths that do not resolve back to the main index.
  • Provides anchor points for WP extensions to insert their own directives (be cautious when re-deploying or updating from wslave).
  • Disables the file/post upload limits imposed by PHP (file and upload limits are regulated by the web server).

Notes on nginx

While WordPress is generally run on Apache, the truth is it can be run on nginx with PHP-FPM, and it runs very quickly, cleanly, and securely when configured properly. We've done our best to provide an auto-generated nginx vhost configuration file, but as nginx does't scatter its configuration across multiple local files [such as .htaccess] combined with a main configuration it's very likely you'll need to customize this file - so please consider the provided file only a reference and be sure to review it before using it in a deployed configuration. If you have any recommendations as to how we could improve the nginx configuration file template please submit an issue or a PR!

For more help

Try the following commands:

wslave --help
rake -T
cap production -T

Extra Functionality

wslave integrates/allows for some other tools to be run comfortably within the wslave installation. Currently these include:

  • Sage theme generation and management helpers from the wslave command.
  • wp-cli from the installation directory.

Sage Themes

wslave has integrated Sage theme helpers.

Existing Sage Themes

If you have an extisting Sage theme you will need to copy the files into public/wp-content/themes/"theme name" (replacing "theme name" with the actual name of your theme). Then create a config file at config/sage.yml with the following content:

---
:theme: "theme name"

(Again, replacing "theme name" with the name of your theme).

New Sage Themes

You can create a new sage theme with wslave sage new theme_name, repacing theme_name with the name of the theme you wish to create.
CAUTION When/if asked Do you want to remove the existing VCS (.git, .svn..) history? answer n. Otherwise you will have to re-add all the wslave project files to git. We have no idea why Sage does this...

Updating/Installing Packages

You can of course work directly with yarn/composer in your theme directory, or use wslave sage update.

Building Theme Files

You can of course work directly with yarn/composer in your theme directory, or use wslave sage build

Produciton Build and Deployment

For static deployment we don't want to commit the vendor and dist directories to SCM, so there's an extra task in Capistrano that's been added to do this for us if you're using the wslave deployment chain. The command to do this would be cap staging deploy:sage or cap production deploy:sage for staging or production respectively.

To do a production build without Capistrano simply use wslave sage production and assets will be compiled and placed in the appropriate locations in your theme directory.

wp-cli

wp-cli is not bundled with wslave, and must be installed separately, but you can use your local wp-cli installation from within the root directory of your site as long as the development containers are running (wslave server start). A settings file and extra handling is set up for the development environment so no extra settings should be necessary. Keep in mind however that wp-cli is not actually aware of the development environment and anything installed or updated with wp-cli will not have :www-data group ownership (*we hope to add automatic group ownership/file monitorning in the future). You may need to run wslave sync or sudo chown -R :www-data public/wp-content to fix/adjust permissions.

Caution

  1. Even though Capistrano is bening used for deployment, many files such as the wordpress installation and Sage static build (production) files are not committed. Other files are generated dynamically. Because of this, some files are deployed that are not/never included in the git repository, and you should be careful that your repository and local development files appropriately match. This also, unforunately, can make automated deployments somewhat complicated.
  2. URL replacement: WordPress doesn't use relative paths and hard-codes URLs in the database (which is a terrible way to manage URLs and should be refactored in WP core...). Because of this we need to replace URL entries depending on where we are running the site. During development this is localhost:8000, so all the Staging and Production URLs get changed to localhost:8000, and then these localhost:8000 entries are converted to production or staging URLs on deployment. If you have an article or something that explicitly used localhost:8000 this would end up getting changed to the production or staging URL during deployment.
  3. Sage version: Due to Sage v9 being massively outdated, wslave Sage theme creation currently defaults to the upstream dev version (Sage 10 beta). If you would like to use version 9, please create the theme manually with composer and add the "sage.yml" file as described above in the Sage Theme section.
  4. Permissions: When working on themes or extensions you may encounter many permission issues with files not being read due to them not being owned by the www-data group. You may have to periodically run wslave sync or manually chown files (EG: chown -R :www-data ./ in the directory you are working in).

Known Issues

There are a few known issues which can't really be fixed by the wslave package. While we can't provide support for any of these (please don't report system specific issues not directly related to wslave) here are some notes on common problems we've dealt with and what solutions we've found.

PHP 8

At the time of this update WordPress and most extensions do not function properly and you will essentially need to be running a PHP version in the 7 series. The problem being here that on some operating system versions and package managers the default for PHP has already been raised to 8, and as PHP doesn't have language backward-compatibility modes you may need to either supply your build of PHP/PHP-FPM. If you're using a compatible OS we recommend phpenv with php-build. The phpenv repository with install instructions for phpenv and php-build can be found here.

gyp

One of the requirements for a variety of node packages used in Sage theme development and in various other JavaScript/node based tools is gyp [node-gyp], which is notorious for having installation issues which will break a lot of setup scripts and can cause issues. If you encounter an issue during a gyp installation we've found the following commands seem to fix it fairly often (*YMMV):

npm install --global node-gyp@latest
npm config set node_gyp $(npm prefix -g)/lib/node_modules/node-gyp/bin/node-gyp.js

wp-admin Forwards to HTTPS even on localhost

Newer versions of WordPress have embedded forwards to HTTPS for wp-admin and login pages. While you can attempt to diffuse these this doesn't always work, and you may experience a forward to port 443 or to https:// even when you're working on a local wslave development server. We have attempted to prevent this but different browsers and the Apache container at :8000 and the nginx container at :8001 all behave differently and you are likely to encounter this issue. Simply fixing your wp-admin url to something like http://localhost:8001/wordpress/wp-admin after being forwarded should bring you to the admin login page and there should be no issues after that.

Suggestions or patches to more cleanly avoid this issue would be much appreciated.

License

AGPL version 3 and GPL version 3. Please contact info@phantom.industries if you want a commercial or closed source license.

Support

If you find a bug or want to add an improvement we strongly recommend you provide a patch and a Pull Request. This project came from a need for a very specific tool, and as long as that tool fulfills our needs there's no incentive for us to add extra features or improvements; but we will check Pull Requests and merge what looks good. If you'd like to fund development of a specific feature we'll include you as a sponsor.