fix(gatsby-transformer-remark): fix excerpt generation - strip excessive white spaces, extract alt from images

[macklinu]

Description

This is a failing test case to reproduce the issue stated in #12398. Here is a screenshot of test failure output:

Related Issues

Relates to #12398

[wardpeet]

@macklinu mind fixing this as well? Thanks for creating the test, if you don't really know how we can take over from here.

[macklinu]

@wardpeet I'll take a look at patching this and ping you if I come up with a solution or need to ask for another contributor's help. Thank you!

[macklinu]

Would it make sense to use strip-markdown as a remark plugin instead of writing our own markdown processor in order to parse out the excerpt content? Or maybe we don't need all the functionality of strip-markdown and just need to account for a few more node types? Any implementation suggestions?

gatsby/packages/gatsby-transformer-remark/src/extend-node-type.js

Lines 412 to 449 in 15ef178

	async function getExcerpt(
	markdownNode,
	{ format, pruneLength, truncate, excerptSeparator }
	) {
	if (format === `html`) {
	const excerptAST = await getExcerptAst(markdownNode, {
	pruneLength,
	truncate,
	excerptSeparator,
	})
	const html = hastToHTML(excerptAST, {
	allowDangerousHTML: true,
	})
	return html
	}
	if (markdownNode.excerpt) {
	return markdownNode.excerpt
	}
	const text = await getAST(markdownNode).then(ast => {
	const excerptNodes = []
	visit(ast, node => {
	if (node.type === `text` || node.type === `inlineCode`) {
	excerptNodes.push(node.value)
	}
	return
	})
	if (!truncate) {
	return prune(excerptNodes.join(` `), pruneLength, `…`)
	}
	return _.truncate(excerptNodes.join(` `), {
	length: pruneLength,
	omission: `…`,
	})
	})
	return text
	}

[wardpeet]

strip-markdown strips markdown away, we still want to show the link or bold text right? That plugin just makes the text unstyled.

So if not mistaken it's not really what we're after

[macklinu]

Ah, I see. I think my failing test assertion is incorrect then. We don't want to necessarily strip markdown, but we just want to preserve markdown as is and not remove any content?

[wardpeet]

yes, the issue is that an extra space is added between the link & the ! symbol. It might be something that we have to fix upstream.

[frinyvonnick]

Hi @macklinu 👋,

First thanks for your failing test it really helped me diving in this bug 👏

I'm not sure to understood all you said with @wardpeet.

I found out that the function getExcerpt in extend-node-type.js wasn't handling image and the last link was classified as an image and in contrary of links image doesn't have a text node in it. That was causing the missing parts. The second problem of extra spaces was caused by a join of excerpt parts with a space.

Was that the right thing to do ?

Sorry to not asking before pushing on your branch 😊

[pieh]

Nice! Just one nitpicky comment/question about value of snapshot in the test (as most value comes from node.excerpt assertion)

[frinyvonnick]

Thanks for your review @pieh 👍

[wardpeet]

Sweet! thanks for getting this PR through the door @frinyvonnick! I'll fix the unit test.

[frinyvonnick]

Thanks @wardpeet 🙏 !

[wardpeet]

pretty sure we can ignore danger here.

[sidharthachatterjee]

Published in gatsby-transformer-remark@2.3.12