Replace cheerio with one dependency (parse5)

[fregante]

addons-linter uses cheerio to validate the HTML, currently there are two rules¹ and cheerio is unreasonably large just for that (2.80MB!)².

Cheerio wraps parse5 to parse the AST and addons-linter could use it directly, saving almost 2 megabytes.³ The issue is that parse5 only returns an object⁴, no querying ability.

💡 I'm not sure whether this is actually possible, but would you be open to a PR that passes current tests?

Given the number of rules (2!) would it make sense to try and walk the AST directly? Maybe a simple walker would be as simple as:

function walk(dom, filter) {
	if (filter(dom)) {
		return dom;
	}

	if (!dom.childNodes) {
		return;
	}

	for (const node of dom.childNodes) {
		const result = walk(node);
		if (result) {
			return result
		}
	}
}

And then the rules would look something like:

const invalid = walk(parse5(source), node => node.tagName === 'script' && node.attrs.some(attr => attr.type === 'etc etc'))
if (invalid) {
	rules.push(invalid, ...details)
}

Footnotes

1. https://github.com/mozilla/addons-linter/tree/5c608c102ed926099f6ed0a6b5d53c93ff9c8ae3/src/rules/html ↩

2. https://packagephobia.com/result?p=cheerio ↩

3. https://packagephobia.com/[email protected] ↩

4. https://astexplorer.net/#/1CHlCXc4n4 ↩