We actually use the original node.js parser code (http_parser), which was originally based _I believe_ on the nginx parser. The Node people have since moved on to generating the C parser code with Typescript because the original code was too difficult to maintain (llparse).

Like I said, human-readable HTTP needs to die, but it's far too late.

llhttp different from the original http-parser. It has lesser bugs because it's written in Typescript which transpiled to C and it's faster than the original http-parser