Categories
Uncategorized

`wp-now` → Awesome

Check out the cool new tool for doing WordPress development. This uses a WASM version of PHP to serve/host the environment directly in your browser. (It does need node for getting started though.)

Categories
Software Development

WordPress Blocks: Hydrate InnerBlocks From HTML

Quick tip about creating or replacing InnerBlocks from HTML in the WordPress block editor. For instance if your block pulls information from a remote API and you want to display as part of the post, or embed in the post such that it’s editable. The block editor (gutenberg) already handles the job very simply.

Use pasteHandler. It not only includes HTML support, but plain text & markdown also!

import { pasteHandler } from '@wordpress/blocks';

And from a dispatch of the block-editor store, use the replaceInnerBlocks() method.

I had a couple issues with the code that made it somewhat confusing.

  1. I’m still getting used to the concept of React hooks, so managing the useSelect, useDispatch, and useCallback might not be 100% correct — but this incantation worked for me 😆.
  2. Originally I tried updating the first InnerBlock and finding it was slightly tricky. Good news — this is actually fairly simple with useSelect to getBlocksByClientId() and then just get the innerBlocks property of the first block returned.

InnerBlocks Example:

// `clientId` is passed as a prop to the edit component.
const { replaceInnerBlocks } = useDispatch( 'core/block-editor' );

// call `updateInnerBlocks( markdownOrHTMLContent );`
const updateInnerBlocks = useCallback(
	source => {
		// Get an array of inner blocks
		const newInnerBlocks = pasteHandler( {
			HTML: '',
			mode: 'BLOCKS',
			plainText: source,
		} );
		replaceInnerBlocks( clientId, newInnerBlocks );
	},
	[ innerBlocks ]
);

Interested in more? See other posts about WordPress and the block editor.

Categories
Software Development

WordPress Blocks: React Context Cuts Clutter

Using Reacts’ Context in WordPress blocks passes parameters painlessly. Attributes carried by context keep coupling minimal.

Reacts’ Context feature facilitates distributing data down a component tree to any descendant component. Instead of passing attributes between your components manually (e.g. <FooComponent attrib={ attrib }>), which tightly couples the current attributes and components, using a block-level context makes it easy to update and manage attributes (and their changes).

WordPress Blocks (a.k.a. Gutenberg) supports a similar concept called Block Context, but it’s for communicating and sharing attributes between discrete blocks (those registered using registerBlockType).

To someone familiar with React the following is probably obvious and straight-forward, but I found implementing it non-trivial, and also not covered in the WordPress documentation. Regardless, this appears to be so functional it should be a best practice.

Example:

A WordPress block has at minimum 2 top-level components, an Edit and a Save. Each component is passed the block attributes in the first argument. Typically the save component is less-interesting as it specifies the block data saved in the post-content, but if required it would function identically to the following example used with the edit component.

For this example let’s examine a block with 1 attribute, passed to both the toolbar and the sidebar.

export default function Edit( props ) {
	const { foo = '' } = props.attributes;
	const blockProps = useBlockProps();

	const updateFoo = useCallback(
		( _foo ) => {
			props.setAttributes( { foo: _foo } );
		},
		[ foo ]
	);

	return (
		<div { ...blockProps }>
			<MyPluginBlockControls foo={foo} onFooChange={updateFoo}></MyPluginBlockControls>
			<MyPluginInspectorControls foo={foo} onFooChange={updateFoo}></MyPluginInspectorControls>
			...
		</div>
	);
}

While this example is a bit contrived (with only one attribute it is probably not necessary to create subcomponents managing our controls), it should be evident that adding additional attributes cascades changes through the tree. Using a Context instead would resemble code like (differences are highlighted):

import { FooContext } from './context';

export default function Edit( props ) {
	const { foo = '' } = props.attributes;
	const blockProps = useBlockProps();

	const updateFoo = useCallback(
		( _foo ) => {
			props.setAttributes( { foo: _foo } );
		},
		[ foo ]
	);

	const blockContext = {
		foo,
		updateFoo,
	};

	return (
		<FooContext.Provider value={ blockContext }>
			<div { ...blockProps }>
				<MyPluginBlockControls></MyPluginBlockControls>
				<MyPluginInspectorControls></MyPluginInspectorControls>
				...
			</div>
		<FooContext.Provider/>
	);
}

The Context

Creating the context itself is pretty simple:

import { createContext, useContext } from '@wordpress/element';

export const FooContext = createContext( {
	foo: undefined,
	updateFoo: () => {},
} );

export const useFooContext = () => useContext( FooContext );

Using the Context in a sub-component

import { useFooContext } from './context';

export function MyPluginBlockControls() {
	// Use `foo` as normal.  If you need to write it, you can also add `updateFoo`.
	const { foo } = useFooContext();
}

Adding additional attributes

If you modify your block.json to have additional attributes it’s simple to add them to the context (use the property attribute and create a write function at the top-level for every attribute ), then it’s trivial to change the components to import the attribute with the const {foo,bar} = useFooContext() syntax.

Conclusion

This pattern appears to be a good way to decouple attributes within a WordPress block that need to be passed either to re-usable components or just to simplify what code should change when adding/updating the attributes being used. I’m still a React novice and I would love to know if there’s a problem or if it could be improved any way — let me know in the comments!

Categories
Software Development

Debugging Gutenberg Blocks: 1

I’m adding some tags for tips as I find them.

I found this tip helpful when debugging what the WordPress Block Editor (Gutenberg) was actually doing behind the scenes.

Find the post id (in the edit url there will be a field for post (e.g. post.php?post=121&action=edit). Using the WP-CLI tool you can quickly view the source of the post (assuming you have a very basic post with just the block you are developing) by running:

wp post get 121 --field=post_content

This will display the content showing the HTML as it is stored in the database — including the attributes for the block!

Categories
Software Development

Filecasted: Create Podcasts From Local Files

Running isn’t the most enjoyable activity, but it’s great for burning calories. One of the best ways to cope is by listening to podcasts. Podcast apps are generally great and provide innovative features — like speeding up the playback (to get through the content quicker). Switching between podcasts and audiobooks becomes jarring after expecting many of the same behaviors, so I spent a weekend throwing together a quick application to create podcasts from local files.

It’s pushed to github (as open-source) and can be used for turning local mp3’s into deliverable podcast episodes. Use it to produce a podcast or publish audio files to the web.

Filecasted

It’s a simple app — it takes a list of files and creates an RSS (.xml) file. The RSS file can be hosted on something like AWS for general delivery. I threw it together using poetry/python, and development is started by:

  1. pip install poetry
  2. poetry install
  3. poetry shell — the filecast command is now available.
$ filecast --help

Usage: filecast [OPTIONS] INPUT

  Create a podcast file from INPUT.

  INPUT can be a playlist (`.m3u`), stdin (`-`), a directory or a collection
  of audio files.  Playlists and stdin should be a simple text document with
  a list of files to add.

Options:
  --version
  -v, --verbose
  -f, --force             Force the output to an existing file.
  --audio-extensions EXT  Comma separated list of processable extensions.
                          [default: .mp3,.mp4,.wav]

  -a, --append            Append INPUT to the output file.
  -o, --output PATH
  -n, --dry-run           Run as normally, but don't make any permanent
                          changes

  --help                  Show this message and exit.

It’s still a bit buggy. For instance, the album art doesn’t show correctly in the podcasts app I use, and some of the audio files downloaded strangely (but I don’t believe that was related to my code).

Also, there is a lot of room for improvement. It would be nice to manage some of the metadata and update/change feeds as needed. Or maybe support transcoding/speeding files up/adding/editing/splicing the audio files… Or adding support for awscli and delivering both the “episodes” and the RSS file to their space on the internet.

The sad ending

Despite the bugs, Filecasted worked for me in the limited capacity that I needed it to. I was able to create podcasts from local files, publish to AWS and then listen to it with my podcast app. Since I last used it though, I switched to Pocket Casts, which has a feature to play local files without the intermediate steps needed to host and download.

So the bugs will likely remain and I’ll just keep the joyful memory of building and using my creation 😢.

Categories
Software Development

MermaidJS & WordPress

I get confused by a lot of words. The combination of complex ideas and large walls of text short-circuit and crash my brain. I’ll have to read and re-read a text-only passage multiple times to understand exactly what’s being described. Speaking presentations exacerbate the issue.

However the adage “A picture is worth a thousand words” definitely applies to me. Given graphical clues demonstrating the principle helps me begin to make connections and jumpstarting my understanding.

graph TD
   start(Explain an idea) -->|with words|start
   start --> |with pictures| understanding

One of the tools I’ve found very helpful (if only for myself) is MermaidJS. This software lets you create several different types of charts using a markdown-ish syntax. For instance, the above chart was created with:

graph TD
   start(Explain an idea) --> |with words| start
   start --> |with pictures| understanding

This grants other benefits in addition to the quicker comprehension. Quick edits to the text automatically recreates the chart without requiring any manual intervention/tweaking to the rest of the diagram.

MermaidJS supports flowcharts, sequence diagrams, class diagrams, state diagrams, ERD (entity-relationship diagrams), Gantt charts, pie charts, requirement diagrams and User Journey diagrams.

On my last post I actually used it to show the sequence of flows between WordPress, WooCommerce and webhooks.

WordPress Plugins

Working at Automattic, all our internal documentation and institutional knowledge is recorded using WordPress (you should really checkout using p2’s — it’s an awesome tool). Using Mermaid in this environment has been fantastic — Mermaid can be supported anywhere you have HTML (and I can be exported to SVG too), but as WordPress is running 40% of the web, this seems like a pretty good place to start.

In fact MermaidJS has a page dedicated to integrations of their software and mention a couple plugins:

  • WordPress Markdown Editor (seems to have been rebranded to WP Githuber-MD — see below)
  • WP-ReliableMD: The page was in a different language than I understand and I made a snap decision to avoid it 😢. Should I revisit this decision?

Here is my un-asked for 2¢ about some different WordPress plugins supporting MermaidJS.

WP Mermaid (and WP Githuber-MD)

These plugins are both created & maintained by Terry Lin. I think they’re both great. They support Gutenberg blocks. They display the Mermaid text. Give them a shot and see if they float your boat.

Personally, I didn’t need the full markdown editor, and WP Mermaid had a couple small annoyances that I couldn’t solve. I submitted a couple PR’s, but decided I could use something a little simpler, which brings me to my own plugin…

MerPress

I can’t take full credit for this plugin. My coworker Michael Pretty actually did most of the work on this iteration and I submitted it to WordPress.org. I’m giving him full-credit. He did an awesome job with it.

A couple features I really like:

  • MerPress keeps the wptexturize functionality. Posts preserve the capability to convert simple characters like quotes and dashes into the nice curly quotes and dashes (e.g. “”, –, —).
  • In the Gutenberg editor, the block is encased in a code block letting you edit the code with a monotype font.

Some features I would like to see:

  • Ability to change the theme site-wide and per block
  • Add other configuration parameters to the block
  • Support 3rd party CDN (something that WP Mermaid already does)

Conclusion

If you learned nothing else, you now know I need pictures in order to comprehend complex concepts. Hopefully it’s also obvious that Mermaid is a neat tool and can be used to quickly share intention and purpose and thoughts without all the words.

graph TD
  ad(all done) --> te(the end)

Please let me know if I’m missing a plugin that I should try or if there is something you would like to see added to MerPress…

Categories
Software Development

Working with WooCommerce Webhooks

I recently went deep into the WooCommerce Webhooks code. I’m preserving my thoughts here (if anyone needs more help — feel free to comment — I’m happy to share what I know).

Webhooks?

TL;DR; A separate website can be notified when you make a sale by creating a webhook.

WooCommerce provides a method to send announcements of actions that occur on your site to any URL. These announcements (topics) consist of WooCommerce domain objects (orders, products, subscriptions, etc. also generically named resources) and include their life-cycle events (creation, updates, deletion, etc.). The WooCommerce documentation has a pretty good writeup about this.

WooCommerce Webhooks Handling Overview

sequenceDiagram

participant WordPress
activate WordPress
WordPress ->> WooCommerce: load woocommerce
activate WooCommerce
WooCommerce ->> Webhooks: instantiate webhooks
activate Webhooks
loop Every webhook
Webhooks ->> Webhooks: hook topics
Note right of Webhooks: An topic is defined by a <br> resource and an event.<br>-<br>A topic has a mapping to one or <br>more WordPress hooks. When the <br>hook fires (i.e. do_action),<br> the webhook is enqueued.<br>-<br>See $webhook->enqueue()
end
deactivate Webhooks
deactivate WooCommerce
WordPress ->> WordPress: ...
WordPress ->> Webhooks: Webhook Event (e.g. Order created)
Note right of Webhooks: Events are added to a global<br>array with: 1️⃣ the Webhook object<br>and 2️⃣ the args passed to the hook.<br>-<br>See $webhook->process()
WordPress ->> WordPress: ...
Note right of WordPress: shutdown
loop Enqueued Webhooks
WordPress ->> Webhooks: deliver
Note right of Webhooks: Actually 2 different methods to <br>deliver. Defaults to async (using<br> Action Scheduler), but can<br>run synchronously.<br>-<br>See $webhook->deliver()
end
deactivate WordPress

Using Webhooks For Fun and Profit

Topics aren’t limited to just WooCommerce domain objects. You can use any WordPress hook to send a webhook. Using the resource “action”, add the hook name as the event and you will receive the first argument passed to the hook as the parameter in your webhook. Events must start with wc_ or woocommerce_.

Example Payload

{
  "action":"woocommerce_my_custom_event",
  "arg":{
     "test":"something I passed in"
  }
}

Deliver custom topics

Using the action resource is limiting.

  • The hook name needs to have a special prefix
  • The shape of the data returned will be wrapped in the arg parameter.
  • WooCommerce uses ActionScheduler for storing/managing async events, capping the size of the data at 191 characters. (Async handling can be disabled to avoid this limitation).

What I really want is my own resources/events. For this example let’s pretend we needed to add our own bookmark object, and we want to send a webhook whenever a bookmark is created.

First, add our topic to the list so it can be appropriately enqueued. This mapping is from topic → hook. When the hook is called (e.g. do_action( 'send_bookmark_created_webhook', $bookmark->id ); ) The webhook is added to the queue.

/**
 * Add webhook topics.
 *
 * This is used in the startup portion of webhooks, and is called for each webhook.  As the webhook has the reference
 * to the specific topic, just return the WordPress action to listen for this topic.
 *
 * @param array $topic_hooks
 * @param \WC_Webhook $webhook
 */
function add_topics( $topic_hooks, $webhook ) {
	$topic_hooks['bookmark.created'] = 'send_bookmark_created_webhook'; // When we create a bookmark, we'll fire this action.
	return $topic_hooks;
}

add_filter( 'woocommerce_webhook_topic_hooks', 'add_topics', 10, 2 );

Now we need to create our payload (avoiding the 191 character limit).

/**
 * Build a payload for our custom topics.
 *
 * @param $filtered_payload mixed  Passed from the existing build_payload function.
 * @param $webhook_resource string Resource parsed from the webhook topic.
 * @param $arg mixed               Args passed to the trigger for the topic.
 * @param $webhook_id int          Can use this to rehydrate a copy of the webhook (e.g. new WC_Webhook( $id )).
 *
 * @return mixed
 */
function build_payload( $filtered_payload, $webhook_resource, $arg, $webhook_id ) {
	if ( 'bookmark' === $webhook_resource ) {
		return get_bookmark( $arg );
	}
	return $filtered_payload;
}

add_filter( 'woocommerce_webhook_payload', 'build_payload', 20, 4 );

Finally, whitelist the webhook.

/**
 * Before the webhook is sent this hook determines if it should be delivered.
 *
 * Have to override this because the custom types.  Alternately could continually overwrite
 * `woocommerce_valid_webhook_resources' and 'woocommerce_valid_webhook_events` like is done
 * in the store currently.
 *
 * @param $should_deliver bool Existing value
 * @param $webhook \WC_Webhook
 * @param $arg mixed           Delivery body
 */
function should_deliver( $should_deliver, $webhook, $arg ) {
	if ( 'bookmark.created' === $webhook->get_topic() ) {
		return true;
	}
	return $should_deliver;
}

add_filter( 'woocommerce_webhook_should_deliver', 'should_deliver', 10, 3 );

At the end of the day

If this helped you or if something remains confusing let me know, I’m interested in hearing people’s pain points and wish list items. Personally, I’m looking at adding retry functionality to this infrastructure and will add another post when I get that sorted.

Categories
Uncategorized

How I Work :: 2021 Browser Edition

I’ve been surfing the web for almost 30 years and gone through many browsers in that time. I’ve tried configuring them, writing extensions & themes and evaluating them for personal use. In the future I might go more in-depth on my choices, but by way of introduction (to myself, and this site) — here’s a snapshot of how I work with browsers, y rather convoluted way of surfing the Internet.

Chromium Chromium-based browser

Google’s open source browser is ubiquitous. It powers something like 99% of the web (that’s a made-up statistic that may or may not be true). As the engine for many different browsers, this can lead to some interesting trade-offs.

  • It’s by far the most used, so it’s a de facto requirement for any development efforts.
  • I really enjoy Chromium dev-tools.
  • I don’t want to be too deep in the Google eco-system, so if I can avoid Chrome, I will. (Yes, they currently have my email, calendar and maps — but I don’t have to make it easy for them).
  • The less extensions I have to install the better.

For this reason Vivaldi is my chromium browser of choice. It has fantastic tab-management, configuration and even gestures available out of the box in addition to the other tools I need (e.g. profiles, extensions, etc.)

I use a different profile for work, gmail, and personal.

firefox Firefox Developer Edition

My first browser was Netscape Navigator and I have always been rooting for it. When Mozilla spun off of Netscape I tried to use it, but it wasn’t until Phoenix (early version of Firefox) came along that I really loved it. I’ve gone away and returned many times as features have improved or stagnated.

Today, I really enjoy the container feature to sandbox almost every open tab into mini/temporary profiles. And it’s very snappy these days too. Firefox serves as a great daily driver (and is my default browser).

 Finicky

The last browser that I use on a daily basis is Finicky. This is a macos-only application (although there are comparable Windows/Linux options available) to route links to different browsers. If I click on a link in a different application (e.g. slack) it uses a series of filters to decide if it should open the Vivaldi work profile, or send it to Firefox.

Other Browsers

Every so often, I retry different browsers to see if something has changed — if they’re faster or better than what I’m currently using — great! So I have a rotating cast of secondary browsers making guest appearances on my systems. Google Chrome, Chrome Canary, Chromium, Safari, and Brave to name a few. Even other esoteric ones few might have heard of (like Browserosaurus). (Sidenote: I still need to give Edge a closer look, but as it shares the Chromium engine, I haven’t felt a need to kick the tires).

If you have seen a new or interesting browser let me know, there’s always room for another one.

Categories
Uncategorized

Hello world!

Welcome. This is your first post. Edit or delete it, then start writing!