Support htmlparser2

[hansottowirtz]

We need to inline very large svgs (~100MB). Cheerio uses parse5 by default, which is very slow and memory intensive for files of this size. Cheerio also supports (and ships with) htmlparser2, which has a slightly different API. This PR makes using htmlparser2 possible by detecting which API is present.

I'm not an expert on this and I'm not sure if the test I wrote is complete enough, but from my tests the only important API difference was that el.childNodes[0].nodeValue in parse5 corresponds to el.children[0].data in htmlparser2.

[simison]

Why not switch to htmlparser2 for everyone?

[jrit]

Yeah, I have the same question as @simison - why support both? Is there something that parse5 is better at?

[hansottowirtz]

It seems like parse5 is a lot faster while htmlparser2 is more spec-compliant and supports XML (and for our use case, is more memory-efficient).

Cheerio uses parse5 but can fall back to htmlparser2 (e.g. needed for XML). In my opinion, as Juice can receive a Cheerio instance with both parsers, it should also support both parsers.

Cheerio parse5 parser discussion: cheeriojs/cheerio#985 and cheeriojs/cheerio#863
Cheerio supports both parsers: cheeriojs/cheerio#1259 (comment)

[TrySound]

Btw some packages (mjml) use [email protected] which is published with latest tag in npm.
http://unpkg.com/cheerio@latest/package.json