[source-wordpress] Transformer Sharp

[samburgers]

Hi, is there any chance the source-wordpress example could be updated to fetch wordpress uploaded images and then run them through responsive image generator and saved to /public/static similar to the gatsbygram example? Many thanks!

[KyleAMathews]

@sebastienfi is working on this exact thing actually :-)

[sebastienfi]

Yes ! This is still a WIP, I plan to be done on this by 24th July. This would also work with ACF.

[samburgers]

Incredible, looking forward to it!

[giacomoalonzi]

thank you @sebastienfi :)

[sebastienfi]

Please submit your use case so I can extend the functionality to your needs.

My use case is :

• Identify each image node field being an URL for an image (ex. post: {featured_media: http://your-site.com/image.jpg}). This will cover responsivenes of images for many use cases, including images components build in templates or components, or images displayed using background-image for styled-components for example (see note 1).
• Identify each image in HTML node field (ex. post: {content: "<p><img src='http://your-site.com/image.jpg />}). This will cover images inserted using the regular WYSIWYG editor of WP.
• Identify each image in stringyfied JSON node field (ex. post: {json: "<p><img src='http:\/\/your-site.com\/image.jpg \/>"}). This will cover images under ACF Field (like these inserted using an ACF Image component).
• images file ended by jpg, png, gif.
• for an image node field, transform the node to an object with properties {original_url, static_url, sizes { blob, size_1, size_2,...}}
• for an image in HTML or json, replace the img tag url with the static one or blob if smaller than 8kb, and insert the srcset attribute.

[EDITED 23/07/17]

Note 1 : https://aclaes.com/responsive-background-images-with-srcset-and-sizes/

[samburgers]

Hi @sebastienfi

Thanks again for this!

Yes, that all sounds right to me... the images from the wordpress side would be a combination of Featured Image, ACF Image/Gallery uploads, and possibly inline post media attachments too, as you say.

On the Gatsby side, these would be used as such:

• og:image meta tags (usually featured image)
• Various resized thumbnail images with srcset attributes.

Cheers :)

[sebastienfi]

Hi guys !

Just an update, need more time, expect this around 29th.
Will focus first at image node field's example above. Then I'll digg into images contained in JSON stringyfied DOM or "simple" DOM. Building the regex that will extract these contents from the DOM won't be an easy task, any help appreciated.

I also updated my post above.

[KyleAMathews]

Use cheerio! Check out the recent changes to gatsby-remark-images adding support for parsing inline html.

[sebastienfi]

I'll give a go at Cheerio.
I've ended with this regex but it may not be bulletproof : ([<;]img\s*src\s*=\s*")?(.*(gif|jpg|png))[\?"]+

[KyleAMathews]

This is worth a read :-P https://stackoverflow.com/a/1732454

[sebastienfi]

I will also probably dedicate to Worpdress the image sizes management, using the thumbnail-width definitons https://codex.wordpress.org/Post_Thumbnails

[sebastienfi]

LOL @KyleAMathews

The force of regex and HTML together in the same conceptual space will destroy your mind like so much watery putty.

I think the flaw here is that HTML is a Chomsky Type 2 grammar (context free grammar) and RegEx is a Chomsky Type 3 grammar (regular grammar). Since a Type 2 grammar is fundamentally more complex than a Type 3 grammar (see the Chomsky hierarchy), you can't possibly make this work. But many will try, some will claim success and others will find the fault and totally mess you up.

[samburgers]

@sebastienfi would it be cleaner to just replicate/sync the wordpress asset library in its entirety? I imagine this might negate the need to regex the post content for image markup, and could just simply swap out the base path or something. As long as the build task is aware of which assets have already been processed it shouldn't really take up too many resources for each build (I am pretty sure the gatsbygram example behaves this way)

In a perfect world, image_sizes should really be defined on the Gatsby side. For example you could have multiple Gatsby sites reading from a single WordPress content API etc.

Just my 2c :)

[sebastienfi]

would it be cleaner to just replicate/sync the wordpress asset library in its entirety?

Yes! Definitely cleaner. The problem is that on Wordpress.com one cannot query this library entirely without auth. We have 2 models : the localy hosted Wordpress instance (at your hosting provider or wordpress.org for example) and the wordpress instance hosted on wordpress.com. I'm trying to remain generic. Also, scraping the HTML content would also take care of the images which src are not hosted on the Wordpress instance.

You are right about the image size management. Gatsby should hold on this as templating and design is done at this level too.

[sebastienfi]

For an update : being late on this sorry guys. I had to de-prioritize this in favor of some more urgent matters. I've made great progress thought. Still bugged by some call back hell. I hope to have time to solve this on the coming weeks.

WIP is at https://github.com/sebastienfi/gatsby/tree/topics/wordpress-responsive-images

PRs welcomed !

[KyleAMathews]

This is working — see the example site https://using-wordpress.gatsbyjs.org/sample-post-1

[jarekpelczynski]

So content images (uploaded by WYSIWYG) are not supported?

[pieh]

They aren't for now. If someone would like to add support for that then PRs are welcome. It would need to parse html content from WYSIWYG fields (at least for main content field) and replace inline images (somethink like gatsby-remark-images does for markdown).

[lightstrike]

I was just doing a little thinking about this and here's what I came up with to parse images from WYSIWYG fields:

function replaceInlineImages({ post }) => {
    const parser = new DOMParser()
    const contentTree = parser.parseFromString(post.content, "text/html")
    const inlineImages = contentTree.getElementsByTagName('img')
    for (const wordpressImage of inlineImages) {
      const gatsbyImage = getGatsbyImage(wordpressImage) // The hard part
      worpressImage.replaceWith(gatsbyImage)
    }
    return contentTree.body.innerHTML
}

@pieh Seem like a decent start for the parsing side?

[KyleAMathews]

My thought it that past is that we should make a gatsby-transformer-rehype that works the same as gatsby-transformer-remark. Then we could create a gatsby-rehype-images that transforms images inside html sections.

https://github.com/rehypejs/rehype

[KyleAMathews]

Along with plugins doing other HTML manipulation e.g. even syntax highlighting.

[lightstrike]

That makes a lot of sense to me, I'll take a look into rehype 😄

[sebastienfi]

There is also images in the CSS.